Neural and Hybrid Neural Modeling and Control of Fed‐Batch Fermentation for Streptokinase: Comparative Evaluation under Nonideal Conditions

Fermentations involving competition between two or more kinds of cells under nonideal conditions show complex profiles that are sensitive to the extra‐cellular environment. These fermentations therefore require accurate and rapid on‐line data acquisition and control. However, both on‐line measurements and modelling are difficult and expensive for large bioreactors, thus limiting the usefulness of model‐based control. While neural networks offer an alternative, they require extensive training and can be difficult to optimize for large arrays. Hybrid networks combining a few neural networks with some mathematical equations offer a good compromise. The possibility of using a hybrid model for simulation‐cum‐control has been examined here for the fed‐batch production of streptokinase. Under noideal conditions, hybrid neural models outperformed both mathematical models and arrays of neural networks, thus suggesting their viability for large‐scale fermentation monitoring and control.     

Application of artificial neural networks to the analysis of two‐dimensional fluorescence spectra in recombinant E coli fermentation processes

A two‐dimensional (2D) spectrofluorometer was used to monitor various fermentation processes with recombinant E coli for the production of 5‐aminolevulinic acid (ALA). The whole fluorescence spectral data obtained during a process were analyzed using artificial neural networks, ie self‐organizing map (SOM) and feedforward backpropagation neural network (BPNN). The SOM‐based classification of the whole spectral data has made it possible to qualitatively associate some process parameters with the normalized weights and variances, and to select some useful combinations of excitation and emission wavelengths. Based on the classified fluorescence spectra a supervised BPNN algorithm was used to predict some of the process parameters. It was also shown that the BPNN models could elucidate some sections of the process's performance, eg forecasting the process's performance. Copyright © 2005 Society of Chemical Industry     

基于神经网络的发酵时间和最优发酵温度模型

以提高间歇式微生物发酵的产品得率为目标 ,利用BP神经网络和本文提出的傅立叶神经网络 ,提出发酵过程的发酵时间模型和最优发酵温度模型 ;在此基础上 ,提出了针对不同生产批次采用不同的最优发酵温度的新方法 ,此方法使不同生产批次的发酵过程都可以在适合其自身的最优的发酵温度下进行发酵 ,从而提高发酵生产的得率。实践应用表明 ,采用此种生产方案 ,产品平均得率提高 5 %。     

Neural‐Networks‐Based Feedback Linearization versus Model Predictive Control of Continuous Alcoholic Fermentation Process

In this work advanced nonlinear neural networks based control system design algorithms are adopted to control a mechanistic model for an ethanol fermentation process. The process model equations for such systems are highly nonlinear. A neural network strategy has been implemented in this work for capturing the dynamics of the mechanistic model for the fermentation process. The neural network achieved has been validated against the mechanistic model. Two neural network based nonlinear control strategies have also been adopted using the model identified. The performance of the feedback linearization technique was compared to neural network model predictive control in terms of stability and set point tracking capabilities. Under servo conditions, the feedback linearization algorithm gave comparable tracking and stability. The feedback linearization controller achieved the control target faster than the model predictive one but with vigorous and sudden controller moves.     

Non‐Linear Predictive Control of a Three‐Phase Catalytic Reactor

In this work, the predictive control of a three‐phase catalytic reactor is considered. A predictive control algorithm, which has a non‐linear internal model represented by functional link networks, is proposed. This network structure has been shown to have a good non‐linear approximation capability, with the advantage that the estimation of its weight is a linear optimization problem. The results show the potential of the proposed procedure when it is applied to the 2‐methyl‐cyclohexanol production process, which is a non‐linear, distributed parameter and time‐varying process, typical of many important industrial systems.     

An algorithm to determine sample sizes for optimization with artificial neural networks

This article presents an algorithm developed to determine the appropriate sample size for constructing accurate artificial neural networks as surrogate models in optimization problems. In the algorithm, two model evaluation methods—cross‐validation and/or bootstrapping—are used to estimate the performance of various networks constructed with different sample sizes. The optimization of a CO₂ capture process with aqueous amines is used as the case study to illustrate the application of the algorithm. The output of the algorithm—the network constructed using the appropriate sample size—is used in a process synthesis optimization problem to test its accuracy. The results show that the model evaluation methods are successful in identifying the general trends of the underlying model and that objective function value of the optimum solution calculated using the surrogate model is within 1% of the actual value. © 2012 American Institute of Chemical Engineers AIChE J, 59: 805–812, 2013     

