基于VMD-LSTM神经网络的风电功率预测研究OA
Research on Wind Power Prediction Based on VMD-LSTM Neural Network
为了更好的预测风电功率,本研究提出了一种变分模态分解(VMD)与长短期记忆网络(LSTM)相结合的风电 功率预测方法.首先使用 VMD 对原始风电功率数据进行分解,得到一系列相对平滑的本征模态函数以简化风电功率数据;然后将这些数据作为 LSTM 模型输入,利用 LSTM 具有的特殊的门机制,解决梯度消失或者梯度爆炸问题;最后将每个本征模态函数所对应的预测值加权求和得到最终预测结果.在某实际风电场的仿真验证中,相较于单独的 LSTM、BP 网络模型,本研究提出的 VMD-LSTM 组合模型具有更低的均方根误差(RMSE)、平均绝对误差(MAE)以及平均绝对百分比误差(MAPE),提高了风电功率预测的精度及鲁棒性,为电网调度提供了更可靠的技术支持.
To better predict wind power,this paper proposes a wind power prediction method that combines Variational Mode Decomposition(VMD)with Long Short-Term Memory Network(LSTM).Firstly,VMD is used to decompose original wind power data,obtaining a series of relatively smooth intrinsic mode functions to simplify wind power data;then these data are used as the input of the LSTM model,and by leveraging the special gating mechanism of LSTM,the problems of gradient vanishing or gradient explosion are solved;finally,the predicted values corresponding to each intrinsic mode function are weighted and summed to obtain the final prediction result.In the simulation verification of a real wind farm,compared with individual LSTM and BP network models,the VMD-LSTM combined model proposed in this papwer has lower root mean square error(RMSE),average absolute error(MAE),and average absolute percentage error(MAPE),and thus improves the accuracy and robustness of wind power prediction,and provides more reliable technical support for power grid dispatching.
耿运涛;于培龙
邵阳职业技术学院,湖南 邵阳 422004中国水利水电第八工程局有限公司,湖南 长沙 410000
信息技术与安全科学
风电功率预测变分模态分解长短期记忆网络
Wind power predictionVariational Mode DecompositionLong Short-Term Memory Network
《船电技术》 2026 (9)
6-9,4
2025年度邵阳市科技计划指导性项目《基于LSTM神经网络的风电功率预测模型研究》(项目编号:2025GZ6090)
评论