A genetic-algorithm-based approach to optimization of bioprocesses described by fuzzy rules

A new approach to optimization of bioprocesses described by fuzzy rules is introduced in the paper. It is based on genetic algorithms (GA) and allows to determine optimal values or profiles of control variables and to optimize fuzzy rules (parameters of membership functions). The process can be described by linguistic variables and fuzzy rules. An algorithm and related software was developed. The approach was applied to an industrial antibiotic fermentation. The optimal profile of a physical variable of the preculture was determined which leads to an increasing output product concentration in the main culture of about 5%.     

Computer control for continuous cultivation ofSaccharomyces cerevisiae

Summary An on-line respiratory quotient control system has been developed for the continuous cultivation of baker's yeast. This system is based on moving identification of the microbial dynamics. The optimal dilution rate that was selected as the control variable was determined by minimizing a performance index. Without resorting to complicated microbial analysis, a simple and practical moving model is obtained by continually updating the input and output data. The experimental results indicate the satisfactory controllability of the present system and the possible extention of the proposed method to other bioprocesses.     

Stable adaptive algorithm for simultaneous estimation of time-varying parameters and state variables in aerobic bioprocesses

The application of modern model based control algorithms in the bioprocesses is hampered by the lack of accurate and cheap on-line sensors, capable of providing on-line measurements of the main process variables and parameters. In this paper, a new approach for estimation of immeasurable time-varying parameters and state variable is presented for a class of aerobic bioprocesses using only on-line measurements of the oxygen uptake rate. The approach consists in the design of a new parameter estimator of biomass growth rate and yield coefficient for oxygen consumption on the basis of the theory of adaptive estimation. The dynamical equation of the measurable reaction rate, oxygen uptake rate, is presented as a linear one with respect to the biomass growth rate and the yield coefficient for oxygen consumption. In this way, the structure of the proposed estimator becomes linear time-varying one. After some mathematical transformations, that structure is presented in a form, allowing to be derived the stability conditions using some theoretical results concerning the stability of adaptive observers. The estimates of the yield coefficient for oxygen consumption, the biomass concentration and specific growth rate are obtained then on the basis of the generated estimates using well known kinetic models of bioprocesses. With respect to previous similar approaches, the new estimation algorithm gives stable estimates not only of immeasurable state variable and reaction rates but likewise of an yield coefficient. The behavior of the proposed estimator is studied under inexact initial conditions, step changes of dilution rate and in the presence of measurement noise by simulations using a process model, which belongs to the investigated class of bioprocesses.     

Maximization of steady-state bacterial production in a chemostat with pH and substrate control.

This analytical study deals with the steady-state behavior and control of microbial growth in continuous cultures. A second order Haldane-Monod model of continuous cultures is used as a basis for study of the effects of the adjustment of pH by the addition of acidic (or basic) materials. The treatment of a hydrogen ion concentration, in addition to substrate and microbial concentrations as state variables, results in a third order system of equations describing the process. The analysis of the system in equilibrium yields several admissible steady states, that is, steady states which satisfy all constraints. An optimal control problem is formulated and subsequently solved to maximize steady-state microbial production.     

The potential of random forest and neural networks for biomass and recombinant protein modeling in Escherichia coli fed‐batch fermentations

Product quality assurance strategies in production of biopharmaceuticals currently undergo a transformation from empirical “quality by testing” to rational, knowledge‐based “quality by design” approaches. The major challenges in this context are the fragmentary understanding of bioprocesses and the severely limited real‐time access to process variables related to product quality and quantity. Data driven modeling of process variables in combination with model predictive process control concepts represent a potential solution to these problems. The selection of statistical techniques best qualified for bioprocess data analysis and modeling is a key criterion. In this work a series of recombinant Escherichia coli fed‐batch production processes with varying cultivation conditions employing a comprehensive on‐ and offline process monitoring platform was conducted. The applicability of two machine learning methods, random forest and neural networks, for the prediction of cell dry mass and recombinant protein based on online available process parameters and two‐dimensional multi‐wavelength fluorescence spectroscopy is investigated. Models solely based on routinely measured process variables give a satisfying prediction accuracy of about ± 4% for the cell dry mass, while additional spectroscopic information allows for an estimation of the protein concentration within ± 12%. The results clearly argue for a combined approach: neural networks as modeling technique and random forest as variable selection tool.     

Real-time fuzzy-knowledge-based control of Baker's yeast production

A real-time fuzzy-knowledge-based system for fault diagnosis and control of bioprocesses was constructed using the object-oriented programming environment Small-talk/V Mac. The basic system was implemented in a Macintosh Quadra 900 computer and built to function connected on line to the process computer. Fuzzy logic was employed in handling uncertainties both in the knowledge and in measurements. The fuzzy sets defined for the process variables could be changed on-line according to process dynamics. Process knowledge was implemented in a graphical two-level hierachical knowledge base. In on-line process control the system first recognizes the current process phase on the basis of top-level rules in the knowledge-base. Then, according to the results of process diagnosis based on measurement data, the appropriate control strategy is subsequently inferred making use of the lower level rules describing the process during the phase in question. (c) 1995 John Wiley & Sons, Inc.     

