Conditional emissivity prior for improved Bayesian pyrometry estimates of advanced high strength steel

Inferring steel strip temperature through non-contact radiometric measurements involves solving a set of underdetermined equations, so additional information must be specified to obtain a unique and stable solution. The Bayesian pyrometry approach provides steelmakers with a rigorous framework for including this information in a way that accounts for the fact that that the radiative properties are imperfectly known. While Bayesian methods often use fixed priors, this study presents a conditional stochastic prior that captures covariate dependence of the radiative properties on annealing dew point and brightness temperature. The technique is evaluated on dual-phase AHSS steel alloys (DP780 and DP980) and extra deep drawing steel (EDDS) grade IF steel annealed in an N 2 -5% vol. H 2 reducing atmosphere at various dew points. The findings show that defining a conditional spectral emissivity prior improves the precision of the prior information, while accounting for the variability in emissivity caused by changing annealing parameters.