Research And Application Of Two-Step Predictive Control Based On Fuzzy Control And Genetic Algorithm On Three Level System

Predictive Control is widely used in industrial processes, while is often limited to linear systems, however, it can not get good control effects in the actual production for the vast majority of nonlinear systems. Hammerstein model can transform nonlinear system into the form of a static nonlinear part with a dynamic linear part. Through the application of predictive control algorithm on the linear part, the intermediate variable can be calculated, and then the actual control variables can be back-calculated through the non-linear part. In the above two-step predictive control algorithm, the design of controller is still confined in linear categary, but sometimes it is difficult to solve the value of the actual variables impacting on the nonlinear systems as a result of the uncertainty of the static non-linear aspect, and can not get the real solution of the global value of control variables. Control action may be saturated, while the existence of solving errors of non-linear algebraic equations(group) is inevitable.The control parameters of the dynamic link often affect the control effects of systems using the method of GPC, such as the system output, stability and robustness systems, and it lacks of corresponding analytic relationship between the parameters and the control effect. So it often takes the method of trial and error to make the parameters tuned, after which the value of control variables of the system can be calculated using the tuned parameters. Such process not only consumes a lot of time, but also it is difficult to guarantee the control quality.In this thesis, the main focus is on nonlinear systems, using the idea of two-step predictive control. Namely it applies Hammerstein model, takes the use of generalized predictive control (GPC) to control the dynamic part separated according to the Hammerstein model, and solves the value of intermediate variables. In the GPC operation, it optimizes the adjustable parameters combined with the method of genetic algorithm, and then obtains the optimal solution set, the process of which is an off-line adjustment, after which applies the set in in-line calculation; it then establishes the fuzzy controller, makes the intermediate variables solved by GPC operation as the input of fuzzy controller, and makes the real-time control variables impacting on the object as the fuzzy controller's output, and then establishes the fuzzy rules between the fuzzy input and output, makes the solution of the value of the actual control variables, which is equivalent to replace the fuzzy control for nonlinear equations (group), and solves the dilemma that calculation of the input of non-linear is failed and even can not be calculated. It applies the above theoretical study to the actual process of the control object in three-level control system of MPCE-1000 devices, realizes the tracking control of the height of the third level for the equipment using above algorithm, and then analyzes its control nature such as stability and robustness, and through comparison with PID control, indicates the effectiveness and superiority of this method.