Adaptive on-line optimizing control of bioreactor systems

Several on-line optimizing control strategies were proposed and tested by computer simulation for the efficient operation of bioreactors. The control task was divided into two, one of which was to search for the optimal operating point and passed the set point to the lower layer of which task was to make the process output follow the set point as soon as possible. It was shown to be effective for the upper layer to express the objective function as a polynomial with respect to the measurement variable and to make use of it for finding the optimum point. Noting that the major dynamic characteristics of bioreactor system is the time-varying and nonlinear nature, the adaptive type control system is in evitable. It was shown to be quite effective to use discrete type self-tuning PID controller and the optimal controller compensated for the interaction between the control loops.Application was made to the cell recycle system for the production of lactic acid and baker's yeast cultivation. I was found from the former application that the control quality can be significantly improved by incorporating the decoupling strategy into the lower layer closed-loop system. It was also found from the latter application that the initial startup period can be significantly reduced by making use of the rough mathematical model.     

Adaptive control of dissolved oxygen concentration in a bioreactor

A new adaptive DO (dissolved oxygen) concentration control algorithm considering DO electrode dynamics with response time delay has been developed. A system model with two time-varying parameters was used to relate the DO concentration with two control variables: air flow rate and agitation speed. Parameters of this model were estimated on-line using a regularized constant trace recursive least-squares method. An extended Kalman filter was used to remove the effect of noises from the DO concentration measurements and thus to improve control performance. A discrete one-step ahead control scheme was adopted to determine control actions based on the parameter estimation results. Experimental results showed that the new adaptive DO concentration control algorithm performed better than other algorithms tested, a PID controller and adaptive algorithms without the DO electrode dynamics.     

A high-yielding, generic fed-batch cell culture process for production of recombinant antibodies

A fed-batch cell culture process was developed that has general applicability to all evaluated Sp2/0 (n = 8) and NS0 (n = 1) antibody-producing cell lines. The two key elements of this generic process were a protein-free concentrated feed medium, and a robust, metabolically responsive feeding strategy based on the off-line measurement of glucose. The fed-batch process was shown to perform equivalently at the 15 L development scale and 750 L manufacturing scale. Compared to batch cultures, the fed-batch process yielded a 4. 3 fold increase in the average integral of viable cell concentration and a 1.7 fold increase in average specific antibody production rate, equivalent to a 7.6 fold increase in average final antibody concentration. The highest producing cell line reached a peak viable cell concentration of 1.0 x 10(7) cell mL(-1) and a final antibody concentration of 750 mg L(-1) in a 10 day process. For all lines evaluated, reducing bioreactor pH set point from 7.2 to 7.0 resulted in an additional 2.4 fold increase in average final antibody concentration. The optimized fed-batch process consistently yielded a volumetric productivity exceeding 50 mg L(-1) day(-1). This generic, high-yielding fed-batch process significantly decreased development time, and increased manufacturing efficiency, thereby facilitating the clinical evaluation of numerous recombinant antibodies.     

A contribution to optimal control of fed-batch biochemical processes

The problem of looking for high efficient modern control strategies in fermentation technology is very urgent, nowdays. Particular attention should be paid to the processes in fed-batch mode. Both, optimal feedforward and feedback control approaches are suggested. A contribution is considered to have been made in the feedback control where continuous and discrete versions are treated as well. The control laws are carried out by a variation calculus problem and a polynomial pole placement synthesis solution, respectively. All the algorithms result in an optimal substrate feed rate profile. On the basis of recursive least squares identification of the model coefficients an adaptive discrete-time control strategy is proposed. Some satisfying simulation results are dealt with.     

