A Novel Power Cable Fault Location Method Based On On-line Monitoring Of Sheath Current

Power cables have been increasingly adopted in China with the progress of urbanization.The total length of power cables of 35 kV and above have reached 62,760 km at the end of 2017.Along with the rapid growth of power cables,the urban power grid structure(including the laying form and the environment of the power cables)has become more and more diversified,leading to the long fault repair time and increased power outage loss.Being different from overhead lines,power cables have the structure of multiple metal layers,hence the electrical relationship between the metal structures under the short circuit fault is more complicated.Traditional transmission-line based fault location methods could hardly be applied to the case of cable circuits to locate faults in an accurate and effective manner.Based on the analysis of structural characteristics of power cables,combined with transmission line equations,Ampere loop theorem,electric dipole theory and Sommerfeld integral,an accurate calculation model for high voltage single-core cable sheath current has been established.The fault section localization criterion and the precise fault position location criterion of the high voltage cable have been proposed.An improved method for traveling wave fault location based on unsupervised learning has been proposed for high voltage cables with missing line parameters and medium voltage three-core cables.The recognition of the arrival time of the traveling wave has been improved,so that the arrival time could be recognized accurately and effectively even in the presence of noise or with the time deviation of the signals.Finally,a fault location system has been built based on the on-line monitoring of the sheath current.The following research work was completed:(1)The structural characteristics and the functions of each layers of power cables have been analyzed systematically.An accurate calculation model for high voltage single-core cable sheath current has been established,based on transmission line equation,Ampere loop theorem,electric dipole theory and Sommerfeld integral.The variation of the induced sheath currents under fault and non-fault conditions have been studied.Research results revealed the quantitative relationship between electromagnetic coupling of the core conductor and the metal sheath under short-circuit fault conditions.The robustness of the phase difference between the sheath currents at the two ends of the cross-bonded cable section versus the fault position has been presented.(2)Based on the sheath current calculation model,the influencing factors of fault current have been analyzed systematically.The distribution of sheath currents under fault and non-fault conditions have been studied,and a novel fault location method using the phase difference of the power frequency sheath currents at the two ends of the fault section has been proposed.For long cable lines composed of multiple major sections,the fault section can be localized by the phase difference of the power frequency sheath currents at the two ends of the fault section P(section).If P(section)?(90°,270°),the section is the fault section,or it is non-fault section.Finally,the fault position is determined by the relationship between P(section)and the fault distance.(3)For the high voltage cable with missing line parameters and the medium voltage three-core cable,the unsupervised learning method has been introduced to improve the traveling wave arrival time recognition based on the traveling-wave fault location principle,thus achieving the accurate fault location.In order to reduce the random errors such as the actual possible noise and the time delay of synchronization,multiple sets of sheath currents affected by the fault have been monitored,and a high-dimensional data matrix has been formed.The t-SNE algorithm has been introduced to reduce the data dimension according to the global and local characteristics of the data,thereby eliminating errors caused by time deviation of a few signals.The DBSCAN algorithm has been introduced to achieve clustering by the density of the data,thereby reducing the effects of noise.(4)An on-line fault location system has been developed for power cables.The system has been designed to be consisted of a data acquisition module,a communication module,a fault location software installed on a cloud server,and a cable engineer terminal.The fault location test has been carried out on a medium voltage three-core cable line,and the fault location system has been installed on a 110 kV cable.The test results on the medium voltage cable verified the effectiveness of the fault location method proposed by the paper.The online fault location system installed on the 110 kV cable has been successfully triggered by a shortcircuit fault,and made the correct fault section determination.The case analysis of the fault has confirmed the effectiveness of the fault location system.