WO2023193913A1

WO2023193913A1 - Method for configuring model predictive control system, and model predictive control system making use of the method

Info

Publication number: WO2023193913A1
Application number: PCT/EP2022/059272
Authority: WO
Inventors: Shrikant Bhat; Ulaganathan Nallasivam
Original assignee: Abb Schweiz Ag
Priority date: 2022-04-07
Filing date: 2022-04-07
Publication date: 2023-10-12

Abstract

A method for configuring a model predictive control system is provided. The method comprises running a process control of an industrial process, and, while the process control is running, carrying-out in an on-line manner, determining a performance indicator of a performance of the process control, outputting, by an on-site expert feedback component (102), the performance indicator and at least a subset of process control variables to an on-site operator of the process control and providing an input interface to the on-site operator, inputting, by the on-site operator, process optimization information into the input interface of the on-site expert feedback component (102), transmitting, by the on-site expert feedback component (102), the process optimization information to a monitoring and improvement component (103), configuring, via the monitoring and improvement component (103), the process control based on configuration information derived from the process optimization information. Also, a system comprising one or more process components (101) is provided. The process components (101) are controlled by process control involving model predictive control. The system comprises a monitoring and improvement component (103) and an on-site expert feedback, component (102), and the system, is configured to implement the method for configuring the model predictive control system.

Description

METHOD TOR CONFIGURING MODEL PREDICTIVE CONTROL SYSTEM, AND MODEL PREDICTIVE CONTROL SYSTEM MAKING I SE OF THE METHOD

TECHNICAL FIELD

The disclosure relates to a method for configuring a model predictive control system, and to a system comprising one or more process components controlled by process control involving model predict! \e control, and eon figured to implement the method

BACKGROUND

Model predictive control (may be abbreviated as MFC hereinafter) is a method fbr performing process control of a system comprising process components, such as an industrial processing sy stem. MPC makes use ol’a model of the process that the process control is controlling, such as a dynamic process model obtained via system identification When MPG is performed, the process control repeatedly performs a mathematical optimization of the process over a finite time horizon. That is, the optimization is done fbr a process behavior calculated from the process model, considering a finite time period in the future; then, the control thus mathematically optimized is implemented for the current timeslot; and afterwards, the optimization is repealed for the next finite time period in the future, and so on.

In some cases, a process control of a system comprising process components is not equipped or only partially equipped with MPC. For some conventional systems, it is unknown whether equipping the system with MPC. or extending the MPC in die system, is beneficial. In other cases, a process control of a system comprising process components is equipped with MPC. Conventionally, as MPC is a complex approach, the MPC component in a process control system may require tuning and/or improvement due to a deterioration of the MPC' ox er time. For example, the deviation of the process model with respect to the true process behavior increases in the course of time, e.g. because parameters of the process model deviate due to ageing etc. of process components or process component parts. As a consequence, the MPC performance deteriorates.

SUMMARY

A method for configuring an MPC system according to claim 1 is provided. A system comprising one or more process components controlled by process control involving MPC and configured to implement a method as described herein according to claim 18 is provided. Exemplary embodiments arc. among others, defined in the dependent claims.

According to an aspect, a method for configuring an MPC system, or MPC system configuration method, comprises running a process control of an industrial process, and, while the process control is running, carrying out determining a performance indicator, outputting the performance indicator and at least a subset of process control variables, inputting process optimization information, transmitting the process optimization information, and configuring the process control based on configuration information derived from the process optimization information.

According to an aspect, a system comprises one or more process components. The process components are controlled by process control involving MPC. The system comprises a monitoring and improvement component and an on-site expert feedback component. The system is configured to implement a method as described herein, e.g. the MPC system configuration method as described herein.

In an example, determining the performance indicator includes determining a performance indicator of a performance of the process control. In an example, determining the performance indicator includes retrieving the performance indicator. In an example, determining the performance indicator includes determining, or retrieving, a current performance indicator.

In an example, outputting the performance indicator and at least the subset of process control variables includes outputting, by an on-site expert feedback component, the performance indicator and at least a subset of process control variables to an on-site operator of the process control and providing an input interface fc) the on-site operator.

