小波神经网络在短期电力负荷预测中的应用

Application of wavelet neural networks in short-period power load prediction

针对小波网络建模存在难以有效对小波级数进行截断的问题,提出一种基于频带能量的小波级数截断方法,该方法能克服传统方法对个人经验的依赖,确保小波网络的逼近能力;结合正交最小二乘法,将其应用于电力负荷预测中,建立负荷预测的小波网络模型.仿真结果表明:用该方法建立的小波网络模型简洁、明了,结果有效、准确,具有较好的预测能力.

In order to truncate wavelet series effectively in the wavelet network modeling, the paper presents a new method based on frequency energy to truncate wavelet series. The new method insures the approaching ability of wavelet network. Combined with the orthogonal least square (OLS), the method is applied to load prediction, and the wavelet network model of load prediction is established. Simulation shows effectiveness of the method.