This study focuses on modeling the yeast fermentation process using hybrid method. To improve prediction accuracy of model and reduce training time, this paper presents a semi−supervised method based an extreme learning machine for process. The is composed mechanism residual model. built from residuals between real used in parallel with Considering that might be related to inaccurate parameters...

Chemical process engineering and machine learning are merging rapidly, hybrid models have shown promising results in analysis design. However, uncertainties first-principles an adverse effect on extrapolations inferences based models. Parameter sensitivities essential tool to understand better the underlying uncertainty propagation system identification challenges. Still, standard parameter sen...