In an example, inputting the process optimization information includes inputing, by the on-site operator, the process optimization information into the input interface of the on-site expert feedback component.

In an example, transmitting the process optimization information includes transmiting, by the on-site expert feedback component, the process optimization information to a monitoring improvement component.

In an example, configuring the process control based on configuration information derived from the process optimization information includes configuring, v ia the monitoring and improvement component, the process control based on configuration information derived from the process optimization information. Tins summary section is provided merely to introduce certain concepts and not to identify any key or essential features of the claimed subject mater. Other features of the inventive arrangements will be apparent from the accompanying drawings and from the following detailed description.

BRIEF DESCRIPTION OF THE DRAWINGS

The inventive arrangements are illustrated by way of example in the accompanying drawings. The drawings, however, should not be construed to be limiting of rhe inv entic e arrangements to only the particular implementations shown. Various aspects and advantages will become apparent upon rex icw of the following detailed description and upon reference to the drawings.

Fig. 1 illustrates a block diagram show ing an exemplary system according to the disclosure.

Fig. 2 illustrates a block diagram showing a part of the system of Fig. 1.

Fig. 3 illustrates an exemplary method for configuring the system of Figs. 1 and 2, according to an embodiment.

Fig. 4 illustrates an exemplary method for configuring the sy stem of Figs. 1 and 2, according to an embodiment.

Fig. 5 illustrates an exemplary method for configuring the system of Figs. I and 2. according to an embodiment.

DETAILED DESCRIPTION

It is believed that the \ arious features described w ithin this disclosure will be better understood from a consideration of the description in conjunction with the drawings. The processes, systems, methods etc, and any x ariations thereof described herein are provided for the purpose of illustration, and shall be construed as a representative basis for teaching a person skilled in the art to employ the features described herein, including variation thereof

This disclosure relates to techniques employed in connection with process control of an industrial process. The process control may include using one or more industrial control systems for carrying out a control strategy that enables or ensures achieving a certain level of the controlled process, for example a consistent production lex cl in a production process or a consistent machining level in a machining process. For example, without any limitation, an industrial processing system as used herein may include an industrial plain such as a chemical plant or a pharmaceutical plant or a food-manufacturing plant, a manufacturing site such as an automotive manufacturing site having a plurality of manufacturing lines, a steel mill, an oil refinery, etc., or any combination thereof As a consequence, for example, w ithout any limitation, an industrial process as used herein includes a controllable process carried out in an industrial plain, such as a chemical process or a pharmaceutical process or a food-manufacturing process, a controllable process earned out in a manufacturing site such as an automotive manufacturing process, a controllable process carried out in a steel mill such as a steel fabrication process, a controllable process carried out in an oil refinery such as an oil refining process, etc. or any combmation thereof

Fig. 1 shows a block diagram of an exemplary system 100 (a control configuration system). The system 100 includes an industrial processing system 200.

As shown in the block diagram of Fig 2, for example, the industrial processing system 20d comprises a plurality of process components 10, 11, 12, 13 that are coupled, or linked or interlinked, with one another to perform an industrial process of the industrial processing system 200. Note that the number of process components 10, 11, 12, 13 is not limited to that shown in Fig. 2 which is merely an example. In general, the number of process components will be two or more. The coupling between the process components I (). 1 1 . 12, 13 is shown by a solid line. Note that the coupling configuration is not limited to that shown in Fig. 2 which is merely an example. For example, there may be a direct or indirect coupling between any one of the process components 10, 1 1, 12, 13. or certain couplings ma\ be missing, as demanded by the actual industrial processing system 200. Each of the process components 10, 11, 12, 13 may carry out a part of an overall process function of the industrial process, such as w irhont limitation - conveying, stirring, mixing, heating, machining etc. The coupling may be - without limitation - a conveying or exchange of process components, such as reactants/starting materials/precursors, interstage/intennediate products, final products etc.

