Processing of dependability testing data

Aim. The development of the electronics industry is associated with a fast growth of the products functionality, which in turn causes increasing structural complexity of the radioelectronic systems (RES) with simultaneously more pressing dependability requirements. The currently used methods have several shortcomings, the most important of which is that they allow accurately evaluating reliability indicators only in individual cases. This type of estimation can be used for verification of compliance with specifications, but it does not enable RES dependability analysis after the manufacture of the pilot batch of equipment. That is why the task of identification of dependability indicators of manufactured radioelectronic systems is of relevance.

Methods. The paper examines the a posteriori analysis of RES dependability analysis that is performed after the manufacture of the pilot batch of equipment in order to identify its dependability characteristics. Such tests are necessary because at the design stage the design engineer does not possess complete a priori information that would allow identifying the dependability indicators in advance and with a sufficient accuracy. An important source of dependability information is the system for collection of data on product operational performance. There are two primary types of dependability tests. One of them is the determinative test intended for evaluation of dependability indicators. It is typical for mass-produced products. Another type of test is the control test designed to verify the compliance of a system’s dependability indicator with the specifications. This paper is dedicated to the second type of tests.

Results. The question must be answered whether the product (manufactured RES) dependability characteristics comply with the requirements of the manufacturing specifications. This task is solved with the mathematical tools of the statistical theory of hypotheses. Two hypotheses are under consideration: hypothesis H0, mean time to failure t*=T0 as per the specifications (good product); hypothesis H1, mean time to failure t*=T1<T0, alternative (bad product). The hypothesis verification procedure has a disadvantage that consist in the fact that the quality of the solution is identified after the test. Such procedure of hypothesis verification is not optimal. The paper examines the sequential procedure of hypothesis verification (Wald test) that involves decision-making after each failure and interruption of the test if a decision with specified quality is possible. An algorithm is shown for compliance verification of the resulting sample distribution law with the exponential rule or other distribution law over criterion χ2.

Conclusions. It was shown that the test procedure [n, B, r] ensures the quality of decision identical to that of the procedure [n, V, r] provided the testing time t is identical. Under the sequential procedure, if the number of failures r and testing time are not known from the beginning, a combined method is used (mixed procedure), when additionally the failure threshold limit r0 is defined and the decision rule is complemented with the condition: if r < r0, the sequential procedure is used; if r = r0, normal procedure is used. An algorithm is shown for compliance verification of the resulting sample w1(yi) distribution law with the exponential rule or other distribution law over criterion χ2. The paper may be of interest to radioelectronic systems design engineers.

1. Zhadnov VV, Polessky SN. Proektnaya otsenka nadiozhnosti radiotekhnicheskikh system.Nadiozhnost i kachestvo: tr. Mezhdunar. simpoz.: v 2 t. T. 1 / pod red.Yurkova NK[Engineering estimate of dependability of radiotechnical systems. Dependability and quality: proceedings of international symposium: in 2 vol. Volume 1.Yurkov NK, editor]. Penza: Penza State University Publishing; 2006 [in Russian].

2. Zhadnov VV, Sarafanov AV. O upravlenie kachestvom pri proektirovanii teplonagruzhennykh radioelektronnykh sredstv [Quality management in the design of thermally loaded radiolectronics facilities]. Moscow: Solon-Press; 2004 [in Russian].

3. Artiukhova MA, Zhadnov VV, Polessky SN. Metod ouchiota vliania sistemy menedzhmenta nadiozhnosti predpriyatia pri raschiotnoy otsenke pokazatley nadiozhnosti elektronnikh sredstv [Method for accounting of the impact of enterprise dependability management system in estimation of dependability indicators of electronic facilities]. Radioelekpronika, informatika, ouparvlinnia 2013;2:48–53 [in Russian].

4. Filippov BI.Apriorny analiz nadiozhnosti radiotekhnic heskikh sistem bez vosstanovlenia [A priori dependability analysis of radiotechnical facilities without recovery]. IzvestiaVolgGTU, seria Elektronika, izmeritelnaya tekhnika, radiotekhnika i sviaz 2015;11(176):97-103 [in Russian].

5. Levin BR. Teoria nadiozhnosty radiotekhnicheskikh sistem [Dependability theory of radiotechnical systems]. Moscow: Sovietskoye Radio; 1978 [in Russian].

6. Filippov BI. Aposteriorny analiz nadiozhnosti radiotekhnicheskikh sistem [A posteriori dependability analysis of radiotechnical facilities]. Vestnik AGTU, seria Oupravlenie, vychislitelnaya tekhnika i informatika 2015;4:81-91 [in Russian].