基于对向传播神经网络的齿轮箱故障诊断研究

Fault diagnosis to gear box based on counter propagation neural network

介绍了CPN网络的原理、算法,利用神经网络具有任意逼近非线性函数的能力,建立了齿轮箱故障征兆与故障原因的对向传播神经网络模型。并用CPN网络对齿轮箱机械传动系统的故障进行了诊断。实例结果表明,该方法能够准确地诊断齿轮箱故障,同时具有训练速度快、结构简单、精度高等特点,是一种行之有效的诊断方法。

The paper introduces the CPN＇s principle, algorithm, and describes how CPN model for fault symptoms and causes is built by means of the non-linear approximate function from neural network, and how faults of gear box are diagnosed by means of CPN. The practice shows that it features accurate fault diagnosis, easier training and simple structure. It is a practicable diagnosis method.