Bing Li

A new feature extraction and selection scheme for hybrid fault diagnosis of gearbox

A novel feature extraction and selection scheme was proposed for hybrid fault diagnosis of gearbox based on S transform, non-negative matrix factorization (NMF), mutual information and multi-objective evolutionary algorithms. Time-frequency distributions of vibration signals, acquired from gearbox with different fault states, were obtained by S transform. Then non-negative matrix factorization (NMF) was employed to extract features from the time-frequency representations. Furthermore, a two stage feature selection approach combining filter and wrapper techniques based on mutual information and non-dominated sorting genetic algorithms II (NSGA-II) was presented to get a more compact feature subset for accurate classification of hybrid faults of gearbox. Eight fault states, including gear defects, bearing defects and combination of gear and bearing defects, were simulated on a single-stage gearbox to evaluated the proposed feature extraction and selection scheme. Four different classifiers were employed to incorporate with the presented techniques for classification. Performances of four classifiers with different feature subsets were compared. Results of the experiments have revealed that the proposed feature extraction and selection scheme demonstrate to be an effective and efficient tool for hybrid fault diagnosis of gearbox.

A weighted multi-scale morphological gradient filter for rolling element bearing fault detection

This paper presents a novel signal processing scheme, named the weighted multi-scale morphological gradient filter (WMMG), for rolling element bearing fault detection. The WMMG can depress the noise at large scale and preserve the impulsive shape details at small scale. Both a simulated signal and vibration signals from a bearing test rig are employed to evaluate the performance of the proposed technique. The traditional envelope analysis and a multi-scale enveloping spectrogram algorithm combining continuous wavelet transform and envelope analysis (WT-EA) are also studied and compared with the presented WMMG. Experimental results have demonstrated the effectiveness of the WMMG to extract the impulsive components from the raw vibration signal with strong background noise. We also investigated the classification performance on identifying bearing faults based on the WMMG and statistical parameters with varied noise levels. Application results reveal that the WMMG achieves the same or better performance as EA and WT-EA. Meanwhile, the WMMG requires low computation cost and is very suitable for on-line condition monitoring of bearing operating states.

Fuzzy lattice classifier and its application to bearing fault diagnosis

In this work, we present a novel classification scheme named fuzzy lattice classifier (FLC) based on the lattice framework and apply it to the bearing faults diagnosis problem. Different from the fuzzy lattice reasoning (FLR) model developed in literature, there is no need to tune any parameter and to compute the inclusion measure in the training procedure in our new FLC model. It can converge rapidly in a single pass through training patterns with a few induced rules. A series of experiments are conducted on five popular benchmark datasets and three bearing datasets to evaluate and compare the presented FLC with the FLR model as well as some other widely used classification methods. Experimental results indicate that the FLC yields a satisfactory classification performance with higher computation efficiency than other classifiers. It is very desirable to utilize the FLC scheme for on-line condition monitoring of bearings and other mechanical systems.

Classification of time-frequency representations based on two-direction 2DLDA for gear fault diagnosis

Time-frequency representations (TFR) have been intensively employed for analyzing vibration signals in mechanical faults diagnosis. However, in many applications, time-frequency representations are simply utilized as a visual aid to be used for vibration signal analysis. It is very attractive to investigate the utility of TFR for automatic classification of vibration signals. A key step for this work is to extract discriminative parameters from TFR as input feature vector for classifiers. This paper contributes to this ongoing investigation by developing a two direction two dimensional linear discriminative analysis (TD-2DLDA) technique for feature extraction from TFR. The S transform, which combines the separate strengths of the short time Fourier transform and wavelet transforms, is chosen to perform the time-frequency analysis of vibration signals. Then, a novel feature extraction technique, named TD-2DLDA, is proposed to represent the time-frequency matrix. As opposed to traditional LDA, TD-2DLDA is directly conduct on 2D matrices rather than 1D vectors, so the time-frequency matrix does not need to be transformed into a vector prior to feature extraction. Therefore, the TD-2DLDA can reduce the computation cost and preserve more structure information hiding in original 2D matrices compared to the LDA. The promise of our method is illustrated by performing our procedure on vibration signals measured from a gearbox with five operating states. Experimental results indicate that the TD-2DLDA obviously outperforms related feature extraction schemes such as LDA, 2DLDA in gear fault diagnosis.

Bearing fault detection using multi-scale fractal dimensions based on morphological covers

Vibration signals acquired from bearing have been found to demonstrate complicated nonlinear characteristics in literature. Fractal geometry theory has provided effective tools such as fractal dimension for characterizing the vibration signals in bearing faults detection. However, most of the natural signals are not critical self-similar fractals; the assumption of a constant fractal dimension at all scales may not be true. Motivated by this fact, this work explores the application of the multi-scale fractal dimensions (MFDs) based on morphological cover (MC) technique for bearing fault diagnosis. Vibration signals from bearing with seven different states under four operations conditions are collected to validate the presented MFDs based on MC technique. Experimental results reveal that the vibration signals acquired from bearing are not critical self-similar fractals. The MFDs can provide more discriminative information about the signals than the single global fractal dimension. Furthermore, three classifiers are employed to evaluate and compare the classification performance of the MFDs with other feature extraction methods. Experimental results demonstrate the MFDs to be a desirable approach to improve the performance of bearing fault diagnosis.

