基于 BP 神经网络的光伏发电预测模型设计

摘要

结合历史发电量和气象数据分析了影响光伏系统发电功率的各项因素，针对传统光伏发电预测模型预测精度不高的问题，加入了电池板温度信息作为光伏发电预测模型的输入参考量；针对传统 BP 神经网络易陷入局部极值的缺陷，提出了基于改进学习率和权值的弹性自适应规则的 BP 神经网络。采用光伏监控系统历史发电量和气象数据建立了弹性自适应 BP 神经网络预测模型，对训练好的模型进行了测试和评估。预测结果表明，该预测方法较好地解决了传统 BP 算法易陷入局部极值的问题，提高了系统预测结果精度。%Combined with historical power and weather data,all factors which influence the power generation of photovoltaic system are discussed,and the temperature information from the solar panels is added to the forecasting model as the reference input to solve the problems of low prediction accuracy of the traditional forecasting model.Aiming at the defects of the traditional BP neutral network into a local extreme value,this paper proposes the BP neutral network based on the improved learning rate and elastic adaptive rule.The BP neural network forecasting module of elastic adaptive method is trained by histori-cal power and weather data of photovoltaic monitoring system.The trained module is tested and evaluated.The forecasted results show that the prediction method can effectively solve problems of the traditional BP algorithm into a local extreme value and improve the system precision of forecast results.