Artificial Neural Networks as a Methodology for Optimal Location of Static Synchronous Series Compensator in Transmission Systems

This paper aims to optimally locate a static synchronous series compensator (SSSC), using an algorithm based on artificial neural networks to improve the voltage profile in a transmission system. An IEEE test system is used as a base for study cases. The neural network training is carried out in Matlab software, in which an exhaustive search of all the possible scenarios for compensation is performed, where non-linear loads are modified in all the transmission busbars. The implementation of the non-linear loads varies in a range between 50 MVAr and 150 MVAr, for every scenario the data analyzed are voltage profile, reactive power, active power and power factor of each bar. The resulting data is classified according to the target deviation of each variation, where the optimal position of the compensator is obtained and incorporated into the artificial neural network. The results obtained by the artificial neural network show the optimal location for the SSSC compensator, which shows the behaviour of the network in the presence of new unknown data. The error percentage presented by the algorithm analysis is in the range of -1% to 1.5%, which determines the efficiency of the artificial neural network and its accuracy in the face of variations and input of unknown data.