Mechanical fault diagnosis method based on translation invariant fractional wavelet sparse representation

Abstract

The invention discloses a mechanical fault diagnosis method based on translation invariant fractional wavelet sparse representation, and the method comprises the following steps: collecting a vibration signal of a monitored object, and enabling the vibration signal to serve as system input; carrying out translation invariant fractional order wavelet decomposition on the mechanical fault vibrationsignal; processing the wavelet coefficient by using a nonlinear operator so as to enhance the sparsity of the wavelet coefficient, and performing generalized hard threshold noise reduction on the processed wavelet coefficient to eliminate residual noise in the coefficient so as to further enhance the sparsity of the wavelet coefficient; performing translation invariant fractional wavelet inverse transformation on the processed wavelet coefficient to obtain a sparse representation result of the fault shock waveform; and extracting a fault characteristic frequency from the sparse representationresult of the fault characteristic waveform, and carrying out fault diagnosis. The method is higher in denoising capability, can provide a sparser and more accurate fault shock waveform representationresult, achieves the accurate diagnosis of a mechanical fault, and provides a reliable basis for the safe operation of equipment.

Description

technical field [0001] The invention relates to the field of mechanical vibration signal processing and fault diagnosis, in particular to a mechanical fault diagnosis method based on translation invariant fractional wavelet sparse representation. Background technique [0002] Once the key mechanical equipment fails, it will bring huge economic losses and even casualties. Major accidents caused by mechanical equipment failures have occurred frequently at home and abroad. If the fault can be found in time during the operation of the equipment, it is of great significance for the safe operation of the equipment and the avoidance of economic losses and catastrophic accidents. [0003] An arduous task in mechanical fault diagnosis is how to accurately recover periodic fault shock waveforms from noisy vibration signals, which has attracted widespread attention in the industry and academia in recent years. The wavelet analysis method is based on the fact that the wavelet coefficie...

AI Technical Summary

Problems solved by technology

However, most traditional wavelet post-processing methods (such as hard threshold denoising, L1 norm regularization denoising) are not ideal for denoising signals with low signal-to-noise ratio (wavelet coefficient sparsity is not obvious), and it is easy to denoise Artifacts such as noisy sharp points or Pseudo-Gibbs effects are introduced into the signal (that is, the representation of the fault shock waveform is not sparse and accurate enough), which affects the interpretation of fault characteristics and the accurate diagnosis of fault types

Images