WO2002095514A2 - Control of multi-variable processes - Google Patents

Control of multi-variable processes Download PDF

Info

Publication number
WO2002095514A2
WO2002095514A2 PCT/GB2002/002110 GB0202110W WO02095514A2 WO 2002095514 A2 WO2002095514 A2 WO 2002095514A2 GB 0202110 W GB0202110 W GB 0202110W WO 02095514 A2 WO02095514 A2 WO 02095514A2
Authority
WO
WIPO (PCT)
Prior art keywords
variables
variable
values
alarm condition
value
Prior art date
Application number
PCT/GB2002/002110
Other languages
French (fr)
Other versions
WO2002095514A3 (en
Inventor
Robin William Brooks
John Gavin Wilson
Richard John Thorpe
Original Assignee
Curvaceous Software Limited
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Curvaceous Software Limited filed Critical Curvaceous Software Limited
Priority to AU2002255152A priority Critical patent/AU2002255152A1/en
Priority to JP2002591923A priority patent/JP4156934B2/en
Priority to DE60226730T priority patent/DE60226730D1/en
Priority to US10/477,340 priority patent/US7336278B2/en
Priority to EP02724460A priority patent/EP1402324B1/en
Priority to CA2446688A priority patent/CA2446688C/en
Publication of WO2002095514A2 publication Critical patent/WO2002095514A2/en
Publication of WO2002095514A3 publication Critical patent/WO2002095514A3/en

Links

Classifications

    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B23/00Testing or monitoring of control systems or parts thereof
    • G05B23/02Electric testing or monitoring
    • G05B23/0205Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
    • G05B23/0259Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterized by the response to fault detection
    • G05B23/0267Fault communication, e.g. human machine interface [HMI]
    • G05B23/0272Presentation of monitored results, e.g. selection of status reports to be displayed; Filtering information to the user
    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B19/00Programme-control systems
    • G05B19/02Programme-control systems electric
    • G05B19/418Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS], computer integrated manufacturing [CIM]
    • G05B19/41875Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS], computer integrated manufacturing [CIM] characterised by quality surveillance of production
    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B23/00Testing or monitoring of control systems or parts thereof
    • G05B23/02Electric testing or monitoring
    • G05B23/0205Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
    • G05B23/0218Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
    • G05B23/0224Process history based detection method, e.g. whereby history implies the availability of large amounts of data
    • G05B23/0227Qualitative history assessment, whereby the type of data acted upon, e.g. waveforms, images or patterns, is not relevant, e.g. rule based assessment; if-then decisions
    • G05B23/0235Qualitative history assessment, whereby the type of data acted upon, e.g. waveforms, images or patterns, is not relevant, e.g. rule based assessment; if-then decisions based on a comparison with predetermined threshold or range, e.g. "classical methods", carried out during normal operation; threshold adaptation or choice; when or how to compare with the threshold
    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B23/00Testing or monitoring of control systems or parts thereof
    • G05B23/02Electric testing or monitoring
    • G05B23/0205Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
    • G05B23/0259Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterized by the response to fault detection
    • G05B23/0286Modifications to the monitored process, e.g. stopping operation or adapting control
    • G05B23/0294Optimizing process, e.g. process efficiency, product quality
    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00Program-control systems
    • G05B2219/30Nc systems
    • G05B2219/35Nc in input of data, input till input file format
    • G05B2219/35507Spider, radar, parallel axes, multivariate plot
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02PCLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/02Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]

