The superficial defects of even stainless and magnetic steel laminates: Analyses of reasons for their appearance and of remedies to the production installation for their prevention
Project ID: 7210-GB/406Funded under: ECSC-STEELRES 8C
The Centro sviluppo materiali (CSM) developed an inspection system, based on charge coupled device (CCD) camera and hardware intended to allow on time image processing, This system has been used to collect images and data building up an archive of about 2000 defects. The classification statistic method regarding the k phaserpoints (KNN) has been utilized to classify the defects achieving 60% correct classification out of 5 defects groups. In order to improve performance and provide information on defects' causes, an attempt has been made towards a based-on-rules approach. It has been developed on the shell for G2 expert systems. The rules have been made from questionnaires which were submitted to experts on defects. The results show that by elaborating carefully the rules and their parameters it is possible to perform as successfully as by utilizing other classifiers (statistic, neutral), limitedly to 5-10 classes. Although the based-on-rules classifier turns out to be inadequate for actual applications, it seems to be an efficient instrument to process non-deterministic information.
Bibliographic Reference: EUR 16667 IT (1997) 94pp., FS, ECU 16.50
Availability: Available from the (2)
Record Number: 199710876 / Last updated on: 1997-09-02
Original language: it
Available languages: it