As show n in Fig, 2. the industrial processing s\ stem 200 further comprises a controller 20. The controller 20 is communicatively coupled, or linked or interlinked, with one or more of the process components 10, 11, 12, 13 that need to be controlled, and it is typically communicatively coupled with all process components 10, 11, 12, 13. The communicative coupling is indicated by a dashed line. The controller 20 is configured to control one or more of the process components 10, 11, 12, 13 while at least partially making use of model predictive control ( MPG k MPG makes use of a model of the process that the process control is controlling, such as a dynamic process model obtained via system identification When MPC is performed, the process control repeatedly performs a mathematical optimization of the process oxer a finite time horizon. That is, the optimization is done for a process behavior calculated from the process model, considering a finite time period in the future; then, the control thus mathematically optimized is implemented for the current timeslot; and afterwards, the optimization is repeated for the next finite time period in the future, and so on.

As used herein, the controller 20 may ate be referred as an MPC controller 20, and the process control performed with help of the controller 20 may be referred to as MPC control; it is however noted that in addition to certain MFC-related control approaches via the controller 20 in performing the MPC control, also certain non-MPC control approaches may be employed by one and the same controller 20. Also, where one controller 20 is described herein, it is intended that this description also includes a configuration including multiple control dcx iccs ( not show n) that interact at least partially, such as in a distributed control approach, and that formthe controller 20. That is, the process controller 20 may comprise a single control unit or multiple control units that act together to perform the process control The industrial processing system 2W equipped w i th the controller 2D is configured to run. or cany out. the process control of the industrial process. The process control is carried out, or performed, by the controller 20 which forms a process controller. Generally, the MFC is implemented as digital control.

Referring again to Fig. 1, the system 100 farther includes an MPC configuration component 101 , an on-site expert feedback component 102, and a monitoring and improvement component 103. In Fig. 1, the on-sitc expert feedback component 102 is show n as an element of the MPC configuration component 101 ; however, this is not by way of limitation, and the on-site expert feedback component 102 may be separate from the MPC configuration component 101 and. for example, communicatively coupled with the MPC configuration component 101 to interact with the MPC configuration component 101. The on-site expert feedback component 102 is - directly or indirectly - communicatively coupled with the monitoring and improvement component 103. The on-site expert feedback component 102 may also be coupled with an expert interface 104.

The expert interface 104 may include or provide a human machine interface (HMI) such as, without limitation, a graphical user interface (GUI). The HMI of the expert interface 104 may include an input unit for receiving an input by an expert, or expert operator, such as an MPC expert. The input may, for example, include MPC related x allies, such as parameters. The HMI of the expert interface 104 may include an output unit for providing an output to the expert. The output may, for example, a tagger to the expert to prompt, or request, for input, and/or present various operational data to the expert. The H Ml is not limited to a GUL and may. among others, also include a speech recognition engine, a gesture recognition engine etc.

The expert, such as tire MPC expert, as used herein, is an entity that has detailed knowledge and/or experience at least with MPC, in particular with MPC used in the process control of the industrial process. The expert is, for example, a person particularly skilled in MPC related analysis, configuration, and/or specialist counseling.

Figs. 3 through 5 each illustrate an exemplary method for configuring the MPC system 100 of Figs. 1 and 2. First, foe common parts of the exemplary methods of Figs. 3 through 5 are explained below.

Each exemplary method starts in 1001; a process control of an industrial process is run in the industrial processing system 200 (e.g., via controller 20); and the subsequent actions are carried out in an online-manner: In 1002, a performance indicator of a performance of the process control is determined. Following 1002, in 1003, the performance indicator and at least a subset of process control variables are output, by the on-site expert feedback component 102, to an onsite operator of the process control, and an input interface is provided to the on-site operator. Following 1003, in 1004, process optimization information is input into the input interface of the on-site expert feedback component 102 by the on-site operator. Following 1004, in 1005, the process optimization information is transmited to a monitoring and improvement component I u3 b> the on-site expert feedback component. Following 1005. in 1006, the process control is configured via the monitoring and improvement component 103 based on configuration information that is derived from the process optimization information.

