Control of deviations and prediction of surface roughness from micro machining of THz waveguides using acoustic emission signals. (15th February 2017)

• Record Type: Journal Article
• Authors: Griffin, James M.; Diaz, Fernanda; Geerling, Edgar; Clasing, Matias; Ponce, Vicente; Taylor, Chris; Turner, Sam; Michael, Ernest A.; Patricio Mena, F.; Bronfman, Leonardo
• Is Part Of: Mechanical systems and signal processing. Volume 85(2017)
• Page Start: 1020
• Page End: 1034
• Publication Date: 2017-02-15
• Subjects: Acoustic emission -- Precision manufacturing -- Surface finish -- Precision control -- Submillimetre waveguide manufacturing -- CART -- Neural networks and simulations; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2016.09.016
• Languages: English
• ISSNs: 0888-3270

Abstract

Abstract: By using acoustic emission (AE) it is possible to control deviations and surface quality during micro milling operations. The method of micro milling is used to manufacture a submillimetre waveguide where micro machining is employed to achieve the required superior finish and geometrical tolerances. Submillimetre waveguide technology is used in deep space signal retrieval where highest detection efficiencies are needed and therefore every possible signal loss in the receiver has to be avoided and stringent tolerances achieved. With a sub-standard surface finish the signals travelling along the waveguides dissipate away faster than with perfect surfaces where the residual roughness becomes comparable with the electromagnetic skin depth. Therefore, the higher the radio frequency the more critical this becomes. The method of time-frequency analysis (STFT) is used to transfer raw AE into more meaningful salient signal features (SF). This information was then correlated against the measured geometrical deviations and, the onset of catastrophic tool wear. Such deviations can be offset from different AE signals (different deviations from subsequent tests) and feedback for a final spring cut ensuring the geometrical accuracies are met. Geometrical differences can impact on the required transfer of AE signals (change in cut off frequencies and diminished SNR at the interface) and therefore errors have to be minimised to within 1 µm. Rules based on both Classification and Regression Trees (CART) and Neural Networks (NN) were used to implement a simulation displaying how such a control regime could be used as a real time controller, be it corrective measures (via spring cuts) over several initial machining passes or, with a micron cut introducing a level plain measure for allowing setup corrective measures (similar to a spirit level). Highlights: Different levels of each phenomenon in terms of acoustic emission (AE) correlated to deviation were characterised. Physical phenomena using both the time and frequency domain analysis of the AE were gained from the vertical-axis micro milling tests. Significant tool wear of micro milling was characterised after and before the onset of tool malfunction. Different levels of the characterised phenomena were classified for the prediction of surface quality. A real-time simulation displaying changing energy patterns as deviation errors increase/decrease was implemented.