Neural network based transient stability model to analyze the security of Java-Bali 500 kV power system

A transient stability model based on back propagation neural network is used to analyse transient stability of Java-Bali electricity system, especially in calculating the critical clearing time. Inputs used for the neural network is the real and the active variable load power, whereas the target of the neural network is the critical clearing time (CCT). Data of target CCT used for the training was calculated using the concept of One Machine Infinite Bus (OMIB), which is the reduction of multi-machine system in combination with the method of equal area criterion (EAC) through the Trapezoid method and the 4^th Order Runge-Kutta method. The position and type of disturbance is assumed not changed. This model was tested on 500 kv Java-Bali system with type of disturbance is 3 phase ground fault. The results indicate that the approximate value of the CCT calculated by the proposed method has a minimum error of 0% and maximum error of 0.0008% compared with CCT by OMIB.