A Novel Method in the Diagnosis of the Assembly Conditions of Gears

Abstract

In this study, a novel method is proposed to diagnose the assembly failure of a gearbox. Different kinds of failure mode can be simulated through the CAE software firstly. The signals of the failure modes can thus be store in the database for the purpose of diagnosis of the gear faults. The method avoids the complex experimental process to obtain the signals gathered from the actual device. The database thus could be set up effectively. In this paper, the assemblies of the 3-D gear model, grid mesh of the gear pair, material properties, contact/impact model, hourglass control, etc. are built firstly in the HyperMesh software. Then, LS-DYNA is utilized to analyze the dynamic responses of gear transmission in time domain. The FFT (Fast Fourier Transform), power spectrum, EMD (Empirical Mode Decomposition), and wavelet analysis are applied for signal processing. The fault characteristic parameters can thus be extracted and regularized. BPN (Back-propagation Network) and PNN (Probabilistic Neural Network) methods are then applied to develop the diagnosis system.