非线性偏最小二乘回归在电力负荷预测中的应用

Application of non-linear partial least square regression in electricity load prediction

非线性(二次多项式)偏最小二乘既能够解决线性偏最小二乘只能提取线性成分的问题,它又借鉴了偏最小二乘回归方法能够有效地解决自变量集合多重相关性的问题,因而它更具有先进性,其计算结果更为可靠。本文将二次多项式非线性偏最小二乘回归应用于泉州地区的电力负荷预测。文章还将二次多项式偏最小二乘的预测结果并线性偏最小二乘和logistic模型的预测结果进行比较,实例预测结果表明,非线性偏最小二乘具有较高的预测精度,它能满足实际工程的要求。

The non-linear （polynomial） partial least square （NPLS） solves both the problem of multiple correlations among variables and the issue where the partial least square （PLS） could only extract linear components, thus being more advanced and its calculation results being more reliable. The paper applies the quadratic PLS （QPLS） to electricity load prediction of Quanzhou, illustrates the model-building process and proves its better prediction effects and wider application by comparing its prediction result with those of linear PLS and Logistic Model.