Identification and control of a riser‐type FCC unit

This paper addresses the use of feedforward neural networks for the steady‐state and dynamic identification and control of a riser type fluid catalytic cracking unit (FCCU). The results are compared with a conventional PI controller and a model predictive control (MPC) using a state space subspace identification algorithm. A back propagation algorithm with momentum term and adaptive learning rate is used for training the identification networks. The back propagation algorithm is also used for the neuro‐control of the process. It is shown that for a noise‐free system the adaptive neuro‐controller and the MPC are capable of maintaining the riser temperature, the pressure difference between the reactor vessel and the regenerator, and the catalyst bed level in the reactor vessel, in the presence of set‐point and disturbance changes. The MPC performs better than the neuro controller that in turn is superior to the conventional multi‐loop diagonal PI controller.     

Non‐Linear Model Predictive Control: A Personal Retrospective

An overview of non‐linear model predictive control (NMPC) is presented, with an extreme bias towards the author's experiences and published results. Challenges include multiple solutions (from non‐convex optimization problems), and divergence of the model and plant outputs when the constant additive output disturbance (the approach of dynamic matrix control, DMC) is used. Experiences with the use of fundamental models, multiple linear models (MMPC), and neural networks are reviewed. Ongoing work in unmeasured disturbance estimation, prediction and rejection is also discussed.     

Modeling and Optimization of Anode‐Supported Solid Oxide Fuel Cells on Cell Parameters via Artificial Neural Network and Genetic Algorithm

An artificial neural network (ANN) and a genetic algorithm (GA) are employed to model and optimize cell parameters to improve the performance of singular, intermediate‐temperature, solid oxide fuel cells (IT‐SOFCs). The ANN model uses a feed‐forward neural network with an error back‐propagation algorithm. The ANN is trained using experimental data as a black‐box without using physical models. The developed model is able to predict the performance of the SOFC. An optimization algorithm is utilized to select the optimal SOFC parameters. The optimal values of four cell parameters (anode support thickness, anode support porosity, electrolyte thickness, and functional layer cathode thickness) are determined by using the GA under different conditions. The results show that these optimum cell parameters deliver the highest maximum power density under different constraints on the anode support thickness, porosity, and electrolyte thickness.     

Multi‐Objective Optimization of Pseudo‐Dynamic Operation of Naphtha Pyrolysis by a Surrogate Model

A simple pseudo‐dynamic surrogate model is developed in the framework of the state space model with the feed‐forward neural network to replace the complex free radical pyrolysis model. The surrogate model is then applied to investigate the multi‐objective optimization of two key performance objectives with distinct contradiction: the mean yields of key products and the day mean profits. The ?‐constraint method is employed to solve the multi‐objective optimization problem, which provides a broad range of operation conditions depicting tradeoffs of both key objectives. The Pareto‐optimal frontier is successfully obtained and five selected cases on the frontier are discussed, suggesting that flexible operations can be performed based on industrial demands.     

Colour‐appearance modeling using feedforward networks with Bayesian regularization method— Part I: Forward model

In this article, a method of predicting colour appearance (from colorimetric attributes to colour‐appearance attributes, i.e., forward model) using an artificial neural network is presented. The neural network model developed is a multilayer feedforward neural network model for predicting colour appearance (FNNCAM for short). The model was trained by LUTCHI colour‐appearance datasets. The Levenberg–Marquardt algorithm is incorporated into the back‐propagation procedure to accelerate the training of FNNCAM and the Bayesian regularization method is applied to the training of neural networks to improve generalization. The results of FNNCAM obtained are quite promising. © 2000 John Wiley & Sons, Inc. Col Res Appl, 25, 424–434, 2000     

Optimization of inulinase production by solid‐state fermentation in a packed‐bed bioreactor

BACKGROUND: This work is focused on inulinase production by solid‐sate fermentation (SSF) using sugarcane bagasse, corn steep liquor (CSL), pre‐treated cane molasses, and soybean bran as substrates in a 3‐kg (dry basis) packed‐bed bioreactor. SSF was carried out by the yeast Kluyveromyces marxianus NRRL Y‐7571 and response surface methodology was used to optimize the temperature, air flow rate and initial mass of cells. RESULTS: The optimum inulinase activity (436.7 ± 36.3 U g^?1 dry substrate) was obtained at 24 h at an inlet air temperature of 30 °C, air flow rate 2.2 m³ h^?1 and 22 g of cells for fermentation. Inulinase productivity at these conditions was 18.2 U gds^?1 h^?1. Kinetic evaluation at the optimized conditions showed that the maximum inulinase production was verified at 24 h of fermentation. The carbon dioxide and the metabolic heat generation are directly associated with the consumption of total reducing sugars present in the medium. CONCLUSION: The high productivity achieved in this work shows the technical viability of inulinase production by SSF in a packed‐bed bioreactor. Copyright © 2009 Society of Chemical Industry     