Glucose-based optimization of CHO-cell perfusion cultures

Perfusion cultures of CHO cells producing t-PA were performed using acoustic filter cell retention. A robust off-line glucose analysis and predictive control protocol was developed to maintain the process within approximately 0.5 mM of the glucose set point, without the need for a more fallible on-line sensor. Glucose usage (the difference between the inlet and reactor glucose concentrations) provided an easily measured indicator of overall medium utilization for mapping acceptable ranges of operation, including the edge of failure. Earlier onset of perfusion with a ramping glucose set point (1.5 mM/d) resulted in improved growth and consistency during the perfusion culture start-up. At steady state, the t-PA concentration variability increased gradually with increasing glucose usage up to approximately 22 mM, then up to 24 mM the variability increased threefold. Peak t-PA concentrations of over 90 mg/L were obtained by controlling at a glucose usage of approximately 24 mM, but these t-PA levels were not sustainable for more than 3 days. A consistent t-PA concentration of 40 mg/L was obtained at a glucose usage of 21.5 mM.     

Multilayer adaptive control of continuous bioprocesses using optimising control technique. Case study: Bakers' yeast culture

High costs associated with many fermentation processes in an increasingly competitive industry make any prompt application of modern control techniques to industrial bioprocesses very desirable. However, this is often hampered by the lack of adequate mathematical models, on the one hand, and by the absence of continuous, on-line measurement of the most relevant process variables, on the other hand. This paper addresses these problems and offers a new strategy to control continuous bioprocesses using a hierarchical structure such that neither structured process models nor continuous measurement of all relevant variables have to be available. The control system consists of two layers. The lower layer represents a dynamic adaptive follow-up control of a continuously measured output — in our case dissolved oxygen concentration. This variable is supposed to be strongly correlated with the key output variable — in our case cellular concentration which is not continuously available for measurement. The higher layer is then designed to maintain a desired profile of the process key output using a set-point optimising control technique. The Integrated System Optimisation and Parameter Estimation method used operates on an appropriately chosen steady-state performance criterion. A prerequisite for successful application of the proposed approach is an approximate steady-state model, describing the relationship between the measured output and the process key output variable. Furthermore, occasional in situ, off-line or laboratory measurement values of the key output variable are needed. Promising simulation results of the biomass concentration control, by manipulating the air flow-rate in the continuous bakers' yeast culture are presented.     

Computer control of pH and DO in a laboratory fermenter using a neural network technique

In this contribution, the advantages of the artificial neural network approach to the identification and control of a laboratory-scale biochemical reactor are demonstrated. It is very important to be able to maintain the levels of two process variables, pH and dissolved oxygen (DO) concentration, over the course of fermentation in biosystems control. A PC-supported, fully automated, multi-task control system has been designed and built by the authors. Forward and inverse neural process models are used to identify and control both the pH and the DO concentration in a fermenter containing a Saccharomyces cerevisiae based-culture. The models are trained off-line, using a modified back-propagation algorithm based on conjugate gradients. The inverse neural controller is augmented by a new adaptive term that results in a system with robust performance. Experimental results have confirmed that the regulatory and tracking performances of the control system proposed are good.     

表达TNFR-Fc融合蛋白的GS-CHO细胞动态流加培养过程的设计

人肿瘤坏死因子受体Ⅱ-Fc融合蛋白在治疗风湿性、类风湿性关节炎方面拥有广阔的市场前景和巨大的经济价值。本实验以表达TNFR-Fc融合蛋白的GS-CHO细胞为研究对象,结合细胞生长代谢特性和动力学参数分析,以葡萄糖为关键控制参数,通过测定培养上清的葡萄糖浓度对培养过程中的葡萄糖消耗进行及时的预测,调整流加速率,形成了以满足细胞生长代谢需要为基本原则的动态流加培养过程设计模型。在此控制模型指导下,建立了高效的流加培养过程。使最大活细胞密度和最大融合蛋白浓度分别达9.4×106cells/mL和207mg/L,较批次培养分别提高了3.4倍和3倍。本研究所采用的研究方法和控制策略为优化GS-CHO细胞培养过程和TNFR-Fc融合蛋白成功迈向产业化奠定了基础。     

Advanced monitoring and supervision of biological treatment of complex dairy effluents in a full-scale plant

