Identification And Optimization Of Nonlinear Object Of Mechanical And Electronic System Based On Genetic Programming And Bond Graph

This dissertation briefly introduced the development of evolutionary algorithm (EA) and detailedly discussed a branch of EA——genetic programming (GP). This paper is based on the evolutionary algorithm research, proposed the identification method to nonlinear system based on the genetic programming and bond graph.In this paper, the base algorithm and operation flow chart were showed by demonstrations of genetic operators running. The paper introduced the development situation of nonlinear system identification at present. With the example of wireless access points configuration proved the feasibility of nonlinear system identification using genetic programming. Bond graph is a modeling method of dynamics system. The paper proposed the method of identification of mechanical and electronic system based on genetic programming and bond graph. Using the genetic operators into the embryo bond graph, evolved the new bond graph which satisfied the design request. We used the example of eigenvalue optimization to validate the experiment. In the process of genetic programming research, aim at the problem which population multiformity losing. For finding the optimization solution, we need increase the size of population and prolong the evolving generations. But the following problem is individual complicated degree and computation increasing fast, all of them spends lots of evolutional time. This paper presents the method which used automatically defined funct ion (ADF) to improve on the traditional GP. This method is good for improve the speed of evolution. With the example of TSP, we proved the capability of the improved algorithm.Finally, this paper discussed the basic problem of GP, and research in the correlative fields. Had pointed out the future will further study direction.