Voltage stability analysis using artificial neural network / Norol Janah Ahmad

Voltage stability problems have been one of the major concern for electric utilities as a result of heavy loading. This thesis presents the applications of artificial neural network in voltage stability analysis. A multi layer feed forward artificial neural network with error back propagation and memory-based feed forwards networks based on the estimation probability density functions, general regression neural network are proposed for calculation of voltage stability. Back propagation using three layers and four layers i.e input layer, one hidden layer and output layer while four layer i.e input layer, two hidden layer and output layer are use. General regression using four layers i.e. input layer, hidden layer, division & summation layer and output layer are being applied in predicting voltage stability. Both methods used same sets of data in training process and same other sets of data for testing process. All those sets of data are generated by Fast-Decoupled Load Flow Simulation using fifteen bus system. Real and reactive power at all buses and real and reactive power outputs of generator were applied as inputs to artificial neural network. Test are carried out and the results for this two method are compared. From thr result, it shows that artificial neural network can be used to predict voltage stability of power system. The result also shows that General Regression are more accurate compare to Back propagation method. Back propagation using one hidden layer show good value compare to back propagation using two hidden layer.