Applications of Statistical Methods to Nondestructive Evaluation

A Statistical Framework for Improved Automatic Flaw Detection in Nondestructive Evaluation Images

“…3% for the vibrothermography applications and ≤ 1% for the ultrasonic inspections), based on the images corresponding to specimens without any flaws. In this paper, we follow Olin and Meeker (1996)'s definition of the PFA as the probability, for a particular inspection opportunity, of a flaw determination when there is no flaw.…”

“…There is substantial and long-standing interest in the development of statistical methodology and algorithms for the analysis of NDE data (e.g., see Berens and Hovey, 1981, 1982, 1983, 1984Gray and Thompson, 1986;Annis and Erland, 1989;Burkel et al, 1996;Hovey and Berens, 1988;Perdijon, 1988aPerdijon, ,b, 1989Neal and Speckman, 1993;Howard and Gilmore, 1994;Sweeting, 1995;Spencer and Schurman, 1995;Olin and Meeker, 1996;Howard et al, 1998;Aoki and Suga, 1999;Zaki et al, 2001;Legendre et al, 2001;Meyer and Candy, 2002;Zavaljevski et al, 2005;Dogandzic and Zhang, 2007;Hasanzadeh et al, 2008;Li et al, 2010;Gao and Meeker, 2012;Ng et al, 2013). Much of the NDE literature has revolved around the issues of noise reduction and comcomitant increased signal-to-noise ratio (SNR) and development of methodologies (e.g., development of wavelet methods, expectation-maximization algorithm-type methods) for better estimating the extent and probability of detection (POD) of a flaw in different techniques.…”

“…Some other attempts have been in the area of automated inspections. A comprehensive review of statistical issues and development in NDE is provided in the discussion paper of Olin and Meeker (1996). Another review of available methodology and techniques in this research area is provided in MIL- HDBK-1823A (2009.…”

Statistical methods for automatic crack detection based on vibrothermography sequence‐of‐images data

“…Olin and Meeker 2 and Spencer 3 provided an overview of statistical methods for NDE techniques. MIL‐HDBK‐1823A 1 described the standard statistical procedures for scalar NDE data analyses and Annis 4 provided an R package to implement these procedures through the maximum likelihood (ML) method.…”

Contributions to the design and analysis of nondestructive evaluation experiments

“…One can estimate the parameters γ 0 and γ 1 in the generalized linear model specied by (3.2) via maximum likelihood (ML) estimation. We use a ourescent-penetrant hit-miss dataset found in Table 1 of Olin and Meeker (1996) to obtain planning values that we will use as inputs for planning a similar study. The following Figure 3.1 shows POD vs aw size (in inches) for the given values of γ 0 = 7.47 and γ 1 = 2.19 estimated from example hit-miss data (0 ≡ miss, 1 ≡ hit) from Olin and Meeker (1996).…”