改进偏最小二乘回归在电力负荷预测中的应用

多元回归分析是中长期电力负荷预测中的一个重要方法,其中偏最小二乘法因为可以有效解决样本数据偏少以及自变量多重线性的问题而得到采用。探讨了通过Bootstrap方法给偏最小二乘法筛选自变量后进行负荷分析预测的可行性,将计算结果与一般偏最小二乘法及经变量筛选的逐步回归法进行比较。算例表明,应用Bootstrap方法进行参数检验的偏最小二乘方法在变量关系的描述上更简明准确,同时提高了预测精度,具有一定的实用性。

Application of improved partial least square regressive model in power load forecasting

The multiple regression analysis is an important method in medium and long-term load forecasting,in which the partial least squares(PLS) regression analysis that can solve the difficulty of not enough sample numbers and severe multiple correlation of self-variables has been successfully used. This paper approached the feasibility of using partial least square regression with the selected optimal variables by the method Bootstrap and it is compared with the normal PLS and stepwise regression by variable sele...