Gear fault detection using adaptive morphological gradient lifting wavelet

Wavelet transform is one of the most acceptable tools to analyze vibration signals for gear fault detection. However, there are still some limitations of the traditional wavelet transforms due to the utilization of fixed linear filters. This investigation presents an adaptive morphological gradient lifting wavelet (AMGLW) to remedy the shortcomings of traditional wavelet transform schemes. A novel nonlinear filter, named morphological gradient filter, is designed for enhancing the impulsive features of the original signal. Then the adaptability of AMGLW is implemented by selecting between two filters, namely the average filter and the morphological gradient filter, to update the approximation signal dependent upon the local gradient of the analyzed signal. This new scheme is evaluated on a simulated signal and a practical vibration signal measured from a gearbox. Experimental results demonstrate that the presented AMGLW outperforms the traditional linear wavelet (LW) transform obviously for detecting gear defects. Furthermore, the computational cost of AMGLW is much less than the traditional LW. Thus the AMGLW scheme is quite suitable for the online condition monitoring of gears.

Feature extraction and selection for fault diagnosis of gear using wavelet entropy and mutual information

This paper aims to develop an complete system including signal processing, feature extraction, feature selection and classification approaches for fault diagnosis of gear by using the wavelet transform, the entropy, the mutual information and the least-square support vector machine (LS-SVM). Firstly, the vibration signals are decomposed to several wavelet coefficients. The energy of every coefficient and the singularity values (SV) of the coefficient matrix are extracted. Two type entropies means the Shannon entropy and Renyi entropy are calculated of the energy and SV distribution. Secondly, a maximum relevance and minimum redundant (mRMR) method based on the mutual information and the greedy search technique are employed to select the optimal feature subsets for gear fault classification. A cross-validation method based on the LS-SVM is proposed to determine the number of features that the optimal subset contained. Application to practical gear fault diagnosis showed that the proposed techniques provide a more effective and fast approach to gear fault diagnosis.

Engine fault diagnosis based on a morphological neural network using a morphological filter as a preprocessor

Feature extraction and faults classification are the two most significant issues involved in the field of mechanical fault diagnosis problems. In this work, we address these two problems using mathematical morphology and non-negative matrix factorization. In particular, we present a novel engine fault diagnosis scheme utilizing the averaged multi-scale morphological filter to enhance the vibration signals, non-negative matrix factorization to characterize the signals, and a constructive morphological neural network to classify the engine operating states. Eight engine running states including the healthy state and seven defective states are tested in an engine experiment rig to evaluate the presented fault diagnosis scheme. Conventional feature extraction methods as well as classifiers popularly used in the literature are also employed as a comparison. The experimental results indicate the proposed approach to be an effective and efficient scheme for detection of the intelligent faults of engines.

A Two Stage Feature Selection Method for Gear Fault Diagnosis Using ReliefF and GA-Wrapper

This paper presents a novel two stage feature selection method for gear fault diagnosis based on ReliefF and genetic algorithm. Prior to the feature selection, 114 parameters were extracted as the original feature set based on EMD, AR model, statistical methods and entropy. Then the ReliefF was employed to evaluate the quality of every individual feature and a sequential feature sets were obtained according to the marks evaluated by ReliefF. Then the cross validation technique was used to get the candidate feature set from the sequential sets. At the second stage, the genetic algorithm was utilized to search a more compact feature set based on the candidate set. Three different classifiers means the LDC (linear discriminant classifier), KNNC (k nearest neighbors classifier) and NBC (naïve bayes classifier) were employed to evaluate the proposed method. The application results to the real gear fault diagnosis have shown that the proposed method can obtain a higher performance with a small size feature set.

Denoising of Mechanical Vibration Signals Using Quantum-Inspired Adaptive Wavelet Shrinkage

The potential application of a quantum-inspired adaptive wavelet shrinkage (QAWS) technique to mechanical vibration signals with a focus on noise reduction is studied in this paper. This quantum-inspired shrinkage algorithm combines three elements: an adaptive non-Gaussian statistical model of dual-tree complex wavelet transform (DTCWT) coefficients proposed to improve practicability of prior information, the quantum superposition introduced to describe the interscale dependencies of DTCWT coefficients, and the quantum-inspired probability of noise defined to shrink wavelet coefficients in a Bayesian framework. By combining all these elements, this signal processing scheme incorporating the DTCWT with quantum theory can both reduce noise and preserve signal details. A practical vibration signal measured from a power-shift steering transmission is utilized to evaluate the denoising ability of QAWS. Application results demonstrate the effectiveness of the proposed method. Moreover, it achieves better performance than hard and soft thresholding.