George Georgoulas

Asynchronous Machine Rotor Fault Diagnosis Technique Using Complex Wavelets

This paper introduces a novel approach for the detection of rotor faults in asynchronous machines, based on wavelet analysis of the stator phase current. To be more specific, the measured stator phase current is filtered through a complex wavelet. Theoretical analysis validates that the spectrum of the modulus of the result of the filtering is free from the fundamental supply frequency component, and the fault characteristics can be highlighted. This is advantageous, especially if the induction machine operates at low slip values, where the characteristic frequency components of the rotor fault are very close to the fundamental frequency component. At the same time, by matching the wavelet function to the frequencies of the faulty components, a narrow bandpass filter at the frequency region of the fault characteristic spectral components is obtained. Furthermore, in the context of this paper, features extracted using the proposed technique are used as input to a support vector machine classifier that is employed for the detection of the rotor fault. Simulation and experimental results demonstrate the effectiveness of the proposed technique.

Remaining Useful Life Estimation in Rolling Bearings Utilizing Data-Driven Probabilistic E-Support Vectors Regression

We report on a data-driven approach for the remaining useful life (RUL) estimation of rolling element bearings based on ε-Support Vector Regression ( ε-SVR). Lifetime data are analyzed and evaluated. The occurrence of critical faults in every test is located, and a critical operational threshold is established. Multiple statistical features from the time-domain, frequency domain, and time-scale domain through a wavelet transform are extracted from the recordings of two accelerometers, and assessed for their diagnostic performance. Among those features, Wiener entropy is utilized for the first time in the condition monitoring of rolling bearings. A SVR model is trained and tested for the prediction of RUL on unseen data. Special attention is given in the tuning and the optimization of the user-defined hyper-parameters of the e-SVR model. Error bounds are estimated at each prediction point through a Bayesian treatment of the classical SVR model. The results are in good agreement to the actual RUL curve for all the tested cases. Prognostic performance metrics are also provided, and the discussion on the test results concludes with the generic character of the proposed methodology and its applicability in any prognostic task.

Principal Component Analysis of the start-up transient and Hidden Markov Modeling for broken rotor bar fault diagnosis in asynchronous machines

This article presents a novel computational method for the diagnosis of broken rotor bars in three phase asynchronous machines. The proposed method is based on Principal Component Analysis (PCA) and is applied to the stators three phase start-up current. The fault detection is easier in the start-up transient because of the increased current in the rotor circuit, which amplifies the effects of the fault in the stators current independently of the motors load. In the proposed fault detection methodology, PCA is initially utilized to extract a characteristic component, which reflects the rotor asymmetry caused by the broken bars. This component can be subsequently processed using Hidden Markov Models (HMMs). Two schemes, a multiclass and a one-class approach are proposed. The efficiency of the novel proposed schemes is evaluated by multiple experimental test cases. The results obtained indicate that the suggested approaches based on the combination of PCA and HMMs, can be successfully utilized not only for identifying the presence of a broken bar but also for estimating the severity (number of broken bars) of the fault.

FEATURE EXTRACTION AND CLASSIFICATION OF FETAL HEART RATE USING WAVELET ANALYSIS AND SUPPORT VECTOR MACHINES

Since the fetus is not available for direct observations, only indirect information can guide the obstetrician in charge. Electronic Fetal Monitoring (EFM) is widely used for assessing fetal well b...

Automatic Pattern Identification Based on the Complex Empirical Mode Decomposition of the Startup Current for the Diagnosis of Rotor Asymmetries in Asynchronous Machines

This paper presents an advanced signal processing method applied to the diagnosis of rotor asymmetries in asynchronous machines. The approach is based on the application of complex empirical mode decomposition to the measured start-up current and on the subsequent extraction of a specific complex intrinsic mode function. Unlike other approaches, the method includes a pattern recognition stage that makes possible the automatic identification of the signature caused by the fault. This automatic detection is achieved by using a reliable methodology based on hidden Markov models. Both experimental data and a hybrid simulation-experimental approach demonstrate the effectiveness of the proposed methodology.

