加速破壞性衰變試驗常用來評估一次性產品之可靠度資訊。而規劃此類型的計畫中，模型的假設、量測時間的選擇、應力水準的設定及其樣本配置的比例，皆會影響產品可靠度推論的精確性。本文將常用的加速破壞性衰變模型，加入具描述時間相關性之隨機過程，來配適實際資料以增加模型解釋之能力。利用類蒙地卡羅 (qusai-Monte Carlo) 方法來估計未知參數，且提供簡易流程來驗證衰變模型之適合性。最後，在模型參數是可辨別的條件下，提出最少試驗組合之D-最佳化實驗計畫。;Accelerated destructive degradation tests (ADDTs) are commonly used to assess the reliability information of one-shot products. The accuracy of the reliability inference can be affected by the settings of test plan, including model assumptions, measurement times, stress levels and sample size allocations. This thesis takes stochastic processes into consideration in the traditional ADDT model to improve the ability of model interpretation. The qusai-Monte Carlo method is used to estimate the unknown parameters and a simple model-checking procedure is provided to assess the validity of different model assumptions. Finally, the D-optimal test plan with minimum run-size is proposed under the model assumption with identifiability.
