Differential evolution algorithm based on successful history parameters
Kanghua Zhu, Jincheng Xu, Xiaobo Fan · 2025
Differential Evolution (DE) is a straightforward yet powerful method for numerical optimization. The performance of the DE algorithm is greatly affected by its parameter configuration, as a result, a lot of research has been devoted to its adaptive parameter adjustment. In this work, we introduce a novel approach for adapting DE parameters, where the historical success of previous individuals is utilized to inform the generation of the next set of candidates, thereby improving the search for optimal solutions.. Predict and guide the generation of the next generation of individuals through the optimal individuals produced in the previous generation.We compare the proposed algorithms with some classic DE and its improved algorithms on the CEC2013 benchmark set. The experimental results show that the DE algorithm proposed by us based on historical successful individual prediction is highly competitive compared with the advanced DE algorithm..