基于变分模态分解和改进鲸鱼算法优化的神经网络风速预测模型

Neural Network Wind Speed Prediction Based on VMD and Improved Whale Algorithm

风速预测在风电场安全并网和智能化管理中起着决定性作用,针对风速的非线性和不稳定等特点,提出一种基于变分模态分解和改进鲸鱼算法优化的模糊神经网络(VMD-CGWOA-ANFIS)的混合预测模型。该模型首先使用变分模态分解(VMD)技术将原始风速序列分解为一系列子序列,对各子序列分别采用模糊神经网络(ANFIS)建立预测模型。为进一步提高预测精度,克服鲸鱼(WOA)算法容易陷入局部最优和收敛过早的缺点,引入共轭梯度算法(CG)对WOA进行改进,利用改进的CGWOA算法对ANFIS参数进行优化。使用优化后的ANFIS分别对VMD分解后的各子序列进行预测,最后将预测后各子序列叠加得到最终预测结果。为测试模型的有效性,选择宁夏地区三组实际风电数据进行模拟试验,将ANFIS、VMD-ANFIS、VMD-WOA-ANFIS与提出模型进行对比,结果表明,所提出的混合模型预测精确度明显高于其他对比模型。

Wind speed prediction plays a decisive role in the safe gird-connection and intelligent management of wind farms.It is characterized by the nonlinearity and instability of wind speed prediction,and the fact that the whale algorithm is easy to fall into local optimum,resulting in weak generalization and premature convergence.Therefore,the prediction results are not ideal.A hybrid prediction model of VMD-CGWOA-ANFIS was proposed.The model first uses VMD technology to decompose the original wind speed sequence into a series of sub-sequences,and uses ANFIS to establish prediction models for each component,in order to improve the prediction accuracy and avoid the optimization to fall into local optimum.The CGWOA algorithm,which is optimized by the conjugate gradient algorithm(CG),is used to optimize the parameters of ANFIS.The optimized ANFIS predicts each subsequence after VMD decomposition,and finally superimposes each subsequence after prediction to obtain a final prediction result.In order to verify the validity of the model,the data of three wind powers in Ningxia region were selected for experiment,and ANFIS,VMD-ANFIS,VMD-WOA-ANFIS were compared with the proposed model.The results show that the proposed hybrid model has higher prediction accuracy than other models.model.