{"id":{"repo_id":"vt","oai_identifier":"oai:vtechworks.lib.vt.edu:10919/135551"},"canonical_url":"https://search.dev.ndltd.org/etd/vt/oai:vtechworks.lib.vt.edu:10919/135551","repository":{"repo_id":"vt","name":"Virginia Tech","base_url":"https://vtechworks.lib.vt.edu/oai/request"},"display":{"title":"Design and Integration of Machine Learning-Based Vision System for Automated Power Line Inspection Using a Mobile Damping Robot","abstract":"Ensuring the structural integrity of overhead power line conductors is critical for maintaining the safety and reliability of the electrical grid. Environmental stressors such as moisture, dust, and wind-induced vibrations contribute to surface degradation, including corrosion and fretting, which compromise conductor conditions and therefore performance over time. This thesis presents a vision-based, autonomous inspection framework using a modified Mobile Damping Robot, with two distinct levels of health assessment. The first method focuses on a simpler binary classification, where image filtering techniques—such as Sobel, Scharr, and Gray-scale Variance Normalization—are compared to highlight defect patterns, followed by histogram-based feature extraction and classification into healthy or unhealthy categories using traditional supervised machine learning models, including Random Forest, Multi-Layer Perceptron, and Gradient Boosting. Next, the second method provides a more detailed assessment by classifying conductors into four categories: Healthy, Minor Corrosion, Pollution-Induced Corrosion, and Pollution-Induced Fretting. To facilitate this classification, real-world conductor images collected via the MDR were preprocessed using segmentation models, such as U-Net and the Segment Anything Model, to isolate the conductor from the background. Two deep learning models, a custom-designed Convolutional Neural Network and a ResNet-50 transfer learning model, were trained for multi-class classification. Experimental results validate the effectiveness of the ResNet-50 model, demonstrating the potential of vision-based inspection for enabling proactive, data-driven, and scalable power line maintenance. By automating condition assessment through image-based analysis, this approach facilitates early detection of degradation, reduces reliance on manual inspections, and supports enhanced operational planning across transmission infrastructure.","abstract_html":"Ensuring the structural integrity of overhead power line conductors is critical for maintaining the safety and reliability of the electrical grid. Environmental stressors such as moisture, dust, and wind-induced vibrations contribute to surface degradation, including corrosion and fretting, which compromise conductor conditions and therefore performance over time. This thesis presents a vision-based, autonomous inspection framework using a modified Mobile Damping Robot, with two distinct levels of health assessment. The first method focuses on a simpler binary classification, where image filtering techniques—such as Sobel, Scharr, and Gray-scale Variance Normalization—are compared to highlight defect patterns, followed by histogram-based feature extraction and classification into healthy or unhealthy categories using traditional supervised machine learning models, including Random Forest, Multi-Layer Perceptron, and Gradient Boosting. Next, the second method provides a more detailed assessment by classifying conductors into four categories: Healthy, Minor Corrosion, Pollution-Induced Corrosion, and Pollution-Induced Fretting. To facilitate this classification, real-world conductor images collected via the MDR were preprocessed using segmentation models, such as U-Net and the Segment Anything Model, to isolate the conductor from the background. Two deep learning models, a custom-designed Convolutional Neural Network and a ResNet-50 transfer learning model, were trained for multi-class classification. Experimental results validate the effectiveness of the ResNet-50 model, demonstrating the potential of vision-based inspection for enabling proactive, data-driven, and scalable power line maintenance. By automating condition assessment through image-based analysis, this approach facilitates early detection of degradation, reduces reliance on manual inspections, and supports enhanced operational planning across transmission infrastructure.","abstract_has_math":false,"creators":["Kang, Hyun Myung"],"institution":"Virginia Tech","degree_name":"Master of Science","degree_level":"masters","degree_discipline":"Mechanical Engineering","degree_department":"Mechanical Engineering","school":null,"contributors":[],"advisors":[],"committee_chairs":["Barry, Oumar"],"committee_members":["Sandu, Corina","Southward, Steve C."],"year":2025,"date_issued":"2025-06-20","date_published":"2025-06-20","updated_at":"2026-07-22T22:20:41Z","subjects":["Machine Learning","Robotics","Neural Network","Image Segmentation","Power Line"],"languages":["en"],"rights":["In Copyright"],"rights_urls":["http://rightsstatements.org/vocab/InC/1.0/"],"identifier_entries":[{"key":"dc:identifier.other","label":"Dc Identifier Other","values":["vt_gsexam:44255"],"render_values":[{"text":"vt_gsexam:44255","href":null,"code":true}]}]},"links":{"outbound_url":"https://hdl.handle.net/10919/135551","outbound_label":"Handle","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.committeechair","label":"Committee Chair","values":["Barry, Oumar"]},{"key":"dc:contributor.committeemember","label":"Committee Member","values":["Sandu, Corina","Southward, Steve C."]