Mahalanobis Distance Criterion Research Articles

A Robust Parameter Identification Approach with Anti-Outlier Characteristics for Lithium-Ion Batteries

Accurate acquisition of model parameters of lithium-ion batteries (LIBs) is imperative for precious estimation state of charge. However, due to the interference of noise, the inevitable uncertainty may lead to outliers appearing in measurement data, which often leads to inaccurate parameter identification. To solve the problem, a robust Kalman filter (RKF) that uses the Mahalanobis Distance Criterion (MDC) is proposed for the parameter identification of LIBs. Firstly, the one-order equivalent circuit model of the LIBs is created. Subsequently, the gain of robust factor is implemented into the Kalman filter to decrease filter gain and swell the measurement noise covariance when facing the outlier measurements. Finally, the dynamic stress test condition of the electric vehicle is used to prove the effectiveness of the RKF-MDC approach. The simulation reveals that the RKF-MDC can reduce the influence of outlier values for the parameter identification of LIBs.

Demarcating Z-line and Gastric Folds Boundary Based on the Segmentation of the Lower Esophageal Sphincter Images.

The endoscopic diagnosis of pathological changes in the gastroesophageal junction including esophagitis and Barrett's mucosa is based on the visual detection of two boundaries: mucosal color change between esophagus and stomach, and top endpoint of gastric folds. The presence and pattern of mucosal breaks in the gastroesophageal mucosal junction (Z line) classify esophagitis in patients and the distance between the two boundaries points to the possible columnar lined epithelium. Since visual detection may suffer from intra- and interobserver variability, our objective was to define the boundaries automatically based on image processing algorithms, which may enable us to measure the detentions of changes in future studies. To demarcate the Z-line, first the artifacts of endoscopy images are eliminated. In the second step, using SUSAN edge detector, Mahalanobis distance criteria, and Gabor filter bank, an initial contour is estimated for the Z-line. Using region-based active contours, this initial contour converges to the Z-line. Finally, by applying morphological operators and Gabor Filter Bank to the region inside of the Z-line, gastric folds are segmented. To evaluate the results, a database consisting of 50 images and their ground truths were collected. The average dice coefficient and mean square error of Z-line segmentation were 0.93 and 3.3, respectively. Furthermore, the average boundary distance criteria are 12.3 pixels. In addition, two other criteria that compare the segmentation of folds with several ground truths, i.e., Sweet-Spot Coverage and Jaccard Index for Golden Standard, are 0.90 and 0.84, respectively. Considering the results, automatic segmentation of Z-line and gastric folds are matched to the ground truths with appropriate accuracy.

An operating condition information-guided iterative variational mode decomposition method based on Mahalanobis distance criterion for surge characteristic frequency extraction of the centrifugal compressor

As a significant high-speed rotating machinery being widely used in modern industry, the centrifugal compressor is subject to a potentially damaging phenomenon called surge, which may lead to property damage, casualties and even catastrophic accidents. Therefore, it is necessary to identify the early surge state of centrifugal compressors to prevent such accidents. In this paper, an operating condition information-guided iterative variational mode decomposition (OCIGIVMD) analysis method is proposed to extract the surge characteristic frequencies (SCFs) from the vibratory signals of centrifugal compressors under early surge state. The variational mode decomposition (VMD) is an effective methodology to decompose a multi-component signal into a series of amplitude-modulated-frequency-modulated (AM-FM) sub-signals with limited bandwidth, from which the characteristic frequencies can be extracted. However, when implementing VMD, the optimal determination of the decomposition mode number is still a challenging task, especially in face of complex signals containing a lot of noise and unknown interference components. Coping with the complex vibration signals of the centrifugal compressor, the VMD in an iterative form based on operating condition information is developed to extract the SCFs, while the decomposition process is controlled by Mahalanobis distance criterion to avoid under-decomposition or over-decomposition and it is needless to select the sensitive intrinsic mode function (IMF) from the multiple IMFs generated by OCIGIVMD. Finally, comparisons with the conventional VMD, empirical mode decomposition (EMD), spectral kurtosis (SK), local mean decomposition (LMD) and two improved VMD methods are conducted using both simulation and experimental data, which demonstrates that the proposed OCIGIVMD is a superior approach in extracting SCFs of the centrifugal compressor under different operating conditions.

