Considering the complex operation mechanism of the accelerometer and the small number of test samples, this paper proposes a health state assessment model based on belief rule base(BRB) and ballistic simulation. Firstly, an initial health state assessment model is established based on BRB. Due to the subjectivity of qualitative knowledge, the initial health state assessment model is difficult to provide accurate assessment results, so it needs to be optimized. However, the real value of health state required for optimization is generally given by experts, which are fuzzy and imprecise. Therefore, according to the physical model of the accelerometer in the navigation process, the real value of its health state is accurately calculated based on the trajectory simulation method. Based on the covariance matrix adaption evolution strategy(P-CMA-ES), the real value of health state is used to optimize the initial evaluation model, so as to accurately and quickly evaluate the health state. Finally, the effectiveness of the proposed method is verified by the health assessment of an accelerometer.