Research On Optimization And Control Approaches For A Class Of Biochemical Processes

With the rapid development of bioengineering technology, biotechnology industry is playing a more and more important role in national economy. But it is difficult to optimize and control a biochemical process due to its inherent nonlinearity, time variation, uncertainty and so on. Thus, it is necessary to develop and establish those control and optimization techniques that are suitable for this class of systems. This will help to enhance the production rate of target products and conversion rate of raw materials. Hence, the whole production level of a biochemical process can be improved.In this dissertation several optimization and control approaches for a class of biochemical processes are studied and applied to some real plants, especially in tryptophan biosynthesis and bio-dissimilation process of glycerol to 1, 3-propanediol. The attained results can not only realize the optimum operation and process control of biochemical systems, but also have important academic meaning and applying value in the research on nonlinear optimization and control algorithms. The main contributions and obtained results are summarized as follows:(1) To achieve the steady-state optimization of tryptophan biosynthesis in Escherichia coli, a steady-state optimization model that maximizes the accumulative term accounting for both consumption and secretion is established by using a novel three-dimension nonlinear dynamic system of tryptophan biosynthesis. Based on the Indirect Optimization Method (IOM), the nonlinear model describing tryptophan biosynthesis is firstly transformed into an S-system. Then the original nonlinear optimization problem is approximately reduced to a linear optimization problem. The results show that it is possible to attain a stable and robust steady-state with a rate of tryptophan production increased about 9 times. Compared with the existing findings, both the higher rate of tryptophan production and better robust optimum steady-state are obtained.(2) By the study on the steady-state optimization of tryptophan biosynthesis, it can be seen that the standard iterative IOM approach hardly yields the true optimum solution. To overcome this difficulty, a new algorithm is proposed to solve the steady-state optimization problem of biochemical systems. An additional equality constraint to account for the consistency of solutions between the S-system and the original model is introduced into the existing linear optimization problem of the direct IOM method. And using the general Lagrangian multiplier method, the resulting optimization problem is modified as an equivalent linear optimization problem. Finally, the modified iterative IOM algorithm is applied to the steady-state optimization of several different types of biochemical systems. The results show that the presented algorithm can obtain the real optimum solutions of biochemical systems.(3) The H_∞control of biochemical processes is studied. According to the material balance equations of continuous biochemical processes, a uniform framework for mathematical modeling of this class of processes is presented. A robust controller is designed by using the bilinear transformation and H_∞mixed sensitivity method for biochemical processes. Under the controller a biochemical system can work near an optimal steady-state for the volumetric productivity of some desired product attaining its maximization. The design procedure is carried out by tuning the transformation parameter and DC gain of the performance weighted function. The former is the key parameter in placing the dominant closed-loop poles at the desired locations. And the latter determines the steady-state tracking error of the system. Finally, the proposed H_∞control strategy is applied to bio-dissimilation process of glycerol to 1, 3-propanediol. Simulation results are given which show that the designed robust controller not only ensures the robust stability of the system in face of the parametric variations in the model, but also makes the system has a favourable robust tracking performance. The validity of the presented H_∞controller has been tested.(4) The on-line steady-state optimizing control of biochemical processes is studied. An iterative optimization strategy is proposed and applied to the steady-state optimizing control of biochemical processes in the presence of model-plant mismatch and input constraints. The scheme is based on the augmented Integrated System Optimization and Parameter Estimation (ISOPE) technique, but a linearization of some performance function is introduced to overcome the difficulty of determining an appropriate penalty coefficient. When carrying out the iterative optimization, the penalty coefficient is updated at each iteration, which can promote the evolution rate of the iterative optimization. The effects of measurement noise, measured and unmeasured disturbances on the presented algorithm are also investigated. Simulation studies illustrate that the proposed approach has a self-optimizing ability to rapidly improve the real-time performance.(5) To characterize the nonlinear nature of real biochemical systems, a nonlinear process model with the structure of power-law functions is usually used to describe the steady-state characteristic of true plants. A modified Integrated System Optimization and Parameter Estimation technique (ISOPEN2) for determining the steady-state optimizing control of biochemical processes in logarithmic space is proposed. In the optimization scheme, using the logarithmical transformation, the original problem is firstly transformed into an equivalent optimizing control problem that can be solved in logarithmic space. Then using the linearizations of both the objective and constraint functions, the original problem is further reduced to a quadratic programming problem, which can overcome the difficulty of solving a non-convex optimization problem to choose an appropriate penalty coefficient. Simple filter approaches are emplyed to improve the algorithm performance in the presence of noise. The simulation results show that the performance of the proposed algorithm is superior to that of the traditional ISOPE approach in convergence rate and computation time.