基于相似日理论和IPSO-Elman模型的短期光伏发电功率预测

为提高光伏发电功率预测精度，提出一种基于相似日理论和改进的IPSO-Elman神经网络模型的短期光伏发电功率预测方法。将历史数据细分为不同季节不同天气类型的多个子集，通过灰色关联度和余弦相似度组合而成的综合关联度指标筛选相似日。针对标准粒子群算法的缺陷，提出一种改进的自适应混沌变异粒子群算法(IPSO)来优化Elman神经网络，将优化得出的最优权值和阈值作为初始值建立IPSO-Elman神经网络模型，对3种不同季节和天气类型条件下的光伏发电功率分别预测。选用甘肃省某光伏电站2014年数据进行实例分析，结果表明，IPSO-Elman模型在不同天气类型条件下的功率预测效果都有明显提高。

Short-Term Photovoltaic Power Forecast Based on Similar Day Theory and IPSO-Elman Model

In order to improve the accuracy of photovoltaic power generation prediction,a short-term photovoltaic power generation prediction method based on similar day theory and improved IPSO-Elman neural network model is proposed.The historical data is subdivided into multiple subsets of different weather types in different seasons,and the comprehensive correlation degree index that is combined gray correlation degree and cosine similarity is used to screen similarity day.In order to eliminate the defects of the standard particle swarm optimization (PSO) algorithm,an improved particle swarm optimization (IPSO) algorithm with adaptive chaotic mutation was proposed to optimize the Elman neural network.The optimal weights and thresholds were used as initial values to establish the IPSO-Elman neural network model to separately predicte photovoltaic power generation under three different seasons and weather conditions.Based on the analysis of photovoltaic power plant in Gansu Province in 2014,the results show that compared with the Elman model and the PSO-Elman model,the IPSO-Elman model has better prediction effects under different weather types.