Modeling of High-Temperature Superconducting Cable via Time Domain Reflectometry and General Regression Neural Network

When a fault occurs in the high-temperature superconducting (HTS) cable, a fault current consisting of extremely high frequency components is generated and propagated based on the transient response of the HTS cable. Therefore, a simulation model based on high frequency characteristic of the HTS cable is essential to obtain accurate simulation results in the transient fault analysis. In this paper, the result of time domain reflectometry is used to design the model of the HTS cable. To determine model parameters, a general regression neural network based on the kernel density estimation is utilized. After the modeling procedure, the accuracy of the model is evaluated by time-frequency domain reflectometry, whose response depends on the high frequency characteristic of the cable. It is expected that the proposed modeling method can be applied to various application area of HTS cable, such as fault analysis, protection, and diagnostics in the future.