基于VMD-IBWO-BiLSTM的短期风电功率预测

Short-Term Wind Power Forecasting Based on VMD-IBWO-BiLSTM

中文关键词:风电功率预测  变分模态分解  北方苍鹰优化算法  改进白鲸优化算法  双向长短期记忆神经网络  深度学习

keywords:wind power prediction  variational mode decomposition  northern goshawk optimization  improved beluga whale optimization  bi-directional long short-term memory neural network  deep learning

中文摘要:准确预测风电功率对实现风电场稳定运行和电网优化调度具有重要意义。为了提高风电功率预测的稳定性和精准性，提出一种基于变分模态分解（variational modal decomposition，VMD）、融合Logistics混沌映射、折射反向学习策略的改进白鲸优化算法（improved beluga whale optimization，IBWO）和双向长短期记忆（bi-directional long short-term memory，BiLSTM）神经网络的组合模型。首先，利用模糊熵为适应度函数的北方苍鹰优化算法（northern goshawk optimization，NGO）优化VMD的核心参数，通过NGO-VMD对采集到的原始风电功率数据分解，得到模态分量。然后，利用改进白鲸优化算法IBWO对双向长短期记忆BiLSTM神经网络中的超参数进行寻优，再使用IBWO-BiLSTM模型对各模态分量预测。最后，将各模态分量的预测值叠加得到风电功率的预测值。实验表明，该组合模型较其他普通组合模型在预测精度上有较大提高。

Abstract:Accurate prediction of wind power is of great significance for the stable operation of wind farms and the optimal dispatching of power grids. To enhance the stability and accuracy of wind power prediction， an integrated model based on Variational Mode Decomposition （VMD）， Bi-directional Long Short-Term Memory （BiLSTM） neural network， and an Improved Beluga Optimization Algorithm （IBWO） that combines logistic chaotic mapping and refracted opposition-based learning is proposed. Firstly， the Northern Goshawk Optimization Algorithm （NGO） with fuzzy entropy as the fitness function is employed to optimize the core parameters of VMD， and the modal components are obtained through the decomposition of the original wind power data collected by NGO-VMD. Secondly， the Improved Beluga Optimization Algorithm （IBWO） is utilized to optimize the hyperparameters in BiLSTM， and subsequently， the IBWO-BiLSTM model is applied to predict each modal component. Finally， the predicted value of wind power is obtained by superimposing the predicted values of each modal component. Experiments demonstrate that this combined model significantly improves prediction accuracy compared to other common combination models.