A Vibration Feature Extraction Method Based on Time-Domain Dimensional Parameters and Mahalanobis Distance

To accurately describe the characteristics of a signal, the feature parameters in time domain and frequency domain are usually extracted for characterization. However, the total number of feature parameters in time domain and frequency domain exceeds twenty, and all of the feature parameters are used for feature extraction, which will result in a large amount of data processing. For the purpose of using fewer feature parameters to accurately reflect the characteristics of the vibration signal, a simple but effective vibration feature extraction method combining time-domain dimensional parameters (TDDP) and Mahalanobis distance (MD) is proposed, i.e., TDDP-MD. In this method, ten time-domain dimensional parameters are selected to extract fault features, and the distance evaluation technique based on Mahalanobis distance criterion function is also introduced to calculate the feature vector, which can be used to classify different failure types. Finally, the proposed method is applied to fault diagnosis of rolling element bearings, and experimental analysis results show that the proposed method can recognize different failure types accurately and effectively with only ten time-domain dimensional parameters and a small quantity of training samples.

Improving Person Re-Identification Rate in Security Cameras by Orthogonal Moments and a Distance-based Criterion

Surveillance and security cameras help security forces in public places such as airports, railway stations, universities and office buildings to perform high-level surveillance tasks such as detecting suspicious activity or anticipating undesirable events. Re-Identification (Re-ID) is defined as the process of communicating between images of the person in different cameras in a surveillance environment. Changing the field of view of any camera presents challenges such as changing body posture, changing brightness, noise and blockage. This article focuses on extracting the most distinctive features to overcome these challenges. The features of Hu moment, Zernike moment in 9th order and Legendre moment in 9th order for each image are extracted and merged into a single feature vector to form a single feature vector for each image. Principal Component Analysis (PCA) was used to reduce the vector dimensionality and finally the Mahalanobis distance criterion was used for identification. The proposed method in the VIPeR database has achieved a re-ID rate of 96.5. Although the presented method is simple, the outcome has been superior compared to many of the state-of-the-art methods.

Phonocardiography-based mitral valve prolapse detection with using fractional fourier transform

Mitral Valve Prolapse (MVP) is a common condition among people, which is often benign and does not need any serious treatment. However, this doesn’t mean that MVP can’t cause any problems. In malignant conditions, MVP can cause mitral failure and also heart failure. Early diagnosis of MVP is significantly important to control and reduce its complications. Since the phonocardiogram signal provides useful information about heart valves function, it can be used for MVP detection. To detect MVP, the signal was denoised and segmented into heart cycles and constant three-second pieces in the first and second approaches, respectively. Next, based on the Fractional Fourier Transform (FrFT), the desired features were extracted. Then, the extracted features were windowed by a Moving Logarithmic Median Window (MLMW) and optimum features were selected using Mahalanobis, Bhattacharyya, Canberra, and Minkowski distance criteria. Finally, using the selected features, classification was performed by using the K-Nearest Neighbor (KNN) and the Suppor Vector Machine (SVM) classifiers to find out whether a segment is prolapsed. The best results of the experiment on the collected database contain 15 prolapsed and 6 non-prolapsed subjects using the A-test method show 96.25 ± 2.43 accuracy, 98.5 ± 3.37 sensitivity, 94.0 ± 5.16 specificity, 96.0 ± 3.44 precision, 92.5 ± 4.86 kappa, and 96.6 ± 2.34 f-score with the SVM classifier.

Thermal Image Based Fault Diagnosis of Gears using Support Vector Machines

Condition monitoring and fault diagnosis of working machines have gained significant attention due to their prospective benefits, such as enhanced productivity, decreased repair and maintenance costs and enhanced machine operation. In this paper, a thermal image based non-contact methodology has been proposed to diagnose the gear faults using support vector machines (SVM). The thermal images acquired from gearbox simulator were preprocessed using 2D-discrete wavelet transform to decompose the thermal images. The relevant features were extracted from converted thermal gray-scaled images followed by selecting the strongest feature using Mahalanobis distance criteria. Finally, the selected features were given to a SVM classifier for classifying the different gear faults. The experimental findings indicate that fault diagnosis using thermography for rotary machinery can be put into practice to industrial fields as a new smart fault diagnostic method with excellent prediction performance.

