Hybrid fault diagnosis algorithms for transmission lines

This paper presents a fault location method for transmission lines with the application of a mono-objective optimization technique using the ellipsoid algorithm with voltage and current data of both terminals. The fault detection is performed using the stationary wavelet transform and Parseval’s theorem, and the classification was conducted with the application of artificial neural networks. The minimization of the objective function defined for the short and long transmission line models provides not only the distance to the fault point, but also the fault resistance value. Many short-circuit situations simulated in the alternative transients program are tested with variations in the fault type, adjustments in the distance to the fault point, and fault resistance. The results of the algorithm applied to real faults in the electrical system of Brazil are also presented and compared to the values obtained with a classic fault location algorithm. According to the observations, the adopted formulation achieves the pre-established objectives, with mean errors of fault location for the real cases lower than 2% of the line length.