5 - A Bayesian Statistical Inference Approach to the Non-Destructive Inspection of Composite Material

Abstract

This chapter focuses on the usage of an electromagnetic nondestructive testing (NDT) method for the inspection of composite materials such as composite panels and helicopter rotor blades. Infrared thermography, electromagnetic testing, radiography, and visual examination have been carried out on different samples provided by the aerospace industry. Data fusion and integration have been performed using the Bayesian statistical theory, thresholding, and ensemble averaging to help in decision making in regard to the presence and location of defects, and to improve the signal-to-noise ratio. The theory of Bayesian inference is described through Bayes' rule and an example of binary decision making in regard to disbond detection in a helicopter rotor blade, using multiple eddy current sensors is presented. The results of binary decision analysis show that combining information from multiple sensors can be used to help in decision making by reducing uncertainty and verifying the degree of support of a hypothesis. The final outcome improves the reliability of information and facilitates defect location and characterization in a statistical and probabilistic manner.