ON THE PLANNING OF THE SCOPE OF NEW TECHNOLOGY TESTING

This paper is a follow-up to [1]. It examines the matters of planning of the scope of highly dependable objects testing. The process of new technology development and manufacture involves determining its dependability indicators. The most objective method of identifying dependability characteristics of products is field testing. One of the widely used testing plans is the [N,U,T] plan. This plan that involves testing N nonreparable samples within the time interval between 0 and a certain T. It is assumed that during the tests k objects fail, while N-k objects successfully pass the tests. Thus, at the outcome of the experiment we have a mixed sample that includes k times to failure and N-k right censored observation. If the tested object is highly dependable it is quite possible that within the time period [0,T] failures will not happen, i.e. k will be equal to 0, therefore the probability of failure within this time period is extremely low and the number of tested objects is limited. Nevertheless even in this situation it would be desirable to be able to be in control of the accuracy of the estimation obtained during such experiments. It is clear that the accuracy of such estimation will depend not only on the number of tested objects N, but also on the experiment duration. For a fixed N, as the observation time T grows the estimation accuracy increases due to the increasing proportion of complete times, while the proportion of censored ones goes down. It should be noted that when we talk about identifying the dependability characteristics of complex and costly objects we cannot test large batches of finished products. Therefore the problem consists in defining testing duration and size of the product batch to be tested subject to specified requirements for the accuracy of estimation of dependability characteristics obtained as the result of the tests. The scope planning is based on the manufacturer’s requirement to validate the lower bound of the probability of no failure P0 with a specified confidence level at a certain time point t0. The aim of the paper is to identify the test scope of a batch of finished products N(T) under the condition of fulfilment of the manufacturer’s requirement for compliance with the lower confidence bound of the probability of no failure with a specified confidence level 1 – α. Three failure distributions are under examination: exponential distribution law, Weibull distribution and distribution with linear rate function. The considered types of distribution law enable the research of objects with decreasing, constant and increasing failure rate function. Methods. In this paper the authors deduce formulas for calculation of the scope of experiment for a number of experiment durations. The estimates are obtained using the maximum likelihood method and methods of researching asymptotic properties of estimates through the Fisher information quantity. Conclusions. The findings allow for a substantiated approach to planning the scope of highly dependable objects testing. It is shown that the longer is the experiment duration the fewer products must be supplied for testing. The dependence is non-linear, close to hyperbolic and is conditioned by both the input parameters and the parametrization of the failure rate function.

test scope planning, experiment duration, probability of no-failure, failure rate, lower bound estimate of probability of no-failure, confidence level

1. Antonov AV, Chepurko VA, Chekhovich VE, Ukraintsev VF. Regarding the planning of testing scope for new equipment samples. Dependability 2016;(3):3-7. DOI:10.21683/1729-2640-2016-16-3-3-7.

2. Antonov AV, Nikulin MS, Nikulin AM, Chepurko VA. Teoria nadiozhnosty. Statisticheskie modeli: Uchebnoie posobie [Dependability theory. Statistical models: A study guide]. Moscow: INFRA-М; 2015 [in Russian].

3. Gnedenko BV, Beliaev YuK, Soloviev AD. Matematicheskie metody v teorii nadiozhnosti [Mathematical methods in the dependability theory]. Мoscow: Nauka; 1965 [in Russian].

4. Beliaev YuK, Bogatyrev VA, Bolotin VV et al. Ushakov IA, editor. Nadiozhnost tekhnicheskikh system: Spravochnik [Dependability of technical systems: Reference book]. Moscow: Radio i sviaz; 1985 [in Russian].

5. Antonov AV, Nikulin MS. Statisticheskie modeli v teorii nadiozhnosti; Ucheb. posobie [Statistical models in the dependability theory: A study guide]. Moscow: Abris; 2012 [in Russian].

6. Zarenin YuG, Stoyanova II. Opredelitelnye ispytania na nadiozhnost [Determinative dependability testing]. Moscow: Izdatelstvo standartov; 1978 [In Russian].

7. Cramer H. Matematicheskie metody statistiki [Mathematical methods of statistics]. Moscow: Mir; 1975 [In Russian].

8. Antonov АV. Sistemny analiz; Uchebnik dlia vuzov [System analysis. Textbook for higher educational institutions]. Moscow: Vysshaya Shkola; 2008 [in Russian].

9. David H. Poriadkovye statistiki [Order statistics]. Мoscow: Nauka. Main office of physics and mathematics literature; 1979 [in Russian].