基于负荷分解和实时气象因素的短期负荷预测  

Short-Term Load Forecasting Based on Load Decomposition and Hourly Weather Factors

作　　者：刘旭[1] 罗滇生[1] 姚建刚[1] 贺辉[2] 张凯[2] 刘霏 

机构地区：[1]湖南大学电气与信息工程学院,湖南省长沙市410082 [2]湖南省电力调度通信中心,湖南省长沙市410007 [3]衡阳市电业局,湖南省衡阳市421001

出　　处：《电网技术》2009年第12期94-100,共7页Power System Technology

基　　金：电子信息产业发展基金(信部运[2004]479)

摘　　要：根据地区气象与负荷的相关关系,从总负荷中分解出对气象不敏感的基础负荷和受气象因素影响的气象敏感负荷,并分别采用灰色系统GM(1,1)模型和基于LMBP(Levernberg-Marquardt back propagation)算法的多层前馈神经网络对二者进行建模预测。在对实时气象因素、日特征气象因素与气象敏感负荷相关性分析的基础上,重点把握某些气象因素与气象敏感负荷之间的联系。通过合理选择神经网络的输入变量,实现了基于实时气象因素的短期负荷预测。实际应用证明了所提出方法的有效性。According to the correlation of area weather and power load, the basic load that is not sensitive to weather and weather-sensitive load influenced by weather factors are decomposed from total load, and these two decomposed parts are modelled and forecasted by GM（1,1） model and multi-layer feed-forward neural network based on Levernberg-Marquardt back propagation （LMBP） algorithm respectively. On the basis of analyzing the correlation among hourly weather factors and daily characteristic meteorological factors with weather- sensitive load, the correlativity of some of these weather factors with weather-sensitive load are taken as the focus to be grasped. By means of choosing input variables of neural network reasonably, the short-term load forecasting based on hourly weather factors is implemented. Practical application shows the proposed method is effective.

关 键 词：短期负荷预测 实时气象因素 负荷分解 气象敏感负荷 神经网络 

分 类 号：TM715[电气工程—电力系统及自动化]