基于PSO-GRU的光伏电站短期功率预测研究

Research on Short-term Power Prediction of Photovoltaic Power Plant Based on PSO-GRU

为了提高光伏电站短期功率预测的准确性,提出一种结合注意力机制(Attention)与GRU网络的功率预测模型。首先对历史数据集分析并进行归一化处理,在此基础上将Attention机制与GRU网络相结合搭建预测模型;Attention机制通过对GRU的输入特征赋予不同的权重,使得预测模型对长时间序列输入的处理更为有效;最后,使用PSO优化GRU的超参数。采用新疆某光伏电站实际运行数据,仿真结果表明,PSO-Attention-GRU模型具有更好的预测性能,预测精度更高。

In order to improve the accuracy of short-term power prediction of photovoltaic power plants,a power prediction model combining Attention mechanism and GRU network is proposed in this paper.Firstly,historical data sets are analyzed and normalized,and then a prediction model is built by combining Attention mechanism with GRU network.The Attention mechanism gives different weights to the input features of GRU,which makes the prediction model more effective in processing long-time series input.Finally,PSO is used to optimize GRU super-parameters.Using the actual operation data of a photovoltaic power station in Xinjiang,the simulation results show that PSO-Attention-GRU model has better prediction performance and higher prediction accuracy.