关键词: 差分进化算法, 双档案, 多样性度量, 自适应种群大小, 数值优化

Abstract: Addressing the inadequacy of population size improvement methods in existing Differential Evolution (DE) algorithms when dealing with decreased population diversity and local optima challenges, an improved Differential Evolution algorithm (APDE) based on a dual-Archive Population Size Adaptive method (APSA) was proposed in this article. First, two archives were involved to record individuals that had been discarded in previous evolutions and trial individuals. Then, diversity changes were measured based on the variations in the population distribution state. When population diversity decreased, the individuals from the archives were selected and added to the population to enhance diversity and the ability to escape local optima. Finally, an improved Differential Evolution algorithm based on the APSA method, named APDE, was proposed. It has been demonstrated by extensive tests on the CEC2017 test suite and the Lennard-Jones potential problem that the APDE algorithm is outperformed by none among five other differential evolution algorithms in the Friedman test across 30 benchmark functions, achieving the highest ranking. Significant improvements have been attained in at least 20% of these functions, and the best performance has also been achieved in solving potential energy minimization problems.

Key words: Differential Evolution(DE), dual-archive, diversity measure, adaptive population size, numerical optimization