Prediction of surface roughness in magnetic abrasive finishing using acoustic emission and force sensor data fusion

Abstract

The configuration of automated polishing systems requires the implementation of monitoring schemes to estimate surface roughness. In this study, a precision polishing process - magnetic abrasive finishing (MAF) - was investigated together with an in-process monitoring set-up. A specially designed magnetic quill was connected to a CNC machining center to polish the surface of Stavax (S136) die steel workpieces. During finishing experiments, both acoustic emission (AE) signals and force signals were sampled and analyzed. The finishing results show that MAF has nanoscale finishing capability (up to 8nm in surface roughness), and the sensor signals have strong correlations with parameters such as the gap between the tool and workpiece, feed rate, and abrasive size. In addition, the signals were utilized as input parameters of artificial neural networks (ANNs) to predict generated surface roughness. To increase accuracy and resolve ambiguities in decision making/prediction from the vast amount of data generated, sensor data fusion (AE + force)-based ANN and sensor information-based ANN were constructed. Among the three types of networks, the ANN constructed using sensor fusion produced the most stable outcomes. The results of this analysis demonstrate that the proposed sensor (fusion) scheme is appropriate for monitoring and prediction of nanoscale precision finishing processes.