Fault detection in rotor system by discrete wavelet neural network algorithm

This study identifies a method for detection of irregularities such as open cracks or grooves on rotating stepped shaft with multiple discs, based the wavelet transforms. Cracks are represented reduction in diameter (groove) small width. Single well considered at locations stress concentration. Translational rotational response curves/mode shapes extracted from finite element analysis rotors and without grooves. Discrete continuous 1D transforms applied resultant curve mode shapes. The results show that curves more sensitive to key contributors identify location than translation accurate enough locate groove smaller size. Effectiveness by wavelets is analysed single increase depth. Increase depth can be quantified coefficient, it an indicator. White Gaussian noise low signal-to-noise ratio added crack identification. Intelligent techniques artificial neural networks used quantify crack. coefficients provided input network. Feed forward trained Levenberg–Marquardt back propagation algorithm. Trained able accurately.