The Research Of Predictive Control Algorithm Based On Variable Gain And Application In Cement Rotary Kiln System

Nowadays, in industrial production process, there are numerous variables, and amongwhich the relationship is complex and changeable with strong coupling, and the controlledsystems are often nonlinear, time-varying. So it is difficult to establish an accurate modelby the traditional mechanism modeling, and which lead to process control hardly. With theimprovement of control technology, the model predictive control technology based on theprinciple of predictive model, rolling optimization and feedback correction has become amain way to solve this problem. This paper based on the idea of combination model andvariable gain, through the establishment of RBF-ARMAX variable gain predictive controlmodel to deduce the predictive output, and using the idea of rolling optimization to solvethe problem of predictive control. Specific research works are as follows:First of all, studying the identification of combination model, the idea that the steadystate model is separately identified from the dynamic model is used, the steady state modelof the system with nonlinear is identified by RBF(Radial Basis Function) neural networkalgorithm, the linear ARMAX(autoregressive moving average with exogenous inputs)dynamic model is identified by the push augmented least squares method, the parametersof the ARMAX model obtained on the idea of discrete identification are adjusted online bythe time-varying gain provided from the RBF steady-state model, finally these two modelscombine together and we obtain the RBF-ARMAX variable gain prediction model.Secondly, based on the RBF-ARMAX variable gain model to deduce the modelpredictive control algorithm, and the output variables of the objective function arepredicted in real time through using the sequential quadratic programming optimizationalgorithm, and finally achieve the goal of stable control. the model predictive controlalgorithm is verified by simulation.Finally, according to the object-oriented C#language develop variable gain modelpredicted control software which is based on the RBF-ARMAX variable gain modelpredictive control theory, in order to verify the feasibility of the predictive controlsoftware, we study the process and mechanism of the cement rotary kiln calcinations system, and according to the predictive control software design the cement rotary kilncalcination system predictive controller, and apply it in the cement rotary kiln calcinationsystem.