Definitions

  • This invention relates to methods and systems for control of multi-variable processes.
  • a method for control of a multi-variable process comprises deriving a multi-dimensional representation of the process variables according to individual coordinate axes, defining a feasible region of the process-variables based on sets of values for the process-variables accumulated respectively in relation to multiple operations of the process, defining for each process variable a range due to the values of the other process variables, within which the value of that variable is to lie in order for operation to remain within the feasible region, detecting in respect of each of the variables whether an alarm condition exists in which the current value of that variable lies outside the range defined for it, responding to the existence of said alarm condition to calculate change of the value or values of one or more of the variables required to rectify the alarm condition, and implementing the calculated change.
  • a system for control of a multi-variable process comprises means for deriving a multi-dimensional representation of the process variables according to individual coordinate axes, means for defining a feasible region of the process-variables based on sets of values for the process-variables accumulated respectively in relation to multiple operations of the process, means defining for each process variable a range due to the values of the other process variables, within which the value of that variable is to lie in order for operation to remain within the feasible region, means for detecting in respect of each of the variables whether an alarm condition exists in which the current value of that variable lies outside the range defined for it, and means responding to the existence of said alarm condition to calculate change of the value or values of one or more of the variables required to rectify the alarm condition, and means for displaying the calculated change.
  • convex hulls or any other closed envelope embracing the data points represented respectively by the accumulated sets of values.
  • a convex hull in orthogonal coordinates is a closed polygon that encloses all relevant data points of the two-dimensional space, whereas in parallel coordinates it is a pair of spaced linear curves that as between corresponding parallel axes, bound the region occupied by the lines that represent (in the parallel- coordinate space) those data points.
  • the current values of at least some of the process variables and said ranges defined individually for them, may be displayed in said multi-dimensional representation .
  • the invention is applicable to optimisation and efficient on-line operation of multi-variable processes, and in this respect the calculated change required to rectify the alarm condition may be implemented by control exercised by an operator in accordance with display of the change. Alternatively, the change may be communicated to a process controller for automated implementation .
  • Figure 1 is a schematic representation of a system according to the invention in the context of collection and utilisation of data derived from operation of a multi-variable processing plant;
  • Figure 2 is illustrative of a plot in multi-dimensional space defined by parallel coordinate axes, of operation of the multi-variable processing plant of Figure 1;
  • Figure 3 shows in part a multiplicity of plots corresponding to that of Figure 2 resulting from successive operations of the multi-variable process
  • Figure 4 is illustrative of the display of convex hulls in the system of Figure 1;
  • FIGS. 5 to 7 are illustrative of successive displays provided according to the invention in connection with operation of the method and system of the invention.
  • Figure 8 is illustrative of the derivation of certain coefficients utilised in the method and system of the invention.
  • Figure 9 shows an alternative form of display representation that may be used in accordance with the invention.
  • the example of method and system to be described is related to the control of operation of a multi-variable process carried out by a simple, notional processing plant. Details of the plant and its purpose are not of consequence, and indeed the method and system of the invention are related more specifically to operation of the plant as an example of a multi-variable process rather than to the purpose of the process performed, being applicable in the generality to any situation involving a multi-variable process.
  • the method and system of the invention are related more specifically to operation of the plant as an example of a multi-variable process rather than to the purpose of the process performed, being applicable in the generality to any situation involving a multi-variable process.
  • the remaining three variables are quality variables in the sense that their values define, or more especially are defined by, that outcome.
  • the plant 1 has an input 2 and an output 3 between which there are, for example, a multiplicity of processing stages 4.
  • the processing within each stage 4 is carried out in accordance with one or more variables that, in this example, are regulated by eleven controllers 5.
  • the values of these variables for each operation or 'run' of the process are communicated to a data collection unit 6 to be accumulated in a store 7.
  • the term ⁇ run' in this context may refer to a discrete operation of the process, but it may also refer to what applies at a discrete point in time within continuous operation.
  • the outcome at the output 3 of each run of the process is submitted to a unit 8 for analysis in respect of its quality as determined according to three variables.
  • the values of these three quality variables are accumulated in a store 9, so that each run of the process and its outcome is defined by an accumulated set of fourteen values, eleven in the store 7 and three in the store 9, for the fourteen variables respectively.
  • the fourteen values are plotted on fourteen equally-spaced, parallel axes X01-X14 representing the fourteen variables respectively.
  • the first three axes, X01-X03, are used for the quality variables, and the plots are joined up to form a polygonal line L that is representative of the single fourteen-value operating point of the process.
  • the other sets of process-values are each correspondingly plotted against the same axes X01-X14 resulting in a multiplicity of polygonal lines corresponding to the line L; this is illustrated in part in Figure 3.
  • Each polygonal line is representative of an individual operating point or run of the process from the historical record.
  • convex hulls H for all pairs of adjacent variables of the parallel-axis system are calculated in the unit 11. Between each pair of adjacent axes X01-X14 there will be an upper and lower hull H defining upper and lower limiting boundaries between those two axes, of the operating-point lines. The upper and lower hulls H of the successive pairs of adjacent axes join together to define top and bottom bounds or chains TC and BC respectively. The convex hulls applicable to all other pairings of variables are also calculated. Once the calculation of all the convex hulls has been completed, the upper and lower hulls H between adjacent axes, restricted for simplicity to those parts lying within the range 0 and 1, are communicated to the unit 12 for display as shown in Figure 4.
  • the upper and lower hulls H between adjacent axes are represented in the display as joining up together as top and bottom chains TC and BC respectively, which define (for example, in colour red) the upper and lower boundaries of a region within which feasible operation of the process can take place.
  • top and bottom chains TC and BC respectively, which define (for example, in colour red) the upper and lower boundaries of a region within which feasible operation of the process can take place.
  • the system also operates to determine appropriate warning alarm levels on plant variables during process operation, and to display these alarm levels and the current values of the corresponding variables to the process operator.
  • This is achieved as illustrated in Figure 1, using a ' further electronic display unit 13, however, it is possible for the functions of the units 12 and 13 to be combined into one unit.
  • the display unit 13 is driven from an algorithms unit 14 in accordance with data from the unit 11 and the values of the process variables in real time, supplied from the unit 6.
  • the unit 14 determines which, if any, of the variables have values lying outside a region or zone defined intermediate the top and bottom chains TC and BC.
  • the top and bottom chains TC and BC are defined by the convex hulls applicable between the variables of adjacent axes, but the convex hulls between all the other pairings of variables calculated by the unit 11, overlap them so that there is in general a narrower region or zone within which successful operation can be expected to lie.
  • This narrower zone nominally the 'best operating zone' ('BOZ') defines in relation to each variable the range of value which that variable may have due to the value of each other variable.
  • the unit 14 determines in dependence upon data received from the unit 11, the upper and lower limits of these ranges, and whether the current value of any of the processing variables signalled from the unit 6 is outside the BOZ. If the current value of any variable is outside the BOZ, warning is given by indicating this condition in the display of unit 13 or otherwise, and action is taken to determine what change or changes are appropriate to correct the situation.
  • the display unit 13 provides representation of warning alarm limits for all variables simultaneously. These limits are always calculated using the current values of all the other variables; no model-fitting or statistical assumptions are required, and it is to be understood that reference to 'current' values does not preclude the use of time-lagged values of some variables.
  • the current values Qa-Qh of the process variables derived from the unit 6 are plotted in the display of unit 13 against respective axes Xa-Xh of a system of ten parallel axes Xa-Xj .
  • upper and lower limits for the individual variables a-h calculated by the unit 14 are plotted on their respective axes Xa-Xh and joined up to provide polygonal, upper- and lower-limit lines UL and LL.
  • the lines UL and LL delineate the applicable BOZ.
  • Chains Tc and Be (corresponding to chains TC and BC respectively of Figure 4) and representing bounds defined by the convex hulls between the variables of adjacent axes are shown plotted in Figure 5 (and also in Figures 6 and 7) , but are optional .
  • the values of the quality variables i and j are assumed not to be available at the relevant run-time and so no values for them are plotted or shown in the display. However, upper and lower limits for each of them are calculated by the unit 14 from the current values of the process variables a-h . Although the requirement for the values of the process values a-h to be inside the BOZ sets ranges on the quality variables i and , the ranges due to the current values of the process variables a-h , may be narrower than those specified in selecting the BOZ and hence give useful information.
  • the plotted values Qa-Qh are all within the current best-operating zone defined between top and bottom chains Tc and Be respectively.
  • Upper and lower current limits calculated for the individual variables and plotted on the respective axes Xa-Xj are joined up to provide polygonal lines UL and LL.
  • the lines UL and LL define the zone, namely the BOZ, within which the values of the variables are to be retained.
  • all values Qa-Qh are within the zone, but this is not so in the circumstances of the display illustrated in Figure 6.
  • an upwardly-directed caret UC is displayed on the axis Xg where the line intersects that axis, to signify that the value Qg is either on or below the line UL.
  • the value Qg had been on or above the line UL a similar, but downwardly-directed, carat would have been displayed on the intersection of the line UL with the axis Xg.
  • the process operator can interact with the display unit 13 to adjust one or more of the fixed values Qa-Qc up or down their respective axes experimentally, to see the effect this has on the limits of the other variables.