The operation of a wastewater treatment plant treating effluents from a dairy laboratory was monitored by an advanced system. This plant comprises a 12 m(3) anaerobic filter (AF) reactor and a 28 m(3) sequential batch reactor (SBR) coupled in series and is equipped with the following on-line measurement devices: biogas flow meter, feed and recycling flow meters, temperature sensor, dissolved oxygen analyzer, and redox meter. Other parameters such as chemical oxygen demand (COD), volatile fatty acids (VFA), etc. were determined off-line. The plant has been in operation for 634 days, the influent flow rate being 6-8 m(3)/d. COD concentration of the influent ranged between 8 and 12 kg COD/m(3), resulting in COD values in the effluent around 50-200 mg/L. The behavior of the system was studied using the set of measurements collected by the data acquisition program especially developed for this purpose. Monitoring of variables such as anaerobic reactor temperature permitted the detection and prevention of several failures such as temperature shocks in the AF reactor. Besides, off-line measurements such as the alkalinity or the VFA content, together with the on-line measurements, provided immediate information about the state of the plant and the detection of several anomalies, such as organic overloads in the SBR, allowing the implementation of several fast control actions.     

Evaluation of startup and operation of four anaerobic processes treating a synthetic meat waste

Two continuous stirred tanks reactors (CSTR) and four anaerobic fluidized bed reactors (AFBR) were used to study the treatment of a synthetic meat waste during single-and two-stage anaerobic treatment. Four configurations were investigated; a single-stage CSTR and AFBR and the two-stage systems CSTR-AFBR and AFBR-AFBR. Startup of the anaerobic reactors was achieved within 50 days by use of a regime that included stepped increases in influent COD, methanol substitution of the substrate, and addition of essential trace metals such as cobalt and nickel. Two-stage reactors removed up to 85% of influent COD concentrations of 5000 mg/L, whereas the single-stage AFBR and CSTR removed 76 and 9%, respectively. The proportion of methane in the effluent gases increased as the influent COD concentration was increased. Volumetric production of methane was greatest for the first stage of the AFBR-AFBR system. Solids retention times calculated for the AFBRs ranged from 7 to 12 days, sufficient to support methanogenesis. The AFBRs and two-stage systems were more resistant to an influent pH shock from the operating value of pH 6.8 down to pH 3 than the CSTRs and single-stage reactors. It was concluded that high-rate anaerobic treatment systems were applicable to meat industry wastewaters and that two-stage digestion produced a better quality effluent.     

Design of methanol Feed control in Pichia pastoris fermentations based upon a growth model

The methylotrophic yeast Pichia pastoris is an effective system for recombinant protein productions that utilizes methanol as an inducer, and also as carbon and energy source for a Mut(+) (methanol utilization plus) strain. Pichia fermentation is conducted in a fed-batch mode to obtain a high cell density for a high productivity. An accurate methanol control is required in the methanol fed-batch phase (induction phase) in the fermentation. A simple "on-off" control strategy is inadequate for precise control of methanol concentrations in the fermentor. In this paper we employed a PID (proportional, integral and derivative) control system for the methanol concentration control and designed the PID controller settings on the basis of a Pichia growth model. The closed-loop system was built with four components: PID controller, methanol feed pump, fermentation process, and methanol sensor. First, modeling and transfer functions for all components were derived, followed by frequency response analysis, a powerful method for calculating the optimal PID parameters K(c) (controller gain), tau(I) (controller integral time constant), and tau(D) (controller derivative time constant). Bode stability criteria were used to develop the stability diagram for evaluating the designed settings during the entire methanol fed-batch phase. Fermentations were conducted using four Pichia strains, each expressing a different protein, to verify the control performance with optimal PID settings. The results showed that the methanol concentration matched the set point very well with only small overshoot when the set point was switched, which indicated that a very good control performance was achieved. The method developed in this paper is robust and can serve as a framework for the design of other PID feedback control systems in biological processes.     

