{"id":{"repo_id":"uiuc","oai_identifier":"oai:www.ideals.illinois.edu:2142/129250"},"canonical_url":"https://search.dev.ndltd.org/etd/uiuc/oai:www.ideals.illinois.edu:2142/129250","repository":{"repo_id":"uiuc","name":"University of Illinois - Urbana-Champaign","base_url":"https://www.ideals.illinois.edu/oai-pmh"},"display":{"title":"Effective and efficient online process monitoring and control of ultrasonic metal welding","abstract":"Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2025-10-19 without embargo terms","abstract_html":"Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2025-10-19 without embargo terms","abstract_has_math":false,"creators":["Lu, Kuan-Chieh"],"institution":"University of Illinois Urbana-Champaign","degree_name":"Ph.D.","degree_level":"Dissertation","degree_discipline":"Mechanical Engineering","degree_department":null,"school":null,"contributors":["Shao, Chenhui","Ferreira, Placid M.","Salapaka, Srinivasa M","Wang, Pingfeng"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2025,"date_issued":"2025-04-25","date_published":"2025-04-25","updated_at":"2026-07-22T22:25:04Z","subjects":["Ultrasonic metal welding","Cost-efficient monitoring","Unknown fault detection","Continuous learning","Real-time control"],"languages":["en","eng"],"rights":["Copyright 2025 Kuan-Chieh Lu"],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://hdl.handle.net/2142/129250","outbound_label":"Handle","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Shao, Chenhui","Ferreira, Placid M.","Salapaka, Srinivasa M","Wang, Pingfeng"]},{"key":"dc:creator","label":"Author","values":["Lu, Kuan-Chieh"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2025-04-25","2025-05"]},{"key":"dc:type","label":"Dc Type","values":["text","Thesis"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Mechanical Engineering"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Dissertation"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Ph.D."]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["University of Illinois Urbana-Champaign"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Ultrasonic metal welding","Cost-efficient monitoring","Unknown fault detection","Continuous learning","Real-time control"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["en","eng"]},{"key":"dc:rights","label":"Dc Rights","values":["Copyright 2025 Kuan-Chieh Lu"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://hdl.handle.net/2142/129250"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2025-10-19 without embargo terms","The student, Kuan-Chieh Lu, accepted the attached license on 2025-04-24 at 22:41.","The student, Kuan-Chieh Lu, submitted this Dissertation for approval on 2025-04-24 at 22:42.","This Dissertation was approved for publication on 2025-04-25 at 15:50.","DSpace SAF Submission Ingestion Package generated from Vireo submission #21971 on 2025-10-19 at 18:10:59","Ultrasonic metal welding (UMW) is a solid-state welding process with widespread industrial applications such as automotive body construction, lithium-ion battery manufacturing, and electronic packaging. UMW has various advantages compared to conventional fusion welding methods, such as environmental friendliness, low energy consumption, and short process cycle. However, UMW is sensitive to uncontrollable factors (e.g., tool conditions, surface conditions of workpieces), leading to deteriorated joint quality. While some progress has been made in quality control for UMW, there is a lack of effective, efficient, and generalizable online monitoring and control algorithms. To address these challenges, this dissertation develops integrated learning, monitoring, and control capabilities for UMW, with a focus on not only performance but also cost-effectiveness and efficiency. The contributions of this dissertation are summarized as follows. Hardware costs and decision-making responsiveness are two primary considerations when designing and implementing an online UMW monitoring system. Within the short welding cycle (typically under 1 s), the time needed to detect abnormalities, i.e., responsiveness, is inversely correlated to the reserved control window for in-process parameter adjustment. High-frequency sensing increases hardware costs due to the necessity of a high-end data acquisition (DAQ) system and the computation costs induced by large data sets. Chapter 1 demonstrates the feasibility of using the initial signal segment from selected sensors while maintaining high monitoring accuracy in the presence of mixed tool and surface conditions. High-frequency sampling has been adopted in existing online monitoring systems in order to capture high-frequency information in UMW, the vibration of which operates at 20 kHz. Consequently, a massive amount of data points is collected during each UMW cycle, which induces substantial computational burden and limits the feasibility of real-time control. Therefore, extracting and selecting a parsimonious set of low-dimensional features from high-density sensing signals are crucial. To this end, Chapter 2 formulates a feature and sensor selection problem, where a cost-related fitness function is defined to quantify the monitoring performance and cost associated with data acquisition and processing. We further develop a genetic algorithm (GA)-based feature selection method to select information-rich features structured by time and frequency. Additionally, the separability between unknown and existing process conditions in the feature space is leveraged to more efficiently accommodate new, unknown fault types. In industrial applications of UMW, it is not uncommon to encounter new process anomalies that have not been learned by monitoring algorithms. In such cases, the risk posed by incorrectly recognizing process anomalies can be substantial. As such, it is important to develop new monitoring capabilities that are robust to the occurrence of new, unknown process anomalies. Chapter 3 develops a novel process monitoring algorithm that can (1) identify unknown abnormal operating conditions and (2) continuously learn from incoming data and enrich class labels. The effectiveness of the proposed algorithm is demonstrated using a large-scale experimental dataset consisting of nine mixed process conditions. Drawing on the new decision-making capabilities created in previous chapters, Chapter 4 presents an integrated learning, monitoring, and control system that can automatically adjust welding pressure, which is an influential UMW process parameter, to overcome undesirable process anomalies and improve joint quality. Specifically, the system allows for a non-constant pressure profile including single or multi-step adjustments, tailored to compensate for undesired disturbances. The integration between learning, monitoring, and control enables process adjustments that are adaptive to various process anomalies that may occur on the factory floor. The effectiveness of the integrated system is validated through experiments, demonstrating that the welding success rate can be improved by 92% under the disturbance of surface contamination."]