  • adjusting the value of even one of them may be found to move the limit lines UL and LL outwardly from one another sufficiently to relieve the alarm condition on the others. Accordingly, using the on-line display of unit 13 in this mode, the operator can not only monitor the current settings and results of the process, but can also be made aware of alarm situations and receive guidance in focussed investigation of the remedial action necessary.
  • the system is operable in a mode in which the unit 14 calculates a set of changes in the process variables that can be manipulated, in this example the variables a-c, which will clear the alarm condition.
  • the variables a-c which will clear the alarm condition.
  • These changes can be displayed immediately to the process operator so that he/she may implement them; they may also (or alternatively) be communicated to a process controller for direct, automated implementation.
  • Figure 7 shows the result of automatic operation of the system to calculate changes in the variables a-c which will clear the alarm condition represented in Figure 6.
  • the calculation carried out by the unit 14 in this example determines that the optimum changes required to clear the alarm condition resulting from the current value of Qg outside the BOZ, is change in the values of variables a and c. More particularly, the changes required are reductions of the values of both variables a-c, and such changes are shown in the display by corresponding movements of the dot-representations of the values Qa and Qc to new locations on the respective axes Xa and Xc positions.
  • the unit 14 in response to the change of values re-calculates the upper and lower limits of each variable and indicates them with lines UL and LL in the display of unit 13.
  • the former values of the changed variables are retained as open-centre dots OQa and OQc, and the former limits applicable are shown as lines OUL and OLL, for reference purposes until the changes indicated have been implemented. It may even be useful to show the operator information about earlier time-steps. For example, the lines UL and LL could be faded rather than erased each time a new set of data is received, so that the display of the limits for one time-step would vanish over a pre- selected number of time-steps.
  • the area enclosed by the lines fJ UL and LL might be filled in with solid colour (for example, lime green) that would be faded at each time-step, with the intensities due to a set number of time-steps up to the latest being additive so that the part of the feasible region which has been feasible throughout the largest number of time-steps would have the most intense colour.
  • solid colour for example, lime green
  • the data collected comprises process data accumulated in store 7 and quality data accumulated in store 9.
  • This data as merged in the merge unit 10 is examined to make a selection of all data points which are satisfactory in terms of the values of all the quality and process variables (a "data point” in this regard consists of a set of measurements of all process and quality variables made at one time; some variables may be time-shifted by a constant amount with respect to others) .
  • This selection gives the best operating zone (BOZ), and the envelope (in this example represented by the two-dimensional convex hull) of the BOZ (which does not change during process operation) is used to derive limits on the operating variables. These limits are different at every time-step, and are displayed by the display unit 13 in the manner of Figures 5 to 7.
  • the algorithms unit 14 uses only the process variables whose values are available in real time.
  • the envelope of the BOZ in the multi-dimensional space defined by all the variables is projected into the multi-dimensional subspace defined by the process variables, and the process is required to remain within this projected envelope.
  • This projected envelope is the envelope of the values of the process variables for all historical data- points for which the values of the process variables and the values of the quality variables were satisfactory.
  • the algorithms unit 14 in the present context operates according to a control algorithm as well the alarm algorithm referred to above.
  • the control algorithm is operative to determine first whether any variables are outside their absolute limits. Absolute limits on variables apply regardless of the values of any other variables; absolute limits are shown by upper and lower horizontal lines '1' and '0' respectively of Figures 5 to 7. If any manipulated variables (a-c in the example of Figures 5 to 7) are outside their absolute limits, they will have to be moved within those limits as part of the alarm rectification, but if any non-manipulated variables are outside their absolute limits, the alarm cannot be fully rectified.
  • the main work of the control algorithm is to calculate the best set of moves of manipulated variables to rectify alarms where variables are within their absolute limits but outside the limits due to the values of the other variables.
  • the control algorithm proceeds by calculating the total infeasibility, that is the sum of the sizes of all violations of limits, and the coefficients of this total infeasibility with respect to each of the manipulated variables.
  • the control algorithm derives a suggested set of new values for the manipulated variables from these coefficients. A suitable algorithm for this purpose will now be described.
  • the algorithm to be described uses a planar convex hull as described above, but it would also be possible to derive a control algorithm from the constraints obtained by fitting any convenient closed hyper-geometric figure to the BOZ.
  • Total infeasibility is the sum of the sizes of the differences between the values of all variables that are in alarm, and the limits that they respectively violate. If, for the purposes of illustration, the values of Qa and Qd, are less than the lower-limit values LLa and LLd respectively, calculated for the variables a and d, and the values of Qe and Qh are more than the upper-limit values ULe and ULh respectively calculated for the variables e and , due to the current values of the feasible variables - that is, b , c , f , g .
  • the total infeasibility in this case is equal to:
  • the coefficient of the total infeasibility with respect to a manipulated variable is the sum of the coefficients of the infeasibilities of all alarms with respect to that variable. If the manipulated variable is itself in alarm, the coefficient of the infeasibility of that alarm is +1 if the upper limit is violated and -1 if the lower limit is violated. For all other alarms the coefficient is the amount by which the violated limit will change for a unit change in the manipulated variable, with a positive sign if the limit moves away from the value of the alarmed variable and a negative sign if it moves towards it.
  • n is not equal to m
  • the variable of axis Xm is a manipulated variable. Accordingly, if in the illustration above, a is a manipulated variable, the coefficient of the total infeasibility with respect to the value of the variable a is:
  • Each of the coefficients dLLn/dQm is calculated from the two-dimensional convex hull between axes Xm and Xn.
  • Figure 8 shows how the coefficient is calculated.
  • a convex hull between axes Xm and Xn of parallel co-ordinates has upper and lower chains UCmn and LC n respectively, each comprising an ordered set of points or nodes connected by straight lines; nodes of the upper chain are referenced N1-N5 in Figure 8.
  • the value of variable m plotted on the axis Xm is indicated as Qm, and ULn is the upper limit on the axis Xn due to Qm.
  • ULn is the intercept on the axis Xn of the tangent TU from Qm to UCmn, which touches UCmn at node N4 of chain UCmn.
  • ULn due to Qm is the most restrictive (lowest) upper limit on axis Xn due to any other variable
  • ULn is the overall upper limit on axis Xn.
  • the coefficient dULn/dQm is non-zero, and is given by:
  • x4 is the horizontal distance of node N4 from the axis Xm
  • L is the distance between the axes Xm and Xn.
  • the coefficient dULn/dQm is valid for only a finite range of values of Qm.
  • Figure 8 shows the limiting values VLm and VUm for variable m at which the tangent to the chain UCmn from the point representing value Qm on axis Xm, would touch UCmn at node N5 and node N3 respectively.
  • the limits on the validity of the coefficient of the total infeasibility with respect to a value Qm are the most restrictive limits on the validity of the coefficients of the infeasibilities of the individual alarms. These limits are further restricted if the limits on any other variable, feasible or infeasible, would reach the current value of that variable, or the limit due to Qm ceases to be the most restrictive.
  • control algorithm proceeds iteratively as follows:
  • control algorithm described above which is one of many that could be used, requires the minimum total movement from the current values of the manipulated variables, but other criteria could be used. Constraints and coefficients could be derived from fitting any convenient closed figure to the points of the BOZ instead of the linear convex hull.
  • the recommended new values of the manipulated variables are displayed on display unit 13 ( Figure 1) , and, as referred to above, may also (or alternatively) be used for real-time control of the plant.
  • the control algorithms implemented in known plant-control systems commonly operate by linearising a set of process equations about an operating point, and then solving the resulting set of simultaneous equations to calculate the next variable; this requires constraints (limits) on each variable.
  • Consresponding constraints on each variable are also required by process control systems (sometimes known as 'real-time optimisers') using large sets of equations solved by other means.
  • the constraints in each case are set manually by process operators because no method for calculating the correct values for the limits on the variables, is currently available to them in practice.
  • the envelope method described above can be used to calculate values for the relevant constraints, using the current operating point. These constraints are then used by the control algorithm to calculate a new operating point, and the method then re-applied to calculate new constraints related to this new operating point. This two-phase use of the method and the control algorithm is continued iteratively until convergence to the best solution is obtained.
  • the two-phase method described above has significant advantages. More especially, the control algorithm is made to obey the product quality and other requirements built into the BOZ envelope, not just the requirements on variables that are measured in real time. Secondly, the operator is not left to assign values to variable constraints without a defined logic, thus avoiding the effect of driving the process into a constraint 'corner' from which a satisfactory operating point cannot be found. Thirdly, the closer relationship between the control algorithm connection and the envelope calculation allows the control to achieve a smaller operating region, which is equivalent to more nearly optimal operation of the process.
  • Alarms and recommended process movement may also, or as an alternative, be displayed on a circular plot (sometimes known as a 'radar plot' or 'spider diagram') rather than in the linear forms illustrated in Figures 5 to 7.
  • the axes are arranged as the spokes of a wheel with equal angles between them, as illustrated in Figure 9 for a circular plot of twelve variables Xa-Xl. All calculations are performed using the parallel co-ordinate system, it being the display alone that is transformed to circular form.
  • polygonal lines UL and LL (for example in green, represented here in dashed line) show the current upper and lower limits respectively on the variables. In this case the lines UL connecting the upper limits form a closed figure, as do the lines LL connecting the lower limits.
  • Dots Qa-Ql representing the current values of the process variables are joined by a solid (blue) line to assist the operator to recognise the 'shape' of the current operating point.
  • a caret for example in red
  • a caret UC (as illustrated for axes Xe and Xi) where the upper limit UL is violated
  • a caret LC (as illustrated for axes Xa and Xd) where the lower limit LL is violated.

