Robust Fault Diagnosis of Rolling Bearings Using Multivariate Intrinsic Multiscale Entropy Analysis and Neural Network Under Varying Operating Conditions

Rui Yuan,Yong Lv,Gangbing Song,Hewenxuan Li

doi:10.1109/access.2019.2939546

Abstract

olling bearings are crucial components in mechanical, civil and aerospace engineering. The practical working conditions of rolling bearings are complex and tough, hence fault diagnosis of rolling bearings under varying operating conditions is very challenging. This paper proposes a robust fault diagnosis approach of rolling bearings using multivariate intrinsic multiscale entropy analysis and neural network under varying operating conditions. The proposed approach deals with multivariate signal collected from multi-sensor acquisition system to capture much dynamical characteristic information. Multivariate intrinsic multiscale entropy analysis consists of adaptive projection intrinsically transformed multivariate empirical mode decomposition with adaptive noise (APIT-MEMD-AN) and improved multivariate multiscale sample entropy (IMMSE) with smoothed coarse graining process. Intrinsic mode functions (IMFs) obtained by APIT-MEMD-AN depict dynamical properties of multivariate signals. IMMSE of certain orders IMFs are adopted as input values of back propagation (BP) neural network to achieve fault classification of rolling bearings. APIT-MEMD-AN and IMMSE endow the proposed approach with the underlying adaptivity and robustness, making the proposed approach a fully data driven and robust method. Theoretical derivations, numerical simulations and experimental results verify the effectiveness and superiority of the proposed approach. The research work demonstrates the proposed approach is promising in fault diagnosis of rotary machinery under varying operating conditions.

Highlights

Rolling bearings are crucial components in mechanical, civil and aerospace systems, and their working conditions have great influence on the entire system [1]–[3]
Multivariate intrinsic multiscale entropy analysis consists of APIT-multivariate empirical mode decomposition (MEMD)-AN and improved multivariate multiscale sample entropy (IMMSE)
Multivariate signals are decomposed by APIT-MEMD-AN to get multiple sets of Intrinsic mode functions (IMFs), and IMMSE of certain IMFs are adopted as input values of back propagation (BP) neural network to achieve fault diagnosis of rolling bearings

Summary

INTRODUCTION

Rolling bearings are crucial components in mechanical, civil and aerospace systems, and their working conditions have great influence on the entire system [1]–[3]. Bivariate empirical mode decomposition (BEMD) was proposed to process binary signal [17]–[19], whose local means are obtained by projecting in multiple directions. APIT-MEMD-AN can alleviate adverse effect of power imbalances among multiple channels of the collected multivariate signal and utilize its underlying filter bank property. APIT-MEMD alleviates adverse effect of power imbalances among multiple channels by large amounts of adaptive projection vectors. 5) Use adaptive projection direction vectors xθk and xθk o1 to conduct iterative decomposition, and local mean estimation is achieved based on MEMD, obtain (n1 + n2) sets of IMFs of s (t). The advantages of APIT-MEMD-AN against APIT-MEMD in fault diagnosis of rotary machinery lie in two aspects as follows: 1) APIT-MEMD-AN utilizes the intrinsic filter bank property in the presence of Gaussian white noise with frequency uniformly distribution property to alleviate mode mixing problem. If different orders IMFs were generated in different decompositions, it could cause confusion if certain orders IMFs were adopted in subsequent analysis

IMPROVED MULTIVARIATE MULTISCALE SAMPLE

NUMERICAL SIMULATIONS

SIMULATION RESEARCH OF APIT-MEMD-AN

EXPERIMENTAL VERIFICATION

DISCUSSION AND CONCLUSION