OPTIMAL CLASSIFICATION OF HIGHLY RELIABLE PRODUCTS WITH A LINEARIZED DEGRADATION MODEL

In manufacturing a product, some parts (or components) of this product may be purchased from vendors. For the parts (or components), classifying the better and worse designs among several competing designs is helpful for the manufacturer to enhance the product's reliability. It is a great challenge for the manufacturer if these completing designs are highly reliable, since there are few (or even no) failures can be obtained by using traditional life tests or accelerated life tests. In such a case, degradation tests will be effective techniques to assess the products' reliability if degradation measurements relating to reliability can be observed. This paper proposes a systematic approach to the classification problem where the products' degradation paths satisfy a linearized model. An intuitively appealing classification rule is proposed. Then, with respect to the objective of minimizing the total experimental cost, the optimal test plan (including the sample size, inspection frequency, and the termination time needed by the classification rule for each of competing designs) is derived by solving a nonlinear integer programming with a minimum probability of correct classification and a maximum probability of misclassification. Finally, an example is provided to illustrate the proposed method.