基于改进SLIQ算法及多粒度气象信息匹配的短期负荷预测  

Short-Term Load Forecasting Based on Improved SLIQ Algorithm and Multi-Granularity Meteorological Information Matching

作　　者：李滨[1] 覃芳璐 李倍存 吴茵[2] 李佩杰[1] 

机构地区：[1]广西电力系统最优化与节能技术重点实验室(广西大学),广西壮族自治区南宁市530004 [2]广西电网公司电力调度控制中心,广西壮族自治区南宁市530023

出　　处：《电网技术》2018年第1期291-300,共10页Power System Technology

基　　金：国家自然科学基金资助项目(51407036);国家重点基础研究发展规划项目(973项目)(2013CB228205)~~

摘　　要：短期负荷预测容易受到气象等多种因素共同作用的影响,找到关键影响因素是提高短期负荷预测精度的必要手段。电力系统海量数据包含了巨量的运行信息,为挖掘有用信息,提高数据利用效率,提出了一种基于改进SLIQ算法及多粒度气象信息匹配的短期负荷预测方法。采用改进的SLIQ决策树算法对气象负荷信息进行聚类,提取同等气象条件下决定负荷波动的关键因素。由动态灵敏度方法建立短期负荷拐点预测模型,再由熵权法选择最佳预测参考日并预测曲线拐点,并在此基础上提出多粒度气象信息匹配算法进行负荷曲线预测。通过对我国南方某地区的多季节负荷进行仿真预测,计算结果表明在任意气象条件下曲线预测精度均能满足电网要求,证明了所提方法的正确性及普适性。Short-term load forecasting is vulnerable to impacts of various factors. Necessary measure of improving its accuracy is to identify the critical factors. A huge amount of operation information lays in massive power system data collection. To dig up useful information and improve efficiency of data utilization, a short-term load forecasting method based on improved SLIQ algorithm and multi-granularity meteorological information matching is proposed in this paper. The meteorological load information is clustered with improved SLIQ decision tree. In return, the critical meteorological factors determining load fluctuation come out. A short-term load inflection point forecasting model is established based on dynamic sensitivity method. The best reference day of forecasting is selected by applying entropy method, with which forecasted inflection point is calculated. Finally, based on above research, a multi-granularity meteorological information matching method is proposed to perform load curve forecasting. Simulations are carried out on multi-seasonal load of a southern China region. Simulation results show that accuracy of load curve forecasting fully meets requirements of power grid on any meteorological condition, proving correctness and universality of the proposed method.

关 键 词：短期负荷预测 大数据挖掘 改进SLIQ气象分类器 动态灵敏度 多粒度气象信息匹配 

分 类 号：TM715[电气工程—电力系统及自动化]