},{"key":"dc:contributor.department","label":"Department","values":["Mechanical Engineering"]},{"key":"dc:creator","label":"Author","values":["Kang, Hyun Myung"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2025-06-21T08:00:43Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2025-06-21T08:00:43Z"]},{"key":"dc:date.issued","label":"Date","values":["2025-06-20"]},{"key":"dc:publisher","label":"Institution","values":["Virginia Tech"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Mechanical Engineering"]},{"key":"thesis:degree_level","label":"Degree Level","values":["masters"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Master of Science"]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["Virginia Polytechnic Institute and State University"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Machine Learning","Robotics","Neural Network","Image Segmentation","Power Line"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en"]},{"key":"dc:rights","label":"Dc Rights","values":["In Copyright"]},{"key":"dc:rights.uri","label":"Rights URI","values":["http://rightsstatements.org/vocab/InC/1.0/"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.other","label":"Dc Identifier Other","values":["vt_gsexam:44255"]},{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/10919/135551"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Ensuring the structural integrity of overhead power line conductors is critical for maintaining the safety and reliability of the electrical grid. Environmental stressors such as moisture, dust, and wind-induced vibrations contribute to surface degradation, including corrosion and fretting, which compromise conductor conditions and therefore performance over time. This thesis presents a vision-based, autonomous inspection framework using a modified Mobile Damping Robot, with two distinct levels of health assessment. The first method focuses on a simpler binary classification, where image filtering techniques—such as Sobel, Scharr, and Gray-scale Variance Normalization—are compared to highlight defect patterns, followed by histogram-based feature extraction and classification into healthy or unhealthy categories using traditional supervised machine learning models, including Random Forest, Multi-Layer Perceptron, and Gradient Boosting. Next, the second method provides a more detailed assessment by classifying conductors into four categories: Healthy, Minor Corrosion, Pollution-Induced Corrosion, and Pollution-Induced Fretting. To facilitate this classification, real-world conductor images collected via the MDR were preprocessed using segmentation models, such as U-Net and the Segment Anything Model, to isolate the conductor from the background. Two deep learning models, a custom-designed Convolutional Neural Network and a ResNet-50 transfer learning model, were trained for multi-class classification. Experimental results validate the effectiveness of the ResNet-50 model, demonstrating the potential of vision-based inspection for enabling proactive, data-driven, and scalable power line maintenance. By automating condition assessment through image-based analysis, this approach facilitates early detection of degradation, reduces reliance on manual inspections, and supports enhanced operational planning across transmission infrastructure."]},{"key":"dc:description.abstractgeneral","label":"General Abstract","values":["Overhead power lines are essential for delivering electricity, but they are constantly exposed to harsh weather conditions like wind, moisture, and dust, which can cause damage over time. To help detect early signs of wear and prevent failures, this thesis introduces an automated inspection system using a Mobile Damping Robot. The system uses computer vision, a field of artificial intelligence that enables computers to understand images, to assess the physical condition of the power lines. In the first part of the research, machine learning techniques were used to analyze these images and classify power lines as either healthy or damaged based on patterns detected in the surface texture. In the second part, a more detailed analysis was developed using deep learning to identify four specific conditions, including different types of corrosion and pollution-related wear. To focus the analysis on just the cable, advanced image segmentation tools were used to isolate the power line from its background image. This research demonstrates that combining robotic inspection with AI-powered image analysis can support safer, more efficient maintenance of electrical infrastructure and help prevent costly outages."]