Optimization of Series Compensation in Transmission Networks Using Artificial Neural Networks

Abstract

To respond to the ever-increasing power demand of load centers, power is transmitted at extrahigh voltages. However, an increase in power transfer level should be supported by an enhanced level of security. Flexible AC Transmission System (FACTS) devices present an economical and efficient alternative to consider for achieving higher power transfer level with enhanced security instead of introducing new transmission facilities, to maintain a large stability margin of power in transmission line. This project aims to optimize the level of series compensation in transmission networks using Artificial Neural Network (ANN). Series compensation enables higher level of power to be transferred by reducing (reactive) losses. Among the series FACTS controllers, Thyristor Controlled Series Compensator (TCSC) has been chosen to be optimized on an SMIB system. Lead-Lag (LL) based TCSC remain the controller of choice due to the favorable performance to cost ratio. Nevertheless, modeling the highly non-linear power system with a linear controller, limits the system’s performance during adversities. ANN being non-linear per se and possessing high generalization capabilities, offers more versatility in modeling the power system. Indeed, an ANN based TCSC was designed and the performance during contingencies was compared to that of the LL based TCSC. As expected, the ANN based TCSC demonstrated a damping capability twice as fast and offering the SMIB system with a higher robustness as well as better resistance to fault condition. To increase the accuracy and reliability of the proposed controller, the investigation can be performed on a multi-machine system with different loading conditions as well as determining the optimal location of the TCSC module.