Interval estimation of reliability indices from the results of complex system component tests

The goal of this work is to find the lower estimate no-failure probability (NFP) a complex monotonic nonrecoverable system from results independent binomial tests its components. Using general-and-probabilistic method, NFP considered as function polynomial, which linear homogeneous polynomial in each S variables where number component types. Based on method confidence sets, found minimum an unknown multidimensional parameter at aggregate test (failure-free operation) components equal one minus guaranteed coefficient. paper reports equations, for two types relates reliability derivatives (and more equation and coefficient). Conditions are initial guess numerical solution above nonlinear equations (the conditions one; condition like sign functions depends particular type probability). In some specific cases, program dimension can be reduced due simple structure polynomial. presented gives with coefficient that cannot serial-parallel or parallel-serial consist arbitrary failure time distribution. allows get small failures their absence, especial importance high-reliability systems.