数据降维和特征分析的GOA-LSSVM短期负荷预测  

Study on Short Term Load Prediction Based on GOA-LSSVM and Data Lowering Dimensions Analysis

作　　者：邹仕富 李嘉周 毛启均 宋明军 ZOU Shifu;LI Jiazhou;MAO Qijun;SONG Mingjun(State Grid Sichuan Electric Power Company,Chengdu 610041,China;State Grid Sichuan Information and Communication Company,Chengdu 610041,China;Ao star Information Technology Co.,LTD.,Chengdu 610041,China)

机构地区：[1]国网四川省电力公司,成都610041 [2]国网四川省电力公司信息通信公司,成都610041 [3]四川中电启明星信息技术有限公司,成都610041

出　　处：《实验室研究与探索》2019年第11期38-41,55,共5页Research and Exploration In Laboratory

基　　金：国网四川省电力公司科技项目(SGTYHT/16-JS-198)

摘　　要：提出一种基于数据降维和特征分析的GOA-LSSVM负荷预测方法.运用主成分分析对负荷的影响因素进行数据降维和特征表示,确定影响负荷的主要影响因素;针对LSSVM预测结果易受参数组合C和g影响,运用蝗虫优化算法对LSSVM模型参数进行自适应选择.研究结果表明,与GOA-LSSVM和LSSVM相比,本文算法可以有效提高负荷预测精度,通过数据降维和特征表示可以消除数据冗余和影响因素之间的相关性,不但简化了负荷预测模型的复杂程度,而且提高了预测模型的速度.In order to improve the accuracy of load forecasting, eliminate the correlation between multi-meteorological factors and eliminate redundancy, a load forecasting method based on GOA-LSSVM is proposed. Firstly, the principal component analysis is used to express the data dimension reduction and characteristics of load influencing factors, and the main influencing factors are determined. Then, based on the influence of parameter combination of C and G, the LSSVM model parameters are selected by Grasshopper optimization algorithm. The results show that compared with GOA-LSSVM and LSSVM, this algorithm can effectively improve the accuracy of load prediction, and can eliminate the correlation between data redundancy and influencing factors through data reduction and feature representation. It simplifies the complexity of load prediction model, and also the speed of prediction model is improved, which proves the validity and reliability of this algorithm.

关 键 词：数据降维 主成分分析 蝗虫优化算法 最小二乘支持向量机 

分 类 号：TN702[电子电信—电路与系统]