Prediction of total transfer capability using ANN in restructured power system

The work proposed embodies prediction of total transfer capability in a restructured power system using artificial neural network under normal and single line outage conditions. A suitable feed forward network with 14 hidden layer neurons is designed to predict transfer capability in a modified IEEE 14-bus test system. Line status, initial voltage magnitude at all the 14 buses and loading in buyer area are taken as input variables while total transfer capability is taken as output of neural network. A novel approach to introduce line stability index as one of the constraints along with voltage magnitude, reactive power limits and angular stability is presented in this work. Predicted results from artificial neural network (ANN) are compared with conventional repeated power flow method to determine relative error between the predicted and calculated results. Maximum relative error obtained is 1.698651% which is quite acceptable considering speed of prediction.