In modular plants, quality of methods and models must be specifiable, automatically testable and certifiable, if they are increasingly integrated into equipment as a product and an added value. Since the underlying processes are often not completely understood, hybrid and data-driven methods are a promising approach to combine process knowledge and process data for more reliable simulation models. In this paper, the conducted quality assessment utilizes the framework proposed by Mädler et al. (2021) for quality assurance applied to hybrid models. The quality model is revised to include quality factors, criteria and metrics for dynamic, hybrid semi-parametric simulation models. A state observer for the estimation of key process parameters during fermentation is presented as a use case. For this three hybrid models of the fermentation with differing levels of detail are identified and coupled with an extended Kalman filter (EKF). It was found that the quality model can successfully be used to assess quality differences in different types of state observers. The quality model allows a structured and quick assessment and is therefore able to show e.g. the performance improvement of the different hybrid models coupled with an EKF. With the transfer of the quality model to hybrid state observers a broader range of simulations models can be assessed within the framework.