摘要
针对高校公寓现用的电能表计量方案不具备负载自动识别的功能,提出了一种基于小波变换与BP神经网络相结合的非线性负载功率预测的方法。先采用Dmeyer小波函数对用户负载电流波形分解,提取表征非线性负载类型的参数值。然后建立三层BP神经网络模型,并采用L-M算法进行网络训练与预测,实现公寓的负载识别功能。研究结果表明,小波BP神经网络公寓负载识别方法具有可靠性和实用性,实现了学生公寓的用电管理现代化,对消除校园火险隐患具有重大意义。
A method of forecasting for non-linear load power based on combination of Wavelet with BPNN is presented in the paper, because the current electric energy meter in campus apartment can＇t recognize load automatically. At first, Dmeyer wavelet is used to decompose user＇s load current wave so that feature parameter of non-linear load is picked up. Secondly, three layer BP model is established. In the end, network is trained and forecasted by Levenberg-Marquardt algorithm. Results show that the load recognition method for campus apartment based on wavelet and ANN is reliable and practical, which can realize modem and automatic management about student＇s apartment, and is important to avoid fire troubles.
出处
《电测与仪表》
北大核心
2008年第12期8-11,共4页
Electrical Measurement & Instrumentation
基金
河北省科技厅科研计划资助项目(06213507D)
关键词
小波分析
BP神经网络
L-M算法
非线性负载
wavelet analysis, BP Neural network, Levenberg-Marquardt algorithm, nonlinear load
