| Modeling and control of the complicated process is the main research subject in controltheory and control engineering. It is also the motivity of industrial automation. In this paper, apolymerization process is chosen as the studying object.Since the chemical process is highly non-linear, time-varying, huge-lag, uncertain andcomplicated, it is difficult to build the mathematic model. The model based on reactionmechanism can' t satisfy the traditional control because of ignoring many factors. So to modeland control these chemical reactions, the advanced theory and methods including neuralnetworks, fuzzy control and advanced control should be used.In this paper, the propylene polymerization is used as a research object and the input andoutput datum of the object are used to model. If one neuron network is used to map the wholesample space, it is hard to describe the nonlinear region precisely. In order to solve the problemof the single neuron network that it can' t fit in the huge range perfectly, a distributed fuzzynetworks is constructed by using fuzzy categorizer and ANN. ANN is used here to build thenonlinear sub-process to the chemical reaction modeling issue. ANFIS modeling is discussedfurther in this paper and applied to simulate chemical reaction datum of the real industryproduction.To deal with a water level control problem in processing control, fuzzy controller andfuzzy adaptive PID controller are designed respectively based on fuzzy controlling methods.The approaches are realized in a process control experimental device. Compare the fuzzycontroller with average PID controller, the results of experiments show that the fuzzy controlhas more of superiority.Active disturbances rejection control (ADRC) has many merits such as no overshootingand steady error. This paper introduces each components of ADRC and summarizes the issuesof parameter adjusting. ADRC algorithm is used to control complex chemical processes. Theresults of simulation show that the method is much better.
|
PDF Full Text Request