Bearing fault diagnosis based on wavelet transform and fuzzy inference

doi:10.1016/S0888-3270(03)00077-3

Mechanical Systems and Signal Processing

Volume 18, Issue 5, September 2004, Pages 1077-1095

https://doi.org/10.1016/S0888-3270(03)00077-3 Get rights and content

Abstract

This paper deals with a new scheme for the diagnosis of localised defects in ball bearings based on the wavelet transform and neuro-fuzzy classification. Vibration signals for normal bearings, bearings with inner race faults and ball faults were acquired from a motor-driven experimental system. The wavelet transform was used to process the accelerometer signals and to generate feature vectors. An adaptive neural-fuzzy inference system (ANFIS) was trained and used as a diagnostic classifier. For comparison purposes, the Euclidean vector distance method as well as the vector correlation coefficient method were also investigated. The results demonstrate that the developed diagnostic method can reliably separate different fault conditions under the presence of load variations.

Introduction

Condition monitoring of rotating machinery is important in terms of system maintenance and process automation. Rolling element bearing failures are one of the foremost causes of failures in rotating machinery. This necessitates the development, implementation, and deployment of on-line diagnostic monitoring systems that are independent of operating conditions.

In most machine fault diagnosis and prognosis systems, the vibration of the rotating machine (motor, gearbox, etc.) is directly measured by an accelerometer, in some few cases, by an acoustic pickup. Some techniques use the stator currents of the electrical motor as the input signals for fault detection [1]. Fault signal detection and recognition are often accomplished by pattern recognition using a neural network [2], [3], RBF network [4], Gaussian mixture model network [5], [6], fuzzy logic network [5], Bayesian classifier [7], vector correlation or vector distance measure [8]. Commonly used feature generation methods include the short-time Fourier transform (STFT) [2], wavelet time-scale decomposition [2], [9], [10], cumulant spectrum [8], etc.

The discrete wavelet transform (DWT) provides an efficient method for generating feature vectors. The DWT coefficients can be used to generate statistical parameters from each resolution level of the transform. This method of feature extraction has been used to recognise signals from RF transmitters with a back propagation neural network [9] and ground vehicles with vector correlation and distance pattern matching [11]. Acoustic analysis methods have been developed to detect and classify underwater objects using wavelets with a neural network and quadratic Bayesian classifiers [12]. The discriminative feature extraction recogniser, which combines a feature extractor and classifier, is presented in [13]. This network optimises both a feature extraction process and a classification process by pattern production and adaptation. As an alternative to the back propagation neural network, a supervised radial basis function network is used. A new network type called “wave-net” [14] adapts the RBF network concept, and uses wavelets as the basis functions for the network. This network has been used for speaker identification [15]. Liu and Ling have applied the principle of mutual information to the identification of wavelets that carry significant information of machinery faults, instead of the “best matching” criterion used in matching pursuit [16]. Altmann and Mathew have used ANFIS for automated selection of wavelet packets containing bearing fault related features [17]. Peng et al., proposed a fusion fault diagnosis method based on the wavelet transform, genetic algorithms and neural networks [10]. Xu and Chan have done very similar work [18].

In this paper, a new technique for localised bearing fault diagnosis is developed using the discrete wavelet transform (DWT). In this method, experimental vibration signals for normal and faulty bearings are pre-processed to obtain a (0,1) normal distribution where the wavelet transform was used to process the normalised data. Then a feature vector is defined using the components from the DWT. By using selected segments from the available experimental data, typical sample feature vectors are generated for both normal bearings and bearings with different types of faults under different load conditions. Then different pattern classification methods have been studied in the decision making stage, including the neural-fuzzy inference system, which is believed to be most suitable for complex situations due to its adaptability and the capability of the network to realise a non-linear approximation.

