Robust Analysis And Parallel Optimization Of Nonlinear Hybrid System In Microbial Continuous Fermentation

This dissertation, on the basis of microbial production of1,3-propanediol using glycerol as substrate by Klebsiella pneumoniae, in the context of the existing uncertain metabolic mech-anisms and the lack of intracellular experimental results in continuous cultures,studied the modeling, robustness analysis,system identification and parallel optimization of a class of non-linear enzyme-catalytic hybrid systems.The research is supported by the National Natural Science Foundation of China, the Major State Basic Research Development Program of China and the National High Technology Research and Development Program of China. The research can not only develop the study of the nonlinear hybrid dynamical system, system identification, parallel computing and parallel optimization algorithm, but also be helpful for deeply under-standing of metabolic pathways of glycerol fermentation by Klebsiella pneumoniae.The main results obtained in this dissertation are summarized as follows:1.The process of glycerol bioconversion to1,3-propanediol by Klebsiella pneumoniae can be characterized an intricate network which includes biochemical fluxes,metabolic com-pounds,key enzymes and genetic regulatory. According to different metabolic mecha-nism of microbial fermentation process in the continuous culture,we present two kinds of nonlinear hybrid systems,with parameters related or irrelated to the control variables (initial glycerol concentration and dilution rate).And then the Lipschitz continuity, linear growth,the existence and uniqueness of the solutions to the nonlinear hybrid systems are explored, as well as the continuity of the solution with respect to the parameters.2.Since there still exists some uncertain factors in this complex network and the limitation in bio-techniques,based on the idea of biological robustness,we present three different quantitative definitions of robustness index of the intracellular substances to deal with the system identification problem.And then,taking the robustness index of the intracellular substances and relative enzyme-catalytic a group of nonlinear hybrid systems or multi-group of hybrid systems,respectively, as performance index and constraint, two kind of system identification models are formulated and the existence of the optimal solution is proved.3.The identification problems,which are nonlinear, contain implicit dynamic constraints, and the continuous and discrete two variables,are unable to get the analytical solutions.In the process of numerical calculation, the numerical solution of differential equations need to be solved for thousands times.That is difficult to achieve for ordinary PC machine. So, we construct parallel optimization algorithms to solve the parameter identification models and to get the most possible metabolic pathway and the optimal parameters of the corresponding systems.