Nonlinear feedforward control of bioreactors with input multiplicities

A steady-state nonlinear feedforward controller (FFC) for measurable disturbances is designed for a continuous bioreactor, which is represented by Hammerstein type nonlinear model wherein the nonlinearity is a polynomial with input multiplicities. The manipulated variable is the feed substrate concentration (S_f) and the disturbance variable is the dilution rate (D). The productivity (Q=DP) is considered as the controlled variable. The desired value of Q=3.73 gives two values of feed substrate concentration. The nonlinearity in the gain is considered for relating output to the manipulated variable and separately for the relation between output to disturbance variable. The FFC is also designed for the overall linearized system. The performance of the FFC is evaluated on the nonlinear differential equation model. The FFC is also designed for the model based on a single nonlinear steady-state equation containing both D and S_f. This nonlinear FFC gives the best performance. The nonlinear FFC is also designed by using only linear gain for the disturbance and nonlinear gain for the manipulated variable. Similarly, nonlinear FFC is also designed by using linear gain for the manipulated variable and the nonlinear gain for the disturbance variable. The performances of these FFC schemes are compared.     

Simulation of L/A control in fed-batch culture of baker's yeast

L/A controllers have extended their use from continuous to fed-batch fermentation where the control is applied from the start of an initial batch phase. As opposed to proportional integral derivative (PID) controllers where even a startup procedure is recommended prior to fed-batch, the L/A controller is not upset by an early connection. It is easily retuned continuously by means of ethanol measurements and can cope with a large range of output conditions. The performance of an L/A algorithm, which uses biomass concentration as the controlled variable, is assessed through simulation. The self-contained algorithm is relatively simple with no greater intrinsic complexity than modern PID stand alone controllers.     

Control strategies of continuous bioprocesses based on biological activities

Based on the material balance principle applied to microbial reactions in continuous bioprocesses, the concept of reaction rate control has been developed theoretically. This concept provides a more direct way of controlling biological activities than the control of physical or chemical parameters in practice today. From an analysis of dynamic and steady-state experiments, two control systems for carbon dioxide production rate control during the continuous culture of baker's yeast have been designed and evaluated experimentally. In these control methods, intracellular NADH concentration is used as an immediate indication of the onset of glucose repression. A more sophisticated master controller based on the respiratory quotient can be combined with these control methods. The resulting control system provides a means to indirectly optimize biomass production while preventing ethanol formation in the continuous culture of baker's yeast.     

Enhanced production of L-(+)-lactic acid in chemostat by Lactobacillus casei DSM 20011 using ion-exchange resins and cross-flow filtration in a fully automated pilot plant controlled via NIR

Due to the lack of suitable in-process sensors, on-line monitoring of fermentation processes is restricted almost exclusively to the measurement of physical parameters only indirectly related to key process variables, i.e., substrate, product, and biomass concentration. This obstacle can be overcome by near infrared (NIR) spectroscopy, which allows not only real-time process monitoring, but also automated process control, provided that NIR-generated information is fed to a suitable computerized bioreactor control system. Once the relevant calibrations have been obtained, substrate, biomass and product concentration can be evaluated on-line and used by the bioreactor control system to manage the fermentation. In this work, an NIR-based control system allowed the full automation of a small-scale pilot plant for lactic acid production and provided an excellent tool for process optimization. The growth-inhibiting effect of lactic acid present in the culture broth is enhanced when the growth-limiting substrate, glucose, is also present at relatively high concentrations. Both combined factors can result in a severe reduction of the performance of the lactate production process. A dedicated software enabling on-line NIR data acquisition and reduction, and automated process management through feed addition, culture removal and/or product recovery by microfiltration was developed in order to allow the implementation of continuous fermentation processes with recycling of culture medium and cell recycling. Both operation modes were tested at different dilution rates and the respective cultivation parameters observed were compared with those obtained in a conventional continuous fermentation. Steady states were obtained in both modes with high performance on lactate production. The highest lactate volumetric productivity, 138 g L(-1) h(-1), was obtained in continuous fermentation with cell recycling.     

Analyzing and understanding the robustness of bioprocesses

The robustness of bioprocesses is becoming increasingly important. The main driving forces of this development are, in particular, increasing demands on product purities as well as economic aspects. In general, bioprocesses exhibit extremely high complexity and variability. Biological systems often have a much higher intrinsic variability compared with chemical processes, which makes the development and characterization of robust processes tedious task. To predict and control robustness, a clear understanding of interactions between input and output variables is necessary. Robust bioprocesses can be realized, for example, by using advanced control strategies for the different unit operations. In this review, we discuss the different biological, technical, and mathematical tools for the analysis and control of bioprocess robustness.     

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.     