Particle filter based anomaly detection for aircraft actuator systems

This paper describes the background, simulation and experimental evaluation of an anomaly detector for Brushless DC motor winding insulation faults in the context of an aircraft Electro-Mechanical Actuator (EMA) application. Results acquired from an internal Failure Modes and Effects Analysis (FMEA) study identified turn-to-turn winding faults as the primary mechanism, or mode, of failure. Physics-of-failure mechanisms used to develop a model for the identified fault are provided. The model was implemented in Simulink to simulate the dynamics of the motor with a turn-to-turn insulation winding fault. Then, an experimental test procedure was devised and executed to validate the model. Additionally, a diagnostic feature, identified by the fault model and derived using Hilbert transforms, was validated using the Simulink model and experimental data for several fault dimensions. Next, a feature extraction routine preprocesses monitoring parameters and passes the resulting features to a particle filter. The particle filter, based on Bayesian estimation theory, allows for representation and management of uncertainty in a computationally efficient manner. The resulting anomaly detection routine declares a fault only when a specified confidence level is reached at a given false alarm rate. Finally, the real-time performance of the anomaly detector is evaluated using LabVIEW.

A Symbolic Representation Approach for the Diagnosis of Broken Rotor Bars in Induction Motors

One of the most common deficiencies of currently existing induction motor fault diagnosis techniques is their lack of automatization. Many of them rely on the qualitative interpretation of the results, a fact that requires significant user expertise, and that makes their implementation in portable condition monitoring devices difficult. In this paper, we present an automated method for the detection of the number of broken bars of an induction motor. The method is based on the transient analysis of the start-up current using wavelet approximation signal that isolates a characteristic component that emerges once a rotor bar is broken. After the isolation of this component, a number of stages are applied that transform the continuous-valued signal into a discrete one. Subsequently, an intelligent icon-like approach is applied for condensing the relative information into a representation that can be easily manipulated by a nearest neighbor classifier. The approach is tested using simulation as well as experimental data, achieving high-classification accuracy.

Examining cross-database global training to evaluate five different methods for ventricular beat classification.

The detection of ventricular beats in the holter recording is a task of great importance since it can direct clinicians toward the parts of the electrocardiogram record that might be crucial for determining the final diagnosis. Although there already exists a fair amount of research work dealing with ventricular beat detection in holter recordings, the vast majority uses a local training approach, which is highly disputable from the point of view of any practical-real-life-application. In this paper, we compare five well-known methods: a classical decision tree approach and its variant with fuzzy rules, a self-organizing map clustering method with template matching for classification, a back-propagation neural network and a support vector machine classifier, all examined using the same global cross-database approach for training and testing. For this task two databases were used-the MIT-BIH database and the AHA database. Both databases are required for testing any newly developed algorithms for holter beat classification that is going to be deployed in the EU market. According to cross-database global training, when the classifier is trained with the beats from the records of one database then the records from the other database are used for testing. The results of all the methods are compared and evaluated using the measures of sensitivity and specificity. The support vector machine classifier is the best classifier from the five we tested, achieving an average sensitivity of 87.20% and an average specificity of 91.57%, which outperforms nearly all the published algorithms when applied in the context of a similar global training approach.

A simulation based Decision Support System for logistics management

Abstract This paper deals with designing and developing a Decision Support System (DSS) that will be able to manage the flow of goods and the business transactions between a port and a dry port. An integrated DSS architecture is proposed and specified and the main components are designed on the basis of simulation and optimization modules. In order to show the use and implementation of the DSS, this work tests and analyzes the case of the area of the Trieste port and manages the export flows of freights between a dry port and a seaport. An integrated approach is designed mainly at tactical and operational decision level exploiting simulation and optimization approaches and especially metaheuristic approaches.

Real-time fault detection and accommodation for COTS resolver position sensors

Resolver sensors are utilized as absolute position transducers to control the position and speed of actuators in many flight critical applications where robustness, accuracy and ability to operate in extreme environmental conditions are required. To ensure these requirements, several designs for self-diagnosing sensors were proposed in the past. However, such designs require additional instrumentation leading to further development and certification costs. To mitigate such costs, a methodology was sought to achieve similar self-diagnosis capabilities that can be retrofitted to already existing commercial off-the-shelf (COTS) resolver sensors. To achieve this goal, this paper proposes a new approach for real-time tracking of resolver faults with the ability to perform real-time fault detection. However, the additional benefit of this approach is the ability to accommodate faults in real-time once a fault is detected. The primary fault studied in this paper is resolver channel mismatch. A formulation based on the physical operating principles for a resolver sensor is provided. Diagnostic measures are identified for use in statistical-based fault-detection routines. Then, estimates for the resolver mismatch are tracked using a time-varying Kalman filter. Fault accommodation is achieved by applying the tracked estimates to adjust for the resolver mismatch. Finally, the fault detection and accommodation routines are evaluated in Simulink for an electro-mechanical actuator (EMA).