Quality Pre-Assesment in Steel Industry Using Data Based Estimators

In this paper we present a comprehensive framework enabling data based quality pre-assessment for the quality of metallurgical products, and report on the main experiences gained in cooperation with an industrial partner. The proposed approach is based on a sequential structure including data pre-processing to obtain a well-conditioned problem as well as nonlinear modelling approaches to determine dependencies. As we document here, this approach works correctly for approximately 90% of the cases, but hints are given on which essential modifications (associated with the problem setup) should be taken to re-formulate the problem in order to increase its performance.