M. M. Manjurul Islam

Reliable bearing fault diagnosis using Bayesian inference-based multi-class support vector machines

This letter presents a multi-fault diagnosis scheme for bearings using hybrid features extracted from their acoustic emissions and a Bayesian inference-based one-against-all support vector machine (Bayesian OAASVM) for multi-class classification. The Bayesian OAASVM, which is a standard multi-class extension of the binary support vector machine, results in ambiguously labeled regions in the input space that degrade its classification performance. The proposed Bayesian OAASVM formulates the feature space as an appropriate Gaussian process prior, interprets the decision value of the Bayesian OAASVM as a maximum a posteriori evidence function, and uses Bayesian inference to label unknown samples.

Multi-fault Diagnosis of Roller Bearings Using Support Vector Machines with an Improved Decision Strategy

This paper proposes an efficient fault diagnosis methodology based on an improved one-against-all multiclass support vector machine (OAA-MCSVM) for diagnosing faults inherent in rotating machinery. The methodology employs time and frequency domain techniques to extract features of diverse bearing defects. In addition, the proposed method introduces a new reliability measure (SVMReM) for individual SVMs in the multiclass framework. The SVMReM achieves optimum results irrespective of the test sample location by using a new decision strategy for the proposed OAA-MCSVM based method. Finally, each SVM is trained with optimized kernel parameters using a grid search technique to enhance the classification accuracy of the proposed method. Experimental results show that the proposed method is superior to conventional approaches, yielding an average classification accuracy of 97 % for five different rotational speed conditions, eight different fault types and two different crack sizes.

Reliable multiple combined fault diagnosis of bearings using heterogeneous feature models and multiclass support vector Machines

Abstract This paper proposes a reliable multiple combined fault diagnosis scheme for bearings using heterogeneous feature models and an improved one-against-all multiclass support vector machines (OAA-MCSVM) classifier. Distinct feature extraction methods are simultaneously applied to an acoustic emission (AE) signal to extract unique fault features for diagnosing bearing defects. These fault features are composed of time domain, frequency domain statistical parameters, and complex envelope spectrum analysis. Generally, a high-dimensional feature vector is used to train the standard OAA-MCSVM classifier for diagnosis and identification of bearing defects. However, this classification method ignores individual classifier competence when results from multiple classes are agglomerated for the final decision, and therefore, yields undecided and overlapped feature spaces where classification accuracy is severely degraded. To solve this unreliability problem, this paper introduces a dynamic reliability measure (DReM) technique for individual support vector machines (SVMs) in the one-against-all (OAA) framework. This DReM accounts for the spatial variation of the classifiers performance by finding the local neighborhood of a test sample in the training samples space and defining a new decision function for the OAA-MCSVM. The efficacy of the proposed OAA-MCSVM classifier with DReM is tested for identifying single and multiple combined faults in low-speed bearings. The experimental results demonstrate that the proposed classifier technique is superior to three state-of-the-art algorithms, yielding 6.19–16.59% improvement in the average classification performance.

A Hybrid Feature Selection Scheme Based on Local Compactness and Global Separability for Improving Roller Bearing Diagnostic Performance

This paper proposes a hybrid feature selection scheme for identifying the most discriminant fault signatures using an improved class separability criteria—the local compactness and global separability (LCGS)—of distribution in feature dimension to diagnose bearing faults. The hybrid model consists of filter based selection and wrapper based selection. In the filter phase, a sequential forward floating selection (SFFS) algorithm is employed to yield a series of suboptimal feature subset candidates using LCGS based feature subset evaluation metric. In the wrapper phase, the most discriminant feature subset is then selected from suboptimal feature subsets based on maximum average classification accuracy estimation of support vector machine (SVM) classifier using them. The effectiveness of the proposed hybrid feature selection method is verified with fault diagnosis application for low speed rolling element bearings under various conditions. Experimental results indicate that the proposed method outperforms the state-of-the-art algorithm when selecting the most discriminate fault feature subset, yielding 1.4% to 17.74% diagnostic performance improvement in average classification accuracy.

