Application of continuous potential function stochastic resonance in early fault diagnosis of rolling bearings

Xueping Ren,Jian Kang,Zhixing Li,Jianguo Wang

doi:10.1177/0020294019890633

Abstract

The early fault signal of rolling bearings is very weak, and when analyzed under strong background noise, the traditional signal processing method is not ideal. To extract fault characteristic information more clearly, the second-order UCPSR method is applied to the early fault diagnosis of rolling bearings. The continuous potential function itself is a continuous sinusoidal function. The particle transition is smooth and the output is better. Because of its three parameters, the potential structure is more comprehensive and has more abundant characteristics. When the periodic signal, noise and potential function are the best match, the system exhibits better denoise compared to that of other methods. This paper discusses the influence of potential parameters on the motion state of particles between potential wells in combination with the potential parameter variation diagrams discussed. Then, the formula of output signal-to-noise ratio is derived to further study the relationships among potential parameters, and then the ant colony algorithm is used to optimize potential parameters in order to obtain the optimal output signal-to-noise ratio. Finally, an early weak fault diagnosis method for bearings based on the underdamped continuous potential stochastic resonance model is proposed. Through simulation and experimental verification, the underdamped continuous potential stochastic resonance results are compared with those of the time-delayed feedback stochastic resonance method, which proves the validity of the underdamped continuous potential stochastic resonance method.

Highlights

A rolling bearing is a frequently used part in rotary machinery
In order to illustrate the advantages of the underdamped continuous potential stochastic resonance (UCPSR) method, we investigated the effectiveness of UCPSR and time-delayed feedback stochastic resonance (TFSR)
Based on simulation and experimental verification, the following conclusions are as follows: 1. Because the traditional Stochastic resonance (SR) method is a first-order differential equation, the first-order model is equivalent to a filter link, whereas the proposed method is a second-order under-damped stochastic resonance method, and the second-order differential equation is equivalent to a second-order filter, so the second-order stochastic resonance model offers improved filtering effect

Summary

Introduction

A rolling bearing is a frequently used part in rotary machinery. If a problem within this part occurs, production accuracy can be affected slightly, resulting in both the scrapping of a product, and seriously endangering the safety and property of machinery operators. After solving the problem of small parameter limitation, scholars applied different potential function SR to the field of mechanical fault diagnosis. Lu et al.[21] applied the Woods potential to the early fault diagnosis of bearings and improved the output signal-to-noise ratio (SNR) by adjusting the Woods-Saxon potential parameters. Shi et al.[28] applied tristable potential SR to fault diagnosis, finding that the energy of a fault signal can be increased and that the fault frequency can be identified with appropriate delay feedback parameters All these methods have achieved good results in the field of early fault diagnosis of bearings. In this paper, an underdamped continuous potential stochastic resonance (UCPSR) method is proposed and applied to the early fault diagnosis of bearings.

Introduction of UCPSR method

D2 a 2p2 exp

Conclusion