讨论了矢谱融合技术和Levenberg—Marquardt(L-M)神经网络的相关理论,提出了基于矢谱和L-M神经网络的旋转机械故障诊断方法,建立了基于矢谱的旋转机械常见故障诊断L—M神经网络模型。模拟实验结果表明:与基于单通道数据的诊断结果对比,将矢谱数据融合应用于旋转机械常见故障诊断,可有效提高故障诊断的准确率。
The vector spectrum and the Levenberg-Marquardt (L-M) neural network were discussed ; the fault diagnosis method for rotary machinery based on vector spectrum and the L-M neural network was put forward, and the L-M neural network model was established. Engineering practice indicated that the fault diagnosis accuracy based on vector spectrum was higher than that based on the information of single channel and the L-M neural network.