Optimisation of Sparse Linear Arrays Based on Enhanced Differential Evolution Algorithm

Abstract

To address the challenge of reducing the peak sidelobe level (PSLL) in linear sparse array antennas, a differential evolution (DE) algorithm that integrates feasible solution repair and dynamic subpopulation selection is proposed. The algorithm initiates by generating an initial population through Logistic chaotic mapping, exploiting the ergodic and stochastic characteristics of chaos to enhance the diversity of initial solutions. Subsequently, a triple constraint handling and contribution-driven elimination to repair the initial feasible solutions is applied, thereby improving the PSLL performance of the linear array antenna while adhering to physical constraints. Additionally, the scaling factor (F) and crossover rate (CR) are dynamically adjusted in response to the iteration process and variations in fitness values, thus enhancing the algorithm's performance. Finally, through dynamic subpopulation selection, the premature convergence is mitigated, facilitating a global optimal search. Simulation results indicate that the PSLL of the antenna is effectively reduced across all three scenarios. In comparison to existing algorithms, the proposed approach demonstrates reductions in optimal PSLL of 0.99, 6.55 and 4.90 dB across 50 Monte Carlo experiments, with corresponding average PSLL reductions of 0.58, 6.82 and 6.03 dB, respectively.