Study On Predictive Control Based On Soft Measurement In Cold Rolling Mill Agc Systems

Due to the existence of various interferences, the actual industrial processes arecomplex processes of multivariable, strong coupling, time-variant and constraint, so it isdifficult to establish accurate mathematical model using the classical control theory, theapplication of traditional control has been limited. Intelligent control strategy, which hasquickly developed in recent years, can be a good solution to this problem. Predictivecontrol as a kind of intelligent control method absorbs optimization theory in moderncontrol theory, the main idea of predictive control is using the model prediction, rollingoptimization and feedback correction to realize local optimum; In complex industrialprocess, some important parameters impossible or very difficult to directly measure by thesensors, but it must be strictly monitored and controlled. This article will use a softmeasuring method, which by collecting the easily obtained auxiliary variables to achieveoptimal estimation of the dominant variables. In this paper, we will study the cold rollingautomatic gauge control system with large hysteresis characteristics, the main workfollows:1、Based on deeply understanding of the basic concepts, the causes of thicknessdifference and principles of conduct of AGC, this paper describes several forms of AGCand their control theory which are most widespread in engineering practice, and thenestablish a brief mathematical model of hydraulic AGC system of cold rolling mill.2、 Because of the thickness gauge installation location and using the feedback formthickness control system, thickness gauge feedback AGC has a large hysteresischaracteristic. Traditional control over-reliance on precise model, this paper introduces apredictive control strategy combined with soft measurement, which doesn’t need accuratemodel. Firstly, we use fuzzy RBF neural network to establish a cold rolling strip exitthickness soft measurement model; then in terms of a large quantity online calculation ofGPC, not suitable for nonlinear system control, this paper introduces a generalizedpredictive control strategy model combined with soft measurement. Lastly, this papershows how this strategy apply in cold rolling strip exit thickness of thickness gaugeFeedback AGC control system.3、The simulation study of soft measurement model and generalized predictive controlstrategy model combined with soft measurement. The simulation results show that the softmeasurement model can realize optimal estimation of dominant variables, and achieve highaccuracy; generalized predictive control strategy model combined with soft measurementcan have a good control performance in exit thickness of cold mill strip, the model canreduce computation, improve the real-time and anti-interference ability of control system, improve large hysteresis characteristics of the system.This paper has some reference value for the study of large-hysteresis system controlstrategy.