A Two-Stage Method for Parameter Identification of a Nonlinear System in a Microbial Batch Process

Abstract

This paper deals with the parameter identification of a microbial batch process of glycerol to 1,3-propanediol (1,3-PD). We first present a parameter identification model for the excess kinetics of a microbial batch process of glycerol to 1,3-PD. This model is a nonlinear dynamic optimization problem that minimizes the sum of the least-square and slope errors of biomass, glycerol, 1,3-PD, acetic acid, and ethanol. Then, a two-stage method is proposed to efficiently solve the presented dynamic optimization problem. In this method, two nonlinear programming problems are required to be solved by a genetic algorithm. To calculate the slope of the experimental concentration data, an integral equation of the first kind is solved by using the Tikhonov regularization. The proposed two-stage method could not only optimally identify the model parameters of the biological process, but could also yield a smaller error between the measured and computed concentrations than the single-stage method could, with a decrease of about 52.79%. A comparative study showed that the proposed two-stage method could obtain better identification results than the single-stage method could.