摘要
以光伏发电系统的输出功率为研究对象,通过分析光伏发电功率的影响因素,利用相似日原理生成训练样本,将混沌搜索和自适应变异思想引入粒子群算法中,提出混沌搜索的自适应变异粒子群优化BP神经网络的预测模型。该模型较好地克服了BP网络初始化的随机性问题,提高了模型的泛化能力、收敛速度与预测精度。利用光伏电站与气象观测站的数据进行仿真分析与验证,结果表明:优化后模型的预测精度高于优化前,且混沌搜索的AMPSO的优化效果好于单纯PSO的优化效果。
Based on the analysis of influential factors of output power in photovoltaic systems, the training samples are generated according to the similar day principle. The chaos search theory and adaptive mutation theory are introduced to improve the particle swarm optimization algorithm, then the prediction model based on chaos search and AMPSO-BP neural network is proposed. The model can preferably overcome the randomness of BP network initialization, and im- prove the generalization ability, convergence speed and prediction accuracy. Different models are trained and verified with the data of photovohaic power station and meteorological observation station. Results show that the optimization model has higher prediction accuracy, and the optimization results of adaptive mutation particle swarm optimization based on chaos search is better than that of simple particle swarm optimization.
出处
《华北电力大学学报（自然科学版）》
CAS
北大核心
2014年第4期15-21,共7页
Journal of North China Electric Power University：Natural Science Edition
基金
国家自然科学基金资助项目(51177047)
关键词
光伏发电系统
相似日
混沌搜索
自适应变异
粒子群算法
photovohaic system
similar days
chaos search
adaptive mutation particle swarm optimization （AMP-SO）
particle swarm algorithm
