REACTIVE POWER COMPENSATION ASSESSMENT USING SELF-ORGANIZING FEATURE MAP

In the present work a self-organizing neural network is used for reactive power compensation assessment. The optimal reactive power sources planning are solved using a genetic algorithm. A genetic algorithm is proposed to optimize an objective function which is a linear combination of several factors, such as, overall system active power loss, cost incurred from the installation and use of VAR devices, enhance voltage profile. A self-organizing feature map is trained to compare the power system state once without optimization, then with optimization, first taking account of voltage constraints, and secondly without voltage constraints.