Investigate The Problem And Solutions Co-evolutionary Optimization Method

With the researching and improving of biology and evolutionary algorithm,researchers concluded that, nature of individuals of the population or between species not only has a between individuals or species competition between the natural environment to survive, but also has a very obvious and common interaction cooperative co-evolutionary rules. Co-evolution, first proposed by Ehrlich and Raven,refers to the interaction between the various populations through the establishment of association, and improve the performance of the optimization of the population. In fact, there are two kinds of collaborative evolution: collaborative co-evolution and competitive co-evolution.It can be understood that the final result of the evolution of species is the result of the evolution of species in the final environment. The environment is keeping evolution, and ultimately evolved into current environment, and now the species is to experience the evolved environment, and has been adapted to the final environment.The co-evolutionary method proposed in this paper is to describe such a method. Our co-evolutionary is similar with that species change is the evolution of solution, while the evolution of the environment is the evolution of the solved problem. The problem is evolving and ultimately evolving into the solved problem, this is the problem and solution and ultimately evolving into the solved problem, this is the problem and solution co-evolution. This method can also be used for reference in agricultural application.At present, many improvements of genetic algorithm are devoted to improving its convergence speed and premature convergence problem. A lot of improvement is just trying to improve the global searching capability of the algorithm in order to reduce the possibility of falling into local optimum. In this paper, we investigate the problem and solution evolutionary algorithm, which aims to discuss the problem of local optimization of genetic algorithm.In this paper, Salesman Problem Traveling(TSP)is sued to test the problem and solution evolutionary algorithm. Based on the traditional genetic algorithm to solvethe TSP problem, the influence of the problem and solution method on the local solution of genetic algorithm on TSP. The traveling salesman problem(TSP)are designed to gradually evolve from a three-dimensional space of TSP problem to the traditional two-dimensional TSP problem. We focus on the initialization and how changing of the third dimensional coordinate Z with the evolutionary process. The problem and solution co-evolutionary algorithms will be investigated from three directions while the jumping out from local solution of genetic algorithm is focused:first, the four classical TSP problem are selected from classic TSPLIB(kroA100,TSP, pr1002, att532, pr2392)and are tested; secondly, how to setup the initial value of Z is explored for the impact of the genetic algorithm; the last, how to change of the Z values is explored.Our method generated the better solutions than the traditional GA for some cases,the similar results for some cases, and slightly bad results for some case. In general,the solutions obtained for our method are not the same as the traditional GA, so the problem and solution co-evolutionary algorithm has the capability of jumping out from local solution of traditional GA.