Differential Evolution Algorithm Based on Spatial Partition Perturbation Mechanism

Differential Evolution (DE) algorithm, as a state-of-the-art population-based stochastic optimizer for continuous non-convex search spaces, can adaptively balance exploration and exploitation through difference vectors among individuals. However, as iterations proceed, population diversity diminishes, leading to reduced exploration capability and susceptibility to local optima, especially in multimodal or complex landscapes. To address this limitation, this paper proposes a Differential Evolution Based on Spatial Partition Perturbation Mechanism (SPMDE). The application of KD-Tree enables SPMDE to dynamically manage population distribution, update stagnant individuals, and maintain diversity, thereby strengthening its global search capability in the evolutionary process. CEC2017 benchmark problems is used to evaluate the overall performance of SPMDE. Experimental results indicate that SPMDE achieves highly competitive optimization performance when compared to both state-of-the-art DE variants and enhanced versions of classic evolutionary algorithms.