Oscillation is one of the most important stability problems of modern power system. In a power system, multiple types of oscillation crossing a broad-band may occur at a same time. In this paper, the mode identification for broad-band oscillation signal is proposed. First, the signal is decomposed to signals of different frequency bands by band pass filters. Then, the mode signals of the decomposed signals are extracted by using Variational Mode Decomposition (VMD) method. The mode parameters can be identified from the mode signals by Prony algorithm. The simulation and case study show that this method can distinguish and extract different kinds of oscillation modes of signals effectively and identify each mode’s information accurately. It can accurately identify mode information from both oscillation signal and damped oscillation in ambient noise. This work is supported by National Natural Science Foundation of China (No.51520105011) and Science and Technology Project of State Grid Hunan Electric Power Company Limited (No. 5216A5170012).