Author/Editor		Grabec, I; Sachse, Wolfgang; Grabec, D
Title		Intelligent processing of ultrasonic signals for process control applications
Type		članek
Source		Mater Eval
Vol. and No.		Letnik 51, št. 10
Publication year		1993
Volume		str. 1174-82
Language		eng
Abstract		We review here some of the approaches that have been used to apply intelligent, neureal-like signal processing procedures to solve a number of acoustic emission (AE) and active ultrasonic (micro T) process control measurement problems which can be expected to have important process control applications. Characteristic of these approaches is the use of a set of learning signals from an array of sensors to developa memory containing prototypical pattern vectors composed of acoustic data and process characteristics. This memory can subsequently be utilized to process signals to optimally recover parameters of the manufacturing process. Approaches applicable for linear and non-linear problems have been developed. For the latter, algorithms implementing a multi-layered, feed-forward neural network or alterantively, a non-parametric, multidimensional regression approach called an automatic modeler have been developed. Here we illustrate this modeler by its application to characterize a dilling process and to recognize the finish of surfaces from the signals generated using a tactile sensor.
Descriptors		SIGNAL PROCESSING, COMPUTER-ASSISTED NEURAL NETWORKS (COMPUTER) PATTERN RECOGNITION ACOUSTICS MATERIALS TESTING