基于差分进化粒子群神经网络的用户侧储能容量优化配置  

Optimal Configuration of User-side Energy Storage Capacity Based on Differential Evolution-particle Swarm Optimization Neural Network

作　　者：兰国龙 陈佳桥 叶恒志 陈旭海 廖宝文 Lan Guolong;Chen Jiaqiao;Ye Hengzhi;Chen Xuhai;Liao Baowen(Sinohydro Engineering Bureau 4 Co., Ltd., Beijing 100071, China;China Power Construction Group Fujian Electric Power Survey and Design Institute Co., Ltd., Fuzhou Fujian 350003, China)

机构地区：[1]中国水利水电第四工程局有限公司,北京100071 [2]中国电建集团福建省电力勘测设计院有限公司,福建福州350003

出　　处：《电气自动化》2021年第2期53-56,共4页Electrical Automation

基　　金：中国电力建设股份有限公司《百兆瓦时级大规模锂电池储能站关键技术研究及示范应用》项目(DJ-ZDXM-2018-43);福建省科学技术厅《百兆瓦时级大规模锂电池储能站关键技术研究及示范应用》引导性项目(2019H01010204)。

摘　　要：为提高用户侧储能系统的经济性及使用寿命,提出一种基于差分进化-粒子群BP神经网络的储能优化配置算法。以提高系统可靠性和降低系统投资成本为目标,以储能电池充放电功率限制以及系统最大装机容量等为约束条件,搭建了用户侧储能容量优化配置模型。提出的差分进化和粒子群的混合算法,保留粒子群算法早期收敛速度快的优点,加入差分进化算法以此保障种群的多样性,避免算法过早陷入局部最优,并相应减少粒子群控制参数数量,提高算法的性能,加快算法收敛速度,优化BP神经网络初始权重和阈值。算例结果表明,相比单一的差分进化法和粒子群算法,采用混合算法能够更快更好地得到优化结果。仿真结果表明了算法的有效性,在保证系统运行可靠性的同时,有效降低了系统的投资成本,使系统的经济性得以提升。In order to improve the economy and service life of the user-side energy storage system,an energy storage optimization and configuration algorithm based on differential evolution-particle swarm optimization BP neural network was proposed.Aiming at raising system reliability and reducing system investment cost,taking energy storage battery charge/discharge power limit and maximal system installed capacity as constraints,a user-side energy storage capacity optimization and configuration model was built.The proposed hybrid algorithm of differential evolution and particle swarm optimization retained the advantage of quick early-stage convergence of the particle swarm optimization,and added the differential evolution algorithm to ensure population diversity so that the algorithm might not fall into local optimization too early.Furthermore,the number of particle swarm control parameters was reduced correspondingly,algorithm performance was improved,algorithm convergence speed was raised,and the initial weight and threshold of the BP neural network were optimized.Results of calculation examples showed that,compared with single differential evolution or particle swarm optimization,the proposed hybrid algorithm could achieve quicker,better optimization effect,and simulation results indicated the validity of the algorithm.While ensuring operating reliability of the system,the proposed algorithm could effectively reduce system investment cost and improve system economy.

关 键 词：用户侧储能 经济性 容量优化 差分进化算法 粒子群算法 

分 类 号：TM732[电气工程—电力系统及自动化]