Optimal sequencing of inspection of dependent characteristics

We address the problem of sequentially inspecting the dependent characteristics of a product, where the dependency is expressed in terms of the joint probabilities of the fitness of the characteristics. We show that, even when the inspection has classification errors, the joint probability mass function of the observed fitness of the characteristics is independent of the sequence of inspection. Using this result, a dynamic programming approach is presented for finding the optimal sequence that minimizes the expected total cost of inspection. Previously reported policies for independent characteristics are shown to be special cases of the results presented here.