Study And Application Of Fuzzy Model Based Predictive Control

As the continuous development and wide application of system control theory in industry, the systems of the industrial process control become more complex due to multi-input and multi-output, time varying characteristics, coupling, time-delay, nonlinearity, uncertainty, high dimension, etc. So that it is not easy to modeling and control. The main content in this dissertation is about Takagi-Sugeno (T-S) modeling and taking the algorithm of T-S model based generalized predictive control (GPC) as a valid approach in solving nonlinear system control problems, and it is applied to pH neutralization process control simulation successfully.Main research work and achievements are summarized as following:1. T-S fuzzy modeling algorithm is researched, including making use of fuzzy cluster algorithm for antecedent part modeling and least square algorithm for the rule part modeling. Studied fuzzy cluster algorithm, an improved sub-cluster algorithm based on space density of dataset is proposed, with fewer parameters needed to be adjusted, it can quickly determine the satisfactory cluster centers to achieve T-S fuzzy model, and is also applied to soft-sensor in industrial project, as a result it can well predict the output.2. Predictive control theory of single variable and multi-variable is studied; succeed in making true of generalized predictive control (GPC) based on Takagi-Sugeno (T-S) model in nonlinear system control. T-S model based GPC control is applied to pH neutralization process, and the modeling of T-S model achieves a satisfied approximation precision to the process. Multiple model predictive control (MMPC) strategy based on T-S model is researched, that is to study the problem of model error within predictive horizon, and control simulation shows that single-step linear model predictive control gains better real-time control results than MMPC. Comparisons of control results show that T-S model based GPC control achieved better performance than conventional PID control over different operating regimes, such as fast and consistent response, quick disturbance rejection, and it turns out to be controlled well even when system is close coupled or with disturbance.Control simulation results of pH neutralization process show the effectiveness and promising performance of T-S model based GPC control for nonlinear systems, such as nonlinear modeling, control optimization computation, coupled system control, disturbance rejection, etc. The application to soft-sensor in advanced control project of hydrogenation cracking process has proved the effectiveness of the improved sub-cluster algorithm.