For the sensor fixed on the equipment to be detected with the inadequacies of fault diagnosis based on vibration signal. Fault diagnosis based on the sound signal has a non-contact acquisition, the signal to obtain convenient and easy to control, low-cost advantages.
To be seized equipment failure, the actual sound signal acquisition is often a mixture of multiple failures of the source signal and its reflection, mixing a variety of ways, such as linear mixed convolution mix, this article discusses the instantaneous linear mixing. How to obtain the real fault source signal equipment to be attending a key step in the diagnosis. Blind source separation technology has developed rapidly in recent years to provide a new solution ideas.
Blind source separation ¨ 1 is to study the case of the transfer function of the unknown, the unknown source signal, mixed-signal from the observed isolated independent of the source signal technology. Usually use a certain number of sensors of several sound sources at the same time measurements, mixed-signal measured by each sensor are several sound sources, but does not know the mixing matrix. Separation method using independent component analysis, without any prior knowledge, just from the observation signal to recover the source signal is extremely difficult, so the ICA model requires that the source signals are statistically independent.
INA Bearing failure sound signal acquisition in engineering applications not only contains a number of fault sources, and statistical correlation between the source signal, the traditional application of independent component analysis method is restricted. In order to resolve between the source signal statistics-related issues, this paper, the Wavelet Transform and Independent Component Analysis of the Shaanxi-speed multi-resolution sub-band decomposition independent component analysis.
INA Bearing failure typical damage class failure, is characterized by high frequency sub-band show a strong modulation characteristics envelopment analysis to diagnose such failures. The method is inspired by the low-frequency impact of high-frequency resonance wave envelope detection, demodulation, and a corresponding low-frequency impact and amplification and broadening of the resonance demodulation wave. Through this resonance demodulation wave spectral analysis to determine the type of fault. Envelopment analysis to law, where the Hilbert resonance demodulation method M3.
Product Model | Inside Diameter | Outside Diameter | Thickness |
24136CK30E4 NSK | 180 | 300 | 118 |
23136CKE4 NSK | 180 | 300 | 96 |