A Research on Entropy of Information Compression Operator-based Multi-stage Genetic Algorithm

Genetictic Algorithm (GA) is a widely-used high-efficiency random search and optimization method, which is geneticrated from evolution theory. This paper aims at overcoming the drawbacks of local convergence and not converging well to the globally-optimal solution, which usually exist in the course of solving optimization problems using Basic Genetic Algorithm(GA),and proposes a modified algorithm- entropy of information compression operator-based multi-stage genetic algorithm(EMC-GA).In every step of the algorithm, the population will make evolution to a given number duplicity. Given the optimal reservation strategy, there will be several individuals of the population selected ,which provides message to design entropy of information-based spatial compression tactics. With the help of the two strategies, the algorithm converges to the globally-optimal solution steadily and quickly. Besides, this article also presents the basic idea and specific implementation strategies of the algorithm, which analyses its convergence under the help of Markov Chain Theory. In order to confirm the practicability and effectiveness of the proposed EMC-GA algorithm, the paper conducts optimizing test on several representative multi-modal functions and it turns out that the global search ability and convergence of the improved GA are highly superior to the standard GA, compared with the analytical solution and the optimal result of standard GA. This algorithm has favorable convergence under the case of the optimal reservation strategy, especially for optimal problems of large scale and high-accuracy.