基于小波去噪和ELMAN神经网络负荷预测

Short-term Electric Load Forecasting of Wavelet De-noising and ELMAN Neural Network

针对短期电力负荷的特点,提出了一种应用小波去噪并采用ELMAN反馈神经网络的短期负荷预测模型。小波变换可以在时域和频域上对信号进行分析,能够较好地区分信号中的噪声。ELMAN反馈型神经网络是一种动态网络,由于反馈的作用,ELMAN神经网络不仅可以存储当前的顺序输入数据,而且可以存储顺序输入数据中的过去的某些信息。仿真计算表明此方法与传统预测模型相比,能有效地提高预测精度。

According to the characteristics of the short-term power load,a kind of short-term load forecasting model is proposed based on wavelet de-noising and ELMAN feedback neural network.Wavelet transform is widely used in the load smoothing process due to its capability of analyzing the signal in time and frequency domain and distinguishing the noise signal preferably.The hidden output of ELMAN will feedback back to forming a dynamic network,which is different from common forward neural network in structure.Due to the feedback effect,not only can ELMAN neural network store the current sequence input data but also some certain input data information of the past.Through the simulation of actual data,it is proved that this method is good at improving the prediction accuracy compared with the traditional forecast model.