Process monitoring for intelligent manufacturing processes - Methodology and application to Robot Assisted Polishing

Process monitoring provides important information on the product, process and manufacturing system during part manufacturing. Such information can be used for process optimization and detection of undesired processing conditions to initiate timely actions for avoidance of defects, thereby improving quality assurance.

This thesis is aimed at a systematic development of process monitoring solutions, constituting a key element of intelligent manufacturing systems towards zero defect manufacturing. A methodological approach of general applicability is presented in this concern.The approach consists of six consecutive steps for identification of product Vital Quality Characteristics (VQCs) and Key Process Variables (KPVs), selection and characterization of sensors, optimization of sensors placement, validation of the monitoring solutions, definition of the reference manufacturing performance and a data driven process validation for each manufactured part.

The concept is based on conscious identification and monitoring of KPVs that are closely related to part VQCs and measurable during manufacturing, thereby enabling in–process quality control (QC). The approach was applied during the development of process monitoring andcontrol strategy for automatic process End Point Detection (EPD) and on the machine surface characterization in Robot Assisted Polishing (RAP) with oscillating tool.

VQCs were identified in terms of surface roughness, defects and gloss. Polishing progression in terms of relative variation in surface roughness was indirectly monitored through identified KPVs in terms of Acoustic Emission (AE), friction forces and power consumption during polishing. A dedicated polishing arm with integrated strain gauge based force sensors and a miniature AE sensor was developed, enabling in–process measurements in RAP with stationary and rotating workpieces.

A commercial scattered light sensor was used for on the machine characterization of polished surfaces. The developed monitoring solutions were validated in a number of experimental tests in coarse stone and fine paste polishing. The results demonstrate the suitability of indirect monitoring of surface generation through AE and friction forces during polishing enabling automatic EPD.

AE signal was found closely related to the Material Removal Rate (MRR). Stabilization in measured friction forces was observed to reflect the stabilization in the mean slope of the surface topography and the overall friction condition in the tool–workpiece interface. Real time AE and force measurements also enable monitoring of the process state, allowing early recognition of process malfunctions and initiation of timely actions to avoid occurrence of defects.

Process control strategy was developed based on an automatic detection of steady-state levels of AE and friction forces, reflecting the stabilization of surface roughness. The on the machine scattered light measurement method was demonstrated to provide high measurement rate allowing 100% QC, recognition and localization of macro as well as nm rage surface defects.

A robust correlation between the scattered light roughness parameter Aq and hybrid roughness parameter Sdq used to describe the surface gloss was found. Also the typical asymptotic trend in surface roughness during polishing was found in a good agreement with the trend in Aq parameter. The developed solutions for in–process EPD, process state monitoring and on the machine characterization of polished surfaces enhance the process efficiency and enable robust methods for automation of RAP process.

The solutions are expected to be implemented in the next generation of RAP machines, resulting in significant quality improvements and cost benefits for industrial users of the system.