{"id":{"repo_id":"uoit","oai_identifier":"oai:ontariotechu.scholaris.ca:10155/2029"},"canonical_url":"https://search.dev.ndltd.org/etd/uoit/oai:ontariotechu.scholaris.ca:10155/2029","repository":{"repo_id":"uoit","name":"Ontario Institute of Technology","base_url":"https://ontariotechu.scholaris.ca/server/oai/request"},"display":{"title":"Real-time process monitoring of Spark-Assisted Chemical Engraving (SACE) machine with data-driven techniques","abstract":"Spark-Assisted Chemical Engraving (SACE) is a promising technique for glass microfabrication, offering good surface quality and machining speed. However, inherent uncertainties arising from gas film instability and the stochastic nature of the process lead to inconsistencies and reduced repeatability in machining quality. Addressing these challenges requires precise online control of the process to improve machining efficiency—an aspect not thoroughly explored in previous studies. This dissertation achieves the groundwork for process control by developing a real-time SACE process monitoring methodology for anomaly detection in glass microchannel fabrication. The first step in the methodology introduces a characterization algorithm that segments the machining current into formation, discharging, and silent phases to extract distinct process signatures. A deep learning-based time series classification model was trained for this segmentation using two architectures—Temporal Convolutional Network (TCN) and Long Short-Term Memory (LSTM)—achieving classification accuracies of 97.18% and 96.44%, respectively. The second step establishes a correlation between extracted process signatures and SACE machining quality through a decision tree-based supervised learning approach, demonstrating an 88.88% sensitivity in classifying machining quality into three categories: uniform, non-uniform, and fractured. This approach provides an alternative to the challenges of direct quality monitoring. The third step integrates the anomaly detection algorithm into a real-time environment, leveraging a preemptive multitasking environment to prevent overrun errors while operating at a high frequency of 62.5 kHz for spark detection. This work establishes a framework for SACE real-time data-driven monitoring, enabling automated defect detection. By demonstrating its effectiveness in real-time process control, this study lays the groundwork for future advancements in glass micromanufacturing, allowing for the control of process parameters to compensate for uncertainties in real-time.","abstract_html":"Spark-Assisted Chemical Engraving (SACE) is a promising technique for glass microfabrication, offering good surface quality and machining speed. However, inherent uncertainties arising from gas film instability and the stochastic nature of the process lead to inconsistencies and reduced repeatability in machining quality. Addressing these challenges requires precise online control of the process to improve machining efficiency—an aspect not thoroughly explored in previous studies. This dissertation achieves the groundwork for process control by developing a real-time SACE process monitoring methodology for anomaly detection in glass microchannel fabrication. The first step in the methodology introduces a characterization algorithm that segments the machining current into formation, discharging, and silent phases to extract distinct process signatures. A deep learning-based time series classification model was trained for this segmentation using two architectures—Temporal Convolutional Network (TCN) and Long Short-Term Memory (LSTM)—achieving classification accuracies of 97.18% and 96.44%, respectively. The second step establishes a correlation between extracted process signatures and SACE machining quality through a decision tree-based supervised learning approach, demonstrating an 88.88% sensitivity in classifying machining quality into three categories: uniform, non-uniform, and fractured. This approach provides an alternative to the challenges of direct quality monitoring. The third step integrates the anomaly detection algorithm into a real-time environment, leveraging a preemptive multitasking environment to prevent overrun errors while operating at a high frequency of 62.5 kHz for spark detection. This work establishes a framework for SACE real-time data-driven monitoring, enabling automated defect detection. By demonstrating its effectiveness in real-time process control, this study lays the groundwork for future advancements in glass micromanufacturing, allowing for the control of process parameters to compensate for uncertainties in real-time.","abstract_has_math":false,"creators":["Seyedi Sahebari, Seyed Mahmoud"],"institution":"University of Ontario Institute of Technology","degree_name":"Doctor of Philosophy (PhD)","degree_level":null,"degree_discipline":"Mechanical Engineering","degree_department":null,"school":null,"contributors":[],"advisors":["Abou Ziki, Jana","Barari, Ahmad"],"committee_chairs":[],"committee_members":[],"year":2025,"date_issued":"2025-04-01","date_published":"2025-04-01","updated_at":"2026-07-24T05:35:39Z","subjects":[],"languages":["en"],"rights":[],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://hdl.handle.net/10155/2029","outbound_label":"Handle","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Abou Ziki, Jana","Barari, Ahmad"]},{"key":"dc:creator","label":"Author","values":["Seyedi Sahebari, Seyed Mahmoud"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2025-10-17T18:05:56Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2025-10-17T18:05:56Z"]},{"key":"dc:date.issued","label":"Date","values":["2025-04-01"]},{"key":"dc:type","label":"Dc