Fuze Arming Distance Mathematical Theory And Simulation Study Of The Statistical Test Methods

Arming distance is an important safty index of the fuze, in order to ensure the index requirement of the fuze, the traditional method to assess fuze arming distance is to fixed target distance for live firing, this method consumes large amount of ammunition, and fuze arming distance value measured is less accurately. The introduction of mathematical statistics test method is an effective solution to this problem. For mathematical statistics test method occurring data processing problems in fuze anning distance test, including the required more sample size, more limit estimation error, the test results distorting due to fuze arming failing, abnormal values being difficult to accurately remove, and the test data distribution model not meeting method assumptions, it was made detailed analysis and computer simulation studies for mathematical statistics test method with examples, such as Probit method, Langlie method, Up and Down method and One-Shot Transformed Response method, etc.Under the condition that confidence level is 95% and variance is known,the mean of sample meets maximum likelihood estimation and mean testing conditions is satisfied, Probit method achieves mean accurate of the minimum sample size, and the size is 75.The minimum sample size of Langlie method tests is related to the mean accuracy, while confidence level is 95% and variance is known and the number of sample is [8,100). the result of variance devises mean is (0.20,0.25),the mean of sample meets maximum likelihood estimation, it ranges from 8 to 24.Up and Down method is special case of the group Up and Down method, analysis for the sequential sensitivity test method, as one of the group Up and Down method, shows that the accuracy of normal distribution mean and variance valuation is greater affected by response probability, the accuracy of estimated value of test probability point by the partly estimated average test is higher than the distribution fitting, but variance estimation system is lower.The response probability with Fuze arming distance One-Shot Transformed Response method is greater impact on test parameter estimation accuracy, the farther deviating from the 50%, the lower for test parameters estimation accuracy. The best threshold width in test is 6 to 18 times than the variance; the most applicable fuze arming distance distribution is lognormal distribution for One-Shot Transformed Response method test, following for the logistic distribution and normal distribution, the adaptability to logistic special distribution is lower.The application for information entropy discrimination method and gray theory discrimination method achieves that mathematical statistics test methods, such as Langlie method etc. can remove abnormal values in the case of small sample, that is the effectiveness determination of the test data.The application for condition probability method achieves the estimation of response probability in the case of One-Shot Transformed Response method existing the failure. By Monte-Carlo method computer simulation and real example, it founds that if response probability less than 0.5 the estimation of response probability decressed.The application for Probit method achieves Up and Down method test data processing for the same type but different batches fuzes arming distance; the application for illustrative diagram predicted test and accurate test achieves the fuze arming distance distribution test.