A Leakage-Current Vectors Fault Location Method for Undersea Observation Networks With Poor Data Quality

For undersea observation networks, accurate fault location is required to reduce repair costs and time. A main technical challenge for this task is the presence of poor-quality data caused by the cascading failure from undersea power networks to the sensors and communication networks. This article proposes a leakage-current vector (LCV) based fault location method for the negative-pole DC network to address this issue. LCVs are obtained through the nodal admittance matrix considering the electrodes which react in the seawater to provide the return path for the negative pole. The attribute of LCVs is used to identify the anomalous, corrupted and missing data. A novel KCL-based data imputation method is proposed to correct the LCVs, which are compared with measurements to obtain the fault distance and fault resistance. The simulation model of north-east pacific time-series undersea networked experiments (NEPTUNE) is established in PSCAD/EMTDC to validate the proposed method. The results show a 23% improvement in fault location accuracy compared with data-interpolation-based methods.