Condition Monitoring of 154 kV HTS Cable Systems via Temporal Sliding LSTM Networks

Abstract

High-temperature superconducting (HTS) cables are expected to be installed in cable tunnels that are already constructed in urban districts. Therefore, the installation of normal joint boxes is inevitable, and it is necessary to develop a diagnostic methodology that considers both the existence of the joints and the electrical characteristics of HTS cables. In this work, temporal sliding long short-term memory (TS-LSTM) is proposed to estimate the locations of the joints that can be hidden by multiple reflections. TS-LSTM includes short-term TS-LSTM and long-term TS-LSTM for analyzing various time dependencies. The reflected signals of the actual joints, which are distinguished from multiple reflections, are analyzed via the chirplet transform (CT) which is one of the time-frequency (TF) analysis methods. The proposed condition monitoring method is applied to an AC 154 kV 600 MVA HTS cable system (1 km) connected to a real power grid network in Jeju, South Korea. For the validation of the proposed methodology, the dielectric and electrical characteristics of the 154 kV HTS cable system are monitored during the cooling process.

Keywords

• Condition monitoring
• long short-term memory (LSTM)
• superconducting cable
• temporal sliding LSTM (TS-LSTM) networks
• time-frequency analysis