Novel Algorithms for Medium/Low-voltage Arcing Faults in Switchgear & Motor Coils: From Concept to Realization

Switchgear arcing faults have been a primary cause for concern for the manufacturing industry and safety personnel alike. The deregulation of the power industry being in full swing and ever-growing competitiveness in the distribution sector calls for the transition from preventive to predictive maintenance. Switchgear forms an integral part of the distribution system in any power system set-up. Keeping in mind the switchgear arcing faults, the transition mentioned above applies most of all to the switchgear industry. Apart from the fact that it is the primary cause of serious injuries to electrical workers worldwide, switchgear arcing faults directly affect the quality and continuity of electric power to the consumers. A great amount of technological advancement has taken place in the development of arc resistant/proof switchgear. However, most of these applications focus on minimizing the damage after the occurrence of the arcing fault. The problem associated with the compromise on the quality and continuity of electric power in such a scenario still awaits a technical as well as economically feasible solution. This dissertation describes the development of a novel approach for the detection of arcing faults in medium/low-voltage switchgear. The basic concept involves the application of differential protection for the detection of any arcing within the switchgear. The new approach differs from the traditional differential concept in the fact that it employs higher order harmonic components of the line current as the input for the differential scheme. Actual arc generating test-benches have been set up in the Power System Simulation Laboratory at Energy Systems Research Center to represent both medium and low voltage levels. Hall-effect sensors in conjunction with Data Acquisition in LabVIEW are employed to record the line current data before, during and after the arcing phenomenon. The methodology is first put to test via simulation approach for medium voltage levels and then corroborated by actual hardware laboratory testing for low voltage levels. The plots provided from the data gathering and simulation process clearly underline the efficiency of this approach to detect switchgear arcing faults. Both magnitude and phase differential concepts seem to provide satisfactory results. Apart from the technical efficiency, the approach is financially feasible considering the fact that the differential protection is already being comprehensively employed worldwide.

Developments spanning a major portion of the previous decade have witnessed the emergence of high/medium/low-voltage arcing fault as one of the more prominent issues confronting the power industry and associated researchers alike. Research over the past decade has been dedicated to the modeling, detection and/or monitoring of arcing faults at various voltage levels and power system apparatuses. Furthermore, the research presented in this dissertation presents and compares the performance of statistical methodologies utilized to classify the severity of low-voltage arcing in a motor coil. The approaches revolve around the utilization of statistical techniques such as Spectral Angle Mapper (SAM), Spectral Information Divergence (SID) and Linear Discriminant Analysis (LDA) to classify the severity of the motor coil arcing fault. Dedicated test-benches are utilized to simulate the arcing phenomenon of varying severity in a motor coil within laboratory environment. Hall-effect sensors in conjunction with the data acquisition module of LabVIEW provide an able means for data collection which is then subjected to off-line analysis. The conceptual approach preceding the classification process revolves around the extraction of pre-decided features associated with the current signal gathered during the arcing process. These features are associated with the higher order harmonic content of the current signal. Comparative analysis of the higher-order harmonic content in the arcing current as obtained from the test-bench and that from contemporary mathematical models for low-voltage arcing faults is presented to validate the choice of parameters for the spectral signature. The extracted features are utilized for classifying the severity of the motor coil arcing fault using the approaches mentioned above. That apart, Support Vector Machines have gained tremendous popularity in the last decade or so in the field of pattern recognition, classification and regression applications. The dissertation also presents the approach, implementation and results associated with the utilization of one-class and multi-class SVM techniques for the classification of motor coil arcing fault severity levels. The data pre-processing, filtering and feature extraction process do remain the same as that utilized for SAM, SID and LDA techniques. However, the Gaussian Kernel function has been utilized to map non-linear scatter spread on to a higher dimension Euclidean space H. The training data is utilized to construct the SVM classifier the parameters of which, such as Lagrangian multipliers, the bias, the penalty factor and the number of support vector machines are utilized to test the accuracy of the SVM classification algorithm on the test data. Detailed elaboration on the technique, the implementation methodology and the results has been provided in relevant sections. The performance of the classification approaches has been evaluated based on the accuracy of classification, robustness and feasibility of implementation of the approach. The classification results seem very promising in terms of accuracy and feasibility of approach for real-time implementation.