基于核偏最小二乘回归分析的线损率预测

Line Loss Rate Forecasting Based on Kernel Partial Least-Squares Regression Analyze

研究了线损率预测问题。由于影响线损率的因素间存在着复杂的非线性和强相关性,一般方法难以得到较高精度的预测结果。针对线损率的特点,为了提高预测精度,本文首次将核偏最小二乘回归算法应用于线损率预测。先以历年来的线损率及其相关数据为样本建立预测模型,然后对预测年线损率进行预测。以某电网为实例进行仿真,并将仿真结果与其他方法所得到的仿真结果进行比较。结果表明基于核偏最小二乘回归分析的线损率预测具有较高精度,能较好地克服变量相关性和非线性因素对预测模型的不利影响,为电力企业制订科学合理的线损率计划提供理论依据。

Studied the line loss rate forecast. In order to improve the prediction accuracy and according to the characteristics of line loss rate, Kernel Partial Least-Squares （KPLS） algorithm was firstly applied in line loss rate forecasting. It used the line loss rate and associated data over years to build predictive models at first, and to forecast the line loss rate of next year. It used a power grid to simulate, and the simulation results was compared with the re- sult obtained with other methods. The results show that the forecast of line loss rate based on KPLS regression analysis is of high accuracy. It can overcome the adverse effects of the variable correlation and nonlinear factors, and provide a theoretical basis for developing scientific and rational line loss rate plan by the power companies.