Assessing Parameter Relative Importance in Bioprocesses Mathematical Models through Dynamic Sensitivity Analysis

摘要

Fast global population growth has imposed an increasing demand for feedstock,food,energy and commodities;therefore,the industries are in constant development to optimize their processes.From the Process System Engineering(PSE)perspective this problem could be addressed employing mathematical models that describe the processes behaviour.However,it is common that mathematical models have parameters that must be estimated from experimental data,leading to the propagation of experimental uncertainty into the model.Several methods have been designed and used for uncertainty quantification and sensitivity analysis over the whole model output.In this work,the Standardized Regression Coefficient method has been applied for quantification of parameter relative importance profiles(PRIP)along the simulation time.The results show that parameter importance changes dynamically and the relevance of one parameter could be different for different state variables,which gives insights of the relevant processes taking place at one specific point in time.This methodology could also be used in experimental design and data selection to improve parameter estimation and parameter interpretability.