野草算法和支持向量机相融合的短期负荷预测  

Short-term load forecast fusing weed algorithm with support vector machine

作　　者：王喆[1,2] 杨华春[3] 

机构地区：[1]天津大学计算机科学与技术学院,天津300134 [2]天津城市建设管理职业技术学院,天津300134 [3]天津理工大学聋人工学院,天津300384

出　　处：《现代电子技术》2016年第19期99-101,106,共4页Modern Electronics Technique

基　　金：天津市高等学校科技发展基金计划项目(20130818)

摘　　要：短期负荷受到天气、季节的综合影响,具有一定的混沌性,为了准确描述短期负荷的变化趋势,以提高预测精度,提出一种野草算法和支持向量机相融合的短期负荷预测模型(WA-SVM)。首先收集大量的短期负荷历史数据,并对它们进行混沌分析和处理,建立支持向量机的训练和测试样本集;然后采用支持向量机建立短期负荷预测模型,并通过野草算法找到支持向量机最优参数;最后采用短期负荷预测仿真对比实验测试其性能。结果表明,WA-SVM获得了比其他模型更高的短期负荷预测精度,为短期负荷建模与预测提供了一种新的研究方法。The short-term load has the chaos characteristic due to the comprehensive influence of weather and seasons. In order to describe the change trend of short-load accurately and improve the prediction accuracy, a short-term load forecasting model fusing weed algorithm with support vector machine （WA-SVM） is proposed. A large number of short-term load historical data is collected, and performed with the chaotic analysis and processing to establish the training and testing data sets of support vector machine. And then the support vector machine is used to establish the short-term load forecasting model, and the weed algo- rithm is used to find out the optimal parameters of support vector machine. The performance of the short-term load forecasting was tested with simulation contrast experiment. The results show that the short-term load forecasting accuracy of WA-SVM model is higher than that of other models, and this model provides a new research method for short term load modeling and forecasting.

关 键 词：电力系统 短期负荷 野草算法 相空间重构 

分 类 号：TN915.34[电子电信—通信与信息系统] TP391[电子电信—信息与通信工程]