云模型优化LSSVR 的短期电力负荷预测.PDF

云模型优化LSSVR的短期电力负荷预测摇 张捍东,等 云模型优化 LSSVR的短期电力负荷预测 Short Term Power Load Forecasting Based onthe LSSVR Optimized by Cloud Model 张捍东 摇 张摇 莉 摇 汤摇 敏1 1 2 (安徽工业大学电气信息学院 ,安徽 马鞍山摇 243000;罗克韦尔自动化(中国)有限公司 ,上海摇 200233)1 2 摘摇 要: 准确的短期电力负荷预测是电力系统安全经济运行的重要依据。 针对短期负荷影响因素的非线性特性,研究了基于最小二 乘支持向量回归机(LSSVR)的短期负荷预测方法。 鉴于LSSVR算法参数选取的难点,提出了云进化算法优化LSSVR 的短期电力负 荷预测模型(CBEA_LSSVR)。 CBEA_LSSVR利用云模型实现LSSVR的参数优化,优化后的模型能够预测下一时刻的电力负荷值。 仿 真结果表明,与其他进化算法相比,云进化算法优化LSSVR模型具有更高的预测精度和鲁棒性。 关键词: 最小二乘支持向量机摇 云模型摇 人工智能技术摇 遗传算法摇 进化算法摇 负荷预测 中图分类号: TP181摇 摇 摇 摇 文献标志码: A Abstract: Accurate short term power loadforecasting is significant basis of safe and economic operation of the power system. In accordance with the nonlinear characteristics of the factors affecting short term load,the long term load forecasting method based on least square support vector regression (LSSVR)isresearched. In view of the difficulties of parameter selection for LSSVR,the short term load forecasting model (CBEA_LSSVR)which is based on the LSSVR optimized with cloud model based evolutionary algorithm (CBEA) is proposed. This model implements parameter optimization forLSSVRbyusingcloudmodel,thepowerloadatnextperiodisrollingforecasted. Theresultof simulation shows that comparing withother evolutionary algorithms;the cloudmodelbasedevolutionary algorithmwelloptimizesLSSVRandmakesit offer higher forecasting accuracy and robustness. Keywords: Least square support vector machine(LSSVM)摇 Cloud model摇 Artificial intelligence technique摇 Genetic algorithm摇 Evolutionary algorithm摇 Loadforecasting 0摇 引言 1摇 短期电力负荷预测概述 短期负荷预测是指在负荷的历史数据基础上,结 国内外学者提出了多种短期电力负荷预测方法。 合影响负荷的多种不同因素,预测出未来一天各时间 20世纪80年代初期,学者们提出了诸如回归分析法、 周期段的负荷值。 准确的短期负荷预测既是电力系统 [1]