Aiming at the problem that the relative important degree of indexes will change, when urban rail power supply equipment is in different running state, this paper proposes a comprehensive degradation assessment model. The model uses the discrete degree of indexes' historical data to describe the relative importance of each indexes, and calculates the initial weights of each indexes by mean variance method (MVM). When the equipment is in normal operation state, the initial weight is corrected by the current detection data of the index, and the comprehensive degradation degree of the equipment is obtained. According to the principle of "more value punishment", when the equipment is in failure state, the weight of the index can be reassigned by the variable weight formula to realize the comprehensive assessment of the degradation degree of the equipment. Finally, the effectiveness of the comprehensive assessment method is verified by an example.