Genetic Algorithm And Its Application In Optimization Of Fuzzy Logic Controller

The Genetic Algorithm(GA) is a kind of parallel,efficient and overall optimal search algorithm based on nature evolution theory.The basic concept of GA is designed to simulate processes in natural system necessary for evolution (via natural selection,crossover,mutation),specifically those that follow the principles first laid down by Charles Darwin of survival of the fittest,then the satisfying solution is obtained.However,there are some shortages in the conventional Genetic Algorithm,for example,the probability of premature convergence caused by its lower search efficiency.Considering the disadvantages of standard Genetic Algorithm,it is of great importance to improve on GA,subsequently many improved methods are proposed. To avoid the limitation of Simple Genetic Algorithm, an improved Genetic Algorithm, Subsection Crossover Genetic Algorithm(SCGA), is presented in the paper.The improved GA divide ordered population into three,and employ different crossover operator on the source of father chromosomes to improve the population diversity in SCGA, gradually leading the search near the true optimum solution.Several complicated function optimization problems illustrate the effectiveness of the proposed improved GA. Furthermore,We have analyzed SCGA's convergence on the base of Markov chains. Fuzzy Control is an important branch of Intelligent Control,and it mainly depends on the human experience but not the mathematical model of controlled-object.Thus,Fuzzy controller can fulfill some human's intelligence and is widely used in complex process and object-model control. The efficiency of Fuzzy Control depends on several key parameters: membership functions and fuzzy control-rule table. Conventional methods that select the parameters mainly reckon on experts' experience and practical modulation, so there exist subjectivity and randomness.GA is used to optimize fuzzy controller parameters to avoid subjectivity and randomness.Previous work using genetic algorithms has mostly focused on the development of rule sets or high performance membership functions,however,the interdependence between these two components suggests a simultaneous design procedure would be a more appropriate methodology.When GA's have been used to develop both,it means code has great long length. In order to solve the above problem,this paper proposed a fuzzy system design method that uses an improved Genetic Algorithm to determine membership functions,the rule table,the quantitative factors and the proportional factor at the same time. A hybrid code which contains both binary and decimal code is designed based on characteristics of FLC, and a method of using rule variable value to change rule table is put forward, which shorten code length effectively. New operators and search parameters are devised to improve search efficiency, for example, put forward double level mutation rate on the base of population era number and individual fitness.The proposed method is applied to a typical industry object and the room temperature control problem in Matlab and has been shown to be more effective through a comparison with another two fuzzy control systems. The result of the simulation and control illustrates that it has good dynamic performance and robust.