HVDC grid fault location method using genetic algorithm on reconstructed frequency-domain voltage profiles

Abstract

This paper presents a novel frequency domain based fault location method for High Voltage Direct Current (HVDC) grids. The method relies only on short data windows (below two milliseconds) of single-ended voltage measurements, thus rendering the technique practically applicable to HVDC grids employing high-speed circuit breakers. Based on the available post-fault voltage traces, a frequency domain profile of voltage transfer function is initially constructed, which is then iteratively evaluated against theoretical voltage profiles, synthesized with the use of travelling wave principles. The evaluation routine is based on a genetic algorithm (GA) regime, which once terminated indicates a close approximation to the feeder’s fault location. The result of this process is further refined using a weighted averaging function to obtain the final fault distance estimation. The performance of the proposed method is demonstrated using comprehensive electromagnetic transient (EMT) simulation studies conducted on a four-terminal meshed HVDC grid using PSCAD/EMTDC. The results revealed that the method can estimate the fault location with high accuracy for all transmission media types, including hybrid lines that comprise of cable and overhead line segments. Moreover, it has been found that the method maintains its accuracy for very long transmission lines, highly-resistive faults and different fault types.