Time-Censored Test Plan Design and Risk Analysis Based on Two-Level ``Compo-nent-System'' Collaboration

In reliability tests, timed censored tests are the most widely used type. How to formulate system-oriented test plans and calculate the test plan risks are current research hotspots. In this paper, a two-level collaborative timed censored test framework of "component-system" is constructed to achieve collaborative search and risk calculation of "component-system" test schemes. Firstly, based on the assumption of exponential distribution, this paper establishes a failure probability model of series systems and derives the system-level risk integral formula considering the prior distribution of components. Secondly, an adaptive dimension hybrid integrator is proposed, which realizes the adaptive switching among three methods: sparse grid, Gauss-Legendre and Monte Carlo, and was used to solve high-dimensional coupled integrals. In addition, this study developed a two-level simulation verification process of "component-system" for the collaborative search of two-level schemes of "component-system". Finally, taking the dual-exponential component series system scenario as an example in this paper, the algorithm has completed the risk curve calculation and the collaborative optimization of the test plan. The results show that the proposed framework can get the two-level test plans while meeting the risk constraints, providing theory and algorithm for system reliability tests.

Keywords: time censored test, risk calculation, Monte Carlo Simulation, series system, exponential distribution.