A process control variable, as used herein, is typically a value of a part of the industrial process controlled by the process control that can be varied (e.g., by manual or automatic intervention such as changing a set point) or that varies in the course of time (e.g. by the running process, and without intcncntioii). For example, the process control v ariables comprise both one or multiple values that can be x aned and one or multiple values that vary in the course of time. Inan example, the process control variables include at least one process variable or process parameter. Non-limiting examples of process control variables (including process variables or process parameters) include pressure, temperature, level, flow rate. For example, the varying process control variables are measured. For example, without limitation, the \ arymg controlvariables include a current temperature of a respective one of the process components 10, 11, 12, 13; a current pressure of a respective one of the process components 10, 11, 12, 13; acurrent level of a medium in a respective one of the process components 10, 11, 12, 13; a current flow rate of a medium in a respective one of the process components 10, 11, 12, 13. For example, the process control variables that can be varied include a set-point of the temperature of a respective one of the process components 10, 1 1, 12, 13; a set-point of the pressure of a respective one of the process components 10, 11, 12, 13; a set-point of the level of a medium in a respective one of the process components 10, 11, 12, 13; a set-point flow of the rate of a medium in a respective one of the process components 10, 11, 12, 13. In some examples, the process control variable includes one or more of measurement value and process value.

A performance indicator, as used herein, is typically a value that represents a performance of the industrial process and/or the process control associated therewith. Typically, the performance indicator is a current performance indicator, i.e. a representation of the performance at the current point in time or at a point in time sufficiently near to the current point in time. For example, the performance indicator is a performance rneasurenient value. In an example, the performance indicator indicates or represents a throughput of the industrial process or a ratio of an actual throughput of the industrial process \\ irh respect to an expected throughput of the industrial process. The throughput may be adjustable or influenceable by parameters of the process control, such as set points and/or control parameters, and thus be representative of the process control performance. In another example, the performance indicator indicates or represents a wear or degeneration of a process component involved in the industrial process or a progression of a wear or degeneration of a process component involved in the industrial process. The wear or degeneration may be adjustable or influenceable by parameters of the process control, such as set points and/or control parameters, and thus be representative of foe process control performance. In examples, the performance indicator is at least in part dependent from the process control variables. Note that these are only examples of performance indicators, and various other performance indicators may be dev ised to reflect, or represent, a performance of the industrial process and or the process control.

The performance indicator may be determined automatically. For example, the performance indicator may be determined by way of a calculation performed in a performance indicator calculator. The on-si tc expert feedback component 102 may be configured to retrieve the performance indicator thus calculated from the performance indicator calculator.

On-site, as used herein, typically refers to the location and/or its vicinity, the location defining where the industrial process is running or carried out. An example for the location is a factory where the industrial process is running, an area of a manufacturing line where the industrial process is running or carried out, etc. This location is also referred to as an on-site location. In contrast, off-site, as used herein, typically refers hr a remote location, i.e. a location at a distance from the location where the industrial process is running or carried out. This location is also referred to as an off-site location. The remote location, i e the off-site location, may be communicatively coupled with the on-site location, e.g. via a communications network such as, but not limited to, the internet.

An on-site operator of the process control, as used herein, is typically a person that can at least partially influence or manipulate the industrial process and/or the process control of the industrial process, from within the site where the industrial process is running or carried out. Typically, the industrial process is running while any such manipulation is carried out by the on-site operator. For example, the industrial process is carried out in a factory; the on-site operator has physical access to the factory and can influence or manipulate the industrial process and/or the process control from within the factory. For example, manipulating the industrial process and/or the process control includes monitoring of one or more proc ess control v ariables, c.g. measured v alues of varying process variables, evaluating the monitored process control variables, typically in a process control context, and manipulating one or more other process control v ariables, such as set points, based on the evaluation. Typically, the on-site operator can also at least partially observe or inspect the running industrial process and/or the process control associated therewith, to obtain an on-site operator observation.

An input interface, as used herein, is any means for enabling the on-site operator to perform an input of one or more values, commands, etc. In particular, the input interface may provide a human machine interface ( HM D such as. w ithout limitation, a GUI. The HMI may include an input unit for receiving an input by the on-site operator. The HMI is not limited to a GUI, and may, among others, also include a speech recognition engine, a gesture recognition engine etc.

