CMStatistics 2017: Start Registration
View Submission - CMStatistics
Title: Adaptive step-stress accelerated life testing models Authors:  Christian Kohl - RWTH Aachen University (Germany) [presenting]
Maria Kateri - RWTH Aachen University (Germany)
Abstract: Under a step-stress accelerated life testing (SSALT) experiment the test items are exposed to stress levels that increase at intermediate time points of the experiment. In standard SSALT experiments the levels of the stresses applied at the various steps of the experiment are prespecified. An adaptive SSALT experiment is proposed, under which only the first stress level is fixed in advance while each subsequent stress level is chosen according to the number of failures on the immediately previous step. Assuming that the lifetimes of the testing items are exponentially distributed under each stress level, we derive the maximum likelihood estimators (MLEs) of the associated parameters and their exact distribution (conditional on their existence) under type-I and type-II censoring. Optimal test planning (i.e. optimal allocation of the stress change time points) in terms of C-, D- and A-optimality is considered for various scenarios. Several designs are compared via extensive simulation studies.