A Study On Fuzzy Modeling And Control Based On Immune Optimization Algorithms For Thermal Processes

Automatic control for thermal processes is an indispensable technical measure to guarantee a safe and economical operation of thermal equipments, and the whole thermal process model in all conditions to be built accurately is the foundation of process optimal control. However thermal processes generally contain nonlinearity and uncertainty, as a result, it is difficult to build the nonlinear models which can express thermal processes accurately and the global optimal control is impossible by traditional modeling and control methods. So it is necessary to make a deep study of advanced nonlinear modeling and control approaches which can be applied to thermal processes to efficiently increase the safety and economical efficiency of power unit.Immune system has been proven to be of immune memory, immune recognition, massive distribution, robust adaptation and so on. Introducing these immune mechanisms to process control field can provide a new approach to resolve the problems such as uncertain system modeling and dynamic adaptive controlling of complex processes. In this paper immune mechanisms are applied to improvement of genetic algorithm (GA), optimizing design of fuzzy control, fuzzy modeling based on clustering, and nonlinear predictive control, thus some difficult problems in nonlinear modeling and control are solved. In addition, simulation studies for thermal processes of power plant such as steam temperature system, unit load system are carried out. The main contents of this paper are as follows:(1) Analyzing the cause of limitations of traditional genetic algorithms, this paper puts forward two improved immune genetic algorithms(IGA) by introducing the immune mechanisms such as antigen recognition, antibody diversity, immune memory and concentration regulation into GA. One is the improved single population IGA, which uses an elite inheritance strategy of antibodies in memory cells to make the algorithm converge to the global optimal point more rapidly and stably; another is multi-population IGA, which separates genetic competition process into two steps, competition among sub populations and competition among individuals in a sub population, and can resolve the conflict between global and local search abilities. Experimental results of some test functions demonstrate that both the two improved IGAs proposed have superior performances in optimizing.(2) A fuzzy controller optimizing design method based on immune genetic algorithm is proposed. With the help of concatenated-encoding method, the parameters of fuzzy rules and membership functions can be simultaneously optimized by the improved IGA. This new design method of fuzzy controller has been applied to a superheated steam temperature system, and simulation result shows that the shortcomings of standard genetic algorithm can be overcome and the optimal fuzzy rule base and membership functions can be obtained more efficiently by using the improved IGA to optimize fuzzy controller, and the fuzzy control system optimized has a satisfactory closed-loop performance.(3) A novel dynamic immune evolutionary clustering algorithm(DIECA) is proposed to overcome the shortcomings of fuzzy modeling methods based on conventional clustering algorithms that fuzzy rule number should be determined beforehand. DIECA can search for the optimal cluster number by using the improved genetic techniques to optimize string lengths of chromosomes; at the same time, the convergence of cluster center parameters is expedited with the help of fuzzy C-means(FCM) algorithm. Moreover, by introducing memory function and vaccine inoculation mechanism of immune system, DIECA can converge to the optimal solution rapidly and stably. The proper fuzzy rule number and exact premise parameters can be obtained simultaneously when using this efficient DIECA to identify fuzzy models. The efficiency of the proposed fuzzy modeling method is proved by some typical simulation examples. When the method isAbstract applied to thermal processes, the nonlinear fuzzy models with high precision can be obtained with less calculation and the fuzzy rule numbers needn't to be determined beforehand, which establish a model foundation for global optimal control of thermal processes.(4) A nonlinear predictive control method based on fuzzy model and immune optimization algorithm is proposed. An accurate global fuzzy model which is obtained by off-line identification serves as the predictor of the future behavior of the process, then the nonlinear rolling optimization problem of predictive control is solved on-line using a specially designed real-coded IGA and the optimal control actions at each sampling point are obtained. Moreover, the method can easily solve the nonlinear optimization problem compliance with input constraints by the carefully designed genetic operators. The method is illustrated via the applications to a superheated steam temperature system and a simplified 500MW boiler-turbine-generating unit load control system, simulation results show that the proposed method is valid, and is a new way to improve the control effect of complex thermal processes.