Nonlinear Predictive Control And Its Applications On The Hardware-in-Loop Simulation Model

There are a large number of nonlinear systems in industrial process, which are nonlinear time-variant and uncertainty. With the development of modern industry and progress of science technology, linear model predictive control could not satisfy the control performance. Therefore, the research on nonlinear predictive control has become an important issue in the control engineering field. In this dissertation, two types of nonlinear model predictive control algorithms were studied and their effectiveness was verified through simulation research.Finally, the nonlinear model predictive control algorithm based on HammersteinWiener model was applied on the Hardware-in-the-loop(HIL) Simulation model.The main contents are as follows:(1) A HIL Simulation model based on real-time control software Quarc and Simulink——Coupled Water Tanks control system was introduced.By using hydrodynamics theory, the mechanistic models of the single-volume water tank and double-volume water tank were obtained and their nonlinear mathematic models were established through Simulink.(2) The nonlinear predictive control based on BP neural network was studied. The multilayer feed-forward neural network was adapted as the predictive model, and the neural network was trained by the highly accurate algorithm——Levenberg Marquardt algorithm. The derivative equations of neural network were deduced, and the Newton-Raphson algorithm was utilized to minimize the performance function. Simulation research was done through Matlab/Simulink combined with S-function for the case of single-volume water tank and double-volume water tank, and the results demonstrated the effectiveness of the proposed scheme.(3) The nonlinear predictive control based on Hammerstein_Wiener model was studied. With a gray-box identification approach, the nonlinear object was described as Hammerstein model and Wiener model. The model is composed of a static nonlinear model and a linear dynamic model with1steady-state gain. Therefore, the static nonlinearity relationship can be described by the static input-output relationship. Then a predictive control algorithm based on this idea was designed and tested on Control Valve based on Wiener model and Heat Exchanger based on Hammerstein model.(4) The nonlinear predictive control based on Hammerstein_Wiener model was applied on the HIL simulation model of a Coupled Water Tanks device. The experiments results validate the algorithms proposed.