基于决策树和逐步回归的气温对负荷影响的定量分析法

Quantitative analysis method of the relationship between temperature and power load based on decision tree and step regression

气象状况尤其是气温是对短期负荷预测精度影响最大的因素。文章提出了一种采用决策树和逐步回归技术对气温和用电负荷之间的关系进行定量分析的方法,并对上海地区气温与负荷之间关系进行分析。该方法首先从原始负荷数据序列中提取受气象因素影响的负荷分量;然后对气温影响负荷的滞后性进行了分析并得出了在上海地区气温对负荷影响的滞后性一般不超过12h的结论;在此基础上运用决策树和逐步回归技术进一步给出了气温对负荷影响的定量分析结果。根据定量分析结果所设计的实际短期负荷预测方法进行负荷预测实践,结果表明该分析结果能够有效反映不同气象条件下气温变化对负荷的影响。

This paper reports some results of a study regarding relationship between temperature and power load of Shanghai electric power system. Weather-influenced load part is picked up from original load series data, with the conclusion that the lagged effect of temperature on load is within 12 hours. Furthermore, decision tree and step regression methods are able to describe the relationship between load and temperature. A short-term load forecasting algorithm is then developed, and its practical implementation shows the quantitative analysis method could reliably reflect the influence of the temperature changes on the load and effectively improve the accuracy of short-term load forecasting.