Carbon fibers are an indispensable material in lightweight construction due to their mass-specific mechanical load capacity. These are produced by means of a time-consuming, thermo-oxidative stabilization process and an inert carbonization process. Due to commercially dominated research, the majority of published optimization approaches for these process steps to date is based on incomplete, empirical or non-parametric, data-driven models. In the present work, hybrid semi-parametric approaches considering the properties of the precursor system and the process parameters are investigated for modeling of the stabilization process.