Modeling And Optimization Method Research And The Software Design For The Fermentation Process

In the fermentation process, some biologic variables are difficult to be measured directlybecause of the strong nonlinear dynamic characteristic and the lack of biosensors. It makesmodeling and optimization control of the process hardly realized. Thus some researches are carriedout, and the modeling and optimization software of the fermentation process has been designed.With the different aims of modeling, there are two modeling methods called "gray-box" and"black-box" modeling technologies in common use in the fermentation processes. Firstly the"gray-box" modeling method is applied for an ethanol fermentation process modeling, good resultshave been obtained. Then the "black-box" modeling technology is mainly studied, which makes useof the neural network in combination with the real product data. Here a new methodology ofdynamic neural network, called recurrent fuzzy neural network with compensatory neurons(RFNNCN), has been proposed. The parameters update learning algorithm is deduced. Then theRFNNCN is applied to model the mycetozoan fermentation process. It has shown that theRFNNCN models can approximate the process and predict the state variables exactly, which is ableto realize the optimization and control of the process.Based on the different models, the optimal target is given. Taking the yield of final product asoptimization goal, the feed rate and pH as well as dissolved oxygen concentration as theoptimization variables, the modified algorithm, genetic algorithm or basic ant algorithm arerespectively applied to optimize the fed-batch fermentation process. The simulation results haveshown that the optimization performance of this improved ant colony algorithm is better, and theoptimization speed is faster. Moreover, the final yield of product of the ethanol process ormycetozoan process is increased by improved ant colony algorithm optimization.With the study of modeling and optimization technology of fermentation processes, a set ofsoftware for modeling and optimization of the fermentation process has been designed by utilizingthe advantages of the strong computational capabilities of Matlab and the interface design of VisualBasic. The software has integrated modeling and optimization as well as dynamic simulationtechnology. The development of the software is not only helpful to study the mechanisms of theprocess, reduce experiments and decrease the research cost, but also helpful to realize thesupervision and optimal control of the process.