by Tiziana Segreto, Sara Karam, Jes Ramsing, Roberto Teti
Abstract
Polishing processes have evolved from being a manual operation to become an automated process based on a robotized arm. Aiming for a more accurate, robust, and reliable polishing process, sensor monitoring would be viable for quality and process control. In this paper, an acoustic emission sensor monitoring system was employed for surface roughness assessment during robot assisted polishing operations on steel bars. Two feature extraction procedures were applied to the detected acoustic emission signals: statistical and wavelet packet transform in order to obtain the relevant features to be inputted in a neural network pattern recognition paradigm for the correlation between the acoustic emission signals and the measured surface roughness.
1-minute pitch
| Paper presenter | |||||||||
 |
|
||||||||
As the aim of the research was to improve the accuracy of the polishing process, what is the measure of this accuracy, was it the surface roughness?
If the answer is yes, have you bench marked your results with other results and what was the outcome?
Very interesting work!
How about influences resulting from the limited repeatability of the robot? Do you see such influences in the AE-signals? Does it affect the process?
Why did you use Daubechies 3 as mother wavelet?
Have you taken into consideration the possibility of using other sensor types in the sensor monitoring of the polishing process?
On what basis were the binary values of the output of the neural network assigned?
On what basis was the 0.07 threshold value chosen?
On what basis were the statistical features calculated from wavelet packet coefficients selected?