Objective
In the project a novel multisensorial-data supported industrial inspection system for metal parts is proposed. It involves new X-ray sensors and conventional light based computer vision modules, for simultaneous detection and correlation of internal and surface quality defects of metal parts, giving non invasively in realtime, a total quality assessment.
Initial research involved the development of 2 high voltage generators to be integrated into the imaging system intended for high thickness material radioscopy. The 225 kV, 13 mA and 420/450 kV, 15 mA generators have been run up to 250 kV, on the dummy load at 0 mA without any high voltage failure.
A new solid state linear detector was developed to detect high energy X-ray photons. Tests demonstrated that the efficiency of the detector was 30 to 300 times higher than that of the screen/low light level camera device.
A process was elaborated upon to identify automatically the defects detected on metal products. The basis for the recognition is the radioscopic image for internal defects and the optical image under structured light for external defects.
Research was then carried out with respect to X-ray image parameter definition.
The structure of the X-ray inspection system for metal products consisted of 4 major components: image acquisition, preprocessing, detection and defect classification. These steps followed the learning phase where the characteristics of a flawless part of the image were determined. The output of this module is a report file containing information on the type and location of each detected defect.
With the purpose of developing an analysis system for visual inspection of aluminium ingot dedicated to external fault detection a structured light system was used. Employing an image self correlation in time technique, a system able to perform a complete real time inspection of aluminium ingot and detect 4 types of fault was built and tested.
Among the manufacturing process plants operated by the ALUMIX Group, which ALURES belongs to, the rod continuous production line was selected as an industrial real case for testing the prototype. The system demonstrated its ability to detect surface irregularities online with a sensitivity complying with the requirement of the aluminium rod production line. Approaches were developed to integrate analysis of the internal and external faults.
The system will be suitable for unattended, as well as for manned operations, in dense factory environments. It will be based on advanced image processing algorithms and multi-sensorial data fusion methodologies, effectively combining images features and high level artifacts derived from them, in order to compile an accurate defect modelling and localization and to fast validate the produced parts, according to their overall quality specifications.
The project aims at a directly exploitable multi-sensorial inspection workstation, to be actively marketed by the consortium members soon after the projects conclusion, both in EEC countries and abroad.
The system will bring high-tech quality inspection tools to a hostile and difficult to operate environment (such as metal parts manufacturing, casting and machining), thus offering a tremendous potential for quality improvement and productivity acceleration.
Fields of science
- engineering and technologyelectrical engineering, electronic engineering, information engineeringelectronic engineeringsensorsoptical sensors
- natural sciencescomputer and information sciencesartificial intelligencecomputer vision
- engineering and technologymechanical engineeringmanufacturing engineeringsubtractive manufacturing
- natural scienceschemical sciencesinorganic chemistrypost-transition metals
- natural sciencesphysical sciencestheoretical physicsparticle physicsphotons
Topic(s)
Data not availableCall for proposal
Data not availableFunding Scheme
Data not availableCoordinator
17674 KALLITHEA ELASSONOS
Greece