},{"key":"dc:format","label":"Dc Format","values":["application/pdf"]},{"key":"dc:title","label":"Title","values":["Effective and efficient online process monitoring and control of ultrasonic metal welding"]}]}],"canonical_facts":{"dc:contributor":["Shao, Chenhui","Ferreira, Placid M.","Salapaka, Srinivasa M","Wang, Pingfeng"],"dc:creator":["Lu, Kuan-Chieh"],"dc:date":["2025-04-25","2025-05"],"dc:description":["Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2025-10-19 without embargo terms","The student, Kuan-Chieh Lu, accepted the attached license on 2025-04-24 at 22:41.","The student, Kuan-Chieh Lu, submitted this Dissertation for approval on 2025-04-24 at 22:42.","This Dissertation was approved for publication on 2025-04-25 at 15:50.","DSpace SAF Submission Ingestion Package generated from Vireo submission #21971 on 2025-10-19 at 18:10:59","Ultrasonic metal welding (UMW) is a solid-state welding process with widespread industrial applications such as automotive body construction, lithium-ion battery manufacturing, and electronic packaging. UMW has various advantages compared to conventional fusion welding methods, such as environmental friendliness, low energy consumption, and short process cycle. However, UMW is sensitive to uncontrollable factors (e.g., tool conditions, surface conditions of workpieces), leading to deteriorated joint quality. While some progress has been made in quality control for UMW, there is a lack of effective, efficient, and generalizable online monitoring and control algorithms. To address these challenges, this dissertation develops integrated learning, monitoring, and control capabilities for UMW, with a focus on not only performance but also cost-effectiveness and efficiency. The contributions of this dissertation are summarized as follows. Hardware costs and decision-making responsiveness are two primary considerations when designing and implementing an online UMW monitoring system. Within the short welding cycle (typically under 1 s), the time needed to detect abnormalities, i.e., responsiveness, is inversely correlated to the reserved control window for in-process parameter adjustment. High-frequency sensing increases hardware costs due to the necessity of a high-end data acquisition (DAQ) system and the computation costs induced by large data sets. Chapter 1 demonstrates the feasibility of using the initial signal segment from selected sensors while maintaining high monitoring accuracy in the presence of mixed tool and surface conditions. High-frequency sampling has been adopted in existing online monitoring systems in order to capture high-frequency information in UMW, the vibration of which operates at 20 kHz. Consequently, a massive amount of data points is collected during each UMW cycle, which induces substantial computational burden and limits the feasibility of real-time control. Therefore, extracting and selecting a parsimonious set of low-dimensional features from high-density sensing signals are crucial. To this end, Chapter 2 formulates a feature and sensor selection problem, where a cost-related fitness function is defined to quantify the monitoring performance and cost associated with data acquisition and processing. We further develop a genetic algorithm (GA)-based feature selection method to select information-rich features structured by time and frequency. Additionally, the separability between unknown and existing process conditions in the feature space is leveraged to more efficiently accommodate new, unknown fault types. In industrial applications of UMW, it is not uncommon to encounter new process anomalies that have not been learned by monitoring algorithms. In such cases, the risk posed by incorrectly recognizing process anomalies can be substantial. As such, it is important to develop new monitoring capabilities that are robust to the occurrence of new, unknown process anomalies. Chapter 3 develops a novel process monitoring algorithm that can (1) identify unknown abnormal operating conditions and (2) continuously learn from incoming data and enrich class labels. The effectiveness of the proposed algorithm is demonstrated using a large-scale experimental dataset consisting of nine mixed process conditions. Drawing on the new decision-making capabilities created in previous chapters, Chapter 4 presents an integrated learning, monitoring, and control system that can automatically adjust welding pressure, which is an influential UMW process parameter, to overcome undesirable process anomalies and improve joint quality. Specifically, the system allows for a non-constant pressure profile including single or multi-step adjustments, tailored to compensate for undesired disturbances. The integration between learning, monitoring, and control enables process adjustments that are adaptive to various process anomalies that may occur on the factory floor. The effectiveness of the integrated system is validated through experiments, demonstrating that the welding success rate can be improved by 92% under the disturbance of surface contamination."],"dc:format":["application/pdf"],"dc:identifier":["https://hdl.handle.net/2142/129250"],"dc:language":["en","eng"],"dc:rights":["Copyright 2025 Kuan-Chieh Lu"],"dc:subject":["Ultrasonic metal welding","Cost-efficient monitoring","Unknown fault detection","Continuous learning","Real-time control"],"dc:title":["Effective and efficient online process monitoring and control of ultrasonic metal welding"],"dc:type":["text","Thesis"],"thesis:degree_discipline":["Mechanical Engineering"],"thesis:degree_level":["Dissertation"],"thesis:degree_name":["Ph.D."],"thesis:institution_name":["University of Illinois Urbana-Champaign"]},"updated_at":"2026-07-22T22:25:04Z"}