Abstract

Control of a multi-variable process involves multi-dimensional representation of the values (Qa-Qh) of the process variables according to individual coordinate axes (Xa-Xh), and response based on sets of values for the process-variables accumulated from multiple, earlier operations of the process. An acceptable range (UL-LL) for each process variable due to the current values of the other variables is defined from the accumulated values, and the existence of an alarm condition in which the current value (Qg) of a variable lies outside the range defined for it, is detected and displayed (UC) on the relevant axis (Xg). The change of the values (Qa, Qb) of one or more of the manipulatable variables (a-c) required to rectify the alarm condition is determined iteratively, and implemented by operator or automated response.

Description

Control of Multi-Variable Processes
This invention relates to methods and systems for control of multi-variable processes.
According to one aspect of the present invention a method for control of a multi-variable process, comprises deriving a multi-dimensional representation of the process variables according to individual coordinate axes, defining a feasible region of the process-variables based on sets of values for the process-variables accumulated respectively in relation to multiple operations of the process, defining for each process variable a range due to the values of the other process variables, within which the value of that variable is to lie in order for operation to remain within the feasible region, detecting in respect of each of the variables whether an alarm condition exists in which the current value of that variable lies outside the range defined for it, responding to the existence of said alarm condition to calculate change of the value or values of one or more of the variables required to rectify the alarm condition, and implementing the calculated change.
According to another aspect of the present invention a system for control of a multi-variable process, comprises means for deriving a multi-dimensional representation of the process variables according to individual coordinate axes, means for defining a feasible region of the process-variables based on sets of values for the process-variables accumulated respectively in relation to multiple operations of the process, means defining for each process variable a range due to the values of the other process variables, within which the value of that variable is to lie in order for operation to remain within the feasible region, means for detecting in respect of each of the variables whether an alarm condition exists in which the current value of that variable lies outside the range defined for it, and means responding to the existence of said alarm condition to calculate change of the value or values of one or more of the variables required to rectify the alarm condition, and means for displaying the calculated change.
The definition of the feasible region and the ranges of the process-variables in the method and system of the invention may be carried out using convex hulls or any other closed envelope embracing the data points represented respectively by the accumulated sets of values. A convex hull in orthogonal coordinates is a closed polygon that encloses all relevant data points of the two-dimensional space, whereas in parallel coordinates it is a pair of spaced linear curves that as between corresponding parallel axes, bound the region occupied by the lines that represent (in the parallel- coordinate space) those data points.
The current values of at least some of the process variables and said ranges defined individually for them, may be displayed in said multi-dimensional representation .
The invention is applicable to optimisation and efficient on-line operation of multi-variable processes, and in this respect the calculated change required to rectify the alarm condition may be implemented by control exercised by an operator in accordance with display of the change. Alternatively, the change may be communicated to a process controller for automated implementation .
A method and system according to the present invention will now be described, by way of example, with reference to the accompanying drawings, in which: Figure 1 is a schematic representation of a system according to the invention in the context of collection and utilisation of data derived from operation of a multi-variable processing plant;
Figure 2 is illustrative of a plot in multi-dimensional space defined by parallel coordinate axes, of operation of the multi-variable processing plant of Figure 1;
Figure 3 shows in part a multiplicity of plots corresponding to that of Figure 2 resulting from successive operations of the multi-variable process;
Figure 4 is illustrative of the display of convex hulls in the system of Figure 1;
Figures 5 to 7 are illustrative of successive displays provided according to the invention in connection with operation of the method and system of the invention;
Figure 8 is illustrative of the derivation of certain coefficients utilised in the method and system of the invention; and
Figure 9 shows an alternative form of display representation that may be used in accordance with the invention.
The example of method and system to be described is related to the control of operation of a multi-variable process carried out by a simple, notional processing plant. Details of the plant and its purpose are not of consequence, and indeed the method and system of the invention are related more specifically to operation of the plant as an example of a multi-variable process rather than to the purpose of the process performed, being applicable in the generality to any situation involving a multi-variable process. In the context of initial description of the present specific example, however, there are fourteen variables involved in plant- operation, and of these, eleven are control variables to the extent that their values determine the outcome of the process. The remaining three variables are quality variables in the sense that their values define, or more especially are defined by, that outcome.
Referring to Figure 1, the plant 1 has an input 2 and an output 3 between which there are, for example, a multiplicity of processing stages 4. The processing within each stage 4 is carried out in accordance with one or more variables that, in this example, are regulated by eleven controllers 5. The values of these variables for each operation or 'run' of the process are communicated to a data collection unit 6 to be accumulated in a store 7. The term λrun' in this context may refer to a discrete operation of the process, but it may also refer to what applies at a discrete point in time within continuous operation.
The outcome at the output 3 of each run of the process, is submitted to a unit 8 for analysis in respect of its quality as determined according to three variables. The values of these three quality variables are accumulated in a store 9, so that each run of the process and its outcome is defined by an accumulated set of fourteen values, eleven in the store 7 and three in the store 9, for the fourteen variables respectively.
As the process is run again and again, a multiplicity of different sets of fourteen values are accumulated, and from these a selection is made to provide a historical record in the stores 7 and 9 of successive runs representing satisfactory operation of the process. This record is used in the method of the present invention as a basis for selection of the values of the various variables appropriate to achieving a particular outcome. More especially, the fourteen values of each individual set, eleven in the store 7 and three in the store 9, are brought together in a merge unit 10 and each scaled to the range 0 to 1. The scaled values are then processed in a unit 11 to plot them in an electronic display unit 12. The scaled values of each set are plotted in fourteen-dimensional space using a system of parallel coordinates as illustrated in Figure 2.
Referring to Figure 2, the fourteen values are plotted on fourteen equally-spaced, parallel axes X01-X14 representing the fourteen variables respectively. The first three axes, X01-X03, are used for the quality variables, and the plots are joined up to form a polygonal line L that is representative of the single fourteen-value operating point of the process. The other sets of process-values are each correspondingly plotted against the same axes X01-X14 resulting in a multiplicity of polygonal lines corresponding to the line L; this is illustrated in part in Figure 3. Each polygonal line is representative of an individual operating point or run of the process from the historical record.
Referring further to Figure 3, convex hulls H for all pairs of adjacent variables of the parallel-axis system, are calculated in the unit 11. Between each pair of adjacent axes X01-X14 there will be an upper and lower hull H defining upper and lower limiting boundaries between those two axes, of the operating-point lines. The upper and lower hulls H of the successive pairs of adjacent axes join together to define top and bottom bounds or chains TC and BC respectively. The convex hulls applicable to all other pairings of variables are also calculated. Once the calculation of all the convex hulls has been completed, the upper and lower hulls H between adjacent axes, restricted for simplicity to those parts lying within the range 0 and 1, are communicated to the unit 12 for display as shown in Figure 4. The upper and lower hulls H between adjacent axes are represented in the display as joining up together as top and bottom chains TC and BC respectively, which define (for example, in colour red) the upper and lower boundaries of a region within which feasible operation of the process can take place. Clearly, the larger the number of historical sets of operational data used, with as wide as possible range of values for the individual variables, the more accurately will this region be defined.