Suspension of solid particles in multi‐impeller three‐phase stirred tank reactors

Solids suspension characteristics in gas—liquid–solid three‐phase stirred tanks with multi‐impellers were experimentally examined. Minimum impeller speeds for ultimately homogeneous solid suspension have been measured stirred tank reactors. Three impellers were installed: two four‐pitched blade downflow disk turbines and one Pfaudler type impeller chosen to provide good gas dispersion and to accomplish off‐bottom suspension of solid particles, respectively. Gas dispersion causes an increase in particle sedimentation associated with a decrease in power consumption and as a result, minimum impeller speeds for ultimately homogeneous solid suspension increase with increasing gas flow rates. A correlation was developed to predict minimum impeller speeds for ultimately homogeneous solid suspension. The proposed correlation, which agrees satisfactorily with the experimental results, is expected to be useful in design and scale‐up.     

Microbial conversion of androst‐1,4‐dien‐3,17‐dione by Mucor racemosus to hydroxysteroid‐1,4‐dien‐3‐one derivatives

BACKGROUND: A large number of bacterial, fungal and microalgal species are able to bio‐transform steroid compounds. Among them, fungi from the Mucor genus have been shown to mediate hydroxylation, oxidation, and desaturation by the double bond formation and epoxidation of various steroid substances. Mucor racemocus has not been studied for its ability to modify androst‐1,4‐dien‐3,17‐dione, a pharmaceutically important steroid precursor. RESULTS: The filamentous fungus M. racemosus was applied for bioconversion of androst‐1,4‐dien‐3,17‐dione (ADD, I ) in a 5‐day fermentation. Microbial metabolites were purified chromatographically and identified on the basis of their spectral data as 17β‐hydroxyandrost‐1,4‐dien‐3‐one ( II ), 14α‐hydroxyandrost‐1,4‐dien‐3,17‐dione ( III ), 15α‐hydroxyandrost‐1,4‐dien‐3,17‐dione ( IV ), 15α,17β‐dihydroxyandrost‐1,4‐dien‐3‐one ( V ), 14α,17β‐dihydroxyandrost‐1,4‐dien‐3‐one ( VI ), and 6β,17β‐dihydroxyandrost‐1,4‐dien‐3‐one ( VII ). CONCLUSION: Observed modifications included hydroxylation at C‐6β, C‐14α, C‐15α positions and 17‐carbonyl reduction. The best fermentation conditions for production of hydroxysteroid‐1,4‐dien‐3‐one derivatives were found to be 25 °C at 150 rpm for 5 days with a substrate concentration of 0.5 g L^?1. Copyright © 2009 Society of Chemical Industry     

Hydrophobic and oil‐resistant coatings based on advanced green polyurethane dispersions

In recent years, there have been significant interests among scientists around the world to design smart coatings that combine the standard desirable mechanical and chemical resistance properties with additional functionalities such as oil and water repellency, hydrophobicity, self‐cleaning and gas barrier properties. In the present research work, advanced aqueous polyurethane dispersion (PUD) systems have been designed and developed using three different types of polyols as soft‐segments. These polyols differ in their chemical structure, functionality, polarity, and interfacial properties. The effects of soy‐based polyol, hydroxy‐terminated perfluoropolyether, and hydroxy‐terminated polydimethylsiloxane—on various mechanical properties of their uncross‐linked and cross‐linked films, and more specifically on their hydrophobicity and oil‐resistance (oleophobicity) have been studied. Hydrophobicity of these coatings has been characterized by their Dynamic Contact Angle measurements and their oleophobicity by n‐octane absorption method. The investigations showed that presence of fluorine and siloxane structures significantly improve the hydrophobicity and oil resistance of these coatings and it is possible to optimize these properties using suitable composition of PUDs. © 2013 Wiley Periodicals, Inc. J. Appl. Polym. Sci. 130: 3874–3884, 2013     

Bis‐(1,2,3,4‐tetrahydroisoquinolinium): A Chiral Scaffold for Developing High‐Affinity Ligands for SK Channels

