Sensitivity analysis of the penicillin fermentation process model based on LHS-EPRCC

The intention of this paper is to evaluate the influences of model parameters on the penicillin fermentation process model. The existing sampling-based global sensitivity analysis (GSA) methods can be used to identify the influences. However, these methods cannot provide quantitative insight into how the uncertainties of model parameters affect the model changes. To address this limitation, a Latin hypercube sampling with the extended partial rank correlation coefficient (LHS-EPRCC) method is proposed in this paper, which incorporates the concept of variance that reflects the degree of change to extend the sensitivity analysis index. First, partial rank correlation coefficient (PRCC) and variance contribution coefficient (VCC) are calculated to reflect the variability of the model to model parameters and the contribution of each model parameter change to the total model change, respectively. Then, importance analysis is carried out to determine the importance parameter set (IPS) that significantly impacts the model. Finally, the proposed method is applied to validate its effectiveness on two types of penicillin fermentation process models.