Abstract: Insulation aging of power cables is a core issue restricting the long-term safe and reliable operation of power distribution systems. Its degradation process is driven by the interactive coupling of multiple factors such as thermal stress and external environment. A single aging stress model is difficult to accurately describe the actual degradation behavior. Aiming at the typical aging modes of cross-linked polyethylene (XLPE) cables, a thermal-electrical composite aging mechanism framework and a treeing degradation path model induced by partial discharge are constructed. A quantitative time-frequency feature extraction procedure for partial discharge, a three-band frequency domain spectroscopy diagnostic rule, and a four-level Bayesian risk-minimization maintenance decision method are established. An 18 month field validation on 132 cable segments shows a 61.7% reduction in unplanned outages and an 18.4% decrease in missed defect rate.