This paper proposes a detection method based on fault identification filtering for the voltage collapse of the Dobson and Chiang's power system model. It is achieved firstly by using the nonlinear state equations to represent the nonlinear systems. Then we apply the fault identification filtering design by treating the difference between the original load voltage and the linearized model's load voltage at a stable operating point as a fault vector. Finally residual signals are used to analyze if the voltage collapse happens. Since the fault identification filtering can be designed with easy calculation and low-pass filtering characteristics, it thus can be employed as a good means for the on-line detection scheme. The illustrated simulation results demonstrate the effectiveness of the proposed application.