Abstract
Purpose . This article describes the issues of planning testing scope for high-reliable objects. The development and manufacture of new samples of equipment is accompanied by a task to define their reliability characteristics. It is based on the fact that there are requirements related to the necessity to specify the above mentioned characteristics in certificates and technical descriptions of the products supplied to the market. The most objective way to define reliability characteristics of the products is a field test. But under the manufacture of complex expensive objects there is no opportunity to introduce a batch with lots of finished products for testing. Thus there is a task to define the duration of field testing and scope of products to be tested, provided there are requirements for the accuracy of estimations related to the objects’ reliability characteristics obtained as the result of testing. Planning of the scope is based on the requirements of a manufacturer related to a necessity to confirm the value of lower bound of reliable operation probability with a predefined confidence level. Two tasks are solved in this work. The first task is to define the scope of testing of a batch with finished products N 0 for a time moment t 0 , for which a customer’s requirement would be fulfilled related to the achievement of the lower bound of probability of reliable operation, specified with a confidence probability 1 - α . This task is solved using a поп-parametric approach. The second task is to define a required scope of test Nt1 of the equipment of this type for the time moment different from the moment of first studies t 1 ≠ t 0 . Here one solves the question: how are N t0 and N t1 correlated? The scope of tests N t1 is defined based on the determination of confidence levels providing with the same accuracy of indices as in point t 0 . This task is solved with a semiparametric approach. When solving the second task, the parameterization of mean time to failure distribution is used. Three types of distribution are studied: exponential law, Weibull distribution and distribution with linear function of a failure rate. The considered types of distribution laws help to study the behavior of the objects with a decreasing, constant and increasing function of failure rate. Methods. The formulas for calculation of test scope for different durations of a test-run are derived. Dependence of scope on the duration of a test-run and on a real level of probability of reliable operation is studied as well. Scope planning and respective studies are carried out for different behavior models of a failure rate of the product. Conclusion . Obtained results give the basis for a well-reasoned approach to the planning of scope of tests of high-reliable objects. The study results showed that the longer a test-run is, the fewer objects are required to be introduced for a test. Dependence is non-linear; it is specified by the parameterization of the failure rate function. Analogous dependence was also obtained for the probability of reliable operation: the higher the PRO is, the fewer objects are required to be tested.
Highlights
Наиболее объективным способом определения характеристик надежности изделий является проведение натурных испытаний
Данная задача решена с помощью непараметрического подхода
Кандидат физико-матема тических наук, ведущий специалист, АО «Государственный научный центр Российской Федерации – Физико – энергетический институт им
Summary
После ознакомления с данными рассуждениями можно поставить первую задачу исследования. Определить объем испытаний партии готовой продукции N0 такой, что для любого значения P0* ≥ P0, заданного с доверительной вероятностью 1 – α, будет выполняться соотношение для требуемого объема испытаний N0* ≤ N0. При решении задачи будем предполагать, что функция интенсивности отказов задается одной из формул [1]: λ(t) = λ;. Выражение (1) (интенсивность постоянна) характерно для экспоненциального распределения наработки до отказа, формула (2) – для функции распределения с линейной интенсивностью отказов и функция (3) – для закона распределения Вейбулла. Где g(t)=1 соответствует экспоненциальному распределению, g(t)=a+bt – распределению с линейной функцией интенсивности отказов,. Функция интенсивности g(t) должна удовлетворять двум основным требованиям: g(t)≥0, при t→∞. Что коэффициенты a, b в (5), (6) известны, неизвестным и оцениваемым по выборке является параметр λ
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have