基于改进灰色马尔科夫预测法的中长期负荷预测  

Mid-long term load prediction based on improved Grey Markov method

作　　者：夏耀杰 程浩忠[2] 

机构地区：[1]上海电力学院电气工程学院,上海200093 [2]上海交通大学电子信息与电气工程学院,上海200240

出　　处：《电力需求侧管理》2015年第1期21-25,共5页Power Demand Side Management

基　　金：国家科技支撑计划课题(2013BAA01B04)

摘　　要：提出了一种提高对可能发生较大畸变序列的负荷预测精度的方法。由于电力负荷受天气、节日、经济等较多因素影响,在时间序列上表现为非平稳的随机过程,在某些年份负荷值可能会出现较大畸变,导致模型预测精度下降。分三步对畸变较大的数据样本进行预测以及误差分析,首先建立灰色预测模型,然后利用残差进行模型修正以增加序列波动性,而后用改进的马尔科夫链进行误差修正以提高精度。采用某市1998年至2013年最大负荷作为样本数据验证算法有效性,算例结果表明:与传统灰色系统预测模型相比,此模型有更高的拟合精度和预测精度。此方法的优势在于:在相同样本情况下,拟合程度高并且可以明显修正畸变数据带来的误差。This paper proposes a method that can improve the accuracy of the load sequence with large distortion. Load is dif- ficult to predict, because it＇ s affected by weather, holidays, econo- my and other factors. Time series of load show a non-stationary ran- dom process. In some years there will be even greater load values distorted that will result in the model prediction accuracy declin- ing. This paper has three steps to forecast the distortion larger data sample, and the error analysis. First, it sets up gray prediction mod- el, then models with residuals model modified to increase the ran- domness. In the end, it uses a modified Markov chain error correc- tion to improve the accuracy of the error. Article sample data using a city from 1998 to 2013 the maximum load is used to verify the re- liability of the algorithm, numerical example results show that： compared with the traditional grey system forecasting model, this model has higher fitting degree. The advantage of this method in the same sample is high degree of fitting and obviously correct dis-tortion error.

关 键 词：负荷预测 灰色模型 残差修正 改进马尔科夫链 数据畸变 

分 类 号：TM715[电气工程—电力系统及自动化]