Md. Rashedul Islam

A Hybrid Feature Selection Scheme for Reducing Diagnostic Performance Deterioration Caused by Outliers in Data-Driven Diagnostics

In practice, outliers, defined as data points that are distant from the other agglomerated data points in the same class, can seriously degrade diagnostic performance. To reduce diagnostic performance deterioration caused by outliers in data-driven diagnostics, an outlier-insensitive hybrid feature selection (OIHFS) methodology is developed to assess feature subset quality. In addition, a new feature evaluation metric is created as the ratio of the intraclass compactness to the interclass separability estimated by understanding the relationship between data points and outliers. The efficacy of the developed methodology is verified with a fault diagnosis application by identifying defect-free and defective rolling element bearings under various conditions.

Reliable Fault Diagnosis of Bearings Using Distance and Density Similarity on an Enhanced k-NN

The k-nearest neighbor (k-NN) method is a simple and highly effective classifier, but the classification accuracy of k-NN is degraded and becomes highly sensitive to the neighborhood size k in multi-classification problems, where the density of data samples varies across different classes. This is mainly due to the method using only a distance-based measure of similarity between different samples. In this paper, we propose a density-weighted distance similarity metric, which considers the relative densities of samples in addition to the distances between samples to improve the classification accuracy of standard k-NN. The performance of the proposed k-NN approach is not affected by the neighborhood size k. Experimental results show that the proposed approach yields better classification accuracy than traditional k-NN for fault diagnosis of rolling element bearings.

An effective approach to improving low-cost GPS positioning accuracy in real-time navigation.

Positioning accuracy is a challenging issue for location-based applications using a low-cost global positioning system (GPS). This paper presents an effective approach to improving the positioning accuracy of a low-cost GPS receiver for real-time navigation. The proposed method precisely estimates position by combining vehicle movement direction, velocity averaging, and distance between waypoints using coordinate data (latitude, longitude, time, and velocity) of the GPS receiver. The previously estimated precious reference point, coordinate translation, and invalid data check also improve accuracy. In order to evaluate the performance of the proposed method, we conducted an experiment using a GARMIN GPS 19xHVS receiver attached to a car and used Google Maps to plot the processed data. The proposed method achieved improvement of 4–10 meters in several experiments. In addition, we compared the proposed approach with two other state-of-the-art methods: recursive averaging and ARMA interpolation. The experimental results show that the proposed approach outperforms other state-of-the-art methods in terms of positioning accuracy.

Reduction of Overfitting in Diabetes Prediction Using Deep Learning Neural Network

Accurate prediction of diabetes is an important issue in health prognostics. However, data overfitting degrades the prediction accuracy in diabetes prognosis. In this paper, a reliable prediction system for the disease of diabetes is presented using a dropout method to address the overfitting issue. In the proposed method, deep learning neural network is employed where fully connected layers are followed by dropout layers. The proposed neural network outperforms other state-of-art methods in better prediction scores for the Pima Indians Diabetes Data Set.

Maximum Class Separability-Based Discriminant Feature Selection Using a GA for Reliable Fault Diagnosis of Induction Motors

Reliable fault diagnosis in bearing elements of induction motors, with high classification performance, is of paramount importance for ensuring steady manufacturing. The performance of any fault diagnosis system largely depends on the selection of a feature vector that represents the most distinctive fault attributes. This paper proposes a maximum class separability (MCS) feature distribution analysis-based feature selection method using a genetic algorithm (GA). The MCS distribution analysis model analyzes and selects an optimal feature vector, which consists of the most distinguishing features from a high dimensional feature space, for reliable multi-fault diagnosis in bearings. The high dimensional feature space is an ensemble of hybrid statistical features calculated from time domain analysis, frequency domain analysis, and envelope spectrum analysis of the acoustic emission (AE) signal. The proposed maximum class separability-based objective function using the GA is used to select the optimal feature set. Finally, k-nearest neighbor (k-NN) algorithm is used to validate our proposed approach in terms of the classification performance. The experimental results validate the superior performance of our proposed model for different datasets under different motor rotational speeds as compared to conventional models that utilize (1) the original feature vector and (2) a state-of-the-art average distance-based feature selection method.

