Dynamic Security Margin Estimation with Preventive Control Using Artificial Neural Networks

Abstract

On-line dynamic security assessment (DSA) is challenging using conventional techniquesbecause most DSA approaches use detailed mathematical models of the system that arecomputationally intensive and time-consuming. In this paper, a method based on Artificial NeuralNetworks (ANN) is developed to estimate the security margin. The security margin for a given powersystem is obtained by applying standard operations criteria for transient response to off-line timesimulations. These simulations then form a database that can be used to train a pattern matchingapproach, such as, ANNs. Feature selection using statistical approaches is applied to overcome thedimensional problem of applying the ANN to larger systems. This method provides a fast and accuratetool to evaluate dynamic security. If the estimated security margin is less than requirements, thenpreventive control actions that guarantee dynamic security of the power system are needed. This isachieved by optimal rescheduling of the generation with given constraints on the network powerflows and system security margins as estimated by the ANN. This requires a modified Optimal PowerFlow (OPF) solution that allows the trained ANN to act as a security objective function. Numericalresults on the New England 39-bus system validate the methodology.