Intelligent soft-sensor technology and automatic model-based diagnosis for improved quality, control and maintenance of mill production lines (Softdetect)
The main goal of this project is the development and application of software sensors for the estimation of hard-to-measure quality and process parameters (e.g. thermal state, mill state and process state) during processing. This builds up the basis for automatic fault-detection and diagnosis of causes. The considered processing stages are hot rolling, tandem cold rolling and temper rolling. Hot strip mill: A virtual thermal sensor system was developed, based on the intelligent elaboration of data coming from the various thermal measurement devices along the line. The system is able to predict instabilities in rolling operation and to detect quality inconsistencies. The virtual global thermal indicator has been implemented to give support to operators by providing intelligent information about the whole thermal characteristics of the slab/bar/strip and the concerned rolling stability. Tandem mill: The developed system extracts new knowledge from the facility, taking advantage of the great automation that is implemented nowadays. Two main topics were treated during the project, both referring to mill quality: rolls evolution and coil quality online assess. The system is able to perform online visual data mining. Temper mill: Several modules for automatic and systematic quality defect detection have been developed and tested: an automatic diagnosis system supervising 39 defect frequencies, automatic identification/differentiation of incoming and locally produced periodic thickness faults, a chatter mark compensation system, a soft-sensor for coil tumble detection, extended methods for control performance monitoring and the 'dynamic fingerprint' method.
Bibliographic Reference: EUR 23893 EN (2009), 132 pp. Euro: 20
Availability: http://bookshop.europa.eu/is-bin/INTERSHOP.enfinity/WFS/EU-Bookshop-Site/en_GB/-/EUR/ViewPublication-Start?PublicationKey=KINA23893 (Catalogue Number: KI-NA-23893-EN-S)
ISBN: ISBN: 978-92-79-11980-4
Record Number: 200910480 / Last updated on: 2009-10-12
Original language: en
Available languages: en