The Partial Discharge Evolution Characteristics of 10kV XLPE Cable Joint

• Published in: IEEE access Vol. 11; pp. 108680 - 108687
• Main Authors: Tian, Fuqiang, Li, Xubin, Zhang, Shuting, Cao, Jinmei
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Accelerated aging; accelerated electrical aging; Acoustics; Back propagation networks; BP neural network; Cable insulation; Cable shielding; Cables; Discharge; Electric potential; Evolution; Insulation; Maintenance; Mathematical models; Neural networks; Parameters; partial discharge; Partial discharges; Power cables; Predictive models; Stable oscillations; state prediction; Voltage; XLPE cable
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2023.3321805

The evolution of partial discharge (PD) with time can provide a deep understanding on the insulation status of power cables. It is of great significance for intelligent operation and maintenance of power cables. In this paper, the PD pulse signal in the 10kV cable joint during accelerated electrical aging under 20kV AC voltage was acquired in the real-time for about 160h. The characteristic parameters of partial discharge-pulse number, average voltage, maximum voltage and energy per second were extracted. The results show that the phase of partial discharge is mainly concentrated at 30°-90° and 200°-270°, which can be characterized as internal discharge. PD characteristic parameters gradually increased after 50h. The pulse number, energy per second and the average voltage of PD pulse reached a peak between 60-80h. Then these parameters reached a steady state between 80-130h and showed a steep rise after 130h. The maximum voltage of PD pulse shows a steep rise at about 70h from 0.1V to 0.3V. It rises sharply from 0.3V to 0.5V after about 120h and then enters a relatively stable oscillation stage. The evolution rules of the PD characteristic parameters comply well with the electrical tree growth states-initiation period, lag period and rapid growth period. Furthermore, model for predicting and evaluating the insulation state based on BP neural network are established and the prediction accuracy is verified. The proposed models can provide early warning for the cable joint before the insulation failure, so as to ensure timely maintenance or replacement.