Power MOSFET is an important part in the main circuit of switching power supply. Its reliability has generated widespread concern. The study indicates that there are over 31% faults by cause of MOSFET in the failure statistics of circuit components. The key indicator of MOSFET failure is on-resistance parameter drift. In order to realize the online identification of MOSFET on-resistance, this paper takes Buck circuit for example, and constructs the hybrid model of Buck based on hybrid system theory. The method is presented to identify the parameters by using the Kalman algorithm. The paper conducts a series of simulation experiments in the Simulink environment. The experimental results show that this method of identifying MOSFET on-resistance is valid and that the on-resistance estimation error is less than 3% compared with the default value. Compared with the recursive least squares method, the Kalman parameter identification has better convergence. The results provide the theory basis and methods for the analysis of the MOSFET parameters failure.