发酵过程混合建模与优化控制新方法

针对当前微生物发酵过程存在因为生物传感器不具备足够的准确性和灵敏性,实验时的菌液和产物浓度等生化指标难以实时监测和控制等缺点,提出了采用量子粒子群优化算法(QPSO)优化最小二乘支持向量机(LSSVM)参数的QPSO-LSSVM混合建模新方法,并用于多粘菌素的发酵过程建模;同时,基于此模型,采用QPSO算法对pH值与溶解氧浓度Do控制轨线进行优化研究;首先,利用LSSVM进行发酵过程的建模,然后采用QPSO对LSSVM建模过程中的重要参数进行优化调整,形成QPSO-LSSVM混合建模与优化控制方法;仿真结果表明,该方法得到的模型能取得更好的预测效果,优化后的pH值与Do浓度控制轨线能够提高最终的产物浓度;该方法用于发酵过程的建模和重要参数的优化控制是可行的、有效的。

A New Method of Modeling and Optimized Controlling of Fermentation Process

In view of the current microbial fermentation process exists because of biological sensors do not have enough accuracy and sensitivity, the experiment of the microbial and biochemical indexes such as difficult to real-time monitor and control product concentration such as faults, puts forward using quantum particle swarm optimization (QPSO) algorithm to optimize the least squares support vector machine (LSSVM) parameters of QPSO - new LSSVM hybrid modeling method, and used in the fermentation process modeling polymyxin; At the same time, based on this model, using QPSO algorithm to control the pH and dissolved oxygen concentration Do trajectory optimization research. First of all, the use of LSSVM for fermentation process modeling, and the important parameters in the process of using QPSO on LSSVM modeling optimization adjustment, formed QPSO LSSVM hybrid modeling and optimization control method. Model of the simulation results show that the approach can obtain better prediction result, the optimal pH and the Do concentration control trajectory can improve the final product concentration. The method used in the fermentation process modeling and optimal control of important parameters are feasible and effective.