Static Security Assessment in Power Systems Using Multi-Class SVM with Parameter Selection Methods

Abstract

 Abstract—Security assessment is a major concern in real time operation of electric power systems. Traditional method of security assessment performed by continuous load flow analysis involves long computer time and generates voluminous results. This paper presents a practical and feasible Support Vector Machine Based Pattern Classification (SVMBPC) approach for static security assessment in power systems. The proposed approach classifies the security status of any given operating condition in one of the four classes - Secure, Critically Secure, Insecure and Highly Insecure based on the computation of a numeric value called security index. The feature selection stage uses a simple and straightforward forward sequential method to select the best feature set from a large set of variables. The static security classifier is designed by a multi-class SVM with different parameter tuning methods. The proposed approach is implemented in New England 39 bus and IEEE 118 bus systems and the results are validated. techniques and expert systems like fuzzy set theory has been proposed for security assessment problem, overcoming the pitfalls of traditional method. Literatures have reported the use of Artificial Neural Network techniques (4), (5), fuzzy logic combined with neural network (6), genetic based neural network (7) for static security assessment process. The performance of all these existing techniques are highly problem dependent and hence its suitability cannot be generalized. Nowadays, pattern classification is gaining more importance in solving many power system problems. In this approach, main bulk of work is done off-line to generate sufficient dataset. The classification function, designed based on the train set, helps to access the system security level in a short period of time. This paper addresses security assessment as a pattern classification problem with the classifier function designed by Support Vector Machine (SVM). SVM is a new and promising tool for learning separating functions in PR system with the capability of handling non-linear separability. The SVM classifier is designed for multi-classification based on the calculation of a term called Static Security Index (SSI), for each specified contingency. In this paper, four-class logic is used for the definition of system security viz., secure, critically secure, insecure, highly insecure. An operator likes to know exactly the severity level of disturbances for a given system operating condition. On-line security assessment allows the operator to know the security status and helps to determine the corrective actions. This paper also addresses different heuristic optimization techniques like Particle Swarm Optimization (8), Real Coded Genetic Algorithm (9) and Differential Evolution (10) used in the selection of SVM parameters globally. The classification approach is implemented in New England (NE) 39 bus system and IEEE 118 bus system and the results are compared.