装备测试性增长过程的贝叶斯验证方法

现有测试性验证试验方案需要较多故障样本,存在测试性验证时间长、结果置信度低的问题,为此提出一种装备测试性增长模型的贝叶斯验证方法,在小子样、异总体的情况下进行测试性验证。建立装备测试性水平动态增长的数学模型,依据已有的研制阶段的数据进行求解,得到测试性增长模型的具体参数值,并对测试性水平进行预测。确定测试性水平的先验分布,使用最大熵法求解分布参数,采用贝叶斯理论融合现场试验数据得到后验分布,实现对装备测试性水平的验证。实例分析表明,该方法在较少样本量的情况下能够提高验证结果的置信度,降低双方风险。

Bayesian Verification Method Based on Equipment Testability Growth

The method currently used to carry out testability verification under the small sample, varying population condition inevitably has a large number of samples, long verification time, and poor confidence interval. Thus, a new Bayesian verification method is proposed based on the equipment testability growth model. The parameters are calculated from its dynamic growth mathematical model using the existing data. According to the prediction value based on the above method, it can determine the prior distribution and calculate the distribution parameter value using the maximum entropy method. The posterior distribution is obtained from field test data using the Bayesian theory so that the evaluation and demonstration of the testability index can be realized. Practical examples indicate that this method can verify the system with fewer samples at higher confidence, which helps reduce risks for both the consumer and producer.