Abstract: The unexpected electric breakdown during the operation of the ion thruster will affect its service reliability when it reaches the impact threshold. That is to say, these unexpected electric breakdowns sometimes cause the beam current of the ion thruster to be interrupted, and sometimes not, depending on the location, frequency and intensity of the unexpected electrical breakdown. On the premise of collecting the fault data of the unexpected electric breakdown in the ground test of the LIPS-200 ion thruster, based on the random process theory, a mathematical model of the unexpected electric breakdown of the ion thruster was established by using the non-homogeneous Poisson process, and the parameter estimation of the model was carried out in the case of interrupted beam current and uninterrupted beam current respectively. According to the AD test results, the failure intensity of the ion thruster satisfies the Weibull distribution when the beam is interrupted, and the failure intensity of the ion thruster satisfies the bathtub curve when the beam is not interrupted. Then, the instantaneous mean time between failures and cumulative mean time between failures were derived from the failure intensity function to characterize the reliability level of the ion thruster under the specified service time. Finally, the robustness of the failure intensity distribution model in two cases was tested based on K-fold cross validation. The results show that the parameter confidence intervals of cross-validation in the two cases are basically coincident, and the model has good stability. The above research results can provide a reference for further improving the reliability and optimization design of ion thrusters, and the methods of model fitting, hypothesis testing and cross-validation involved in the research process can be extended to products with the same data characteristics.