Input-Output Fault Diagnosis in Robot Manipulator Using Fuzzy LMI-Tuned PI Feedback Linearization Observer Based on Nonlinear Intelligent ARX Model

This paper proposes a model-based fault detection and diagnosis (FDD) technique for six degrees of freedom PUMA robot manipulator in presence of noise in actuator and sensor faults. The inverse modeling based on an adaptive method, which combines the fuzzy C-means clustering with the modified autoregressive eXternal (ARX) model, is presented for the system identification. The proposed adaptive nonlinear ARX fuzzy C-means (NARXNF) clustering technique obtains an improved convergence and error reduction than that of the traditional fuzzy C-means clustering algorithm. In addition, proportional integral (PI) feedback linearization observation is used for diagnosing the fault, where the convergence, robustness, and stability are validated by fuzzy linear matrix inequality (FLMI). Experimental results, in presence of 40% noise, show that the rate of root mean square (RMS) error for end-effector position is less than 0.00624. The proposed method also improves the rate of sensors and actuators FDD without additional hardware.

Motor Bearing Fault Diagnosis Using Deep Convolutional Neural Networks with 2D Analysis of Vibration Signal

Bearings are critical components in rotating machinery, and it is crucial to diagnose their faults at an early stage. Existing fault diagnosis methods are mostly limited to manual features and traditional artificial intelligence learning schemes such as neural network, support vector machine, and k-nearest-neighborhood. Unfortunately, interpretation and engineering of such features require substantial human expertise. This paper proposes an adaptive deep convolutional neural network (ADCNN) that utilizes cyclic spectrum maps (CSM) of raw vibration signal as bearing health states to automate feature extraction and classification process. The CSMs are two-dimensional (2D) maps that show the distribution of cycle energy across different bands of the vibration spectrum. The efficiency of the proposed algorithm (CSM+ADCNN) is validated using benchmark dataset collected from bearing tests. Experimental results indicate that the proposed method outperforms the state-of-the-art algorithms, yielding 8.25% to 13.75% classification performance improvement.

Rub-Impact Fault Diagnosis Using an Effective IMF Selection Technique in Ensemble Empirical Mode Decomposition and Hybrid Feature Models

The complex nature of rubbing faults makes it difficult to use traditional signal analysis methods for feature extraction. Various time-frequency analysis approaches based on signal decomposition, such as empirical mode decomposition (EMD) and ensemble EMD (EEMD), have been widely utilized recently to analyze rub-impact faults. However, traditional EMD suffers from “mode-mixing”, and in both EMD and EEMD the relevance of the extracted components to rubbing processes must be determined. In this paper, we introduce a new informative intrinsic mode function (IMF) selection method for EEMD and a hybrid feature model for diagnosing rub-impact faults of various intensities. Our method uses a novel selection procedure that combines the degree-of-presence ratio of rub impact and a Kullback–Leibler divergence-based similarity measure into an IMF quality metric with adaptive threshold-based selection to pick the meaningful signal-dominant modes. Signals reconstructed using the selected IMFs contained explicit information about the rubbing faults and are used for hybrid feature extraction. Experimental results demonstrated that the proposed approach effectively defines meaningful IMFs for rubbing processes, and the presented hybrid feature model allows for the classification of rub-impact faults of various intensities with good accuracy.

Feature selection techniques for increasing reliability of fault diagnosis of bearings

Features selection (FS) techniques have an apparent need in many complex engineering applications especially the bearing fault diagnosis of low-speed industrial motor. The main goal of an FS algorithm is to select the most discriminant features subset from a high-dimension features vector that increases the model performance by reducing the redundant and irrelevant fault features. This paper proposes an efficient fault diagnosis model of bearing by incorporating the optimal feature selection approach for increasing the reliability of fault diagnosis of bearing. Also, this paper investigates the feature selection approaches including sequential forward selection (SFS), sequential floating forward selection (SFFS), and genetic algorithm (GA) for identifying the most discriminant subset. The effectiveness of this discriminant features subset is verified with a low-speed bearing fault diagnosis application for identifying bearing failures. The experimental shows up-to-mark diagnosis performance using GA based optimal feature selection method.