On-line monitoring and control of substrate concentrations in biological processes by flow injection analysis systems

Concentrations of substrates, glucose, and ammionia in biological processes have been on-line monitored by using glucose-flow injection (FIA) and ammonia-FIA systems. Based on the on-line monitored data the concentrations of substrates have been controlled by an on-off controller, a PID controller, and a neural network (NN) based controller. A simulation program has been developed to test the control quality of each controller and to estimate the control parameters. The on-off controller often produced high oscillations at the set point due to its low robustness. The control quality of a PID controller could have been improved by a high analysis frequency and by a short residence time of sample in a FIA system. A NN-based controller with 3 layers has been developed, and a 3(input)-2(hidden)-1(output) network structure has been found to be optimal for the NN-based controller. The performance of the three controllers has been tested in a simulated process as well as in a cultivation process ofSaccharomyces cerevisiae, and the performance has also been compared to simulation results. The NN-based controller with the 3-2-1 network structure was robust and stable against some disturbances, such as a sudden injection of distilled water into a biological process.     

Adaptive fuzzy control of nutrients concentration in fed-batch culture of mammalian cells

An adaptive fuzzy controller was developed to control the glucose and glutamine concentrations in the reactor constant at the desired level. The parameter values of the controller change during the cultivation according to the culture phase which was detected by the lactate concentration. Cultivations with different glucose and glutamine set point concentrations of a recombinant BHK anchorage-dependent cell line were performed in a fed-batch reactor on-line connected with an HPLC system. Glucose and glutamine concentrations were satisfactorily controlled at each set point during all cultivation periods. Ammonia had a determining effect on productivity since it inhibited cell growth and protein specific production. Ammonia production increased with an increase of glutamine or a decrease of glucose set point concentrations, indicating the importance of glucose to glutamine ratio for the optimization of productivity in mammalian cell cultures. This revised version was published online in June 2006 with corrections to the Cover Date.     

An on-line adaptive glucose feeding system incorporating patterns recognition for glucose concentration control in glutamate fermentations

In glutamate fermentation, intermittent feeding is the most widely used glucose feed strategy. This feeding strategy causes severe fluctuations of glucose concentration and osmotic pressure in fermentation broth, which deteriorates the viability of the cell and reduces glutamate production in turn. In order to maintain glucose concentration at stable and constant levels, an on-line prediction and feedback control system based an empiric mass balance model was developed. However, the control system did not work properly and sometimes glucose concentration could even decline to 0 level (glucose exhaustion), as the model parameter varies in different runs. As a result, a novel model-based adaptive feedback control system incoporating with an artificial neural network (ANN) based pattern reconition unit for on-line diagnosizing the fault of glucose exhaustion was proposed and applied for glutamate fermentation. This adaptive control system could accurately detect glucose exhaustion when it occurs, and then immediately updates the control parameter based on some pre-defined rule. With the proposed control system, glucose was automatically fed, and its concentration could be maintained at desired levels constantly. As a result, glutamate concentration was 17 ~ 30% higher than that of the traditional fermentations using the intermittent glucose feed strategy.     

Software sensing for glucose concentration in industrial antibiotic fedbatch culture using fuzzy neural network

In order to control glucose concentration during fed-batch culture for antibiotic production, we applied so called “software sensor” which estimates unmeasured variable of interest from measured process variables using software. All data for analysis were collected from industrial scale cultures in a pharmaceutical company. First, we constructed an estimation model for glucose feed rate to keep glucose concentration at target value. In actual fed-batch culture, glucose concentration, was kept at relatively high and measured once a day, and the glucose feed rate until the next measurement time was determined by an expert worker based on the actual consumption rate. Fuzzy neural network (FNN) was applied to construct the estimation model. From the simulation results using this model, the average error for glucose concentration was 0.88 g/L. The FNN model was also applied for a special culture to keep glucose concentration at low level. Selecting the optimal input variables, it was possible to simulate the culture with a low glucose concentration from the data sets of relatively high glucose concentration. Next, a simulation model to estimate time course of glucose concentration during one day was constructed using the on-line measurable process variables, since glucose concentration was only measured off-line once a day. Here, the recursive fuzzy neural network (RFNN) was applied for the simulation model. As the result of the simulation, average error of RFNN model was 0.91 g/L and this model was found to be useful to supervise the fed-batch culture.     

