Phase Space Similarity as a Signature for Rolling Bearing Fault Diagnosis and Remaining Useful Life Estimation

Fang Liu,Bing He,Yilei Zhao,Siliang Lu,Jiwen Zhao,Yongbin Liu

doi:10.1155/2016/5341970

Fang Liu, Bing He + Show 4 more

Open Access

https://doi.org/10.1155/2016/5341970

Copy DOI

Abstract

Feature extraction from vibration signal is still a challenge in the area of fault diagnosis and remaining useful life (RUL) estimation of rotary machine. In this paper, a novel feature called phase space similarity (PSS) is introduced for health condition monitoring of bearings. Firstly, the acquired signal is transformed to the phase space through the phase space reconstruction (PSR). The similar vibration always exists in the phase space due to the comparable evolution of the dynamics that are characteristic of the system state. Secondly, the normalized cross-correlation (NCC) is employed to calculate the PSS between bearing data with different states. Based on the PSS, a fault pattern recognition algorithm, a bearing fault size prediction algorithm, and a RUL estimation algorithm are introduced to analyze the experimental signal. Results have shown the effectiveness of the PSS as it can better grasp the nature and regularity of the signals.

Highlights

Fault diagnosis is an extremely significant research field in industry; intelligent health monitoring is gradually supplemented in maintenance of the machine to ensure the stability of systems and decrease the downtime
A novel feature extraction method that is based on phase space reconstruction (PSR) and normalized cross-correlation (NCC) is proposed for fault diagnosis and remaining useful life (RUL) estimation
A novel method based on phase space similarity (PSS) is proposed to recognize the fault patterns, predict the fault size, and estimate the RUL of rolling bearing

Summary

Introduction

Fault diagnosis is an extremely significant research field in industry; intelligent health monitoring is gradually supplemented in maintenance of the machine to ensure the stability of systems and decrease the downtime. In the literature [13], the PSR is combined with time-frequency synthesis and utilized in data denoising In these studies, the PSR is introduced for fault diagnosis and RUL estimation, which can extend time series to a high-dimensional phase space and maintain topologically equivalent. On the basis of these features that are extracted from the sensor signals, machine learning is widely used in machine fault diagnosis such as artificial neural networks (ANN) [19] and support vector machine (SVM) [20]. Three applications based on PSS are verified by case studies; the experimental results show that the proposed method can effectively realize the fault diagnosis and RUL estimation for bearing.

Theory Background

Methods and Procedures

Experiment Validation

Conclusions