基于CA-SE-GA-BP的光伏发电功率预测

针对光伏发电功率预测精度低的问题,以澳大利亚爱丽丝泉地区某200kW的光伏电站为例,选用遗传算法(GA)优化BP神经网络,采用相关性分析法(CA)确定太阳辐照度、温度、湿度为影响光伏发电功率的主要因子,结合经样本熵(SE)量化的天气类型作为模型输入量,提出CA-SE-GA-BP神经网络的光伏发电功率预测模型。结果表明,多云天气下CA-SE-GA-BP神经网络均方根误差、平均绝对百分比误差分别为4.48%、2.27%,晴天、雾霾、雨天三种天气类型下的预测误差也基本上不超过10%,相较于SE-GA-BP、CA-GA-BP、GA-BP神经网络,CA-SE-GA-BP神经网络预测误差降低,为解决光伏系统发电功率预测提供了一种高效准确可行的方法。

Photovoltaic Power Generation Prediction Based on CA-SE-GA-BP

Aiming at the low prediction accuracy of photovoltaic power generation, a 200 kW photovoltaic power plant in Alice Springs, Australia was selected as the research object. The BP neural network was optimized by genetic algorithm (GA). Correlation analysis (CA) was used to determine solar irradiance, temperature and humidity as the main factors affecting photovoltaic power generation. Combined with sample entropy (SE) quantification, the weather type is used together as the model input. Thus, this paper proposes a neural network based photovoltaic power prediction model (CA-SE-GA-BP). The results show that the root mean square error and the mean absolute percentage error of the CA-SE-GA-BP neural network are 4.48% and 2.27%, respectively. For fine, fog and rainy day, the prediction errors are below 10%. Compared with SE-GA-BP, CA-GA-BP and GA-BP neural networks, the prediction error is reduced, which provides an efficient and accurate method for solving PV system power generation prediction.