The system also operates to determine appropriate warning alarm levels on plant variables during process operation, and to display these alarm levels and the current values of the corresponding variables to the process operator. This is achieved as illustrated in Figure 1, using a' further electronic display unit 13, however, it is possible for the functions of the units 12 and 13 to be combined into one unit. The display unit 13 is driven from an algorithms unit 14 in accordance with data from the unit 11 and the values of the process variables in real time, supplied from the unit 6.
Whenever a new set of values for the process variables is received, the unit 14 determines which, if any, of the variables have values lying outside a region or zone defined intermediate the top and bottom chains TC and BC. The top and bottom chains TC and BC are defined by the convex hulls applicable between the variables of adjacent axes, but the convex hulls between all the other pairings of variables calculated by the unit 11, overlap them so that there is in general a narrower region or zone within which successful operation can be expected to lie. This narrower zone, nominally the 'best operating zone' ('BOZ'), defines in relation to each variable the range of value which that variable may have due to the value of each other variable. Accordingly, the unit 14 determines in dependence upon data received from the unit 11, the upper and lower limits of these ranges, and whether the current value of any of the processing variables signalled from the unit 6 is outside the BOZ. If the current value of any variable is outside the BOZ, warning is given by indicating this condition in the display of unit 13 or otherwise, and action is taken to determine what change or changes are appropriate to correct the situation.
The display unit 13 provides representation of warning alarm limits for all variables simultaneously. These limits are always calculated using the current values of all the other variables; no model-fitting or statistical assumptions are required, and it is to be understood that reference to 'current' values does not preclude the use of time-lagged values of some variables.
The general form of display provided by the unit 13 is illustrated in each of Figures 5 to 7 , and the method of operation outlined above will now be described in relation to them. For illustrative purposes, a process is assumed involving ten variables a-j , of which variables i and j are assumed to be quality variables and variables a-h process variables. Of the process variables, variables a-c are assumed to be manipulatable in the sense of being directly controllable.
Referring to Figure 5, the current values Qa-Qh of the process variables derived from the unit 6 are plotted in the display of unit 13 against respective axes Xa-Xh of a system of ten parallel axes Xa-Xj . At the same time, upper and lower limits for the individual variables a-h calculated by the unit 14 are plotted on their respective axes Xa-Xh and joined up to provide polygonal, upper- and lower-limit lines UL and LL. The lines UL and LL delineate the applicable BOZ. Chains Tc and Be (corresponding to chains TC and BC respectively of Figure 4) and representing bounds defined by the convex hulls between the variables of adjacent axes are shown plotted in Figure 5 (and also in Figures 6 and 7) , but are optional .
The values of the quality variables i and j are assumed not to be available at the relevant run-time and so no values for them are plotted or shown in the display. However, upper and lower limits for each of them are calculated by the unit 14 from the current values of the process variables a-h . Although the requirement for the values of the process values a-h to be inside the BOZ sets ranges on the quality variables i and , the ranges due to the current values of the process variables a-h , may be narrower than those specified in selecting the BOZ and hence give useful information.
As each new set of current values is received from the unit 6, the display changes, and the limits on all the variables are re-calculated and shown in the display of unit 13.
In the case illustrated in Figure 5, the plotted values Qa-Qh are all within the current best-operating zone defined between top and bottom chains Tc and Be respectively. Upper and lower current limits calculated for the individual variables and plotted on the respective axes Xa-Xj are joined up to provide polygonal lines UL and LL. The lines UL and LL define the zone, namely the BOZ, within which the values of the variables are to be retained. In this example, all values Qa-Qh are within the zone, but this is not so in the circumstances of the display illustrated in Figure 6.
The circumstances of Figure 6 are those in which the current values Qa-Qh have changed resulting in consequential re-location of the lines UL and LL. The changes have, for example, located the value Qb virtually on the line LL, but it together with each of the values Qa, Qc-Qf and Qh remain within the currently-applicable BOZ. The value of Qg plotted on the axis Xg is, however, below the lower limit LL, and the unit 14 in response to this signifies the existence of an alarm condition. In this respect and as at least part of the alarm condition, the unit 14 enters a caret (for example of colour red) in the display indicating where, and in which sense, the BOZ has been violated. More particularly, in this example, an upwardly-directed caret UC is displayed on the axis Xg where the line intersects that axis, to signify that the value Qg is either on or below the line UL. In the event that the value Qg had been on or above the line UL a similar, but downwardly-directed, carat would have been displayed on the intersection of the line UL with the axis Xg.
The process operator can interact with the display unit 13 to adjust one or more of the fixed values Qa-Qc up or down their respective axes experimentally, to see the effect this has on the limits of the other variables. When an alarm condition exists, and several variables are on or beyond their limits, adjusting the value of even one of them may be found to move the limit lines UL and LL outwardly from one another sufficiently to relieve the alarm condition on the others. Accordingly, using the on-line display of unit 13 in this mode, the operator can not only monitor the current settings and results of the process, but can also be made aware of alarm situations and receive guidance in focussed investigation of the remedial action necessary.
As well as identifying alarm conditions, however, the system is operable in a mode in which the unit 14 calculates a set of changes in the process variables that can be manipulated, in this example the variables a-c, which will clear the alarm condition. These changes can be displayed immediately to the process operator so that he/she may implement them; they may also (or alternatively) be communicated to a process controller for direct, automated implementation.
In the latter regard, Figure 7 shows the result of automatic operation of the system to calculate changes in the variables a-c which will clear the alarm condition represented in Figure 6.
Referring to Figure 7, the calculation carried out by the unit 14 in this example, determines that the optimum changes required to clear the alarm condition resulting from the current value of Qg outside the BOZ, is change in the values of variables a and c. More particularly, the changes required are reductions of the values of both variables a-c, and such changes are shown in the display by corresponding movements of the dot-representations of the values Qa and Qc to new locations on the respective axes Xa and Xc positions. The unit 14 in response to the change of values re-calculates the upper and lower limits of each variable and indicates them with lines UL and LL in the display of unit 13.
The former values of the changed variables are retained as open-centre dots OQa and OQc, and the former limits applicable are shown as lines OUL and OLL, for reference purposes until the changes indicated have been implemented. It may even be useful to show the operator information about earlier time-steps. For example, the lines UL and LL could be faded rather than erased each time a new set of data is received, so that the display of the limits for one time-step would vanish over a pre- selected number of time-steps. Alternatively, depending on what the process operator found most useful, the area enclosed by the linesfJ UL and LL might be filled in with solid colour (for example, lime green) that would be faded at each time-step, with the intensities due to a set number of time-steps up to the latest being additive so that the part of the feasible region which has been feasible throughout the largest number of time-steps would have the most intense colour.
It follows from what has been described and explained above that it would be necessary to have available within stores 7 and 9 a significant amount of historical operating data; such data is available from the data bases normally forming part of conventional process- instrumentation. Data collected over a number of different non-contiguous periods may be combined, and in some circumstances data generated by a mathematical model may be used.
Referring further to Figure 1, the data collected comprises process data accumulated in store 7 and quality data accumulated in store 9. This data as merged in the merge unit 10, is examined to make a selection of all data points which are satisfactory in terms of the values of all the quality and process variables (a "data point" in this regard consists of a set of measurements of all process and quality variables made at one time; some variables may be time-shifted by a constant amount with respect to others) . This selection gives the best operating zone (BOZ), and the envelope (in this example represented by the two-dimensional convex hull) of the BOZ (which does not change during process operation) is used to derive limits on the operating variables. These limits are different at every time-step, and are displayed by the display unit 13 in the manner of Figures 5 to 7.
