DOI: 10.1615/ICHMT.1992.IntForumExpSysCompSimEE
ISBN Print: 978-1-56700-486-1
A HYBRID EXPERT SYSTEM FOR EDDY CURRENT TEST OF STEAM GENERATOR TUBES
要約
A hybrid expert system, which means an expert system combined with neural networks, is built to automatically evaluate eddy current (EC) signals generated during steam generator (S/G) tubes inspection. The system consists of three systems, i.e., syntactic pattern recognition subsystem, neural network-subsystem and rule based production subsystem. The syntactic pattern recognition subsystem makes it easy to process the vast EC signal data, screens EC signals and detects event signals such as defect signals and structural signals. The neural network subsystem is useful to classify the event signals which often contain noise signals. The expert system implemented on HP 9000/370 workstation also supplies a good EC test data management function.