Open Access Open Access  Restricted Access Subscription or Fee Access

Observation and Detection of Machining Phases in EDM Microdrilling by Acoustic Emission

QUENTIN LEFEBVRE, MARTIAL PERSONENI, EMMANUEL RAMASSO, SÉBASTIEN THIBAUD

Abstract


Micro Electrical Discharge Machining (Micro-EDM) is a manufacturing process used for drilling of micro-diameter and deep holes whereby electrical discharges emitted by a tool-electrode are used to remove material in a work piece hosed by a dielectric .The process gives the ability to make intricate shapes with great aspect ratio in hard materials not yet feasible with standard drilling process used for instance in turbo-engines. During the process, both the work piece and the tool-electrode get damaged. It is highly difficult and time-consuming to evaluate precisely and in real-time their degree of wear. Moreover, the process is generally managed by a proprietary software which controls both the feed rate of the tool-electrode and the sequence of discharges necessary to remove pieces of material to reach a reference in depth. We demonstrate for the first time that the acoustic emission (AE) technique is able to highlight the different phases in the manufacturing process through the use of an advanced and unsupervised pattern recognition method where each phase is interpreted as an AE source. The task is quite difficult since we are in presence of continuous acoustic emissions on small work pieces. Results concern the drilling of hard material. AE streaming have been collected on multiple AE sensors with different characteristics (for comparison purpose), processed by wavelets for reliable AE hit detection, and then interpreted using a chronologybased consensus clustering with temporal coherence. This is a recent approach used to evidence a chronology of AE sources whereby AE interpretation is focused on the timeline of AE events. The algorithm relies on an unsupervised data fusion process of partitions obtained by many parameterizations (such as subsets of features) with a quantification of both the uncertainty and the robustness of the results obtained. Results show the effectiveness of the approach, in particular in presence of continuous acoustic emissions.


DOI
10.12783/shm2019/32099

Full Text:

PDF