IDENTIFICATION OF THE SHAFT ORBIT FOR ROTATING MACHINES USING WAVELET MODULUS MAXIMA

Abstract

In this paper, a novel approach for the identification of the shaft centre orbits is proposed, which is based on wavelet modulus maxima. For this approach, continuous wavelet transform is utilised upon the Θ–S diagram of the shaft centre orbits, and then, maxima lines are extracted and Lipschitz exponents α of every maxima line are calculated by non-linear least-squares method. The number of maxima lines with positive α, the mean of the positive α, the number of maxima lines with negative α and the mean of the negative α are used as features of the shaft orbits. In succession, the four features are used as the inputs of the back-propagation network to classify the shaft orbits for the fault diagnosis for the rotating machines. The experimental data of four classes of faults (rub-impact, oil whirl, coupling misalignment and unbalance) are used to test the performance of the proposed method. The test result indicates that the proposed method is very effective with the advantages of fewer features used and faster training of the network, and can fulfil classification of the shaft centre orbit with satisfactory accuracy.