Determination of vulnerable machines for online transient security assessment in smart grid using artificial neural network

Smart grid integrates effective harnessing of control, communication and computer technologies in the operation of present day bulk power system to create several possibilities for meeting challenges of the electricity industry. This paper presents an artificial neural network based approach for fast and accurate power system transient security assessment. Radial basis function (RBF) neural network is employed to assess the transient security status by identifying each vulnerable generating machine that will lose synchronism for a given operating condition. The model can serve as decision making tool for the power planners to take preventive control actions for generation shedding/rescheduling for online applications. A feature selection technique based on the class separability index and correlation coefficient has been employed. The effectiveness of the proposed methodology is demonstrated by overall accuracy of the test results for unknown patterns for IEEE 39-bus New England system.