Section snippets

Experimental system

The ball bearings are installed in a motor driven mechanical system, as shown in Fig. 1. A 2 hp, three-phase induction motor (Reliance Electric 2HP IQPreAlert motor), was connected to a dynamometer and a torque sensor by a self-aligning coupling. The dynamometer is controlled so that desired torque load levels can be achieved. An accelerometer with a bandwidth up to 5000 Hz and a 1 V/g output is mounted on the motor housing at the drive-end of the motor to acquire the vibration signals from the

Preprocessing test data

By examining the magnitude of the vibration data under operating conditions with severe bearing faults, it is possible to distinguish the normal data from different types of fault data. However, this is not always applicable because the signal morphology that results from a fault changes over time as the fault progresses from initiation to failure. Thus, some faults will be undetectable until failure is imminent. Because the early detection and isolation of faults is important for

Brief review of the wavelet theory

The Wavelet Transform is defined as the integral of the signal s(t) multiplied by scaled, shifted versions of a basic wavelet function ψ(t)—a real-valued function whose Fourier Transform satisfies the admissibility criteria [21], [22], [23]: $C(a,b)= ∫ R s(t) 1 a ψ t−b a d t, a∈R^{+} −{0}, b∈R.$ where a is the so-called scaling parameter, b is the time localisation parameter. Both a and b can be continuous or discrete variables.

Multiplying each coefficient by an appropriately scaled and shifted wavelet yields the

Vector distance and correlation coefficient

As seen in the previous tables, two simple pattern classification methods can be used to make diagnostic decisions: (1) the Euclidean vector distance, and (2) the vector correlation coefficient method.

Discussions and conclusions

A new scheme has been developed for the diagnosis of defects in ball bearings. The technique is based on statistical analysis, the discrete wavelet transform, and pattern classification techniques such as neuro-fuzzy inference. By using vibration data collected from an AC motor driven system with different faulted bearings installed, this diagnostic strategy was evaluated. The signals were normalised to (0,1) standard random variables, and then the wavelet transforms were performed using the

Acknowledgements

This work was supported in part by the Office of Naval Research under agreement N00014-98-3-0012 and the National Science Foundation, Grant ECS-9906218.

References (25)

D.–M. Yang et al.
Third-order spectral techniques for the diagnosis of motor bearing condition using artificial neural network
Mechanical Systems and Signal Processing
(2002)
B. Liu et al.
On the selection of informative wavelet for machinery diagnosis
Mechanical Systems and Signal Processing
(1999)
J. Altmann et al.
Multiple band-pass autoregressive demodulation for rolling-element bearing fault diagnosis
Mechanical Systems and Signal Processing
(2001)
R.R. Schoen et al.
Motor bearing damage detection using stator current monitoring
IEEE Transactions on Industrial Applications
(1995)
T.B. Brotherton, T. Pollard, Applications of time-frequency and time-scale representations to fault detection and...
J.A. Leonard, M.A. Kramer, Radial basis function networks for classifying process faults, IEEE Control Systems Magazine...
M. Chow et al.
On the application and design of artificial neural network for motor fault detection—II
IEEE Transactions on Industrial Electronics
(1993)
L.P. Heck, K.C. Chou, Gaussian mixture model classifier for machine monitoring, Proceedings of the IEEE world Congress...
E. Meyer, T. Tuthill, Bayesian classification of ultrasound signals using wavelet coefficients, Proceedings of the IEEE...
K.W. Baugh, On Parametrically Phase-Coupled Random Harmonic Processes, Proceedings of the IEEE Signal Processing...

H.C. Choe et al.

Novel Identification of Intercepted Signals from Unknown Radio Transmitters

Proceedings of the SPIE Wavelet Applications

(1995)

T. Peng, W. Gui, M. Wu, Y. Xie, A fusion diagnosis approach to bearing faults, Proceedings of the International...

Cited by (615)

