DESCRIPTION OF GROUND SURFACES BASED UPON AE ANALYSIS BY A NEURAL-NETWORK

Information regarding the surface properties produced by grinding can be obtained on-line by processing the acoustic emission (AE) signals g enerated during grinding. To demonstrate this, an experimental set-up including an adaptive measurement system with an artificial neural net work structure was developed and rested. Surface properties could be p redicted with low error during grinding with used, but not worn out gr inding wheels, while for sharp and worn out wheels, a greater predicti on error was calculated. (C) 1997 Elsevier Science B.V.