ABSTRACT
Nicolae Popoviciu & Floarea Baicu Hyperion University of Bucharest, Bucharest, Romania
ABSTRACT: The work has five sections. After a short description of χ2 test of goodness of fit (section 1), section 2 describes step by step two algorithms of χ2 test to verify the hypothesis of normal distribution: algorithm 1 in specified case (m and σ2 are known values) and algorithm 2 in unspecified case (m∗ and σ2∗ are estimated values). Both algorithms are validated by simulated data. The section 3 gives the algorithm 3 for χ2 test of lognormal distribution in unspecified case. The log-normal distribution is appropriate for estimation of reliability parameters for semiconductor devices. Application of algorithm 3 is the aim of section 4, where experimental data are used. The log-normal hypotheses H is accepted. The conclusion shows that the risk can be managed trough designing of accelerated ageing based on precise input data and a very good mathematical model.
