使用改进混沌神经网络的母线负荷预测

Bus load forecasting using improved chaotic neural network

针对母线负荷非线性、冲击性波动、有较多"毛刺"、含有较多坏数据等特点,提出了一种基于小波变换和混沌神经网络的母线负荷预测方法。该方法通过消除坏数据和噪声对负荷混沌特性分析的影响,能有效提高母线负荷预测的精度。首先对历史数据进行改进的小波阈值去噪,然后对其进行混沌特性分析,重构相空间形成训练样本,最后采用改进的混沌学习算法对网络进行训练。通过对某省某地220kV母线负荷算例分析,显示该方法能显著提高母线负荷预测的精度。

The bus load is nonlinearity, impact fluctuation, more burrs and contain more bad data, the prediction algorithm based on the wavelet transform and chaotic neural network is proposed to set up the bus load model. This method can effectively im prove the accuracy of the load forecast by eliminating bad data and noise to the influence of bus load chaotic characteistic analysis. Firstly, it de-noises data of load using improved wavelet threshoht, then uses the reconstruction of phase space to fore1 the training sample through chaotic characteristic analysis of it. Finally it uses the improved chaotic learning algorithm to train the neural network. Through the example analysis of the 220 kV bus load of provincial somewhere, this method can significantly improve the bus load forecasting accuracy.