Abstract
The global transition toward sustainable energy has intensified research into biofuels, with bioprocess optimization playing a central role in achieving decarbonization goals. Biobutanol, in particular, is a high-value molecule for sustainable fuel applications due to its superior energy density and compatibility with existing infrastructure. However, model-based optimization of its production is hindered by traditional semi-structured kinetic models that often suffer from limited predictive robustness. To address this challenge, within this study we developed a hybrid modeling framework for Clostridium saccharoperbutylacetonicum that integrates mechanistic mass-balance equations with Gaussian Processes (GPs) aiming to describe the biobutanol formation rate. Here, we investigate the effect of data normalization techniques on hybrid model's prediction capabilities comparing min-max normalization, z-score normalization, and no transformation. For each data treatment strategy, 8, 000 hybrid models were trained using experimental fermentation datasets, and ensemble predictors were constructed to mitigate variability and enhance generalization. Results demonstrate that hybrid models significantly outperform parametric baselines, with the ensemble approach leveling the performance differences across all pretreatment techniques. Ensembles achieved validation RMSE reductions of approximately 10.1% to 10.4%, which are two times higher than the best individual models regardless of the scaling method used. Despite similar RMSE reductions, prediction envelopes varied along data transformations revealing tradeoffs between uncertainty and coverage. Notably, targeted hybridization of the butanol reaction rate propagated performance improvements to other states, such as glucose and biomass. These findings suggest that for hybrid modeling, ensemble design provides greater benefits for robustness than any data normalization technique.