Thyristor-controlled series compensator and unified Power-flow controller

Increased demand of bulk power transfer in the modern power network has led to an increased focus on transmission constraints and alleviation. Flexible ac transmission systems (FACTS) devices offer a versatile alternative to conventional reinforcement methods. Among them, the thyristor-controlled series compensator (TCSC) and unified power-flow controller (UPFC) are important FACTS devices, which are used extensively for improving the utilization of the existing transmission system. The presence of TCSC in fault loop not only affects the steady-state components but also the transient components. The controllable reactance, the metal–oxide varistors (MOVs) protecting the capacitors, and the air-gap operation make the protection decision more complex and, therefore, the conventional relaying scheme based on fixed settings finds limitations. UPFC offers new horizons in terms of power system control. Recent techniques based on neural networks , find limitations, since they require a large number of neurons to model the structure of the network involving large training sets and training time. A hybrid technique using a wavelet transform combined with support vector machine (SVM) has been proposed for fault-zone identification in the TCSC line. The aforementioned work finds limitations since the wavelet transform is highly prone to noise and provides erroneous results even with a signal-to-noise ratio (SNR) of 30 dB. The computational time of SVM is higher compared to the proposed ensemble DTs-based data-mining model, which puts constraints on the online realization of SVM-based relays for distance relaying applications, where speed and accuracy are prime considerations. A strong motivation to build up an accurate and faster data-mining model for fault-zone identification in FACTS-based transmission lines. The proposed technique provides a data-mining model, such as ensemble decision trees (RFs), for fault-zone identification in a FACTS-based transmission line with an accuracy and reliability of more than 99%. RF, the data-mining algorithm, was found to be faster (3/4 cycles) and accurate compared to the existing machine-learning technique, such as SVM for fault-zone identification.