Application of Sequential Testing to Motor Vehicle Emission Certification
Abstract
A new structure for economically-based sequential inspection plans is developed for use in motor vehicle emission testing. The structure allows for multiple attributes, multiple sources of randomness, misclassification of inspected items, and adjustment of cost parameters to produce desirable classical properties. Specific sequential plans are developed for the Environmental Protection Agency's motor vehicle certification program and are applied retrospectively to vehicles tested in the past. The use of such plans in the past could have saved resources while resulting in no significant degradation in performance. Both the structure and computational methods used have potential application to other sampling inspection problems.