A scaled-down model for the translation of bacteriophage culture to manufacturing scale

Therapeutic bacteriophages are emerging as a potential alternative to antibiotics and synergistic treatment of antimicrobial-resistant infections. This is reflected by their use in an increasing number of recent clinical trials. Many more therapeutic bacteriophage is being investigated in preclinical research and due to the bespoke nature of these products with respect to their limited infection spectrum, translation to the clinic requires combined understanding of the biology underpinning the bioprocess and how this can be optimized and streamlined for efficient methods of scalable manufacture. Bacteriophage research is currently limited to laboratory scale studies ranging from 1–20 ml, emerging therapies include bacteriophage cocktails to increase the spectrum of infectivity and require multiple large-scale bioreactors (up to 50 L) containing different bacteriophage–bacterial host reactions. Scaling bioprocesses from the milliliter scale to multi-liter large-scale bioreactors is challenging in itself, but performing this for individual phage-host bioprocesses to facilitate reliable and robust manufacture of phage cocktails increases the complexity. This study used a full factorial design of experiments approach to explore key process input variables (temperature, time of infection, multiplicity of infection, agitation) for their influence on key process outputs (bacteriophage yield, infection kinetics) for two bacteriophage–bacterial host bioprocesses (T4 – Escherichia coli; Phage K – Staphylococcus aureus). The research aimed to determine common input variables that positively influence output yield and found that the temperature at the point of infection had the greatest influence on bacteriophage yield for both bioprocesses. The study also aimed to develop a scaled down shake-flask model to enable rapid optimization of bacteriophage batch bioprocessing and translate the bioprocess into a scale-up model with a 3 L working volume in stirred tank bioreactors. The optimization performed in the shake flask model achieved a 550-fold increase in bacteriophage yield and these improvements successfully translated to the large-scale cultures.     

Nonlinear control for algae growth models in the chemostat

This paper deals with output feedback control of phytoplanktonic algae growth models in the chemostat. The considered class of model is of variable yield type, meaning that the ratio between the environmental nutrient absorption rate and the cells’ growth rate varies, which is different from classical bioprocesses assumptions. On the basis of weak qualitative hypotheses on the analytical expressions of the involved biological phenomena (which guarantee robustness of the procedure toward modeling uncertainties) we propose a nonlinear controller and prove its ability to globally stabilize such processes. Finally, we illustrate our approach with numerical simulations and show its benefits for biological laboratory experiments, especially for ensuring persistence of the culture facing classical experimental problems.     

On-line measurement in biotechnology: techniques.

Bioprocesses are generally ill controlled. This is due to the fact that the measurement of relevant variables is difficult. Therefore, fundamental knowledge of metabolic interrelations is, at least in vivo, limited. In this article, some of the most important measurement techniques are reviewed in order to provide an evaluation of their current state. Emphasis is given to the underlying principles and on-line capability which allow to judge their importance and potential for exploitation resulting in well (maybe entirely) controlled bioprocesses in the future.     

Loss of control biomechanics of the human arm-elbow system

An experimental study was conducted to determine whether external disturbance oscillations, such as those that could be created by hand held tools, alter the dynamic response characteristics of the human arm-muscle system. A special arm-test frame was used to induce external sinusoidal torque oscillations of various amplitudes and frequencies, while the reaction force and angular displacement were monitored. Two different output variable frequency responses were determined using input/output cross-spectrum analysis. The angular displacement of the test frame and a component of hand reaction force were the output variables used, while the test frame torque was the input. Test results from one subject are presented in this paper. Changes in the magnitude and phase angle of the frequency responses were observed for different frequencies of the disturbance torque. These changes indicate that the stability margin and response amplitude of the human arm-muscle system do change as a function of the frequency and amplitude of external disturbance oscillations. This suggests that at certain operating frequencies hand held tools can induce large reaction amplitudes or even loss of control.     

Use of at-line spectrophotometry for the rapid definition of pilot-scale flocculation processes

Traditionally most downstream bioprocesses have been operated without real-time knowledge of product and key contaminants, yielding little confidence in their operation and the impact on subsequent operations. A rapid UV-vis spectral prediction technique has been successfully demonstrated for the at-line characterization of a large scale continuous flocculation process in terms of RNA, key protein contaminants, and cell debris. A comparison was made between the spectral predictions and retrospective wet chemical assays, and a highly linear correlation was obtained. The spectral analysis technique allowed for real-time system information, which was applied to control the flocculation process to maintain satisfactory process performance, even when subjected to given possible process disturbances.     

Theory of diseases of steady-state proportional control systems

In many diseases of a semi-stationary nature (chronical diseases) the level of the regulated variable of the diseased system either is too high (a hyper-state disease) or too low (a hypo-state disease). In this paper the steady-state behavior of proportional control systems is, hence, analysed with regard to a longterm pathological change of each single variable or parameter. Each pathological change has its own pattern of changes of the system variables, which also depends upon the system class (left- or right-regulating). Both the block diagram and the regulation characteristic, when used together, allow one to easily derive the pathological behaviour of the system in the steady-state.     

Disturbance-accommodating control approach applied for the control of a continuous stirred tank reactor

This paper describes an experimental study of linear adaptive control to achieve the monitoring of a continuous stirred tank reactor. The practical control objective was the regulation of the substrate concentration at a pre-specified value in the process effluent despite local changes and/or culture physiology variations. The substrate concentration and the dilution rate have been selected as the controlled and the control variable respectively. The results obtained confirm that this approach offers the possibility to combine simplicity and effectiveness in bioprocess control.