Quantitative Rust-under-paint Detection Utilizing Near-field Microwave Nde Techniques

Near-field microwave NDE systems utilizing open-ended rectangular waveguides constitute a competent candidate to detect and evaluate planner rust layers under paint coatings. Basically, the waveguide illuminates the specimen with microwave signals and monitors the reflected waves. Minute variations in the structure reflect in measurable variation in the reflection coefficient at the waveguide aperture. The functional dependence of reflection coefficient on the rust layer physical properties—i.e. thickness and depth—is exploited in the detection scheme. Upon measuring the reflection coefficient, the inverse problem of rust thickness and depth determination should be solved. This problem is ill-posed in nature and requires sophisticated algorithm to be inverted quantitatively. In this paper, we introduce a Maximum-Likelihood algorithm to be applied in conjunction with multi-frequency measurements to solve the inverse problem. As it will be shown, the multi-frequency measurements provide diversity gain over the uncertainties embedded in the system. The practical potential of the proposed algorithm will be demonstrated in real life rust under-paint detection problem. Finally, the performance of the algorithm in noisy environment will be simulated and analyzed. It will be shown that the proposed algorithm provides significant accuracy with high sensitivity in determining the rust layer’s thickness and depth. Introduction: As far as the ultimate purpose of any NDE device is concerned, detection alone is not enough to achieve that purpose. Today, the NDE device is ought to provide the inspector with quantitative assessment of the specimen under inspection [1]. In the rust detection problem, there are two parameters of interest for assessment; the depth and thickness of the rust layer. Rectangular waveguide-based Near-field microwave NDE systems have shown promising results in detecting corrosion layers under paint coatings [2] [3] [4]. In this paper, a general algorithm to determine the rust layer’s depth and thickness from system measurements is presented. Problem Description: Given a certain specimen to inspect, we need to determine the rust layer’s thickness t and depth d . The assessment of both parameters should live up to axial resolution of R . Furthermore, let’s assume that we are interested in detecting both parameters in finite ranges. Then, given the required resolution, there are finite sets of possible rust thicknesses and depths of interest for detection. Mathematically, this is described as follows. , ,... 2 , 1 , ,... 2 , 1 , , | | and | | } ,..., , { and , } ,..., , { 2 1 2 1 M j N i j i Where R d d R t t d d d d t t t t