Process optimization information, as used herein, ty pically includes information about which elements or parts of the industrial process and or the process control associated therew ith ma\ need improvement or optimization. For example, the on-site operator may identify the process optimization information by judging, or evaluating, the performance indicator, possibly making additional use of the on-site operator observation. In embodiments, the process optimization information is based on the performance indicator and the process control variables. For example, the performance indicator and the process control variables are weighted and/or assessed to obtain the process optimization information. In embodiments, the process control includes MPC. The process optimization includes a suggestion for improvement of the MFC, Typically, the monitoring and improx cmcnt component 103 is an MPC monitoring and improvement component.

The process control is configured based on configuration infonnation. The configuration information is derived from the process optimization information. That is, at least a subset of the process optimization is used or considered for determining the configuration information. In other words, the configuration information is taken from the process information and/or calculated on the basis of the process mfomumon v. hi 1c taking into account at least a subset of the process information. Configuring the process control, as used herein, includes one or more of configuring by the monitoring and improvement component and configuring by the expert, or expert operator. Typically, the expert, or expert operator, is located off-site. Configuring the process control may include altering, or tuning, the process control variables (parameters, set points etc. ) In general, the configuring intends to improv e or optimize the process control. Deriving the configuration information may include automatically or manually calculating a parameter change of foe process control based on foe process optimization information, and including the parameter change in the configuration information. Deri \ mg the configuration information may also include automatically or manually calculating a model parameter of foe process model used in connection v ith the M PC.

In an example, configuring via foe monitoring and improvement component 103 includes configuring by the monitoring and improx ement component 103. for example automatically. For example, the monitoring and improvement component 103 is configured to perform a machine-assisted or machine-operated (e,g., a computer-assisted or computer-operated) configuration. In another example, configuring \ ia the monitoring and impro\ ement component includes configuring, using the monitoring and improvement component by the expert, the process control based on configuration infomation derived from tire process optimization information.

In embodiments, configuring the process control includes judging, or assessing, whether to use or to discard foe configuration information. For example, the on-site operator may judge to use or not to use the configuration information that are generated, or derived, with the help of the expert. Using, in this context, may ate include using parts of the configuration information, using and adapting parts of the configuration information, or both, including discarding remaining parts of the configuration information. In embodiments, when it is judged to discard the configuration information, a reason for the judgment maybe provided, and the reason and associated operational data of the process control may be saved alongside, particularly at least the subset of control variables and the reason may be saved alongside. The saved reason and associated operational data of the process control may be included into the process optimization information, and one or more elements may be repeated, in particular one or more of determining the performance indicator, outputring the performance indicator and at least the subset of process control v ariables. inputting rhe process optimization information, transmitting the process optimization information, and configuring the process control based on configuration information derived from the process optimization information.

The techniques described herein enable an improvement, or tuning, of the MFC system. In absence of any such improvement or tuning, the MPC component in industrial process control systems often deteriorate in the course of time, e.g, because of wear or degradation of the process components, resulting in process models used in the MPC becoming gradually inconsistent w ith the actual process behav ior. Thus, v arious kinds of trouble arise When the inconsistency exceeds a certain level, it is often found that turning off the MPT component entirely (and, for example, have the process control run on classical process control approaches only) reduces the trouble. The MPC performance degradation and its probable ultimate discontinuation is not a sudden process but occurs gradually. With the improvement or tuning of the MPC via the techniques described herein, a continuous improv ement or tuning of the MPC system via an expert is facilitated. Continuous improvement or tuning may include an improvement or tuning in predetermined or settable time interv als and/or responsive to an automatic or manual trigger.

For example, the system triggers an evaluation of a disturbance handling with the offrsite expert and captures comments from the expert in the system. The expert may communicate his/her input to the monitoring and improvement component 103. For example, the expert’s input includes suggestions on known and/or unknown disturbance handling of the MPC. In an example, the performance indicator explicitly compares the MPC performance with expert performance. In an example, the expert inputs information on probable causes of disturbances, be thev measured or unmeasured, that could hav e caused the performance degradation. In an example, the expert may suggest addition of at least one of a measurement value, a manipulated variable, a controlled variable, and a known disturbance in the MPC formulation (e.g., the process model) that was beyond the scope of MPC formulation during an earlier MPC implementation and considered relevant subsequently.

