Abstract: Temperature and ampacity were considered as two key parameters characterizing the operating state and thermal stability of power cables. As a critical factor affecting the calculation of cable temperature fields and ampacity, the soil thermal conductivity was strongly influenced by the environments and operating conditions, which might lead to noticeable deviations in conductor temperature and ampacity calculation when nominal parameters were adopted in engineering design. To address this issue, a mathematical model and a solution methodology for multi-physics inverse problems of real-time soil thermal conductivity inversion for cables were developed in this paper. First, a distributed optical fiber temperature measurement system was designed and implemented to acquire the real-time temperature of the cable outer sheath during operation. Then, based on the measured temperatures at sample points on the cable outer sheath, an inverse model for soil thermal conductivity was proposed for cables laying in the duct. Finally, an improved particle swarm optimization algorithm was proposed to solve this inverse problem. It was demonstrated by the numerical results that the real-time ampacity calculated by real-time soil thermal conductivity could be increased by approximately 41.85% compared with the nominal ampacity, and the effectiveness and engineering applicability of the proposed method were validated.