A Robust Cubature Kalman Filter with Abnormal Observations Identification Using the Mahalanobis Distance Criterion for Vehicular INS/GNSS Integration.

INS/GNSS (inertial navigation system/global navigation satellite system) integration is a promising solution of vehicle navigation for intelligent transportation systems. However, the observation of GNSS inevitably involves uncertainty due to the vulnerability to signal blockage in many urban/suburban areas, leading to the degraded navigation performance for INS/GNSS integration. This paper develops a novel robust CKF with scaling factor by combining the emerging cubature Kalman filter (CKF) with the concept of Mahalanobis distance criterion to address the above problem involved in nonlinear INS/GNSS integration. It establishes a theory of abnormal observations identification using the Mahalanobis distance criterion. Subsequently, a robust factor (scaling factor), which is calculated via the Mahalanobis distance criterion, is introduced into the standard CKF to inflate the observation noise covariance, resulting in a decreased filtering gain in the presence of abnormal observations. The proposed robust CKF can effectively resist the influence of abnormal observations on navigation solution and thus improves the robustness of CKF for vehicular INS/GNSS integration. Simulation and experimental results have demonstrated the effectiveness of the proposed robust CKF for vehicular navigation with INS/GNSS integration.

Support vector machines based non-contact fault diagnosis system for bearings

Bearing defects have been accepted as one of the major causes of failure in rotating machinery. It is important to identify and diagnose the failure behavior of bearings for the reliable operation of equipment. In this paper, a low-cost non-contact vibration sensor has been developed for detecting the faults in bearings. The supervised learning method, support vector machine (SVM), has been employed as a tool to validate the effectiveness of the developed sensor. Experimental vibration data collected for different bearing defects under various loading and running conditions have been analyzed to develop a system for diagnosing the faults for machine health monitoring. Fault diagnosis has been accomplished using discrete wavelet transform for denoising the signal. Mahalanobis distance criteria has been employed for selecting the strongest feature on the extracted relevant features. Finally, these selected features have been passed to the SVM classifier for identifying and classifying the various bearing defects. The results reveal that the vibration signatures obtained from developed non-contact sensor compare well with the accelerometer data obtained under the same conditions. A developed sensor is a promising tool for detecting the bearing damage and identifying its class. SVM results have established the effectiveness of the developed non-contact sensor as a vibration measuring instrument which makes the developed sensor a cost-effective tool for the condition monitoring of rotating machines.

Modelling Fault Detection Using SRGM in Agile Environment and Ranking of Models

Today's software applications deployed in an enterprise to cater to the complex business processes, integrate various business units and address requirements of a global customer base. The traditional methodology of software engineering succumbs to the changing need of customer and technology advancement. On the behest of the customer, a software system should be designed in a way that it goes in concert with the present user needs. Agile methodology targets complex systems with its iterative, incremental, and evolutionary approach. There are numerous factors attributing towards the successful implementation of agile methodology. This led to adopting an approach of agile based on ‘lean' principles over the traditional software development life cycle (SDLC) approach. Collaborative work is done with the project team on a priority list. The implementation is done through “SCRUM” an empirical framework for learning. It has multiple sprints which are deliverable products. This idea has substantially reduced the ‘time to market' as the customer can decide which features of the software they would like to be delivered on a priority basis. To model trends of fault detection in each sprint, a growth model of software reliability is used. This research article presents a framework to analyze and measure the cumulative errors in an Agile Testing Process, the authors have applied modeling on various SRGMs to prove acceptability in an agile development process and finally compares these models using the Mahalanobis Distance Formula for Model ranking. The Mahalanobis distance criteria is easy to compute and that can be utilized to get the ranks and select the best model in view of an arrangement of contributing criteria.

Asymptotic Inference for Optimal Rerandomization Designs

AbstractRecently a computational-based experimental design strategy called rerandomization has been proposed as an alternative or complement to traditional blocked designs. The idea of rerandomization is to remove, from consideration, those allocations with large imbalances in observed covariates according to a balance criterion, and then randomize within the set of acceptable allocations. Based on the Mahalanobis distance criterion for balancing the covariates, we show that asymptotic inference to the population, from which the units in the sample are randomly drawn, is possible using only the set of best, or ‘optimal’, allocations. Finally, we show that for the optimal and near optimal designs, the quite complex asymptotic sampling distribution derived by Li et al. (2018), is well approximated by a normal distribution.

