AUTOMATIC MODELING OF ULTRASONIC PHENOMENA

This paper describes a procedure for the modeling of ultrasonic phenomena. The ultrasonic system is treated as an information processing system resembling a neural network comprised of an array of sensors, a memory, an estimator and an array of actuators. Using a conditional average and data which had been memorized in previous measurements with the system, the modeler can predict missing data components from incomplete input data. This procedure is demonstrated by processing simulated AE signals to solve the source location and characterization problem and to process the ultrasonic signals scattered by a fluid-filled inclusion in an elastic solid to recover the characteristics of the scatterer.