Research On Differential Evolutionary Algorithm And Its Application

Along with the micro-electronics technique and the computer technique permeating each technique realm, the mankind is following a new period with a fast fierce development of technique. The computer science crossed with other academics to produce many new academics, pushing science technique toward wider realm development, which is bringing about profound influence on mankind's society. The development of science and technology and the application of technology, especially the demand of compute speed and artificial intelligence, people expect to look for a kind of efficiently of intelligence computer way. Evolution algorithms can solve these problems by imitating some one natural phenomena or process, which have the characteristics of self-organization, self-adaptation and self-study etc, so they get more and more concerns gradually. In recent years, a kind of novel evolution algorithm-the Differential Evolution algorithm(DE) emerges in various evolution algorithm, this algorithm was put forward by Rainer Storn and Kenneth Price for solving Chebyshev polynomial, it have got an extensive application in the restraint optimization computing, the blurry controller optimization design, the nerve network optimization and the filter design etc. Compared with the Genetic Algorithm (GA) and Particle Swarm Optimization (PSO), DE has many advantages, such as stronger stability, easy to realize, the faster speed of getting the approximation solution and so on, it gets extensively application in the optimization of nonlinear functions. But similar to the other evolution algorithms, the standard DE is easy to trap into local minimum point and cause to algorithm precocious.Aiming at the shortage of standard DE, the main work in the article is to improve the effectiveness of standard DE; some modified DE algorithms were put forward and applied them in the problem of numberical computer, mainly including finding the roots of polynomial, train neural network and estimate parameters of the dynamics model etc. Finally through numberical experiments and computer imitation, the results show that the algorithms mentioned in this paper can overcome the shortage of traditional DE algorithm .The obtained results have bigger theories value and application value.