利用小波神经网络对单相三电平逆变器谐波检测  

Harmonic detection to single-phase three-level inverter by wavelet neural network

作　　者：马辉[1] 王玉华[1] 刘朋朋[1] 

机构地区：[1]长春工业大学,吉林长春130000

出　　处：《电气传动自动化》2013年第5期24-28,共5页Electric Drive Automation

摘　　要：采用小波神经网络(Wavelet Neural Network,WNN)算法对时变谐波信号进行检测。利用Harr小波对时变幅值和相角进行逼近;将小波对信号的自适应时频分割特性引入神经网络,提高神经网络的逼近和收敛性能;给出网络参数的选定方案;确定网络的训练算法;在matlab/Simulink环境下进行仿真,验证该方法的可行性和精确性。The time-varying harmonic signal is detected by wavelet neural network algorithm. Using Harr wavelet approximation time-varying amplitude and phase angle, wavelet on the signal adaptive time-frequency characteris- tics of neural network is introduced to improve segmentation, neural network approximation and convergence per- formance. The selected scheme of network parameters is proposed, and the network training algorithm is also de- termined. The feasibility and accuracy in the Matlab/Simulink environment simulation are proved.

关 键 词：小波神经网络 时变谐波检测 Harr小波变换 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TM935[自动化与计算机技术—控制科学与工程]