导弹落点散布的Bayes试验鉴定优化设计

Optimal design for bayesian testing assessment of the impact point dispersion of missile

针对工程实际中所普遍存在的小子样问题 ,研究了落点样本量、验前信息可信度以及 Bayes决策风险三者之间的关系 ,得到了有意义的结论 ,即 :在给定 Bayes决策风险上限的条件下 ,试验次数越少 ,则对“验前信息可信度”的要求就越高。在此基础上 ,通过算例说明了“最优样本量”概念在实际工程应用中的不适应性 ,为尽量减少试验次数 ,提出了“次优样本量”的概念 。

Considering the problem of small samples in engineering,the relation of impact point sample size,prior information credibility and Bayesian decsion adventure was studied.A valuable result has been obtained.It is that the fewer of the test number,the higher of the prior information credibility should be asked under the condition of given the limitation of Bayesian statistical decsion adventure.Based on the result,the engineering suitability of optimal sample size was discussed by using calculation example.In order to decrease the test number,a new concept,we call it secondary optimal sample size,is suggested,and a new method of determining droppoint sample size is presented.