Research Of Improved Evolutionary Resolve Algorithm For TSP

Genetic Algorithm (GA) is a simulation of natural biological evolution of the search algorithm and Particle Swarm Optimization (PSO) is the embodiment of swarm intelligence which also imitates the searching behavior of fishes and birds for food. Both of them are evolutionary algorithms. There are a lot of similarities between the two theories which are both easy to operate and have strong robustness. They are widely applied because they do not need other specialized knowledge in areas but fitness function as the standard in the searching process, which attracts great concerns among experts. And this leads to lots of prominent achievements.TSP is a typical NP and widely use in daily life and everyday activities. How to improve the efficiency of solving the problem has practical significance. At present, there is no optimal solution to the NP problem in polynomial time. Thus, it is very realistic to design the corresponding algorithm to get approximate and satisfactory solution. The application of intelligent algorithm in the TSP is under the guidance of this theory.This paper makes a study on the theories of GA and PSO and their application. After introducing the principles and the application of these two theories, the thesis analyzes the research background, mathematical models and solutions of TSP. And then it introduces the application and development of the two theories in the process of solving the problem of TSP. And on this basis, this paper provides a loop insertion method, used in the generation of the initial population (Particle Swarm) of the two algorithms. Then analyses the validity of the population towards algorithms from the Theoretical foundation of GA and PSO and then makes an experiment of the data in the standard UCI database under MATLAB. After verify the reasonability of this method with which generating the initial population and makes a comparative on the algorithms to produce initial population (Particle Swarm) and other initial methods, this essay finally illustrates the feasibility and effectiveness of such algorithms.