Erroneous measurement detection in substation automation system using OLS based RBF neural network

With the development of communication and information technology over the past decades, Electronic Instrumental Transducer (EIT) and broadband communication network have been prevalent within Substation Automation System (SAS) and power utilities. Since mal-function of EIT and broadband communication network within SAS can produce dangerous erroneous measurements, the risk for the protection system to receive these erroneous measurements and thereafter to mis-operate increase. Pattern identification can be utilized to detect erroneous measurements. In order to achieve satisfying pattern identification precision within time limit imposed by protection systems, Radial Basis Function Neural Network (RBFNN) are investigated in the paper. Orthogonal Least Square (OLS) learning algorithm is used to prune network scale in order to mitigate contradictory requirements of high precision and low time delay. Simulation results show OLS based RBFNN can achieve satisfying performance within limited time.