Assessment and management of pesticide pollution at a river basin level part II: Optimization of pesticide monitoring networks on surface aquatic ecosystems by data analysis methods

The high cost of extensive pesticide monitoring studies, required for the protection of water resources, and the necessity of early identification of environmental threats, highlighted the need for prioritization of pesticides and sampling sites to be monitored. The aim of this study was to develop an optimum surface water monitoring network at a catchment scale including only the sites of a catchment vulnerable to pesticide pollution. The identification of sampling sites vulnerable to pesticide pollution (VPS) was based on the data of an intensive monitoring survey of 302 pesticides in 102 stationary sampling sites located on the surface water network of a river basin. In the proposed methodology the left-censored data of the analytical results derived from the above mentioned monitoring campaign were included in the statistical analyses by transforming all the raw data into categorical variables and arranging them in ordinal scales based on ecotoxicological thresholds derived from pesticide toxicity tests on aquatic non-target organisms. The categorized data were subjected to Categorical Principal Component Analysis with Optimal Scaling. For the identification of the VPS, the Squared Mahalanobis Distance criterion was applied on the extracted values (scores) of the significant principal components. With this methodology a 46% reduction in the number of the monitoring stations was achieved. This approach will be valuable in establishing more cost effective monitoring schemes in the future in other basins and in developing targeted measures to eliminate or limit the effect of critical pollution sources in surface aquatic systems. Moreover, by applying the proposed methodology, historical monitoring data can be used to initiate more efficient pesticide monitoring campaigns in the future.

Fault Diagnosis of Rolling Element Bearing Using Multi-Scale Lempel-Ziv Complexity and Mahalanobis Distance Criterion

A new fault diagnosis technique for rolling element bearing using multi-scale Lempel-Ziv complexity (LZC) and Mahalanobis distance (MD) criterion is proposed in this study. A multi-scale coarse-graining process is used to extract fault features for various bearing fault conditions to overcome the limitation of the single stage coarse-graining process in the LZC algorithm. This is followed by the application of MD criterion to calculate the accuracy rate of LZC at different scales, and the best scale corresponding to the maximum accuracy rate is identified for fault pattern recognition. A comparison analysis with Euclidean distance (ED) criterion is also presented to verify the superiority of the proposed method. The result confirms that the fault diagnosis technique using a multi-scale LZC and MD criterion is more effective in distinguishing various fault conditions of rolling element bearings.

An improved particle filtering-based approach for health prediction and prognosis of nonlinear systems

Health monitoring of nonlinear systems is broadly concerned with the system health tracking and its prediction to future time horizons. Estimation and prediction schemes constitute as principle components of any health monitoring technique. Particle filter (PF) represents a powerful tool for performing state and parameter estimation as well as prediction of nonlinear dynamical systems. Estimation of the system parameters along with the states can yield an up-to-date and reliable model that can be used for long-term prediction problems through utilization of particle filters. This feature enables one to deal with uncertainty issues in the resulting prediction step as the time horizon is extended. Towards this end, this paper presents an improved method to achieve uncertainty management for long-term prediction of nonlinear systems by using particle filters. In our proposed approach, an observation forecasting scheme is developed to extend the system observation profiles (as time-series) to future time horizon. Particles are then propagated to future time instants according to a resampling algorithm instead of considering constant weights for the particles propagation in the prediction step. The uncertainty in the long-term prediction of the system states and parameters are managed by utilizing dynamic linear models for development of an observation forecasting scheme. This task is addressed through an outer adjustment loop for adaptively changing the sliding observation injection window based on the Mahalanobis distance criterion. Our proposed approach is then applied to predicting the health condition as well as the remaining useful life (RUL) of a gas turbine engine that is affected by degradations in the system health parameters. Extensive simulation and case studies are conducted to demonstrate and illustrate the capabilities and performance characteristics of our proposed and developed schemes.

