A fast faulted phase detection relay for high voltage transmission lines using the fir neural networks

This paper presents a new approach to on line faulted phase detection relay for high-speed protection of high voltage transmission lines using artificial neural networks. The phase detector shows its effectiveness using one relay connected to a transmission line of a transmission circuit, which is composed of three phase 400 kV transmission lines. The type of the neural network used in this approach is the Finite Impulse Response Artificial Neural Network (FIRANN). The FIRANN is trained using temporal backpropagation training algorithm. The training patterns are obtained by simulating different fault types, fault locations and loads for a certain power network using the ATP (Alternative Transient Program) version of the EMTP (Electro-Magnetic Transient Program). Also, dynamic arc faults and current transformer saturation had been taken into account. This type of neural networks shows a very great insensibility against lost and bad data and post-fault and pre-fault noise.