Show simple item record

dc.contributor.authorGomes-Mota, Joao
dc.date.accessioned2019-07-24T12:37:17Z
dc.date.available2019-07-24T12:37:17Z
dc.date.issued2019-06-03
dc.identifier.isbn978-2-9602415-0-1
dc.identifier.issn2032-9644
dc.identifier.urihttps://cired-repository.org/handle/20.500.12455/104
dc.identifier.urihttp://dx.doi.org/10.34890/205
dc.description.abstractThe classification of visible issues on Over-Head Line [OHL] inspection for preventive maintenance is a valuable and time consuming task. Traditional image processing algorithms have been used for the automatic detection, either assisted or complemented by human experts for severity classification. In recent years, machine learning has been proposed as a fast-track alternative to automate this task at the expense of huge data volumes (more than 1GB per km of OHL track). The proposed contribution compares the results of both kinds of algorithms for detection and classification of four visible Points of Interest [PoI] of different natures: stork nests on towers, aerial warning spheres, number plates and worn fittings on suspension towers (suspension ball eyes and shackles).The author works on the development of technology solutions across all the inspection process, from the data collection stage to data analytics and asset management information systems. It reviews his team results on the particular issues of image based inspection, considering success rates for both computer vision and machine learning solutions.The comprehensive data to information chain is discussed from data collection, analytics, information management and later, maintenance actions, auditing and long term maintenance planning. In addition to the challenges of data management in such volumes and improving the efficiency of inspection, it is essential to measure the actual benefits utilities across the asset maintenance cycle. A discussion on the multi-variable risk-index associated with severity and exposure, the combination of issues and overall condition assessment is introduced.
dc.language.isoen
dc.publisherAIM
dc.relation.ispartofseriesCIRED Conference Proceedings
dc.titleAutomated visual inspection – comparing computer vision to machine learning
dc.typeConference Proceedings
dc.description.conferencelocationMadrid, Spain
dc.relation.ispartProc. of the 25th International Conference on Electricity Distribution (CIRED 2019)
dc.contributor.detailedauthorGomes-Mota, Joao, Albatroz Engineering, Portugal
dc.date.conferencedate3-6 June 2019
dc.description.peerreviewedYes
dc.title.number52
dc.description.openaccessYes
dc.contributor.countryPortugal
dc.description.conferencenameCIRED 2019
dc.contributor.affiliationAlbatroz Engineering
dc.description.sessionNetwork components
dc.description.sessionidSession 1


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record