Protection of uncompensated and compensated transmission lines problems and solutions

Abstract

This thesis presents artificial intelligence (AI) based efficient fault classification and section identification schemes in TCSC Lines. The features derived from instantaneous and RMS currents are given as input to AI classifiers for classifying faults and identifying faulty section. The input features proposed in this work can be derived with relative ease and imposes lower computational burden. The performance of Support Vector Machine (SVM), Random Forest (RF), Naïve Bayes (NB), and Naïve Bayes Tree (NBTree) classifiers is evaluated for different features. Wide variations in system and fault conditions are selected to generate fault cases. The parameters selected to generate train and test cases are completely different. Binary-class and multi-class classification approach is also considered for the classifiers. The results suggest that the performanceof SVM and RF classifiers is found to be remarkable with different set of features. newlineParallel transmission lines are widely adopted in modern power systems due to their bulk power transfer capabilities. However, the possibilities of inter-circuit and simultaneous faults and zero sequence mutual coupling effects make protection of parallel transmission lines more challenging as compared to single-circuit lines. In this thesis, fundamental current-phasor based technique is proposed for fault classification in parallel lines. The faulty phase(s) are identified based on indices exceeding pre-defined threshold. Also, the technique is validated with recorded data of real time faults in an existing Indian power system network. The performance of the phasor-based technique is impeccable for classifying faults in parallel lines. newline