This article examines if machine learning (ML) and signal processing can be used for online condition monitoring to reveal interturn short circuit (ITSC) fault in the field winding of salient pole synchronous generators (SPSG). This was done by creating several ML classifiers to detect ITSC faults. A dataset for ML was built using power spectral density of the air-gap magnetic field extracted by fast Fourier transform, discrete wavelet transform energies, and Time Series FeatuRe Extraction based on Scalable Hypothesis tests (TSFRESH) to extract features from measurements of SPSG operated under several different severities of ITSC fault. Using this dataset, a wide range of classifiers were trained to detect the presence of ITSC faults. The classifiers evaluated were logistic regression, K -nearest neighbors, radial basis function support vector machine (SVM), linear SVM, XGBoost decision tree forest, multilayer perceptron, and a stacking ensemble classifier including all of the aforementioned. The classifiers were optimized using hyperparameter grid searches. In addition, some feature selection and reduction algorithms were assessed such as random forest feature selection, TSFRESH feature selection, and principal component analysis. This resulted in a classifier capable of detecting 84.5% of samples containing ITSC fault, with a 92.7% chance that fault detections are correct.
13 April 2021 in IEEE Transactions on Industry Applications
This article provides a novel approach for discriminating the interturn short circuit (ITSC) fault in a salient pole synchronous generator (SPSG) based on the use of noninvasive sensors. A stray magnetic field is used as a signal that can provide valuable data regarding the health condition of the SPSG. Generally, the captured signal under an ITSC fault is examined using a classical signal processing tool based on fast Fourier transforms (FFTs). The amplitude of the side-bands to the frequency spectrum is increased by ITSC, eccentricity, and broken damper-bar faults. Therefore, determining the type of fault is not possible by analyzing the FFT side-bands. A unique feature is introduced using a short-time Fourier transform to identify the ITSC in its early stage. The proposed feature can locate and recognize the ITSC fault with high accuracy in either the no-load or the full-load operation of the SPSG. This novel methodology is verified by applying an ITSC fault to a 100 kVA custom-made SPSG.
13 October 2021 in IEEE Transactions on Industrial Electronics
Many methods used for precise fault detection in salient pole synchronous generators (SPSGs) often require a priori knowledge of the healthy case, but this requirement impedes application of the methods since an accurate analysis of different machine quantity waveforms is not usually carried out during commissioning. The inspection and maintenance processes in SPSGs are also costly and time consuming; therefore, reliable methods that can detect and discriminate between different faults without comparison with the healthy condition are highly desirable. This article proposes a precise method for detection and discrimination between different fault types in the SPSG. The method does not require healthy machine data and is applied to diagnose both interturn short circuits (ITSCs) in the field winding and dynamic eccentricities (DEs). The proposed nonintrusive detection algorithm is based on advanced signal analysis of stray magnetic field data and can be applied during SPSG operation. The method is highly precise for monitoring the condition of the rotor field winding and yields a unique pattern for diagnosing possible ITSC faults. Moreover, a distinctive pattern for the DE fault enables the discrimination between both considered failures, even if they are present at the same time. The proposed method is validated through finite-element modeling and experimentally on 100-kVA and 22-MVA SPSGs to demonstrate its applicability in real power plants.
7 May 2021 in IEEE Transactions on Industry Applications
Signal processing plays a crucial role in addressing failures in electrical machines. Experimental data are never perfect due to the intrusion of undesirable fluctuations unrelated to the investigated phenomenon, namely so-called noise. Noise has disturbing effects on the measurement data and, in the same way, could diminish or mask the fault patterns in feature extraction using different signal processors. This article introduces various types of noise occurring in an industrial environment. Several measurements are performed in the laboratory and power plants to identify the dominant type of noise. Fault detection in a custom-made 100-kVA synchronous generator under an interturn short-circuit fault is also studied using measurements of the air-gap magnetic field. Signal processing tools such as fast Fourier transform, short-time Fourier transform (STFT), discrete wavelet transform, continuous wavelet transform (CWT), and time-series data mining are used to diagnose the faults, with a central focus on additive noise impacts on processed data. Two novel patterns are introduced based on STFT and CWT for interturn short-circuit fault detection of synchronous generators that do not need a priori knowledge of a healthy machine. Useful methods are presented for hardware noise rejection.
Although synchronous generators are robust and long‐lasting equipment of power plants, consistent electricity production depends on their health conditions. Static and dynamic eccentricity faults are among the prevalent faults that may have a costly effect. Although several methods have been proposed in the literature to detect static and dynamic eccentricity faults in salient pole synchronous generators (SPSGs), they are non‐sensitive to a low degree of failure and require a predefined threshold to recognise the fault occurrence that may vary based on machine configuration. This article presents a detailed magnetic analysis of the SPSGs with static and dynamic eccentricity faults by focusing on the external magnetic field. The external magnetic field was measured using two search coils installed on the backside of the stator yoke. Also, advanced signal processing tools based on wavelet entropy were used to analyse the induced electromotive force ( emf ) in search coils to extract the fault index. The proposed index required no threshold to recognise the starting point of fault occurrence and was sensitive to a low degree of fault. It was also non‐sensitive to load variation and noise that may induce a false alarm.
13 April 2021 in IEEE Transactions on Industrial Electronics
Reliable operation of synchronous generators in hydroelectric power plants is crucial for avoiding unplanned stoppages that can incur substantial costs. The damper winding of salient pole synchronous generators (SPSGs) contributes to machine operation only during transient periods; however, it is a critical component that preserves the dynamic stability and protects the rotor in case of a fault. Consequently, detection of a broken damper bar (BDB) fault is vital for safe operation. Current methods for the BDB detection depend on visual inspection or offline tests. However, most of the recently proposed approaches have used invasive sensors that can detect BDB faults only during transient operation. In this article, a novel method is proposed based on a noninvasive sensor with high sensitivity to BDB faults that can identify a BDB fault either during transient operation or in the steady-state (SS) period. The effectiveness of the proposed method is validated by finite-element modeling and by experimental results from a 100-kVA custom-made SPSG. The proposed method is confirmed to provide a reliable and sensitive diagnosis of BDB faults during transient or SS operation, even in noisy environments.