| Until recently, ultrasonic inspection has relied upon some very basic concepts, making use of, for example, a pulser unit, a piezoelectric transducer, and an oscilloscope display. These components were excellent for carrying out thickness measurements, and for determining defect locations. There exists today, however, a great need to do work in flaw classification, a topic that can only be covered by using computer technology, in particular utilizing aspect of signal processing and pattern recognition. Signal processing and pattern recognition have just been introduced to the ultrasonic inspection field and development has just entered a research and development phase. With this marriage, a large number of problems in both software and hardware and in demonstrated utility must still be solved. Current work in this area is presently being carried on in only a very few laboratories.;The thesis consists of two major parts, the first describes the availability of suitable tools for analysis, and the second in describing two sample problems that can be handled with this integrated technology. The two problems are related to weld inspection, the first on welded plates, and the second for K-joint damage analysis in an offshore structure. This work culminates in the development of a system software package called GANPUI, an acronym for a Generalized Approach to New Problems in Ultrasonic Inspection.;Tools for carrying out this research have been embedded in other technologies for many years. The purpose of this thesis is to draw upon these different technologies that utilize signal processing and pattern recognition in an attempt to use these desciplines to advance the state of the art in flaw and material classification in ultrasonic inspection. |
