基于VMD和PSO-GWO混合算法优化支持向量机的电力负荷预测

Load Forecasting Based on VMD and Support Vector Machine Optimized by Hybrid PSO-GWO

为了快速且准确地利用历史负荷数据实现电力负荷预测且提高模型的预测精度,提出一种基于变分模态分解(variational mode decomposition,VMD)和粒子群优化-灰狼优化混合算法(particle swarm optimization-grey wolf optimization,PSO-GWO)优化支持向量机回归模型(support vector machine regression,SVR)的电力负荷预测方法.该方法首先对电力负荷数据进行VMD,提取各模态分量进行特征分析;然后基于PSO-GWO混合算法优化支持向量机回归模型中的惩罚系数和核参数,进而分别对各模态分量进行预测;最后对所预测的模态分量进行重构获取最终预测结果.与此同时,将VMD-PSO-GWO-SVR方法与Spark平台的优秀计算能力相结合,节省计算所需要的时间.算例结果表明,VMD-PSO-GWO-SVR方法可有效实现电力负荷的预测,且预测结果明显优于传统SVR算法.

In order to quickly and accurately realize power load forecasting and improve the forecasting accuracy of the model,a power load forecasting method based on hybrid particle swarm optimization(PSO)and grey wolf optimization(GWO)optimized support vector machine regression model(SVR)is proposed.Firstly,the power load data is decomposed by variational mode decomposition(VMD),and the characteristics of each modal component are analyzed.Then,the penalty coefficient and kernel parameter in the support vector machine regression model are optimized based on the hybrid PSO-GWO algorithm,and then the modal components are predicted respectively.Finally,the predicted modal components are reconstructed to obtain the final prediction results.Meanwhile,VMD-PSO-GWO-SVR is combined with the excellent computing performance of the spark platform to save computing time.The results show that the VMD-PSO-GWO-SVR method can effectively realize the power load forecasting,and the forecasting result is better than the traditional SVR algorithm.