Adaptive Control Based Harvesting Strategy for a Predator–Prey Dynamical System

This paper deals with designing a harvesting control strategy for a predator–prey dynamical system, with parametric uncertainties and exogenous disturbances. A feedback control law for the harvesting rate of the predator is formulated such that the population dynamics is asymptotically stabilized at a positive operating point, while maintaining a positive, steady state harvesting rate. The hierarchical block strict feedback structure of the dynamics is exploited in designing a backstepping control law, based on Lyapunov theory. In order to account for unknown parameters, an adaptive control strategy has been proposed in which the control law depends on an adaptive variable which tracks the unknown parameter. Further, a switching component has been incorporated to robustify the control performance against bounded disturbances. Proofs have been provided to show that the proposed adaptive control strategy ensures asymptotic stability of the dynamics at a desired operating point, as well as exact parameter learning in the disturbance-free case and learning with bounded error in the disturbance prone case. The dynamics, with uncertainty in the death rate of the predator, subjected to a bounded disturbance has been simulated with the proposed control strategy.     

Enhancement of alkaline polygalacturonate lyase production in recombinant Pichia pastoris according to the ratio of methanol to cell concentration

Polygalacturonate lyase (PGL) production by Pichia pastoris GS115 was used as a model to study the mechanism and strategy for enhancing heterologous protein production. It was found that the ratio of methanol to cell concentration had a significant influence on PGL production. In this study, an advanced glycerol exponential feeding strategy was developed for biomass accumulation in cell growth phase, by which cell concentration reached 140 g L(-1) after 19 h glycerol feeding. In subsequent production phase, a methanol feeding profile was proposed according to the optimal ratio of methanol to cell concentration at a range of 0.163-0.171 g g(-1), and PGL activity and productivity reached 430 U mL(-1) and 4.34 U mL(-1)h(-1), respectively. The strategy for enhancing PGL production by controlling the optimal ratio may provide an alternative approach to enhance heterologous protein production with P. pastoris.     

木质纤维素燃料乙醇蒸馏废水的高温厌氧处理及菌群结构分析

【目的】为开发高效的高浓度木质纤维素燃料乙醇蒸馏废水厌氧处理及资源化利用工艺,以活性炭为载体,在实验室规模上对高温厌氧流化床反应器处理木质纤维素燃料乙醇蒸馏废水进行研究。【方法】反应器经65 d梯度驯化后启动,对工艺参数进行一系列优化,并通过基于16S rRNA基因的分子生态学技术分析厌氧污泥中的优势菌群。【结果】实验获得了最优的反应条件和处理效果:厌氧流化床反应器(Anaerobic fluidized bed reactor,AFBR)在温度55±1°C、有机负荷率(OLR)13.8 g COD/(L·d)及水力停留时间(HRT)48 h操作时,COD去除率达到90%以上,同时甲烷产率达到290 mL/g COD;菌群鉴定分析结果显示高温厌氧活性污泥中Clostridia所占比例最大,产甲烷菌属以Methanoculleus和Methanosarcina为主,其它功能菌群主要为Alphaproteobacteria等。【结论】AFBR反应器可高效降解木质纤维素燃料乙醇蒸馏废水并产生生物能源甲烷,其反应体系内微生物种类丰富。     

On-line monitoring and control of methanol concentration in shake-flask cultures of Pichia pastoris

The methylotrophic yeast Pichia pastoris can be used to express recombinant genes at high levels under the control of the methanol-inducible alcohol oxidase 1 (AOX1) promoter. Accurate regulation of the methanol concentration in P. pastoris cultures is necessary to maintain induction, while preventing accumulation of methanol to cytotoxic levels. We developed an inexpensive methanol sensor that uses a gas-permeable silicone rubber tube immersed in the culture medium and an organic solvent vapor detector. The sensor was used to monitor methanol concentration continuously throughout a fed-batch shake-flask culture of a P. pastoris clone producing the N-lobe of human transferrin. The sensor calibration was stable for the duration of the culture and the output signal accurately reflected the methanol concentration determined off-line by HPLC. A closed-loop control system utilizing this sensor was developed and used to maintain a 0.3% (v/v) methanol concentration in the culture. Use of this system resulted in a fivefold increase in volumetric protein productivity over levels obtained using the conventional fed-batch protocol. (c) 1997 John Wiley & Sons, Inc. Biotechnol Bioeng 56: 279-286, 1997.