Research On Adaptive Differential Evolution Based On Memory Driving Mechanism

With the rapid progress of society, science and technology, the development of group intelligence and artificial intelligence has become more and more rapid, and both of them have been widely applied to all aspects of social life. The group intelligent optimization algorithm has the advantages of strong distribution, high robustness and non-direct communication. Typical swarm intelligence algorithms include differential evolution algorithm (DE), particle swarm optimization algorithm (PSO), bee colony algorithm, ant colony algorithm, genetic algorithm, and so on. With the gradual development in recent years,differential evolution algorithm of group intelligent optimization algorithm has been widely used to solve practical problems due to its own high robustness, low complexity, simple and efficient performance.This paper aims to analyze and verify the vital status of the adaptive parameter search mechanism with the Top-Bottom strategy from the theoretical and experimental aspects to the swarm intelligence algorithm.This paper summarizes and analyzes the basic knowledge and algorithm flow of differential evolution algorithm, and also analyzes the shortcomings and drawbacks of the current algorithm. Based on the former background, this paper proposes an improved differential evolution optimization algorithm with the adaptive parameter mechanism and introduces the selection and storage mechanism which is based on the superior individuals and the inferior individuals. In this algorithm, we select several representative better and poorer individuals to linearly explore along the specific direction information, returning the better location to the group, and then guide the whole population to develop a more favorable space.In addition, an improved differential evolution algorithm based on the second-order difference vector idea is proposed in this paper, which aims to make full use of the search direction of difference vector in the current population. The simulation results prove the superiority of the proposed second-order differential evolution algorithm.At the same time, due to the traditional fixed parameters of the differential evolution algorithm, this paper introduces an idea of the adaptive parameter, which is to reserve the more favorable parameters and discard the unprofitable parameters, so that individuals can keep the effective exploration step and mutation probability. In the simulation experiment, we select the classical and authoritative test functions to verify the algorithm's performance and prove superiority and effectiveness of the improved differential evolution algorithm over the classical differential evolution algorithm.