While the process is operating, the values of the quality variables accumulated in store 9 will in many cases not be available, as they are obtained from analysis of the product of the process. In this case the algorithms unit 14 uses only the process variables whose values are available in real time. Conceptually, the envelope of the BOZ in the multi-dimensional space defined by all the variables is projected into the multi-dimensional subspace defined by the process variables, and the process is required to remain within this projected envelope. This projected envelope is the envelope of the values of the process variables for all historical data- points for which the values of the process variables and the values of the quality variables were satisfactory.
The algorithms unit 14 in the present context operates according to a control algorithm as well the alarm algorithm referred to above. In the event that one or more alarms are identified from the alarm algorithm at a time-step, the control algorithm is operative to determine first whether any variables are outside their absolute limits. Absolute limits on variables apply regardless of the values of any other variables; absolute limits are shown by upper and lower horizontal lines '1' and '0' respectively of Figures 5 to 7. If any manipulated variables (a-c in the example of Figures 5 to 7) are outside their absolute limits, they will have to be moved within those limits as part of the alarm rectification, but if any non-manipulated variables are outside their absolute limits, the alarm cannot be fully rectified. The main work of the control algorithm is to calculate the best set of moves of manipulated variables to rectify alarms where variables are within their absolute limits but outside the limits due to the values of the other variables. In this regard, the control algorithm proceeds by calculating the total infeasibility, that is the sum of the sizes of all violations of limits, and the coefficients of this total infeasibility with respect to each of the manipulated variables. The control algorithm derives a suggested set of new values for the manipulated variables from these coefficients. A suitable algorithm for this purpose will now be described.
The algorithm to be described uses a planar convex hull as described above, but it would also be possible to derive a control algorithm from the constraints obtained by fitting any convenient closed hyper-geometric figure to the BOZ.
Total infeasibility is the sum of the sizes of the differences between the values of all variables that are in alarm, and the limits that they respectively violate. If, for the purposes of illustration, the values of Qa and Qd, are less than the lower-limit values LLa and LLd respectively, calculated for the variables a and d, and the values of Qe and Qh are more than the upper-limit values ULe and ULh respectively calculated for the variables e and , due to the current values of the feasible variables - that is, b , c , f , g . The total infeasibility in this case is equal to:
(LLa - Qa) + (LLd - Qd) + (Qe - ULe) + (Qh - ULh)
The coefficient of the total infeasibility with respect to a manipulated variable is the sum of the coefficients of the infeasibilities of all alarms with respect to that variable. If the manipulated variable is itself in alarm, the coefficient of the infeasibility of that alarm is +1 if the upper limit is violated and -1 if the lower limit is violated. For all other alarms the coefficient is the amount by which the violated limit will change for a unit change in the manipulated variable, with a positive sign if the limit moves away from the value of the alarmed variable and a negative sign if it moves towards it.
The coefficient of limit LLn with respect to value Qm can be expressed as:
dLLn/dQ
where n is not equal to m, and the variable of axis Xm is a manipulated variable. Accordingly, if in the illustration above, a is a manipulated variable, the coefficient of the total infeasibility with respect to the value of the variable a is:
-1 + dLLd/dQa - dULe/dQa - dULh/dQa.
Each of the coefficients dLLn/dQm is calculated from the two-dimensional convex hull between axes Xm and Xn. Figure 8 shows how the coefficient is calculated.
Referring to Figure 8, a convex hull between axes Xm and Xn of parallel co-ordinates, has upper and lower chains UCmn and LC n respectively, each comprising an ordered set of points or nodes connected by straight lines; nodes of the upper chain are referenced N1-N5 in Figure 8. The value of variable m plotted on the axis Xm is indicated as Qm, and ULn is the upper limit on the axis Xn due to Qm. In this regard, ULn is the intercept on the axis Xn of the tangent TU from Qm to UCmn, which touches UCmn at node N4 of chain UCmn. If ULn due to Qm is the most restrictive (lowest) upper limit on axis Xn due to any other variable, ULn is the overall upper limit on axis Xn. In these circumstances, the coefficient dULn/dQm is non-zero, and is given by:
dULn/dQm = -L/x4
where x4 is the horizontal distance of node N4 from the axis Xm, and L is the distance between the axes Xm and Xn.
In the case shown in Figure 8, reducing the value Qm of variable m will reduce the infeasibility on Xn by increasing ULn towards Qn.
The coefficient dULn/dQm is valid for only a finite range of values of Qm. Figure 8 shows the limiting values VLm and VUm for variable m at which the tangent to the chain UCmn from the point representing value Qm on axis Xm, would touch UCmn at node N5 and node N3 respectively.
The limits on the validity of the coefficient of the total infeasibility with respect to a value Qm are the most restrictive limits on the validity of the coefficients of the infeasibilities of the individual alarms. These limits are further restricted if the limits on any other variable, feasible or infeasible, would reach the current value of that variable, or the limit due to Qm ceases to be the most restrictive.
Using this calculation of the coefficients of the total infeasibility and their limits of validity, the control algorithm proceeds iteratively as follows:
(1) choose the manipulated variable that most reduces the total infeasibility and move it to its limit of validity; (2) recalculate the total infeasibility and all coefficients and limits; and
(3) repeat until the total infeasibility is zero.
When the total infeasibility is zero, the alarm situation has been rectified, and the values of the manipulated variables at this point are the recommended process movement .
The control algorithm described above, which is one of many that could be used, requires the minimum total movement from the current values of the manipulated variables, but other criteria could be used. Constraints and coefficients could be derived from fitting any convenient closed figure to the points of the BOZ instead of the linear convex hull.
The recommended new values of the manipulated variables are displayed on display unit 13 (Figure 1) , and, as referred to above, may also (or alternatively) be used for real-time control of the plant. The control algorithms implemented in known plant-control systems commonly operate by linearising a set of process equations about an operating point, and then solving the resulting set of simultaneous equations to calculate the next variable; this requires constraints (limits) on each variable. Corresponding constraints on each variable are also required by process control systems (sometimes known as 'real-time optimisers') using large sets of equations solved by other means. The constraints in each case are set manually by process operators because no method for calculating the correct values for the limits on the variables, is currently available to them in practice. However, the envelope method described above can be used to calculate values for the relevant constraints, using the current operating point. These constraints are then used by the control algorithm to calculate a new operating point, and the method then re-applied to calculate new constraints related to this new operating point. This two-phase use of the method and the control algorithm is continued iteratively until convergence to the best solution is obtained.
The two-phase method described above has significant advantages. More especially, the control algorithm is made to obey the product quality and other requirements built into the BOZ envelope, not just the requirements on variables that are measured in real time. Secondly, the operator is not left to assign values to variable constraints without a defined logic, thus avoiding the effect of driving the process into a constraint 'corner' from which a satisfactory operating point cannot be found. Thirdly, the closer relationship between the control algorithm connection and the envelope calculation allows the control to achieve a smaller operating region, which is equivalent to more nearly optimal operation of the process.
Alarms and recommended process movement may also, or as an alternative, be displayed on a circular plot (sometimes known as a 'radar plot' or 'spider diagram') rather than in the linear forms illustrated in Figures 5 to 7. In this case the axes are arranged as the spokes of a wheel with equal angles between them, as illustrated in Figure 9 for a circular plot of twelve variables Xa-Xl. All calculations are performed using the parallel co-ordinate system, it being the display alone that is transformed to circular form. Referring to Figure 9, polygonal lines UL and LL (for example in green, represented here in dashed line) show the current upper and lower limits respectively on the variables. In this case the lines UL connecting the upper limits form a closed figure, as do the lines LL connecting the lower limits. Dots Qa-Ql (for example in blue) representing the current values of the process variables are joined by a solid (blue) line to assist the operator to recognise the 'shape' of the current operating point. Where a variable is outside its limits a caret (for example in red) appears at the violated limit, a caret UC (as illustrated for axes Xe and Xi) where the upper limit UL is violated and a caret LC (as illustrated for axes Xa and Xd) where the lower limit LL is violated. The operation of the system to calculate and indicate the corrections required to rectify the alarm condition indicated by the display of Figure 9 is unchanged.

