Multiple-model predictive control based on fuzzy adaptive weights and its application to main-steam temperature in power plant

Abstract

In order to solve the uncertainty and time-varying problems of the complex process during variable operations, a novel multiple-model predictive control system based on fuzzy adaptive weighted is investigated in this paper. Firstly, based on the model-set selecting method for multiple model adaptive control, this system preserves the performance of multiple-model approximating nonlinear property of controlled plant at close quarters, therefore the control system is robust. Secondly, fuzzy adaptive weighted control algorithm is proposed to overcome the output disturbance that caused by model switching. The weighted values are obtained by fuzzy decision-making. In this way the corresponding controller can be switched smoothly when the model is selected through switching. Thirdly, because the output predictive value of practical controlled plant is calculated by weighted average of respective sub-models output probabilities, it can rapidly image the variation of plant characteristic. Meanwhile, the output value of controller is optimized by dynamic matrix control algorithm, so the system has better dynamic performance. Furthermore, because of the convenient design and good real-time performance, this algorithm is of great significance and practical engineering value. This algorithm is applied to the main-steam temperature of a supercritical 600MW Once-through boiler; simulation experiments demonstrate the feasibility and good performance of the proposed approach compared to the former approaches.