Fault identification-based voltage sag state estimation using artificial neural network

This paper presents an artificial neural network (ANN) based approach to identify faults for voltage sag state estimation. Usually ANN cannot be used to abstract relationship between monitored data and arbitrarily named fault indices which are not related at all logically in numerical level. This paper presents a novel approach to overcome this problem. In this approach, not only the networks are trained to adapt to the given training data, the training data (the expected outputs of fault indices) is also updated to adapt to the neural network. During the training procedure, both the neural networks and training data are updated interactively. With the proposed approach, various faults can be accurately identified using limited monitored data. The approach is robust to measurement uncertainty which usually exists in practical monitoring systems. Furthermore, the updated fault indices are able to suggest the difference of the impact of various faults on bus voltages.