Projects

Principal Investigator

• 

Team Members - CEB

• 
• 
• 

Abstract

Some valuable products such as recombinant proteins, antibiotics and amino-acids are produced using fermentation techniques. Thus, there is a tremendous economic incentive to decrease the costs of biotechnologically produced products. However, those processes are complex, involving different phenomena, microbial components and biochemical reactions. The nonlinear behavior and time-varying properties make bioreactors difficult to control with traditional techniques. So, there is the need to consider reliable mathematical models to describe the process dynamics and the interrelation among relevant variables. Additionally, robust optimization techniques must deal with the model´s complexity, environment constraints and the inherent noise of the experimental process.\nThe optimization of these processes is traditionally handled by analytical and numerical methods, whose results degrade when the complexity increases. A different approach comes from general purpose optimization algorithms, taken from the field of Evolutionary Computation (EC). \nOn the other hand, the modeling of fermentation processes typically uses white box mathematical models, based on differential equations. Due to difficulties in accommodating the available knowledge or to the lack of information on a particular process, sometimes it is not possible to fully describe process behavior. More complex approaches have been proposed that take into account the nonlinear nature of the process. In this arena, Neural Networks (NNs) have been a focus of attention.\nThis project aims at studying the application of models that combine the use of NNs in the modeling of the process dynamics and EC algorithms, namely Evolutionary Algorithms and Particle Swarm Optimization, as optimization tools. A case study will be used, where a recombinant bacterial fed-batch fermentation process aims at the production of a bio-pharmaceutical product.\nThe project is organized into the following stages:\n- The selection of the best mutant strain capable of achieving a high productivity in the recombinant fermentation process. EC algorithms are used to optimize off-line process productivity in a bi-level optimization methodology where the manipulated variable is the genome of the organism. The behavior of the process is represented by a linear programming problem with the constraints given by the stoichiometry of the metabolism\n- the optimization of the fermentation process, prior to its start. EC algorithms are used to optimize the initial values of some variables, the trajectory of control variables over time and the duration of the process. Fermentation processes are simulated by differential equations based models.\n- the optimization in real time, where the EC algorithm is continuously running, receiving online measurements of variables and updating its best solution.\n- the study of NNs to model the process dynamics, validated by experimental data.\n- the adjustment of the NN based models and simultaneous optimization of relevant variables in real time, i.e., while the fermentation process is under way.\n