Flaw detection on Tainter gate post-tensioned anchorages utilizing gradient boosting through wavelet decomposition feature extraction
Ray, Jason D.
Item TypeGraduate Thesis
AdvisorBall, John E.
CommitteeGurbuz, Ali C.
Netchaev, Anton D.
Embargo TypeComplete embargo for 2 years
Embargo Lift Date2022-12-15
As the nation’s infrastructure continues to age, there is a growing need for methods to safely inspect critical structures, often during operation. The failure of post-tensioned anchor rods in Tainter style flood gates presented an immediate need for new inspection capabilities for U.S. Army Corps of Engineers (USACE) managed flood control gates. In response to this need, the Sensor Integration Branch (SIB) of The U.S. Army Engineer Research and Develop Center (ERDC) developed the capability to non-destructively test (NDT) both greased and grouted cylindrical steel anchor rods using higher order guided wave ultrasonic testing. Understanding the results requires a knowledge of both guided waves and digital signal processing in order to identify the possibility of a defect. In order to both facilitate rapid defect identification and expanding ease-of-use of the equipment, the research in this thesis uses a combination of the discrete wavelet transform (DWT) and gradient boosting machine learning to build a model capable of identifying the dispersive defect responses in the rods.