Aplication Study Of Solution To TSP Problem Of Evolutionary Algorithm

Evolutionary Algorithm (E. A) applies computers to simulate self-adjusting process of biological evolution in nature, and servers as a solution to such intelligence problems as optimization, machinery learning and self-adjusting manipulation. It is characterized by highly paralleling, self-organization, self-adjusting and self-learning, and very suitable for approaching highly sophisticated non-linear problems which are beyond the traditional searching method. Over recent years, E. A has aroused universal attention among people for its achievement in optimization area. It has become a hot issue to solve optimization problem by E. A at present.TSP is a typical combined optimization problem. Solving TSP problem effectively can produce very important theoretical value and very high applied value in computable theory. Furthermore, the application research of E. A on TSP problem solution is of great significance in the aspects of building up appropriate E. A framework and establishing effective evolutionary operation.This paper analyzes several solutions to combined optimization problems systematically, introduces basic principle and properties of E. A, and then discusses its application in TSP problems.The discovery of this article in the process of analyzing TSP problems: In an advantageous route, a majority of cities are connected with one of the most adjacent cities to them. As a result, this paper proposes a close-by visit method: Firstly the author finds out B cities nearest to every city as candidate visiting cities. Secondly, the author puts them into array range[N][B] from close to distant (N is the total of cities), then produces initial'group according to this array, and make use of the array to restrict the range of inversion point of invert-over in the evolutionary process. Experiments indicate that close-by visit method can increase executive efficiency of algorithm to some degree, and improve the quality of travel route.Parameter setting in E. A influences the performance of algorithm greatly. Through experiments, this article concludes some experience in setting parameter, and has some reference value.