Diesel engine fault diagnosis for multiple industrial scenarios based on transfer learning
2024, Measurement: Journal of the International Measurement Confederation
Fault diagnosis based on data-driven intelligence has recently attracted extensive interest owing to the rapid development of big data and deep-learning algorithms. However, when the amount of faulty data is limited, deep learning training is prone to overfitting. When the application scenario is changed, the generalization ability of the trained network is affected. In this study, a fault diagnosis architecture based on deep transfer learning is proposed to work with limited data and transfer between multiple scenarios. A wide convolution kernel convolutional long short-term memory neural network (WCL) was used to improve the feature extraction ability of fault data from a diesel engine with a low signal-to-noise ratio. A multiple transfer learning scheme based on WCL was further adopted to transfer the well-trained diagnostic knowledge of large-scale labeled source domain data to the target domain with limited samples. In addition, for diesel engines for various purposes, the knowledge transferability between different scenarios was studied. The algorithm evaluates the transfer performance of four different domains when the sample is insufficient, including the cross-fault type, cross-equipment type, cross-fault degree, and cross-working conditions. The results show the proposed method is proven with high noise immunity improves the accuracy of small sample cross-domain diagnosis and provides an optimal transfer scheme suitable for diesel engine fault signals.
Zero-shot motor health monitoring by blind domain transition
2024, Mechanical Systems and Signal Processing
Continuous long-term monitoring of motor health is crucial for the early detection of abnormalities such as bearing faults (up to 51% of motor failures are attributed to bearing faults). Despite numerous methodologies proposed for bearing fault detection, most of them require normal (healthy) and abnormal (faulty) data for training. Even with the recent deep learning (DL) methodologies trained on the labeled data from the same machine, the classification accuracy significantly deteriorates when one or few conditions are altered, e.g., a different speed or load, or for different fault types/severities with sensors placed in different locations. Furthermore, their performance suffers significantly or may entirely fail when they are tested on another machine with entirely different healthy and faulty signal patterns. To address this need, in this pilot study, we propose a zero-shot bearing fault detection method that can detect any fault on a new (target) machine regardless of the working conditions, sensor parameters, or fault characteristics. To accomplish this objective, a 1D Operational Generative Adversarial Network (Op-GAN) first characterizes the transition between normal and fault vibration signals of (a) source machine(s) under various conditions, sensor parameters, and fault types. Then for a target machine, the potential faulty signals can be generated, and over its actual healthy and synthesized faulty signals, a compact, and lightweight 1D Self-ONN fault detector can then be trained to detect the real faulty condition in real time whenever it occurs. To validate the proposed approach, a new benchmark dataset is created using two different motors working under different conditions and sensor locations. Experimental results demonstrate that this novel approach can accurately detect any bearing fault achieving an average recall rate of around 89% and 95% on two target machines regardless of its type, severity, and location.
Operational status monitoring of smart grid and power communication network coupling and collaboration based on multi-head attention mechanism
2024, Electric Power Systems Research
Smart grid is an organic combination of traditional power grid and advanced digital communication technology, and power communication network plays a vital role in ensuring its steady operation. However, as the coupling correlation between the smart grid and the power communication network increases, the propagation of coupled network failures may constitute a great threat to the safe and stable operation of the power system. To address this phenomenon, based on the algorithm of the multi-head attention mechanism (MHA) and convolutional neural network (CNN), this paper proposes a framework that integrates the multi-head attention mechanism with the convolutional neural network (MHA-CNN) for the monitoring of the coupling network’s operation status. In the proposed algorithm, Hilbert–Huang transform (HHT) is used to extract the intrinsic features of the operational data, and MHA employs multi-subspace analysis and extracts the key feature information of the fault data to maximize the fault information extraction and help the system to discover the fault as soon as possible. Simulation results demonstrate that the proposed algorithm can effectively realize a better detection speed and higher detection probability compared with CNN and ResNet-50, and maintain certain robustness under poor transmission conditions, ensuring the steady operation of the coupled network.
An evolutionary ensemble convolutional neural network for fault diagnosis problem
2023, Expert Systems with Applications
