Mandal Chaitali, Gudi Ravindra D, Suraishkumar G K
Department of Chemical Engineering, IIT Bombay, 400 076 Powai, Mumbai, India.
Bioprocess Biosyst Eng. 2005 Dec;28(3):149-64. doi: 10.1007/s00449-005-0021-4. Epub 2005 Nov 16.
A multi-objective optimization formulation that reflects the multi-substrate optimization in a multi-product fermentation is proposed in this work. This formulation includes the application of epsilon-constraint to generate the trade-off solution for the enhancement of one selective product in a multi-product fermentation, with simultaneous minimization of the other product within a threshold limit. The formulation has been applied to the fed-batch fermentation of Aspergillus niger that produces a number of enzymes during the course of fermentation, and of these, catalase and protease enzyme expression have been chosen as the enzymes of interest. Also, this proposed formulation has been applied in the environment of three control variables, i.e. the feed rates of sucrose, nitrogen source and oxygen and a set of trade-off solutions have been generated to develop the pareto-optimal curve. We have developed and experimentally evaluated the optimal control profiles for multiple substrate feed additions in the fed-batch fermentation of A. niger to maximize catalase expression along with protease expression within a threshold limit and vice versa. An increase of about 70% final catalase and 31% final protease compared to conventional fed-batch cultivation were obtained. Novel methods of oxygen supply through liquid-phase H2O2 addition have been used with a view to overcome limitations of aeration due to high gas-liquid transport resistance. The multi-objective optimization problem involved linearly appearing control variables and the decision space is constrained by state and end point constraints. The proposed multi-objective optimization is solved by differential evolution algorithm, a relatively superior population-based stochastic optimization strategy.
本文提出了一种反映多产物发酵中多底物优化的多目标优化公式。该公式包括应用ε-约束来生成权衡解决方案，以增强多产物发酵中一种选择性产物，同时将另一种产物在阈值范围内最小化。该公式已应用于黑曲霉的补料分批发酵，黑曲霉在发酵过程中产生多种酶，其中过氧化氢酶和蛋白酶的表达被选为目标酶。此外，该公式已应用于三个控制变量的环境中，即蔗糖、氮源和氧气的进料速率，并生成了一组权衡解决方案以绘制帕累托最优曲线。我们已经开发并通过实验评估了黑曲霉补料分批发酵中多种底物进料添加的最优控制曲线，以在阈值范围内最大化过氧化氢酶表达以及蛋白酶表达，反之亦然。与传统补料分批培养相比，最终过氧化氢酶增加了约70%，最终蛋白酶增加了31%。为了克服由于高气液传输阻力导致的曝气限制，采用了通过添加液相H2O2进行供氧的新方法。多目标优化问题涉及线性出现的控制变量，决策空间受状态和端点约束。所提出的多目标优化通过差分进化算法求解，这是一种相对优越的基于种群的随机优化策略。
