Design of accelerated life testing using proportional hazards-proportional odds

D-optimality is adopted in the optimum design of accelerated life testing in recent years. However, for some cases, the optimization results tend to approach the extreme stress levels and processing accelerated life testing under such stress levels would not obtain much information about lifetime and reliability of the test units. This paper presents an optimum design of constant stress accelerated life testing based on proportional hazards-proportional odds using penalized local D-optimality. It establishes the objective function as the product of the Fisher information matrix as defined in proportional hazards-proportional odds model and penalty functions which describe the `closeness´ of the probability density functions of two specified stress levels. Optimization results are obtained by maximizing the objective function subject to certain constraints. This optimum method avoids obtaining limit stress levels of test planning using D-optimality for some cases. Simulation study is given finally. The comparison of the optimization results by D-optimality and penalized local D-optimality shows that optimization results using penalized local D-optimality is more reasonable.