Automatic fault diagnosis in many systems is performed via the analysis of vibration data in the time or frequency domain. In the literature, many approaches for extracting featured from signals for fault diagnosis have been proposed. In this paper, we apply a variety of transform functions including Fourier, Wavelet, etc. to extract features from the vibration data, which includes the statistical features and some features automatically extracted via Convolutional Neural Networks (CNNs). For each of these feature extraction approaches, a learning algorithm is trained to diagnose the faults and their results are aggregated in an ensemble machine learning algorithm. The weights of the base learner algorithms are optimized via an evolutionary algorithm to achieve the best-weighted voting scheme. The architecture of CNNs has a significant effect on the performance of the algorithm, thus, in this paper, an evolutionary algorithm is proposed to find the best architecture for CNNs in fault diagnosis. The CNNs are trained via gradient descent algorithms which suffer from getting stuck in local optima. To manage this, we propose an evolutionary algorithm that benefits from the speed of gradient descent and the global search of evolutionary algorithms. The proposed algorithm is tested on a number of benchmark problems and the experimental results are presented.
Review on prognostics and health management in smart factory: From conventional to deep learning perspectives
2023, Engineering Applications of Artificial Intelligence
At present, the fourth industrial revolution is pushing factories toward an intelligent, interconnected grid of machinery, communication systems, and computational resources. Smart factories (SF) and smart manufacturing (SM) incorporate a cyber-physical system that employs advanced technologies such as artificial intelligence (AI) for data analysis, automated process driving, and continuous data handling. Smart factories operate by combining machines, humans, and massive amounts of data into a single, digitally interconnected ecosystem. Prognostics and health management (PHM) has become a critical requirement of smart factories to meet production needs. PHM of components/machines in the smart factory is crucial for securing uninterrupted operation and ensuring safety standards. The growing availability of computational capacity has increased the use of deep learning in PHM strategies. Deep learning supports comprehensive PHM solutions, thus reducing the need for manual feature development. This review presents an extensive study of the PHM strategies employed in the smart factory ranging from the conventional perspective to the deep learning perspective. This includes consideration of the conventional methodologies used for health management along with latest trends in the PHM domain in the smart factory.
An improved GNN using dynamic graph embedding mechanism: A novel end-to-end framework for rolling bearing fault diagnosis under variable working conditions
2023, Mechanical Systems and Signal Processing
Traditional deep learning (DL)-based rolling bearing fault diagnosis methods usually use signals collected under specific working condition to train the diagnosis models. This may lead to the lack of domain adaptive ability of these trained models, thus making it difficult to obtain satisfactory diagnosis accuracy when working conditions fluctuate. To address it, a novel fault diagnosis framework based on the graph neural network (GNN) and dynamic graph embedding mechanism (DGE) was proposed in this paper. Firstly, convolutional neural network (CNN) is used to extract the hidden fault features from raw bearing vibration signals. Secondly, DGE module is designed with edge dropout mechanism to transform the features exacted by CNN into higher-level graph-structured features dynamically. Then, GNN is applied to further mine the fault features sensitivity to the fluctuating bearing working conditions. Finally, a novel mechanism named node voters is proposed to replace traditional graph-level attribute update function in GNN to obtain optimal fault pattern recognition results. Experiment results shows that the proposed framework can not only realize the end-to-end fault diagnosis of rolling bearings, but also has excellent domain adaptive ability to obtain better stability and diagnosis accuracy under variable working conditions compared to traditional DL-based methods.

View all citing articles on Scopus

¹: Current affiliation: E&ES Department, the Alstom Power Plant Laboratories, Windsor, CT 06095, USA.

View full text

Bearing fault diagnosis based on wavelet transform and fuzzy inference

Abstract

Introduction

Section snippets

Experimental system

Preprocessing test data

Brief review of the wavelet theory

Vector distance and correlation coefficient

Discussions and conclusions

Acknowledgements

Mechanical Systems and Signal Processing

Mechanical Systems and Signal Processing

Mechanical Systems and Signal Processing

Motor bearing damage detection using stator current monitoring

IEEE Transactions on Industrial Applications

On the application and design of artificial neural network for motor fault detection—II

IEEE Transactions on Industrial Electronics

Novel Identification of Intercepted Signals from Unknown Radio Transmitters

Proceedings of the SPIE Wavelet Applications