基于改进EMD-PSVM的短期负荷预测  

Short-term Load Forecasting Based on Improved EMD-PSVM

作　　者：胡杨 常鲜戎[2] 

机构地区：[1]国网杭州余杭区供电公司,浙江杭州311100 [2]华北电力大学电气与电子工程学院,河北保定071003

出　　处：《陕西电力》2016年第3期29-33,共5页Shanxi Electric Power

摘　　要：电力系统负荷是具有典型周期性和随机性特点的非线性、非平稳时间序列。为了降低负荷序列的非线性,提高预测精度,提出了结合经验模态分解EMD和支持向量机SVM的改进短期负荷预测法。运用EMD将负荷序列分解成若干不同频率的平稳分量,突出原负荷局部特征,并采取极值延拓法减弱端点效应,同时利用PSO寻优,选择合适的参数对各分量构造不同的EMD-PSVM预测模型,将各分量预测结果重构后得到最终预测值。通过算例分析,与EMD-SVM及BP神经网络预测法比较,验证了改进EMD-PSVM模型能够有效提高预测精度,稳定性较强。The power system load is a typical nonlinear and non-stationary time series with periodic and randomness.In order to reduce the nonlinear of load sequence to improve the accuracy of load forecasting,the method combined with empirical mode decomposition（EMD） and support vector machine（SVM） is proposed for short-term load forecasting.EMD can decompose nonstationary signals into some smooth and stationary intrinsic mode functions with different frequency,which can highlight local features of original load series,extreme continuation method is introduced to weaken the end effect.At the same time,PSO is used to select the appropriate parameters to build different EMD-PSVM model respectively to forecast each intrinsic mode functions.Then these forecasting results of each IMF are combined to obtain the final forecasting result.Finally the simulation results in comparison with the conventional EMD-SVM and BP neural network prediction method show that the improved prediction model based on EMDPSVM has higher precision and stability.

关 键 词：负荷预测 极值延拓法 经验模态分解 支持向量机 粒子群优化 

分 类 号：TM715[电气工程—电力系统及自动化]