In embodiments, the performance indicator comprises a Key Performance Indicator (may be abbrcv iaicd as KPI hereinafter). The KPI is related to a variability of a performance of a controller performing foe process control. For example, foe performance of foe controller is a value indicating or representing an actual control behavior of the controller compared with an ideal control behavior. The KPI may indicate, or represent, foe success of foe actual implementation and/or configuration of the controller when referring it to an ideal implementation and/or configuration thereof The variability of the performance may be a \ ariation of the performance o\ er time. In an example, the KPI is based on the operator and a frequency of set-point changes of the process control.

In embodiments, configuring the process control via the monitoring and improvement component 103 includes providing, to the expert or expert operator, an input interface, and inputting, by the expert operator, foe configuration information, into foe interface. For example, foe input interface includes the expert interface 104 described above, and it may include or provide an HMI such as, without limitation, a GUI. The HMI offoe input interface may include an input unit for receiving the input by foe expert operator. The HMI of foe input interface may include an output unit for providing an output to the expert operator. The output may, for example, a trigger to foe expert to prompt, or request, for input, and/or present various operational data to foe expert. The HMI is not limited to a Gl 1, and may, among others, also include a speech recognition engine, a gesture recognition engine etc.

In embodiments, foe monitoring and improvement component 103 is an ofi-site component. That is, the monitoring and improvement component 103 is located at a distance from the site where the industrial process is running, and it may be conmumcatively coupled with the site where the industrial process is running via a computer network, e.g. foe internet. The distance may be 1 km or more, typically 10 km or more.

In embodiments, the monitoring and improvement component 103 is accessed by a service provider of the MPC system. For example, the service provider enables the expert operator to access the monitoring and improvement component 103. Accessing may comprise reading values or other data from foe monitoring and improvement component 103, inputing values or other data into foe monitoring and improvement component 103, or - typically - both. In embodiments, configuring the process control via the monitoring and improvement component 103 includes presenting the configuration information to the on-site operator, and implementing, by the on-site operator, the configuration information in the process control.

Referring back to Fig. 3, following 1006, in 1010, the method in the example of Fig. 3 ends.

In the example of Fig. 4, following 1006, in 1007, it is decided whether to abort the method. Decision may be performed automatically, or. for example, by any one of the on-site operator and the expert. When it is decided in 1007 ter abort the method, following 1007, the method in the example of Fig. 4 ends. When it is decided in 1007 not to abort the method, the method returns to 1002. However, it is not limited to returning to 1002, and returning may also by performed to any one of 1003, 1004, 1005 or 1006, as appropriate.

The example of Fig. 5 differs from the example of Fig. 3 in that one or more triggers TRIG ! , TRIG2, TRIG 3, TRIG 4 are provided, each representing a corresponding trigger condition. In embodiments, one or more of outputting the performance indicator and at least the subset of process control variables, inputting the process optimization information, transmitting the process optimization information, and configuring the process control based on configuration information derived from the process optimization information, is performed responsive to the respective trigger condition.

For example, trigger TRIG ! may be a request, or event, for outputting the performance indicator and the process control variables to the on-site operator. For example, trigger TRIG2 may be a request, or event, for prompting the on-site operator to input the process optimization information. For example, trigger TRIG3 may be a request, or event, for transmitting the process optimization information to the monitoring and improvement component 103, For example, trigger TRIG4 may be a request for configuring the process control.

In embodiments, the trigger condition pertaining to one or more of triggers TRIG!, TRIG2, TRIG3, TRIG4 includes that the performance indicator exceeds a reference performance. In other embodiments, the trigger condition pertaining to one or more of triggers TRJG1, TRIG2, TRIG3, TRIG4 includes that the performance indicator falls short of a reference performance. In each case, the reference performance may be a predetermined or adjustable value.

