基于小波分析的并行膜混合核支持向量机月度负荷预测

Parallel Membrane Hybrid Kernel Support Vector Machine Based on Wavelet Analysis for Monthly Load Forecasting

针对现代电力系统月负荷数据的趋势增长性和波动性的非线性特征,提出了一种基于小波变换的混合支持向量机负荷预测模型。通过小波变换将负荷序列分解为不同尺度的子序列,考虑负荷的季节波动性,将温度因素作为输入变量,构建混合核函数LWPSO-LSSVM。将负荷子序列分别放入膜系统的基本膜中进行并行预测,然后对子序列预测数据进行重构得到预测结果。利用四川省某地区电网负荷数据进行应用研究,结果表明所提出的模型较传统核函数支持向量机预测精度和效率有明显提高。

Aiming at the properties of increase trend and nonlinear seasonal fluctuation of monthly load in modern power system, this paper proposed a load forecasting model of least squares support vector machine with mixed kernel function based on wavelet transform. The monthly load series was decomposed as the superposition of multi frequency components by means of wavelet transform. Considering the seasonal fluctuation, the climate factors were selected as input variables. Then a mixed kernel function LWPSO LSSVM was established to forecast the load sequence respectively. The load subsequences were predicted with the base film of membrane system respectively. The predicted different frequency components were reconstructed to form the load forecasting. Taking load data in a regional power grid of Sichuan Province as application research, the results show that the predicted precision and efficiency of the proposed model is higher than the traditional support vector machine.