Type","values":["Dissertation"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Mechanical Engineering"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Doctor of Philosophy (PhD)"]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["University of Ontario Institute of Technology"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/10155/2029"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Spark-Assisted Chemical Engraving (SACE) is a promising technique for glass microfabrication, offering good surface quality and machining speed. However, inherent uncertainties arising from gas film instability and the stochastic nature of the process lead to inconsistencies and reduced repeatability in machining quality. Addressing these challenges requires precise online control of the process to improve machining efficiency—an aspect not thoroughly explored in previous studies. This dissertation achieves the groundwork for process control by developing a real-time SACE process monitoring methodology for anomaly detection in glass microchannel fabrication. The first step in the methodology introduces a characterization algorithm that segments the machining current into formation, discharging, and silent phases to extract distinct process signatures. A deep learning-based time series classification model was trained for this segmentation using two architectures—Temporal Convolutional Network (TCN) and Long Short-Term Memory (LSTM)—achieving classification accuracies of 97.18% and 96.44%, respectively. The second step establishes a correlation between extracted process signatures and SACE machining quality through a decision tree-based supervised learning approach, demonstrating an 88.88% sensitivity in classifying machining quality into three categories: uniform, non-uniform, and fractured. This approach provides an alternative to the challenges of direct quality monitoring. The third step integrates the anomaly detection algorithm into a real-time environment, leveraging a preemptive multitasking environment to prevent overrun errors while operating at a high frequency of 62.5 kHz for spark detection. This work establishes a framework for SACE real-time data-driven monitoring, enabling automated defect detection. By demonstrating its effectiveness in real-time process control, this study lays the groundwork for future advancements in glass micromanufacturing, allowing for the control of process parameters to compensate for uncertainties in real-time."]},{"key":"dc:title","label":"Title","values":["Real-time process monitoring of Spark-Assisted Chemical Engraving (SACE) machine with data-driven techniques"]}]}],"canonical_facts":{"dc:contributor.advisor":["Abou Ziki, Jana","Barari, Ahmad"],"dc:creator":["Seyedi Sahebari, Seyed Mahmoud"],"dc:date.accessioned":["2025-10-17T18:05:56Z"],"dc:date.available":["2025-10-17T18:05:56Z"],"dc:date.issued":["2025-04-01"],"dc:description.abstract":["Spark-Assisted Chemical Engraving (SACE) is a promising technique for glass microfabrication, offering good surface quality and machining speed. However, inherent uncertainties arising from gas film instability and the stochastic nature of the process lead to inconsistencies and reduced repeatability in machining quality. Addressing these challenges requires precise online control of the process to improve machining efficiency—an aspect not thoroughly explored in previous studies. This dissertation achieves the groundwork for process control by developing a real-time SACE process monitoring methodology for anomaly detection in glass microchannel fabrication. The first step in the methodology introduces a characterization algorithm that segments the machining current into formation, discharging, and silent phases to extract distinct process signatures. A deep learning-based time series classification model was trained for this segmentation using two architectures—Temporal Convolutional Network (TCN) and Long Short-Term Memory (LSTM)—achieving classification accuracies of 97.18% and 96.44%, respectively. The second step establishes a correlation between extracted process signatures and SACE machining quality through a decision tree-based supervised learning approach, demonstrating an 88.88% sensitivity in classifying machining quality into three categories: uniform, non-uniform, and fractured. This approach provides an alternative to the challenges of direct quality monitoring. The third step integrates the anomaly detection algorithm into a real-time environment, leveraging a preemptive multitasking environment to prevent overrun errors while operating at a high frequency of 62.5 kHz for spark detection. This work establishes a framework for SACE real-time data-driven monitoring, enabling automated defect detection. By demonstrating its effectiveness in real-time process control, this study lays the groundwork for future advancements in glass micromanufacturing, allowing for the control of process parameters to compensate for uncertainties in real-time."],"dc:identifier.uri":["https://hdl.handle.net/10155/2029"],"dc:language.iso":["en"],"dc:title":["Real-time process monitoring of Spark-Assisted Chemical Engraving (SACE) machine with data-driven techniques"],"dc:type":["Dissertation"],"thesis:degree_discipline":["Mechanical Engineering"],"thesis:degree_name":["Doctor of Philosophy (PhD)"],"thesis:institution_name":["University of Ontario Institute of Technology"]},"updated_at":"2026-07-24T05:35:39Z"}