The use of MD-CUMSUM and NARX neural network for anticipating the remaining useful life of bearings

The accurate determination of remaining useful life (RUL) of bearings is of immense importance in the condition-based maintenance of any rotating machinery. In this paper, a data driven prognostic approach based on nonlinear autoregressive neural network with eXogenous Inputs (NARX-NN) in combination with wavelet-filter technique is applied to the RUL estimation of bearings. Firstly, the vibration signals generated in an experimental test rig are processed with the proposed wavelet-filter to augment the impulsive characteristics of bearing signals and improve the quality of fault feature extraction. Secondly, a variety of time-domain features are extracted from the processed bearing signals. However, these features exhibit a highly non-monotonic behavior as the bearing condition degrades. To overcome this drawback, a new health indicator (HI) based on Mahalanobis distance (MD) criterion and cumulative sum (CUMSUM) chart is proposed in this paper. Thirdly, the NARX-NN is first designed as a time delay neural network (TDNN). Then, the derived HI and the age of the bearing are used as inputs with life percentage of the bearing as output in order to train the TDNN model, which unlike the usual artificial neural networks (ANNs) performs a one-step ahead prediction of the bearing RUL. The results suggest that the proposed method can effectively predict the RUL of bearings with an acceptable degree of accuracy, and outperforms the use of self-organizing map-based indicator and the traditional FFNNs for RUL inference.

Radar HRRP target recognition with deep networks

Feature extraction is the key technique for radar automatic target recognition (RATR) based on high-resolution range profile (HRRP). Traditional feature extraction algorithms usually utilize shallow architectures, which result in the limited capability to characterize HRRP data and restrict the generalization performance for RATR. Aiming at those issues, in this paper deep networks are built up for HRRP target recognition by adopting multi-layered nonlinear networks for feature learning. To learn the stable structure and correlation of targets from unlabeled data, a deep network called Stacked Corrective Autoencoders (SCAE) is further proposed via taking the advantage of the HRRP's properties. As an extension of deep autoencoders, SCAE is stacked by a series of Corrective Autoencoders (CAE) and employs the average profile of each HRRP frame as the correction term. The covariance matrix of each HRRP frame is considered for establishing an effective loss function under the Mahalanobis distance criterion. We use the measured HRRP data to show the effectiveness of our methods. Furthermore, we demonstrate that with the proper optimization procedure, our model is also effective even with a moderately incomplete training set.

Fault diagnosis of roller bearings based on Laplacian energy feature extraction of path graphs

Feature extraction of roller bearing is always an intractable problem and has attracted considerable attention for a long time. The vibration signal of roller bearing can be treated as the path graph in a manifold perspective. Generally, vibration signals of roller bearings with different faults have different correlation matrices of path graphs which including different adjacency matrices and Laplacian matrices. Therefore, as a complexity feature of the path graph, the Laplacian energy (LE) can be employed to analyze the roller bearing vibration signals. In this paper, LE is introduced as the fault feature of bearing vibration signals from graph spectrum domain and then a new fault diagnosis method based on the LE single feature extraction and Mahalanobis distance (MD) criterion function is proposed and applied to the analysis of roller bearing vibration signals. Experimental analysis results show that the proposed method can identify the roller bearing faults accurately and effectively only with a small amount of sampling points and training samples.

Beer classification by means of a potentiometric electronic tongue

In this work, an electronic tongue (ET) system based on an array of potentiometric ion-selective electrodes (ISEs) for the discrimination of different commercial beer types is presented. The array was formed by 21 ISEs combining both cationic and anionic sensors with others with generic response. For this purpose beer samples were analyzed with the ET without any pretreatment rather than the smooth agitation of the samples with a magnetic stirrer in order to reduce the foaming of samples, which could interfere into the measurements. Then, the obtained responses were evaluated using two different pattern recognition methods, principal component analysis (PCA), which allowed identifying some initial patterns, and linear discriminant analysis (LDA) in order to achieve the correct recognition of sample varieties (81.9% accuracy). In the case of LDA, a stepwise inclusion method for variable selection based on Mahalanobis distance criteria was used to select the most discriminating variables. In this respect, the results showed that the use of supervised pattern recognition methods such as LDA is a good alternative for the resolution of complex identification situations. In addition, in order to show an ET quantitative application, beer alcohol content was predicted from the array data employing an artificial neural network model (root mean square error for testing subset was 0.131 abv).

