AUTOMATIC CRACK DETECTION AND CLASSIFICATION IN COMPLEX STRUCTURES

A NEURAL NETWORK ARCHITECTURE CALLED THE PARAMETRIC AVALANCHE SCHOLASTIC FILTER, WAS DEVELOPED TO SOLVE DIFFICULT PROBLEMS DEALING WITH NONLINEAR DYNAMICAL SYSTEMS. RESEARCHERS ARE APPLYING THIS NETWORK ARCHITECTURE TO AUTOMATIC RECOGNITION OF SPEECH. ADDITIONALLY, RESEARCH IN THE FIELD OF ACOUSTIC EMISSION IN VARIOUS STRUCTURES, PARTICULARLY AS APPLIED TO FATIGUE MONITORING IS BEING CONDUCTED. THE PARAMETRIC AVALANCHE IS BEING APPLIED TO THE PROBLEM OF UNAMBIGUOUS DETECTION AND CLASSIFICATION OF ACOUSTIC EMISSIONS RESULTING FROM CRACK GROWTH IN METAL STRUCTURES SUBJECT TO FATIGUE CYCLING. RESEARCHERS ARE APPLYING THE PARAMETRIC AVALANCHE NEURAL NETWORK TECHNOLOGY DEVELOPED FOR SPEECH RECOGNITION TO NOISY ACOUSTIC EMISSION DATA WITH THE SPECIFIC GOAL OF DEMONSTRATING A PROOF-OF-CONCEPT THAT CAN BE FURTHER DEVELOPED INTO A PROTOTYPE REAL-TIME FATIGUE CRACK DETECTION AND CLASSIFICATION SYSTEM.