Interpretation of Nondestructive Integrity Tests Using Artificial Neural Networks

by Glenn J. Rix, (A.M.ASCE), Georgia Inst of Technology, Atlanta, United States,



Document Type: Proceeding Paper

Part of: Structures Congress XII

Abstract:

A barrier to more frequent use of nondestructive integrity tests on cast-in-place foundations is the difficulty of interpreting the test results to identify the presence or absence of a defect. One possible solution is to identify the presence or absence of a defect. One possible solution is to automate the interpretation by training an artificial neural network to classify foundations as defective or defect-free. The feasibility of this approach was evaluated by using synthetic sonic mobility curves to train a three-layer artificial neural network using the backpropagation learning algorithm. The trained network successfully predicted the presence or absence of a defect in 98 percent of the test cases.



Subject Headings: Neural networks | Foundations | Automation | Training | Feasibility studies | Defects and imperfections | Algorithms | Cast in place

Services: Buy this book/Buy this article

 

Return to search