基于HGWO-SVM的风电功率预测方法研究

Research on Wind Power Forecasting Method Based on HGWO-SVM

风电并网进程越来越快,风电功率随机性、间歇性等特点也对整个电网电压以及频率等带来了一定的影响,电网对风力发电功率预测精度提出了更高的要求。基于清洗后的数据,以风力发电输出功率作为输出,在原始的灰狼优化算法(grey wolf optimization,GWO)中引入交叉、变异和选择过程,为提升全局搜索性能,搭建了混合灰狼优化算法(hybrid grey wolf optimization,HGWO),并用其对支持向量机(support vector machine,SVM)中的核函数系数g和惩罚系数c进行全局寻优,训练得到基于HGWO-SVM风电功率预测模型。预测结果对比传统的算法形式(包括优化SVM、粒子群算法优化SVM、遗传算法优化SVM),所提的方法更具优越性,耗时相对较短,能实现对风电功率精准、快速预测。

The grid connection process of wind power is getting faster and faster.The randomness and intermittency of wind power also affect the voltage and frequency of the whole grid.The power grid demands higher accuracy of wind power prediction.Based on the cleaned data,it takes the output power of wind power generation as the output,and introduces the crossover,mutation and selection process into the original grey wolf optimization algorithm(GWO).In order to improve the global search performance,a hybrid grey wolf optimization algorithm(HGWO)is built,and it is used to globally optimize the kernel function coefficientg and penalty coefficientc in the support vector machine(SVM)to get the wind power prediction model based on HGWO-SVM.Compared with traditional algorithms,the prediction results(including optimized SVM,particle swarm optimization SVM,and genetic algorithm optimization SVM)proposed method having more advantages,relatively short time,and can achieve accurate and fast prediction of wind power.