Bayesian Hit-Miss Data Nondestructive Inspection Test Planning

• Activity Number: 398
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Yew-Meng Koh*+ and William Q. Meeker
• Companies: Iowa State University of Science and Technology and Iowa State University

Abstract

Although some useful general guidelines exist for planning nondestructive evaluation studies, statistical tools provide more definitive guidance and allow comparison among different proposed study plans. It is possible to obtain expressions for estimation precision (e.g., giving the relative width of a confidence interval), providing an assessment and comparison of alternative test plans. One difficulty is that estimation precision depends on the unknown actual underlying Probability of Detection (POD) function. Engineers generally have some information about the true POD function, based on some combination of knowledge of the physics behind the inspection method or previous experience with the inspection method. If such uncertain information can be described by a probability distribution, it is natural to use a Bayesian method to do the test planning. In this paper we present Bayesian methods to find optimum test plans. Although the optimum plans have practical deficiencies, they provide insight for developing statistically efficient compromise plans that are also developed in this work.