成败型武器性能评估中的“淹没”问题研究

武器性能评估中,当真实试验数据样本量不足时,可利用Bayes方法将仿真试验数据作为先验信息.但若先验信息处理不当,就会出现大量仿真试验数据"淹没"真实试验数据的情况,使评估结果失去意义.提出一种基于全概率思想的先验信息综合方法,通过在成败型武器性能评估中的应用,表明该方法能合理考虑仿真系统可信度的影响和仿真试验数据的重要性,较好地解决了真实试验数据的"淹没"问题.

Solution to Obliteration in Assessments of Binary Weapon Performance

Abundant test data are required in assessments of weapon performance. When weapon test data are insufficient, Bayesian analyses should be considered and test data should be extended by simulations. But assessments of weapon performance will be failure when simulation test data are too much to obliterate weapon test data. A Bayesian approach on prior information based on complete probability was brought forward. The applications in assessments of success-or-failure weapon performance demonstrate that the Bayesian approach put forward considers the importance of simulation system confidence and simulation test data reasonably and solutes to obliteration in assessment.