Evolutionary algorithm for detection and localization of faults in HVDC systems

The aim of this work is to analyze evolutionary algorithms in point of application for detection and localization of faults in HVDC systems. Considered faults are localized at DC side. The initial condition for the algorithm is low sampling rate up to 5-100kHz. This condition eliminates application of traditional traveling wave technique due to not sufficient time resolution. It is proposed to base criterion of the fault location on relation between distance of the fault and amplitude of aperiodic decaying component in current during transient. In order to estimate the amplitude of the aperiodic component a genetic algorithm is proposed. Using benchmarking signals a reference function of distance of the fault in relation to the amplitude of aperiodic component is created. Having this function it is possible to calculate fault location using estimated by genetic algorithm amplitude during the fault. This paper presents advantages and disadvantages of proposed technique.