基于VMD-RL-LSTM的短期风功率预测

Short-Term Wind Power Prediction Based on VMD-RL-LSTM

由于风电具有较高的随机性和较低的波动性,为了提高风电的非平稳性对于电力系统运行稳定性的影响,提出一种基于变分模态分解(Variational mode decomposition, VMD)、强化学习(Reinforcement learning, RL)参数寻优和长短时记忆网络(Long short term memory, LSTM)的短期风功率预测。为了降低数据的复杂度,首先通过变分模态分解将风功率原始数据分解为若干子模态。其次,通过强化学习对LSTM模型进行参数寻优,再对每个子模态建立LSTM模型预测。最终把各子模型预测的数据进行叠加,得到结果。对比分析结果显示,上述模型的预测精度较LSTM神经网络和EMD-LSTM预测模型均有不同程度的提高。

Since wind power has high randomness and low volatility,in order to improve the influence of wind power non-stationarity on the operation stability of the power system,a new method based on VMD,Reinforcement learning(RL)and Long Short term memory(LSTM)network is proposed for short-term wind power prediction.In order to reduce the complexity of the data,the original wind power data was decomposed into several sub-modes by variational modal decomposition.Secondly,the parameters of the LSTM model were optimized by reinforcement learning,and then LSTM model prediction was established for each sub-mode.Finally,the data predicted by each submodel were superimposed to obtain the results.Compared with LSTM neural network and EMD-LSTM prediction model,the prediction accuracy of this model is improved to different degrees.