Feasibility study on the current AMRC process monitoring capability during grinding & investigating the repeatability of using AE response features; AE RMS, Skew, MVD, CFAR and ROP for process monitoring grinding of IN718

This thesis investigates the application of Acoustic Emission (AE) monitoring to enhance the reliability and efficiency of grinding processes, specifically focusing on IN718 superalloy using aluminium oxide grinding wheels. The primary objective of this research was to develop a robust methodology for capturing and analysing AE signals to improve process monitoring, fault diagnosis, and predictive maintenance in industrial grinding operations.

An advanced data acquisition system was implemented to capture synchronous signals from various sensors at high sampling rates. This system facilitated the comprehensive analysis of AE features, including AE RMS, Skew, MVD, CFAR, and ROP, across different grinding regimes: roughing, semi-finishing, and finishing. The study involved multiple grinding trials with varying grinding wheel diameters, enabling the assessment of AE signal repeatability and reliability.

Key findings from the research include:

AE RMS and MVD Values: Demonstrated high repeatability across different grinding regimes and wheel diameters, indicating their robustness for continuous process monitoring. The AE RMS values were particularly higher in the roughing regime due to increased material removal rates and grinding forces.
Skew: Showed consistent patterns in roughing and semi-finishing regimes but exhibited variability in the finishing regime due to sensitivity to lower magnitude AE responses. This sensitivity highlighted the need for careful consideration of grinding parameters when using Skew as a diagnostic feature.
CFAR: Maintained stable values across all grinding regimes and wheel diameters, reflecting the consistency of AE event frequency. The roughing regime had the highest CFAR values, indicating more intense interactions between the grinding wheel and the workpiece.
ROP: The prominent frequency bands were in the 100–150 kHz range, consistent across all grinding regimes. This consistency suggests the potential for using ROP values in predictive modeling of grinding wheel wear.

In addition, significant work was done on developing wavelet de-noising techniques to enhance AE signal quality by reducing machine noise. Parameters for the wavelet de-noising process were optimized to ensure maximum signal clarity without compromising the integrity of the data. This development was crucial for improving the accuracy of AE feature extraction and overall signal analysis.

The novelty of this work lies in the comprehensive application of AE monitoring to different grinding regimes and the integration of advanced signal processing techniques. The study's approach to understanding the correlation between AE features and grinding parameters provides valuable insights that can be leveraged for predictive maintenance and improved process control in industrial settings.
In conclusion, this thesis contributes significantly to the field of process monitoring and fault diagnosis in grinding operations. The methodologies and insights developed herein provide a solid foundation for future research and industrial application, driving innovation and continuous improvement in the manufacturing sector.