Bayesian Estimation of Missile Hit Accuracy for Dirichlet Distribution Based on Multiple Stages Growth Tests

Abstract

During the period of research and design of new missiles, the hit accuracy tests are usually carried out in multiple stages according to the task progress and results. The existing estimation methods seldom consider the tests characteristics of multiple stages growth of missile hit accuracy, which bring difficulties to accurately estimate the missile hit accuracy. Considering the different degrees of damage caused by missile hitting the target in different areas, the Dirichlet distribution is selected as the prior distribution of missile hit accuracy parameters based on the Bayesian method. The sequence constraint relationship between the missile hit accuracy parameters of each stage test is established, and the Bayesian estimation model of missile hit accuracy based on the multiple stages growth tests is constructed. The Markov Chain-Monte Carlo (MCMC) method combined with Gibbs sampling are used to overcome the problem of solving the posterior high-dimensional integral of the model, and the finally estimation of missile hit accuracy is completed. The example shows that compared with the existing single stage test method, this method can consider the tests characteristics of multiple stages growth of missile hit accuracy and integrate the test information of each stage, which can provide reference for the estimation of missile hit accuracy during the period of research and development.