Abstract:

Aiming at the problem that it is difficult to evaluate the storage life of informationized ammunition components, a component storage life evaluation method combining natural storage test data and accelerated test zero-failure data is proposed. Firstly, according to the natural storage test data of the component, the upside down problem in the data is solved by the isotonic regression, and the life distribution function of the component is initially determined by the minimum chi-square estimation method and the goodness of fit test. Then, through the optimal confidence limit method, the model parameters under the accelerated stress level are estimated, and the acceleration factor between the accelerated stress level and the conventional stress level is calculated, and the zero-failure data in the accelerated test is converted to the timing censored data under normal stress levels, and the model parameters are re-evaluated based on that data alone. Furthermore, the conversion data and the original natural storage data are combined, and the storage reliability of the components is evaluated again, and the evaluation results are comprehensively compared to determine the distribution function. Finally, taking a computing device as an example, comprehensively comparing and analyzing the evaluation results of natural storage test data, accelerated test no-failure data, and fusion storage test data, the life distribution function of the computing device and the storage life at a given reliability are determined. The effectiveness of this method is proved, which can be used to promote the engineering applications.

Key words: computing device, zero-failure data, natural storage test, storage life, optimal confidence limit