},{"key":"dc:description.degree","label":"Dc Description Degree","values":["Master of Science"]},{"key":"dc:format.medium","label":"Dc Format Medium","values":["ETD"]},{"key":"dc:title","label":"Title","values":["Design and Integration of Machine Learning-Based Vision System for Automated Power Line Inspection Using a Mobile Damping Robot"]}]}],"canonical_facts":{"dc:contributor.committeechair":["Barry, Oumar"],"dc:contributor.committeemember":["Sandu, Corina","Southward, Steve C."],"dc:contributor.department":["Mechanical Engineering"],"dc:creator":["Kang, Hyun Myung"],"dc:date.accessioned":["2025-06-21T08:00:43Z"],"dc:date.available":["2025-06-21T08:00:43Z"],"dc:date.issued":["2025-06-20"],"dc:description.abstract":["Ensuring the structural integrity of overhead power line conductors is critical for maintaining the safety and reliability of the electrical grid. Environmental stressors such as moisture, dust, and wind-induced vibrations contribute to surface degradation, including corrosion and fretting, which compromise conductor conditions and therefore performance over time. This thesis presents a vision-based, autonomous inspection framework using a modified Mobile Damping Robot, with two distinct levels of health assessment. The first method focuses on a simpler binary classification, where image filtering techniques—such as Sobel, Scharr, and Gray-scale Variance Normalization—are compared to highlight defect patterns, followed by histogram-based feature extraction and classification into healthy or unhealthy categories using traditional supervised machine learning models, including Random Forest, Multi-Layer Perceptron, and Gradient Boosting. Next, the second method provides a more detailed assessment by classifying conductors into four categories: Healthy, Minor Corrosion, Pollution-Induced Corrosion, and Pollution-Induced Fretting. To facilitate this classification, real-world conductor images collected via the MDR were preprocessed using segmentation models, such as U-Net and the Segment Anything Model, to isolate the conductor from the background. Two deep learning models, a custom-designed Convolutional Neural Network and a ResNet-50 transfer learning model, were trained for multi-class classification. Experimental results validate the effectiveness of the ResNet-50 model, demonstrating the potential of vision-based inspection for enabling proactive, data-driven, and scalable power line maintenance. By automating condition assessment through image-based analysis, this approach facilitates early detection of degradation, reduces reliance on manual inspections, and supports enhanced operational planning across transmission infrastructure."],"dc:description.abstractgeneral":["Overhead power lines are essential for delivering electricity, but they are constantly exposed to harsh weather conditions like wind, moisture, and dust, which can cause damage over time. To help detect early signs of wear and prevent failures, this thesis introduces an automated inspection system using a Mobile Damping Robot. The system uses computer vision, a field of artificial intelligence that enables computers to understand images, to assess the physical condition of the power lines. In the first part of the research, machine learning techniques were used to analyze these images and classify power lines as either healthy or damaged based on patterns detected in the surface texture. In the second part, a more detailed analysis was developed using deep learning to identify four specific conditions, including different types of corrosion and pollution-related wear. To focus the analysis on just the cable, advanced image segmentation tools were used to isolate the power line from its background image. This research demonstrates that combining robotic inspection with AI-powered image analysis can support safer, more efficient maintenance of electrical infrastructure and help prevent costly outages."],"dc:description.degree":["Master of Science"],"dc:format.medium":["ETD"],"dc:identifier.other":["vt_gsexam:44255"],"dc:identifier.uri":["https://hdl.handle.net/10919/135551"],"dc:language.iso":["en"],"dc:publisher":["Virginia Tech"],"dc:rights":["In Copyright"],"dc:rights.uri":["http://rightsstatements.org/vocab/InC/1.0/"],"dc:subject":["Machine Learning","Robotics","Neural Network","Image Segmentation","Power Line"],"dc:title":["Design and Integration of Machine Learning-Based Vision System for Automated Power Line Inspection Using a Mobile Damping Robot"],"dc:type":["Thesis"],"thesis:degree_discipline":["Mechanical Engineering"],"thesis:degree_level":["masters"],"thesis:degree_name":["Master of Science"],"thesis:institution_name":["Virginia Polytechnic Institute and State University"]},"updated_at":"2026-07-22T22:20:41Z"}