Small sample verification method for initiating explosive device reliability index

Abstract

The invention provides a small sample verification method for the initiating explosive device reliability indexes. The reliability index of the initiating explosive device can be verified only by using the sensitivity distribution mean value of the initiating explosive device, and the defect that reliability evaluation needs to be carried out according to the mean value and the standard deviationat present is overcome. The test sample amount under the working stimulus amount is equivalently converted into the test sample amount under the low stimulus amount by utilizing the sensitivity distribution mean value, so that the method for verifying the reliability of the initiating explosive device by using less sample amount is realized on the premise of keeping the confidence level unchanged.By adopting the method, the sample size required by reliability evaluation of the initiating explosive device with the reliability index of 0.999 can be reduced to be less than 1 / 10 of the sample size required by a classical statistical method, and the test cost is greatly reduced. Test information of a metering method and a counting method is integrated, on the premise that the confidence levelis kept unchanged, the number of test samples is greatly reduced, inference errors are reduced, and the accuracy of a reliability evaluation result is improved.

Description

technical field [0001] The invention belongs to the technical field of reliability index verification, and in particular relates to a small-sample verification method for the reliability index of explosive devices. Background technique [0002] Pyrotechnic products are sensitive products with high reliability indicators. Generally, they are verified by the counting method test specified in GJB376-1987, and the test sample size is relatively high. The estimated value is conservative and the test volume is large. When the reliability of pyrotechnic products is required to be 0.999, it is often impossible to implement it in engineering due to the large amount of tests. For example, under the conditions of confidence level 0.90 and reliability index 0.999, it is necessary to test 2303 rounds without failure to meet the technical index requirements. At this time, measurement methods are generally used, such as the sensitivity test method for pyrotechnic products specified in GJ...

AI Technical Summary

Problems solved by technology

The advantage of the econometric method is that it uses the information of the product sensitivity distribution model, and the amount of experiments required is less. Its disadvantage is that the deviation of the parameter estimation of the characteristic parameters and the wrong choice of the distribution model will cause a large error in the reliability estimation, and these The method can only be used to infer or predict the reliability of pyrotechnics, and cannot be verified by experiments

GJG377A / 94 proposed a sensitivity test method for pyrotechnics, using the sensitivity distribution model to infer the reliability of pyrotechnics. The estimation is unbiased, but the estimation of the scale parameters of the sensitivity distribution is biased. Therefore, this inference method has large errors and cannot be verified by experiments. Strictly speaking, it cannot be used to evaluate the reliability, but can only be used to predict

In the early 1990s, the Americans proposed the "Neyer D-optimization method", which is also a measurement method, and the error is relatively large when it is used to infer the sensitivity distribution scale parameters

[0003] GJB6478-2008 provides a small sample counting-measurement comprehensive evaluation method for pyrotechnics, the test sample size is reduced by about 9 / 10, but this method also needs to use the estimated value of the sensitivity distribution mean and scale parameters, although this method uses is the estimated value of the scale parameter of the sensitivity distribution after correction, but the estimated value of the scale parameter after correction is not the true value, so the accuracy of the reliability evaluation result cannot be guaranteed

Images