Refinement of flat steel quality assessment by evaluation of high-resolution process and product data (EvalHD) : final report

The aim of this project was to develop an advanced product quality supervisory system for flat steel production based on high resolution (HR) measuring data. To achieve this ambitious target new methods have been developed, employing innovative software technologies and modern computation capabilities, tailored to handle the massive amount of data accumulating during the steel production process. Based on a multi-scale data representation over multiple production stages the concept finally developed provides simple and fast HR data access. Realized as a three-tier software architecture including a web-service for a standardized data access, the potential of this new concept could be remarkably proven and further modules were stepwise integrated. At first a module for efficient through process data visualisation that enables the in-coil aggregation of millions of quality and process measures within seconds could be realized to gain knowledge about origin and evolution of quality deficiencies and allow instant interactive data analysis as soon as a quality problem occurs. Furthermore a component for advanced cause-and-effect analysis based on HR data was realized to accelerate root-cause identification of quality problems and decision support rules were implemented based on HR data. It could be shown that this new concept performs far better than state-of-theart data models in terms of query response time. First experiences with this new system showed that it is able to provide precise results about thousands of coils within seconds, whereas previously the in-depth analysis of HR-data could take weeks to generate only rough statements on a few coils.