【摘要】 针对电动汽车集群入网进行无序充电会加剧峰谷差进而造成“峰上加峰”的问题，提出一种基于改进遗传算法的有序充电策略。首先，分析交流充电桩充电特性、电池充电特性和居民出行习惯；其次，以电网日负荷方差最小和用户充电成本最低作为目标函数，然后通过在遗传算法(GA)中对最优个体增加差分扰动并结合精英策略和自适应策略，来提高种群的多样性和算法的收敛速度；最后，对比遗传算法(GA)、精英遗传算法(EGA)和自适应遗传算法(AGA),以验证改进的有效性，同时将改进后的算法搭载到硬件平台进行实时性测试确保在工程中能进行实时响应。实验结果表明，改进后的算法能提高解的质量，并且在硬件平台中对单个充电桩的优化时间小于2 s。更多还原

【Abstract】 To address the issue where uncoordinated charging of electric vehicle clusters connected to the grid exacerbates peak-valley differences and creates “peak-on-peak” phenomena, this paper proposes an optimized charging strategy based on an improved genetic algorithm. First, the charging characteristics of AC charging piles, battery charging behaviors, and residential travel patterns are analyzed. Subsequently, dual objective functions are established to minimize daily grid load variance and user charging costs. The algorithm enhances population diversity and convergence speed through three key improvements: differential perturbation applied to optimal individuals, elite preservation strategy, and adaptive parameter adjustment in the genetic algorithm(GA). Comparative simulations among standard GA, elite GA(EGA), and adaptive GA(AGA) validate the algorithm′s effectiveness. Furthermore, real-time performance testing on hardware platforms demonstrates the algorithm′s engineering applicability, with optimization time per charging pile maintained below 2 seconds. Experimental results confirm that the enhanced algorithm achieves superior solution quality while ensuring real-time responsiveness for practical implementations.更多还原