Abstract:

An integral inference method is presented for the determination of the function relation between the binomial distribution parameter and state variables. As the crosswise information among the failure-and-success data under different states is exploited and used, the number of specimens for determining the binomial distribution parameters under different states by this method is equivalent to that of specimens under one state by the traditional method in the case of the same precision. Based on the function relation, the one-sided confidence limits and two-sided confidence intervals of the binomial distribution parameters under different states can be calculated. This method have been used in the reliability tests of aircraft and missile successfully. It shows that a great number of specimens have been saved.

Key words: binomial distribution, confidence limit, reliability, integral inference