In embodiments, the trigger condition pertaining to one or more of triggers TRIG ! , TRIG2, TRIG3. TR1G4 includes that the performance indicator indicates that at least one of the manipulated variables and the controlled variables is saturated. A variable saturation may occur when any part of the industrial system involved in the process control, one or more of the process components, reach a physical limit, such as a positioner reaching its stop, a motor driver reaching a maximum current, an output of an operation amplifier reaching a supply voltage of the OP, etc. In an example, the configuration information includes a measure for relieving the saturation.

In embodiments, the trigger condition pertaining to one or more of triggers TRIG1, TRIG2, TRIG3, TRIG4 includes that the performance indicator indicates a disturbance or a disturbance completion event. A disturbance event may be an expected or unexpected interference, or interaction, of one ormore of the process components, such as. but not limited to an unexpected environmental vibration, a shortage of supply goods, a power drop, etc. A disturbance completion event may be a resolve of the expected or unexpected interference.

Typical use case scenarios are described below:

A performance indicator such as a KPI and relevant operational MPG data are input. Manipulated variables (MV) and controlled variables (CV) of the process are identified, e.g. using machine learning and or artificial intelligence. On detection of MV and GV saturation, the operational data of MPC are analyzed by the expert, and an opportunity to relax MV/CV limits leading to the saturation is detected to improve MPV performance. A quantity by which the MV/CV needs to be relaxed is calculated by the expert to exploit the opportunity. The suggestion is sent to the on-site operator. The suggestion is validated in real time by the expert and the feedback is captured in a hierarchical manner starting from a macro lex cl going deeper into specific aspects of details Visualization of comparative assessment score is performed In case the on-site operator accepts the suggestion, he configures the MPC limits accordingly and MPC exploits configuration changes made to optimize the operation, in case, operator rejects the suggestion, then system captures reason for operator rejection along with the relevant input data are presented to the on-site operator for comments on quality of data, algorithm inputs, and limitation on an implementation of the output. The on-site operator’s response is tinicstampcd and logged with relevant operator feedback and shared in real time with the expert. The expert reviews this information to rate the on-site operator’s input. The expert proposes a probable MPC improvement.

When the MPC goes through a disturbance rejection scenario, performance monitoring component detects the completion of such event and triggers for on-site operator feedback onperformance of the MPC. The feedback is captured in a hierarchical manner starting from a macro level going deeper into specific aspects of details. Visualization of comparative assessment score is performed. The on-^itc operator's response is timestamped and logged with relevant operator feedback and shared in real time with the expert. The expert reviews this information to rate the on-site operator's input The expert proposes a probable MPC improvement.

For example, the MPC monitoring component detects that performance is degrading at a particular time of foe day. 2. It triggers alarm to the expert operator to seek operator feedback on the probable reason and presents relevant background data (along with the timestamp) for expert to review. I he expert feedback can comprise of t bin not limited to ) follow ing:

- Probable external disturbances (changes to the existing unit operating conditions OR external changes beyond the specific operation under consideration)

- Expert’s assessment of MPC performance and suggestion on better way of handling the disturbance (recommendation on additional disturbance variables - this implies moving unknown disturbance to known disturbance)

- In case of MPC’s performance beter than anticipated by expert, he also reflects his understanding on the probable factors contributing to improx ed performance by MPC. This improves cxplainability and customer confidence (which can be referred by other experts).

All the above observations are logged and shared with th® internal experts team as well as serv ice prov ider to analyze, v alidate and accordingly plan and recommend suitable actions for MPC performance improvement.

In an example, the expert can engage with the system even without a system trigger but based on expert’s anticipation of improved MPC performance compared to current performance .

In an example, all MPC applications in a specific plant can be centrally coordinated for an expert feedback to find common aspects, such as same disturbance affecting multiple M PC’s.and accordingly control strategies to address such disturbances can be implemented.

