基于预测有效度和马尔科夫-云模型的母线负荷预测模型筛选与变权重组合预测  

Bus load forecasting model selection and variable weights combination forecasting based on forecasting effectiveness and Markov chain-cloud model

作　　者：尹星露 肖先勇[1] 孙晓璐[1] 

机构地区：[1]四川大学电气信息学院,四川成都610065

出　　处：《电力自动化设备》2015年第3期114-119,共6页Electric Power Automation Equipment

摘　　要：预测有效度与鲁棒性是母线负荷预测面临的重要课题。基于日特征相关因素选取待预测日的相似日;从待预测日母线负荷真值未知的实际出发,引入模型预测精度和预测有效度概念,研究模型有效度的转移规律,提出基于马尔科夫链和云模型的预测精度定量估计方法;基于模型综合有效度,提出组合模型筛选方法和变权重母线负荷组合预测方法。基于所提方法开发了一套母线负荷预测系统,并应用于我国某地区电网。多条母线、多时段的预测结果表明,所提方法所得预测结果的有效度高且稳定,具有预测结果准确和鲁棒性好的特点。Availability and robustness are two important indexes of bus load forecasting. Similar days of the day to be forecasted are selected based on the daily characteristic factors and the concepts of model forecasting accuracy and model availability are introduced. The transfer rule of model availability is researched and the quantitative forecasting accuracy assessment based on Markov chain and cloud model is put forward. With the comprehensive model availability,a method for selecting the combination forecasting model and a method for forecasting the bus load with variable weights are proposed,based on which,a bus load forecasting system is developed and applied in a regional power grid. The results of multi-bus multiperiod forecasting show that,the forecasting models selected by the proposed methods are stable and effective while the forecasts are accurate and robust.

关 键 词：地区电网 负荷预测 相似日 有效度 马尔科夫链 云模型 模型筛选 变权重组合 

分 类 号：TM715[电气工程—电力系统及自动化]