The Research On Modelling And Optimization Control Of Fermentation Process

Fermentation engineering is the application foundation of biotechnology, and is the core ofthe biotechnology industry.A successful fermentation process was restricted by two factors:First, the genetic characteristics of the bacterium, and second, the fermentation conditions. In recent decades, the fermentation industry has developed rapidly, its scale grow larger and larger. In order to obtain the maximum efficiency of industrial scale, we must ensure that the animalcules grow and synthesize metabolites in the optimal conditions.So the optimization control of fermentation conditions becomes more and more important.An appropriate mathematical model is the basis and prerequisite for process control and optimization, this paper first studies the modeling of fermentation process, and then the process control and optimization.In the modeling, the unstructured model and the. "black box" model of fermentation process wad studied. Aiming at the difficulty of parameters determine, quantum-behaved particle swarm optimization was used for parameter optimization. An appropriate model structure was selected through the analysis of the fermentation process,and then support vector machine was applied to learning fermentation data to create a "black box" model. Further consider the difference between data and model complex degree, the affinity propagation clustering was introduced into modeling. First AP clustering was used for data classification, then SVM was used to regress for each class, at last, outputs of submodels were syncretized to form the final output. This method was applied in the modeling simulation of Nosiheptide fermentation process, the result shows this multi-modeling method has better performance, and it can well describe the relationship between variables in the fermentation process.In optimization control, first a fuzzy-PI dual-mode control method and a error-based multi-region control method were designed for temperature and pH control based on their characteristics, Finally, an optimization simulation was carry out on the base of "black box" model. Through the optimization of operating conditions by QPSO, the product concentration of fermentation process was improve.