By employing the techniques as described herein, beneficially, the expert engages continuously with the operator, which leads to a continuous service opportunity for other MPC performance monitoring and improvement as well as additional MPC implementation. The description of the inventive arrangements provided herein is for purposes of illustration and is not intended to be exhaustive or limited to the form and examples disclosed, The terminology used herein was chosen to explain the principles of the inventive arrangements, the practical application or technical improvement over technologies found in foe marketplace, and/or to enable others of ordinary skill in the art to understand the inventive arrangements disclosed herein. Modifications and variations may be apparent to those of ordinary skill in theart without departing from the scope and spirit of the described inventive arrangements. Accordingly, reference should be made to the following claims, rather than to the foregoing disclosure, as indicating the scope of such features and implementations.

Claims

1. A method for configuring a model predictive control, MPC, system, the method comprising:

- running a process control of an industrial process; and

- white the process control is running, carrying out in an on-line manner: a) determining (1002) a performance indicator of a performance of foe process control; b) outputing (1003), by an on-site expert feedback component (102), the performance indicator and at least a subset of process control variables to an on-site operator of the process control and providing an input interface to the on-site operator; c) inputting ( 1(104). by the on-site operator, process optimization information into the input interface of the on-site expert feedback component (102); d) transmiting (1005), by the on-site expert feedback component (102), the process optimization information to a monitoring and improvement component ( 103 ): e) configuring ( 1006), via the monitoring and improvement component ( 103), foe process control based on configuration information derived from the process optimization information.

2. The method of claim 1, wherein the performance indicator comprises a Key Performance Indicator, KPI, related to a variability of a performance of a controller performing the process control based on the operator and a frequency of set-point changes of the process control.

3. The method of claim 1 or 2, wherein configuring the process control via the monitoring and improvement component (103) includes providing, to an expert operator, an input interface, and inputting, by the expert operator, the configuration information, into the interface.

4. The method of claim 3. w herein the monitoring and improx emem component ( 103 ) is an off-site component.

5. The method of claim 4, v herein the monitoring and improx emenr component ( 103) is accessed by a service provider of the MPC system.

6. 1 he method of any one of the preceding claims, xx herein configuring the process control via the monitoring and improvement component (103) includes presenting the configuration information to the on-site operator, and implementing, by foe on-site operator, the configuration information in the process control.

7. The method of any one of foe preceding claims, wherein the process optimization information is based on foe performance indicator and the process control variables.

8. The method of any one of foe preceding claims, wherein the process control includes MPC. and the process optimization information includes a suggestion for improv ement ofthe MPC.

9. The method of any one of foe preceding claims, wherein configuring foe process control includes judging whether to use or discard the configuration information.

10. The method of claim 9, further comprising:

- when it is judged to discard foe configuration information, providing a reason for foe judgement;

- saving the reason and associated operational data of the process control, particularly at least foe subset of process control variables; including the saved reason and associated operational data of the process control into foe process optimization information; and

- repeating one or more element of foe method according to any one of foe preceding claims, in particular elements a-c.

1 1. The method of any one of the preceding claims, wherein one or more of b), c), d), and e) is performed responsiv e to a trigger condition 12. The method of claim 11, wherein foe trigger condition includes foe performance indicator exceeding or falling short of a reference performance. 13. The method of claim 11 or 12, wherein the trigger condition includes the performance indicator indicating that at least one of the manipulated v ariables and the controlled variables is saturated. 14. The method of claim 13, wherein the configuration information includes a measure for relieving the saturation. 15,. The method of any one of claims 11-14, wherein the trigger condition includes the performance indicator indicating a disturbance or a disturbance completion event. 16.. The method of any one of the preceding claims, wherein the performance indicator is at least in part dependent from the process control variables. 17.. The method of any one of the preceding claims, wherein the process control variables include one or more of manipulated variables, MV, and controlled \ ariablcs, C'V. of ihe process. 18.. A system comprising one or more process components (101), the process components 1 10 ] ) controlled by process control involving model prcdictix c control. MPC. the system comprising a monitoring and improvement component (103) and an on-site expert feedback component (102), the system configured to implement a method according to any one of the preceding claims, the on-site expert feedback component (102) configured to, at least, output the performance indicator and at least the subset of process control variables and provide an interface for inputing the process optimization information, and the monitoring and improx ement component ( 103 ) configured to, at least, configure the process control based on the configuration information derived from the process optimization information.