摘要
利用小波分解和人工神经网络相结合的方法建立了城市供水管网短期水量负荷的组合预测模型。该方法首先利用小波分解技术将时负荷水量分解为相对简单的带通分量信号,然后根据各分量信号的特点分别建立独立的神经网络预测模型,最后将预报结果集成。由于小波分解后各分量的频率相对单一,因而可有效缩短网络训练时间,提高计算速度。仿真计算结果表明,该方法建模合理、计算量适中,可准确预测管网水量。
A forecast model for short-term water consumption in municipal water supply networks was established by the combined method of wavelet decomposition and Artificial Neutral Network （ANN）. Using this method, hourly water consumption is first separated into several simple band-pass signals, the independent ANN forecasting models are then constructed separately according to the characteristics of the decomposed signals. And finally, the ANN models forecasts the integrated results. Because of the simplicity of the decomposed signals frequencies, the network training time can be shortened in effect, and the calculated speed can also be increased. Simulation results show that the ANN forecast method is reasonable, and can rapidly and accurately forecast the water consumption.
出处
《中国给水排水》
CAS
CSCD
北大核心
2006年第17期60-63,共4页
China Water & Wastewater
关键词
供水管网
水量预测
小波分解
人工神经网络
water supply networks
water consumption forecast
wavelet decomposition
artificial neutral network