Claims

Claims :
1. A method for control of a multi-variable process, comprising deriving a multi-dimensional representation of the process variables according to individual coordinate axes, defining a feasible region of the process-variables based on sets of values for the process-variables accumulated respectively in relation to multiple operations of the process, defining for each process variable a range due to the values of the other process variables, within which the value of that variable is to lie in order for operation to remain within the feasible region, detecting in respect of each of the variables whether an alarm condition exists in which the current value of that variable lies outside the range defined for it, responding to the existence of said alarm condition to calculate change of the value or values of one or more of the variables required to rectify the alarm condition, and implementing the calculated change.
2. A method according to Claim 1 wherein the feasible region is defined by convex hulls derived between pairs of the process variables.
3. A method according to Claim 1 or Claim 2 wherein the current values of at least some of the process variables and said ranges defined individually for them are displayed in said multi-dimensional representation.
4. A method according to Claim 3 wherein the existence of an alarm condition in respect of any variable represented in the display is indicated on the axis of that variable.
5. A method according to Claim 3 or Claim 4 wherein the axes of the multi-dimensional display representation are parallel to one another.
6. A method according to Claim 3 or Claim 4 wherein the axes of the multi-dimensional display representation are spaced from one another angularly.
7. A method according to any one of Claims 1 to 6 wherein the calculation of change of the value or values of one or more of the process variables to rectify the alarm condition, is an iterative process.
8. A method according to Claim 7 wherein the calculation involves determining for each of the variables for which the alarm condition exists the difference amount by which the value of that variable lies outside the range defined for it, summing all said difference amounts to derive a total of infeasibility, selecting from said one or more of the variables the variable for which change most reduces said total, recalculating said total using a limiting value for the selected variable and repeating the sequence of selection and re-calculation until the re-calculated total of infeasibility is reduced to zero.
9. A method according to any one of Claims 1 to 8 wherein implementation of said calculated change is by control exercised by an operator in accordance with display of said change.
10. A method according to any one of Claims 1 to 8 wherein said change is communicated to a process controller for automated implementation.
11. A system for control of a multi-variable process, comprising means for deriving a multi-dimensional representation of the process variables according to individual coordinate axes, means for defining a feasible region of the process-variables based on sets of values for the process-variables accumulated respectively in relation to multiple operations of the process, means defining for each process variable a range due to the values of the other process variables, within which the value of that variable is to lie in order for operation to remain within the feasible region, means for detecting in respect of each of the variables whether an alarm condition exists in which the current value of that variable lies outside the range defined for it, and means responding to the existence of said alarm condition to calculate change of the value or values of one or more of the variables required to rectify the alarm condition, and means for displaying the calculated change.
12. A system according to Claim 11 wherein, the feasible region is defined by convex hulls derived between pairs of the process variables.
13. A system according to Claim 11 or Claim 12 wherein the current values of at least some of the process variables and said ranges defined individually for them are displayed in said multi-dimensional representation.
14. A system according to Claim 13 wherein the existence of an alarm condition in respect of any variable represented in the display is indicated on the axis of that variable.
15. A system according to Claim 13 or Claim 14 wherein the axes of the multi-dimensional display representation are parallel to one another.
16. A system according to Claim 13 or Claim 14 wherein the axes of the multi-dimensional display representation are spaced from one another angularly.
17. A system according to any one of Claims 11 to 16 wherein the calculation of change of the value or values of one or more of the process variables to rectify the alarm condition, is an iterative process.
18. A system according to Claim 17 wherein the calculation involves determining for each of the variables for which the alarm condition exists the difference amount by which the value of that variable lies outside the range defined for it, summing all said difference amounts to derive a total of infeasibility, selecting from said one or more of the variables the variable for which change most reduces said total, recalculating said total using a limiting value for the selected variable and repeating the sequence of selection and re-calculation until the re-calculated total of infeasibility is reduced to zero.
19. A system according to any one of Claims 11 to 18 wherein said change is communicated to a process controller for automated implementation.
PCT/GB2002/002110 2001-05-08 2002-05-08 Control of multi-variable processes WO2002095514A2 (en)

Priority Applications (6)

Application Number Priority Date Filing Date Title
AU2002255152A AU2002255152A1 (en) 2001-05-08 2002-05-08 Control of multi-variable processes
JP2002591923A JP4156934B2 (en) 2001-05-08 2002-05-08 Multivariate process control
DE60226730T DE60226730D1 (en) 2001-05-08 2002-05-08 CONTROLLING A MULTIPLE-VARIABLE PROCESS
US10/477,340 US7336278B2 (en) 2001-05-08 2002-05-08 Control of multi-variable processes
EP02724460A EP1402324B1 (en) 2001-05-08 2002-05-08 Control of multi-variable processes
CA2446688A CA2446688C (en) 2001-05-08 2002-05-08 Control of multi-variable processes

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
GBGB0111215.0A GB0111215D0 (en) 2001-05-08 2001-05-08 Multi-variable process
GB0111215.0 2001-05-08

Publications (2)

Publication Number Publication Date
WO2002095514A2 true WO2002095514A2 (en) 2002-11-28
WO2002095514A3 WO2002095514A3 (en) 2003-07-17

Family

ID=9914212

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/GB2002/002110 WO2002095514A2 (en) 2001-05-08 2002-05-08 Control of multi-variable processes

Country Status (9)

Country Link
US (1) US7336278B2 (en)
EP (1) EP1402324B1 (en)
JP (1) JP4156934B2 (en)
AT (1) ATE396442T1 (en)
AU (1) AU2002255152A1 (en)
CA (1) CA2446688C (en)
DE (1) DE60226730D1 (en)
GB (2) GB0111215D0 (en)
WO (1) WO2002095514A2 (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1719603A2 (en) * 2005-05-02 2006-11-08 Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. Controlling a production process for extruded profiles
WO2007015078A1 (en) * 2005-07-30 2007-02-08 Curvaceous Software Limited Multi-variable operations
WO2008142386A1 (en) * 2007-05-17 2008-11-27 Rolls-Royce Plc Machining process monitor
EP2560062A1 (en) * 2011-08-16 2013-02-20 ABB Research Ltd. Methods and control systems for controlling an industrial system

Families Citing this family (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
GB0211836D0 (en) * 2002-05-23 2002-07-03 Curvaceous Software Ltd Multi-variable processes
DE10238831A1 (en) * 2002-08-23 2004-03-11 Siemens Ag Method and device for process optimization
GB0320670D0 (en) 2003-09-04 2003-10-01 Curvaceous Software Ltd Multi-variable operations
US7268782B2 (en) * 2003-10-31 2007-09-11 Sap Aktiengesellschaft Smart radar chart
DE502005003969D1 (en) * 2005-02-08 2008-06-19 Tektronix Int Sales Gmbh Protocol tester for a telecommunication system and method for displaying transmission-relevant information in a telecommunication system
DE102005020352A1 (en) * 2005-05-02 2007-08-16 Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. Chemical product`s production process controlling method for e.g. pipe system, involves providing variables by individual adjustment in production process and measuring variables in process, to generate horizon, with three scale arms
GB0717991D0 (en) 2007-09-15 2007-10-24 Curvaceous Software Ltd Multi-variable operations
CN102405448B (en) 2009-04-20 2015-09-09 Abb研究有限公司 Operator terminal in Process Control System
US9513403B2 (en) * 2009-07-27 2016-12-06 Peck Labs, Inc Methods and systems for displaying customized icons
WO2015047241A1 (en) 2013-09-25 2015-04-02 Schneider Electric Buildings Llc Alarm displaying method and apparatus
US10635083B2 (en) * 2017-07-10 2020-04-28 Honeywell International Inc. Apparatus and method for generating smart visualizations for effective monitoring and control in industrial process control and automation systems
JP6984370B2 (en) * 2017-12-06 2021-12-17 横河電機株式会社 Production support system, production support method and program
EP3796119A1 (en) * 2019-09-23 2021-03-24 Siemens Aktiengesellschaft Extended trend display of process data and secondary alarms
KR102302299B1 (en) * 2020-03-05 2021-09-15 (주)코윈테크 Method for control of multiple controller using integrated control program

