Monitoring Model Predictive Control Systems Using Pattern Classification and Neural Networks

A novel, pattern classification approach is proposed for monitoring the performance of model predictive control (MPC) systems. Current MPC operation is compared to a simulated database of closed-loop MPC system behavior, containing various combinations of disturbances and plant changes. Neural network-based pattern classifiers are used to classify the MPC performance as normal or abnormal and to determine whether an unusual disturbance or significant plant change has occurred. If a plant change is detected, other classifiers are used to diagnose the specific submodel(s) that are no longer accurate. The proposed methodology is successfully demonstrated in a detailed case study for the Wood-Berry distillation column model.

[1]  Sirish L. Shah,et al.  Multivariate Controller Performance Analysis : Methods , Applications and Challenges , 2002 .

[2]  Biao Huang,et al.  Multivariable model validation in the presence of time-variant disturbance dynamics , 2000 .

[3]  Ali Cinar,et al.  MULTIVARIABLE MPC PERFORMANCE ASSESSMENT, MONITORING AND DIAGNOSIS , 2002 .

[4]  Biao Huang,et al.  A case study on performance analysis and trouble shooting of an industrial model predictive control system , 1999, Proceedings of the 1999 American Control Conference (Cat. No. 99CH36251).

[5]  Janos Gertler,et al.  Sensor and actuator fault isolation by structured partial PCA with nonlinear extensions , 2000 .

[6]  Biao Huang,et al.  Minimum Variance Control and Performance Assessment of Time-Variant Processes , 2000 .

[7]  Biao Huang,et al.  An investigation into the poor performance of a model predictive control system on an industrial CGO coker , 2000 .

[8]  S. Joe Qin,et al.  A survey of industrial model predictive control technology , 2003 .

[9]  Manfred Morari,et al.  Propositional logic in control and monitoring problems , 1997, 1997 European Control Conference (ECC).

[10]  Derrick J. Kozub.  CONTROLLER PERFORMANCE MONITORING AND DIAGNOSIS. INDUSTRIAL PERSPECTIVE , 2002 .

[11]  Simon Haykin,et al.  Neural Networks: A Comprehensive Foundation , 1998 .

[12]  R. K. Wood,et al.  Terminal composition control of a binary distillation column , 1973 .

[13]  Thomas F. Edgar,et al.  Performance assessment of constrained model predictive control systems , 2001 .

[14]  Christos Georgakis,et al.  Controller performance assessment based on minimum and open-loop output variance , 2000 .

[15]  Gwilym M. Jenkins,et al.  Time series analysis, forecasting and control , 1972 .

[16]  Michèle Basseville,et al.  Detection of abrupt changes: theory and application , 1993 .

[17]  C. T. Seppala,et al.  A review of performance monitoring and assessment techniques for univariate and multivariate control systems , 1999 .

[18]  S. Joe Qin,et al.  Control performance monitoring — a review and assessment , 1998 .

[19]  Raghunathan Rengaswamy,et al.  A fast training neural network and its updation for incipient fault detection and diagnosis , 2000 .

[20]  Frederick W. Faltin,et al.  Statistical Control by Monitoring and Feedback Adjustment , 1999, Technometrics.

[21]  Sigurd Skogestad,et al.  Simple analytic rules for model reduction and PID controller tuning , 2003 .

[22]  Jose A. Romagnoli,et al.  Real-time MPC supervisory system , 2000 .

[23]  F G Shinskey,et al.  How Good are Our Controllers in Absolute Performance and Robustness? , 1990 .

[24]  Jan M. Maciejowski,et al.  Predictive control : with constraints , 2002 .

[25]  Dale E. Seborg,et al.  Controller performance assessment based on setpoint response data , 1999, Proceedings of the 1999 American Control Conference (Cat. No. 99CH36251).

[26]  S. Qin,et al.  Detection and identification of faulty sensors in dynamic processes , 2001 .

[27]  Biao Huang.  Minimum Variance Control and Performance Assessment of Time-Variant Processes , 2000 .

[28]  Biao Huang,et al.  Model validation for industrial model predictive control systems , 2000 .

[29]  Michael A. Henson,et al.  A performance measure for constrained model predictive controllers , 1999, 1999 European Control Conference (ECC).

[30]  Thomas F. Edgar,et al.  Performance assessment of multivariable feedback control systems , 2001, Autom..

[31]  Sirish L. Shah,et al.  Issues in performance diagnostics of model-based controllers ☆ , 2002 .

[32]  Weibiao Zhou,et al.  PROCESS/MODEL MISMATCH COMPENSATION FOR MODEL-BASED CONTROLLERS , 1989 .

[33]  Biao Huang,et al.  Perspectives on Multivariate Feedforward/Feedback Controller Performance Measures for Process Diagnosis , 1997 .

[34]  Carl G. Looney,et al.  Pattern recognition using neural networks: theory and algorithms for engineers and scientists , 1997 .

[35]  Sirish L. Shah,et al.  Performance Assessment of Control Loops: Theory and Applications , 1999 .

[36]  V. Manousiouthakis,et al.  Best achievable decentralized performance , 1995, IEEE Trans. Autom. Control..

[37]  Biao Huang,et al.  Robust performance assessment of feedback control systems , 2002, Autom..

[38]  Biao Huang,et al.  Feedback and Feedforward Performance Analysis of the Shell Industrial Closed Loop Data Set , 1997 .

[39]  C. T. Seppala,et al.  Recent Developments in Controller Performance Monitoring and Assessment Techniques , 2002 .

[40]  Jay H. Lee,et al.  Diagnostic Tools for Multivariable Model-Based Control Systems , 1997 .

[41]  Sigurd Skogestad,et al.  Limitations of dynamic matrix control , 1995 .

[42]  T. Edgar,et al.  Assessment of achievable PI control performance for linear processes with dead time , 1998, Proceedings of the 1998 American Control Conference. ACC (IEEE Cat. No.98CH36207).

[43]  Rohit S. Patwardhan,et al.  Studies in synthesis and analysis of model predictive controllers , 1999 .

[44]  Biao Huang,et al.  Trouble shooting and performance audit for a multivariable control system: an industrial case study , 1999, Engineering Solutions for the Next Millennium. 1999 IEEE Canadian Conference on Electrical and Computer Engineering (Cat. No.99TH8411).

[45]  T. J. Harris,et al.  Performance assessment of multivariable feedback controllers , 1996, Autom..

[46]  T. Minges,et al.  Use dry gas seals on cracked gas applications : Loss Prevention/Reliability , 1998 .

[47]  Mark A. Kramer,et al.  Diagnosing dynamic faults using modular neural nets , 1993, IEEE Expert.