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.

Distance and Density Similarity Based Enhanced -NN Classifier for Improving Fault Diagnosis Performance of Bearings

An enhanced -nearest neighbor (-NN) classification algorithm is presented, which uses a density based similarity measure in addition to a distance based similarity measure to improve the diagnostic performance in bearing fault diagnosis. Due to its use of distance based similarity measure alone, the classification accuracy of traditional -NN deteriorates in case of overlapping samples and outliers and is highly susceptible to the neighborhood size, . This study addresses these limitations by proposing the use of both distance and density based measures of similarity between training and test samples. The proposed -NN classifier is used to enhance the diagnostic performance of a bearing fault diagnosis scheme, which classifies different fault conditions based upon hybrid feature vectors extracted from acoustic emission (AE) signals. Experimental results demonstrate that the proposed scheme, which uses the enhanced -NN classifier, yields better diagnostic performance and is more robust to variations in the neighborhood size, .

A Centroid-GPS Model to Improving Positioning Accuracy for a Sensitive Location-Based System

This paper proposes a centroid global positioning system (GPS) model to improve the positioning accuracy of low-cost GPS receivers of a sensitive location-based system. The proposed model estimates the precise movement position by a centroid sum of the individual improved positions of three GPS receivers. Each GPS receiver’s position is improved by using a direction and velocity averaging technique based on combining the vehicle movement direction, velocity averaging, and distance between the waypoints of each GPS receiver using coordinate data (latitude, longitude, time, and velocity). Finally, the precise position is estimated by calculating a triangular centroid sum with distance threshold of the improved positions of three GPS receivers. In order to evaluate the performance of the proposed approach, we used three GARMIN GPS 19x HVS receivers attached to a car and plotted the processed data in Google map. The proposed approach resulted in an improved accuracy of about 2–12 m compared to the original GPS receivers. In addition, we compared the proposed approach to two other state-of-the-art methods. The experimental results show that the proposed approach outperforms the conventional methods in terms of positioning accuracy.

A Novel Transfer Learning Approach upon Hindi, Arabic, and Bangla Numerals Using Convolutional Neural Networks

Increased accuracy in predictive models for handwritten character recognition will open up new frontiers for optical character recognition. Major drawbacks of predictive machine learning models are headed by the elongated training time taken by some models, and the requirement that training and test data be in the same feature space and consist of the same distribution. In this study, these obstacles are minimized by presenting a model for transferring knowledge from one task to another. This model is presented for the recognition of handwritten numerals in Indic languages. The model utilizes convolutional neural networks with backpropagation for error reduction and dropout for data overfitting. The output performance of the proposed neural network is shown to have closely matched other state-of-the-art methods using only a fraction of time used by the state-of-the-arts.

Efficient bearing fault diagnosis by extracting intrinsic fault information using envelope power spectrum

Early and efficient fault diagnosis of bearing of industrial motor is a modern demand for reducing unexpected breakdown of industrial process. Extracting the intrinsic fault signature in very early stage is important. In this point of view, this paper proposes a fault diagnosis model of industrial bearing including efficient fault signature extraction technique based on narrow band frequency domain analysis of acoustic emission (AE) signal using envelope power spectrum. To do that, AE signals are collected from defective and non-defective bearings under different rotational speeds from industrial-like experimental environment. Envelope power spectrum is calculated from the AE signal and narrow band root mean square (NBRMS) fault features are extracted from defect frequency ranges of the envelope power spectrum. Finally, the k-nearest neighbor (k-NN) classification algorithm is used for identifying the fault of unknown signal and validating the efficiency of the proposed feature extraction model. The experimental result shows that the proposed model outperforms state-of-art algorithms in terms of classification accuracy.