Research On Opposition-based Differential Evolution Algorithm And Its Applications

Global optimization problems exist in various fields of real life,and with the rise of artificial intelligence research in recent years,the structure of practical problems is nowadays becoming more and more complex.In order to make the algorithm can solve these complex optimization problems efficiently,scholars have proposed nature-inspired intelligence optimization algorithms,which can obtain an optimal solution under the guidance of a metaheuristic strategy.However,these algorithms still have many problems,On the one hand,if the algorithm focuses on speeding up the convergence,the algorithm will fall into a local optimum.On the other hand,if the algorithm aims at finding a high-quality solution,it will lead to the slow convergence speed or non-convergence.To solve these problems,twos differential evolution algorithms are designed in this paper in combination with opposition-based learning strategies,and the main works are as follows.1.To address the shortcomings that the population-based opposition does not consider the redundancy calculation at the later stage of evolution and the jumping rate parameter does not take into account the self-adaptive control,the subpopulation-based opposition is proposed by fully considering the information among individuals.First,in order to consider the influence caused by individuals on the opposition,the jumping rate is assigned to each individual,and if the rate of an individual is lower than a threshold,the individual will become a member of the subpopulation and generate opposite individuals.Second,in order to dynamically control the subpopulation size during the evolutionary process,historical information is used to guide the generation of jumping rates for the next generation of individuals.2.To eliminate the limitation that the opposition strategy fails to adequately balance between exploration and exploitation in the iterative process,dual opposition-based learning is proposed by introducing a diversity-based estimation method.First,the diversity-based estimation method is used to classify the exiting opposition strategies,and the strategies with the strongest exploration ability and exploitation ability are selected from them respectively,and then the subpopulation strategies are combined with these two strategies to generate two completely new opposition strategies.Second,in order to balance these two strategies during the iteration process,a protective mechanism is designed based on the exploration ability percentage of the algorithm,which can effectively select the opposition suitable for the current evolutionary stage.3.To illustrate the performance improvement of the proposed strategy on constrained optimization problems for real-life constrained optimization problems,the dual opposition is combined with the constrained optimization algorithm and the performance of the algorithm is verified on the CEC2020 benchmark test suite.The experimental results show that dual opposition can bring exploration and exploitation to a balanced state when dealing with complex optimization problems,and the algorithm has the best performance when the percentage of exploration and exploitation is 10% and 90%,respectively.