Parameters Identification For Nonlinear Multistage Dynamic System In Microbial Fermentation

Based on a practical research subject in the biologicalchemical fields that glycerin is converted to1,3-propanediol, in this paper,a problem is studied that how to use parameter identification method tosimulate a class of nonlinear dynamic system according to the features anddynamic behaviors in the fermentation process. Through the improvementof batch and fed-batch fermentation model, the relative error betweencomputing data and experimental data is decreased. The error on microbial,glycerol and1,3-propanediol in the models built in this paper is decreasedby2%, compared that in references. So the dynamical system showed inthis paper can describe the process of batch fermentation and fed-batchculture better.The main work and achievements are summed up as follows:1. Based on the existing batch fermentation models, it is considered thathow the materials produced in the process are across the membrane andthey have some influences on cell growth. Regarding fermentationpiecewise periods as the parameters, a dynamic system model with theintracellular material concentration is established. And related propertiesof the system, existing and uniqueness about its solution are researched,including the continuity and differentiability of solution about parameters.According to the distance between experimental data and calculated value,their minimum error sum of squares is performance index. And theidentification model is conducted which is solved by improved particleswarm algorithm.2. According to identified elapsed time, including the initial and finaltime, three-stage dynamic system about microbial batch fermentation isestablished, and it simulates the intermittent process in detail in some way.The properties of multi-stage dynamic system and the existing anduniqueness about its solution are discussed. Then relative parameteridentification model is built and the model is solved by improved particleswarm algorithm.3. An eight-dimensional nonlinear dynamic system is established on the basis of five-dimensional model about fed-batch fermentation, thenestablish parameter identification model according to the error sum ofsquares about the experimental and calculated data. Finally, apply particleswarm algorithm with Euler’s thought which is used to solve thedifferential equation to identify the unknown parameters.In this paper, it is considered that intermediate product has inhibitioneffects on cell specific growth rate. And glycerol and1,3-propanediol crosscell membrane by active and passive transport way. The establishment ofimproved models about batch fermentation dynamic system and fed-batchfermentation system all reduces the relative error about calculating data andexperimental data.