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0453817A2 (en) * 1990-04-25 1991-10-30 Tektronix, Inc. Test and measurement instrument with digital storage
EP0753283A1 (en) * 1995-07-14 1997-01-15 Hewlett-Packard Company Method and apparatus for comparing a sample with a reference using a spider diagram
US5850339A (en) * 1996-10-31 1998-12-15 Giles; Philip M. Analysis of data in cause and effect relationships
US5917500A (en) * 1998-01-05 1999-06-29 N-Dimensional Visualization, Llc Intellectual structure for visualization of n-dimensional space utilizing a parallel coordinate system
US5949678A (en) * 1993-12-22 1999-09-07 Telefonaktiebolaget Lm Ericsson Method for monitoring multivariate processes
WO2000072268A1 (en) * 1999-05-24 2000-11-30 University Of Massachusetts A performance-based representation for support of multiple decisions

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4571685A (en) 1982-06-23 1986-02-18 Nec Corporation Production system for manufacturing semiconductor devices
US4675147A (en) * 1983-04-06 1987-06-23 Westinghouse Electic Corp. Generating an integrated graphic display of the safety status of a complex process plant
US5339257A (en) 1991-05-15 1994-08-16 Automated Technology Associates Inc. Real-time statistical process monitoring system
JPH0792895A (en) * 1993-09-22 1995-04-07 Tokyo Gas Co Ltd Plant simulator for operator training
US5442562A (en) 1993-12-10 1995-08-15 Eastman Kodak Company Method of controlling a manufacturing process using multivariate analysis
EP0893520B1 (en) * 1997-07-25 2002-10-09 Zellweger Luwa Ag Method for displaying the properties of elongated textile sample bodies
ATE454612T1 (en) * 1999-11-26 2010-01-15 Curvaceous Software Ltd MULTIVARIBLE PROCESS
GB0211836D0 (en) * 2002-05-23 2002-07-03 Curvaceous Software Ltd Multi-variable processes

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0453817A2 (en) * 1990-04-25 1991-10-30 Tektronix, Inc. Test and measurement instrument with digital storage
US5949678A (en) * 1993-12-22 1999-09-07 Telefonaktiebolaget Lm Ericsson Method for monitoring multivariate processes
EP0753283A1 (en) * 1995-07-14 1997-01-15 Hewlett-Packard Company Method and apparatus for comparing a sample with a reference using a spider diagram
US5850339A (en) * 1996-10-31 1998-12-15 Giles; Philip M. Analysis of data in cause and effect relationships
US5917500A (en) * 1998-01-05 1999-06-29 N-Dimensional Visualization, Llc Intellectual structure for visualization of n-dimensional space utilizing a parallel coordinate system
WO2000072268A1 (en) * 1999-05-24 2000-11-30 University Of Massachusetts A performance-based representation for support of multiple decisions

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1719603A2 (en) * 2005-05-02 2006-11-08 Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. Controlling a production process for extruded profiles
EP1719603A3 (en) * 2005-05-02 2008-06-25 Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. Controlling a production process for extruded profiles
WO2007015078A1 (en) * 2005-07-30 2007-02-08 Curvaceous Software Limited Multi-variable operations
US7916140B2 (en) 2005-07-30 2011-03-29 Process Plant Computing Limited Method and a system for operating a controllable multi-variable process
WO2008142386A1 (en) * 2007-05-17 2008-11-27 Rolls-Royce Plc Machining process monitor
EP2560062A1 (en) * 2011-08-16 2013-02-20 ABB Research Ltd. Methods and control systems for controlling an industrial system
WO2013023993A1 (en) 2011-08-16 2013-02-21 Abb Research Ltd Methods and control systems for controlling an industrial system

Also Published As

Publication number Publication date
DE60226730D1 (en) 2008-07-03
US7336278B2 (en) 2008-02-26
GB0111215D0 (en) 2001-06-27
US20040113912A1 (en) 2004-06-17
WO2002095514A3 (en) 2003-07-17
EP1402324A2 (en) 2004-03-31
ATE396442T1 (en) 2008-06-15
JP4156934B2 (en) 2008-09-24
GB2378527B (en) 2005-04-06
CA2446688C (en) 2010-01-26
AU2002255152A1 (en) 2002-12-03
GB2378527A (en) 2003-02-12
CA2446688A1 (en) 2002-11-28
EP1402324B1 (en) 2008-05-21
GB0210503D0 (en) 2002-06-19
JP2004527057A (en) 2004-09-02

Similar Documents

Publication Publication Date Title
CA2446688C (en) Control of multi-variable processes
US6442445B1 (en) User configurable multivariate time series reduction tool control method
EP1435023B1 (en) Method and system for analysing control circuit performance in industrial process
EP2208126B1 (en) Apparatus and method for displaying changes in statistical parameters in a process control system
US7916140B2 (en) Method and a system for operating a controllable multi-variable process
US7623932B2 (en) Rule set for root cause diagnostics
US8380328B2 (en) Method and system for monitoring and controlling a multi-variable process throughout a plurality of distinct phases of the process
JP2003504709A (en) Multivariable process matrix display and related methods
CN111736559A (en) Automation management interface
JP6832327B2 (en) Data-driven method for automatic detection of anomalous workpieces during the production process
JP2006518078A (en) Method for automatic configuration of a processing system
WO2009129042A1 (en) Automated system for checking proposed human adjustments to operational or planning parameters at a plant
US5621668A (en) Prediction control method and a prediction control device
US6879325B1 (en) Multi-variable processes
US7443395B2 (en) Multi-variable operations
CN112650080A (en) Control system for a technical installation with trend profile
JP3218868B2 (en) Predictive control method and apparatus
WO2003009072A1 (en) Systems and methods for isobutyl alcohol (iba) recovery
JP2001067117A (en) Device for monitoring equipment application state
Alinejad et al. Statistical Quality Control Based on the Process Capability Index and Control Charts with Fuzzy Approach (Case Study: Water and Wastewater Company of West Azerbaijan Province)

Legal Events

Date Code Title Description
AK Designated states

Kind code of ref document: A2

Designated state(s): AE AG AL AM AT AU AZ BA BB BG BR BY BZ CA CH CN CO CR CU CZ DE DK DM DZ EC EE ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX MZ NO NZ OM PH PL PT RO RU SD SE SG SI SK SL TJ TM TN TR TT TZ UA UG US UZ VN YU ZA ZM ZW

AL Designated countries for regional patents

Kind code of ref document: A2

Designated state(s): GH GM KE LS MW MZ SD SL SZ TZ UG ZM ZW AM AZ BY KG KZ MD RU TJ TM AT BE CH CY DE DK ES FI FR GB GR IE IT LU MC NL PT SE TR BF BJ CF CG CI CM GA GN GQ GW ML MR NE SN TD TG

121 Ep: the epo has been informed by wipo that ep was designated in this application
WWE Wipo information: entry into national phase

Ref document number: 2446688

Country of ref document: CA

WWE Wipo information: entry into national phase

Ref document number: 2002591923

Country of ref document: JP

WWE Wipo information: entry into national phase

Ref document number: 2002724460

Country of ref document: EP

WWE Wipo information: entry into national phase

Ref document number: 10477340

Country of ref document: US

REG Reference to national code

Ref country code: DE

Ref legal event code: 8642

WWP Wipo information: published in national office

Ref document number: 2002724460

Country of ref document: EP

WWG Wipo information: grant in national office

Ref document number: 2002724460

Country of ref document: EP