基于GRA-EGA-LSTM模型的短期光伏功率预测

Short-term Photovoltaic Power Prediction Based on GRA-EGA-LSTM Model

对光伏发电功率进行预测可为电力系统调度提供参考,有利于电网的安全稳定运行。为了提高光伏发电功率预测精度,采用灰色关联度分析法(Grey Relation Analysis,GRA)寻找待预测日的相似日作为训练样本;采用精英保留遗传算法(Elitist Model of Genetic Algorithm,EGA)对长短期记忆网络(Long Short-Term Memory,LSTM)的超参数进行寻优;将相似日的光伏功率和气象因素数据作为训练样本代入超参数寻优后的长短期记忆网络进行预测。通过仿真测试,基于GRA-EGA-LSTM组合预测模型的短期光伏功率预测精度要优于传统的LSTM模型。

The prediction of photovoltaic power generation can provide reference for power system dispatching and is conducive to the safe and stable operation of the power grid.In order to improve the prediction accuracy of photovoltaic power generation,the Grey Relation Analysis(GRA)is used to find similar days to the days to be predicted as training samples.The Elitist Model of Genetic Algorithm(EGA)is used to optimize the hyperparameters of the Long Short-Term Memory(LSTM).The photovoltaic power and meteorological factor data of similar days are substituted as training samples into the long short-term memory network after hyperparameter optimization for prediction.Through simulation tests,the short-term PV power prediction accuracy based on the GRA-EGA-LSTM combined prediction model is better than that of the traditional LSTM model.