Power System Fault Treatment Using Deep Learning

A novel complete protection scheme is proposed in this article that detects and classifies the transmission line faults. Also it estimates the location of fault. The proposed method is implemented in Kundur two area 4 machine 11 bus system over a distance of 220 km. The data is extracted automatically by the LSTM (Long Short Term Memory) networks of deep learning model. Two different LSTM network modules like LSTM (FL) and LSTM (FD) are used for this purpose. LSTM (FD) module is responsible for accurate fault detection and classification. The LSTM (FL) module is responsible for the estimation of exact location of the fault. Both symmetrical and unsymmetrical fault cases are tested under both LSTM fault modules. The protection modules are also tested under different fault locations, fault resistances and noisy signals etc. The features taken into consideration for the operation of the protection modules are different bus voltages, bus currents, zero sequence voltage, zero sequence current etc. The advantages of this method are it works at very low sampling frequency and it also does not require any communication link.