Decision support system for breast cancer detection using mammograms

Mammograms are by far one of the most preferred methods of screening for breast cancer. Early detection of breast cancer can improve survival rates to a greater extent. Although the analysis and diagnosis of breast cancer are done by experienced radiologists, there is always the possibility of human error. Interobserver and intraobserver errors occur frequently in the analysis of medical images, given the high variability between every patient. Also, the sensitivity of mammographic screening varies with image quality and expertise of the radiologist. So, there is no golden standard for the screening process. To offset this variability and to standardize the diagnostic procedures, efforts are being made to develop automated techniques for diagnosis and grading of breast cancer images. This article presents a classification pipeline to improve the accuracy of differentiation between normal, benign, and malignant mammograms. Several features based on higher-order spectra, local binary pattern, Laws' texture energy, and discrete wavelet transform were extracted from mammograms. Feature selection techniques based on sequential forward, backward, plus-l-takeaway-r, individual, and branch-and-bound selections using the Mahalanobis distance criterion were used to rank the features and find classification accuracies for combination of several features based on the ranking. Six classifiers were used, namely, decision tree classifier, fisher classifier, linear discriminant classifier, nearest mean classifier, Parzen classifier, and support vector machine classifier. We evaluated our proposed methodology with 300 mammograms obtained from the Digital Database for Screening Mammography and 300 mammograms from the Singapore Anti-Tuberculosis Association CommHealth database. Sensitivity, specificity, and accuracy values were used to compare the performances of the classifiers. Our results show that the decision tree classifier demonstrated an excellent performance compared to other classifiers with classification accuracy, sensitivity, and specificity of 91% for the Digital Database for Screening Mammography database and 96.8% for the Singapore Anti-Tuberculosis Association CommHealth database.

Fault diagnosis of rolling bearings based on multifractal detrended fluctuation analysis and Mahalanobis distance criterion

Vibration data of faulty rolling bearings are usually nonstationary and nonlinear, and contain fairly weak fault features. As a result, feature extraction of rolling bearing fault data is always an intractable problem and has attracted considerable attention for a long time. This paper introduces multifractal detrended fluctuation analysis (MF-DFA) to analyze bearing vibration data and proposes a novel method for fault diagnosis of rolling bearings based on MF-DFA and Mahalanobis distance criterion (MDC). MF-DFA, an extension of monofractal DFA, is a powerful tool for uncovering the nonlinear dynamical characteristics buried in nonstationary time series and can capture minor changes of complex system conditions. To begin with, by MF-DFA, multifractality of bearing fault data was quantified with the generalized Hurst exponent, the scaling exponent and the multifractal spectrum. Consequently, controlled by essentially different dynamical mechanisms, the multifractality of four heterogeneous bearing fault data is significantly different; by contrast, controlled by slightly different dynamical mechanisms, the multifractality of homogeneous bearing fault data with different fault diameters is significantly or slightly different depending on different types of bearing faults. Therefore, the multifractal spectrum, as a set of parameters describing multifractality of time series, can be employed to characterize different types and severity of bearing faults. Subsequently, five characteristic parameters sensitive to changes of bearing fault conditions were extracted from the multifractal spectrum and utilized to construct fault features of bearing fault data. Moreover, Hilbert transform based envelope analysis, empirical mode decomposition (EMD) and wavelet transform (WT) were utilized to study the same bearing fault data. Also, the kurtosis and the peak levels of the EMD or the WT component corresponding to the bearing tones in the frequency domain were carefully checked and used as the bearing fault features. Next, MDC was used to classify the bearing fault features extracted by EMD, WT and MF-DFA in the time domain and assess the abilities of the three methods to extract fault features from bearing fault data. The results show that MF-DFA seems to outperform each of envelope analysis, statistical parameters, EMD and WT in feature extraction of bearing fault data and then the proposed method in this paper delivers satisfactory performances in distinguishing different types and severity of bearing faults. Furthermore, to further ascertain the nature causing the multifractality of bearing vibration data, the generalized Hurst exponents of the original bearing vibration data were compared with those of the shuffled and the surrogated data. Consequently, the long-range correlations for small and large fluctuations of data seem to be chiefly responsible for the multifractality of bearing vibration data.