Research Based On Evolutionary Algorithm

The Evolutionary algorithm which we used is a self-adaptive probobility searching algorithm with global optimazition. Evolutionary algorithm is simulates the process of inheritance and evolutionary of life in the natural environment.It is widely used by investigator in industry and engineering and have great impact on these practical fields since 1960s.NSGA-II is suggested by Deb.It introduce Concision and transparent Non-dominated Sorting mechanism, don't need share parameter, also have good run efficiency. At the same time, the evolution algorithm holding rather speedy convergence pace. So it used abroad by scholar in the world. But it have much more improve space.The Differential Evolution is suggested by Storn in 1995. In the early years, Differential Evolution is used to solve the problem of chebyshev polynomial. In the after years, Differential Evolution also is a efficiency technique used to solve complexity optimize problem by people. In the current, Differential Evolution is increasingly attention by people as a sort of predominant capability optimize evolution algorithm. It apply domain is more wide than before. But the efficiency of maintenance distributing have to enhance more than it before.In this paper, our main work is to solve the above two problems by using evolutionary algorithm to optimize it .1.In order to maintenance the diversity of solution, Deb compute the individual crowding distance, then define a partial order set by individual's rank and crowding distance. But such method also have shortage, some individual have good diversity will be delete and some individual have bad diversity will be conserve. In this paper, we give a improve evolution algorithm of NSGA-II and suggests a new approach to measure individual crowding distance by hybrid distance and use priority queue to prune the over-plus of non-dominated solution one by one according hybrid distance for preserving the diversity of solution.Experimental results that the HD-NSGA-II can obtain reasonable distributing solution and diversity of this algorithm are more efficient than NSGA-II.2. In this paper, we suggest a new evolution algorithm is DSMODE. It is the same as NSGA-II. They are select solution randomly and put they into Mating pool. Than, we put the individual who is better are go to next generation. Due to random select is not instability, this method has lead the next generation individual is not ascertain. But we hope the evolution algorithm to find the instability individual into next generation. So in this paper, DSMODE suggest a new elitist selection base on directional information and use a heap to prune the over-plus of non-dominated solution in external achive one by one according crowding distance. Experimental results that the DSMODE has a good diversity.