Non-Destructive Assessment of Gamma Radiation Aging in Nuclear Cables via New Dielectric Spectroscopy Markers and Machine Learning Algorithm

Abstract

Low-voltage instrumentation and control (I&C) cables in nuclear power plants are continuously exposed to gamma (γ) radiation within containment areas, leading to cumulative degradation of their polymer insulation over decades of operation. Since conventional mechanical aging assessments are destructive, this study establishes a non-destructive diagnostic framework using high-frequency dielectric spectroscopy. Cable samples with ethylene propylene rubber (EPR) insulation and chlorosulfonated polyethylene (CSPE) jackets were subjected to controlled γ-irradiation at doses up to 1200 kGy. The broadband dielectric response was analyzed along with derived novel diagnostic parameters from capacitance and loss tangent spectra and a machine learning AI algorithm. The results show a strong, material-dependent relationship between radiation dose and dielectric indicators. For EPR insulation, the central capacitance (CC) and (C × F × LF) exhibit high positive sensitivity for Black and White EPR materials, respectively, whereas for CSPE jackets, the central frequency (CF) shows a pronounced monotonic decrease with the radiation exposure. These findings enable a straightforward, transparent interpretation of dielectric data and implement a new, accurate method of irradiated cables diagnosis.