N‐Methyl‐bis‐(1,2,3,4‐tetrahydroisoquinolinium) analogues derived from AG525 (1,1′‐(propane‐1,3‐diyl)‐bis‐(6,7‐dimethoxy‐2‐methyl‐1,2,3,4‐tetrahydroisoquinoline)) stereoisomers and tetrandrine, a rigid bis‐(1,2,3,4‐tetrahydroisoquinoline) analogue with an S,S configuration, were synthesized and tested for their affinity for small‐conductance calcium‐activated potassium channel (SK/K_Ca2) subtypes using radioligand binding assays. A significant increase in affinity was observed for the quaternized analogues over the parent 1,2,3,4‐tetrahydroisoquinoline compounds. Interestingly, the impact of stereochemistry was not the same in the two groups of compounds. For quaternized analogues, affinities of S,S and R,R isomers for SK2 and SK3 channels were similar and in both cases higher than that of the meso derivative. Among the bis‐tetrahydroisoquinoline compounds, the S,S isomers exhibited high affinity, while the R,R and meso isomers had similarly lower affinities. Furthermore, the SK2/SK3 selectivity ratio was slightly increased for quaternized analogues. Bis‐(1,2,3,4‐tetrahydroisoquinolinium) represents a new scaffold for the development of high‐affinity ligands for SK channel subtypes.     

Optimization of α‐amylase production in solid substrate fermentation

Response surface methodology (RSM) was used to optimize the medium components of α‐amylase production using solid substrate fermentation (SSF). Hazelnut cake (HC), peptone, yeast extract (YE), and (NH₄)₂SO₄ were selected as independent variables for optimization. Central composite design (CCD) was used in design experiments and analysis results. This procedure limited the number of actual experiments performed while allowing possible interactions between the independent variables. By using CCD, 30 experiments were performed for determining the interaction of independent variables and optimization of fermentation medium. The P‐value of the coefficient of linear effect of (NH₄)₂SO₄ concentrations, which was obtained as 0.0001 has shown that this parameter has the greatest effect on the production of α‐amylase. Model F‐value (5.62) implies that the model is significant. The highest α‐amylase activity (4895 IU) was measured when the HC, peptone, YE, and (NH₄)₂SO₄ concentrations in the medium were 22.62, 5.20, 1.62, and 6.81 g L^?1, respectively.     

Shear dispersion in combined pressure‐driven and electro‐osmotic flows in a capillary tube with a porous wall

An analytical expression is derived for the shear dispersion during transport of a neutral nonreacting solute within a coupled system comprised of a capillary tube and a porous medium under the combined effects of pressure‐driven and electro‐osmotic flows. We use the Reynolds decomposition technique to obtain a dispersion coefficient by considering a sufficiently low wall or zeta potential that accounts for the combined flows. The coupled dispersion coefficient depends on the Debye–Hückel parameter, Poiseuille contribution fraction, and Péclet number. The developed model also provides a shear dispersion coefficient for an impervious capillary tube (noncoupled system). The ratio of the coupled (porous wall) and noncoupled (impervious) dispersion coefficients reveals that it is essential to include the transport of chemical species from the tube to the porous medium in several important physical situations. These findings have implications for design of chemical species transport in porous microfluidic networks and separation of emulsions in microchannel‐membrane systems. © 2015 American Institute of Chemical Engineers AIChE J, 61: 3981–3995, 2015     

On the use of the generalized autocatalytic models: The thermal decomposition of 3,5‐dinitro‐4‐methylbenzoic acid

The thermal decomposition of 3,5‐dinitro‐4‐methylbenzoic acid is studied by means of differential calorimetric techniques (DSC). Its autocatalytic behaviour has been highlighted and the decomposition process has been described considering the generalized expression of the ?esták–Berggren model. A new procedure for the optimization of the initiation parameter along with the other Arrhenius constants and kinetic exponents starting from the knowledge of the classic ?esták–Berggren model is illustrated. Encouraging results point out the validity of the approach which has been verified considering both a series of numerical and real experiments. © 2015 American Institute of Chemical Engineers AIChE J, 61: 1300–1308, 2015     

Performance evaluation of an ANN–GA aided experimental modeling and optimization procedure for enhanced synthesis of marine biosurfactant in a stirred tank reactor

BACKGROUND: An improved resilient back‐propagation neural network modeling coupled with genetic algorithm aided optimization technique was employed for optimizing the process variables to maximize lipopeptide biosurfactant production by marine Bacillus circulans. RESULTS: An artificial neural network (ANN) was used to develop a non‐linear model based on a 2⁴ full factorial central composite design involving four independent parameters, agitation, aeration, temperature and pH with biosurfactant concentration as the process output. The polynomial model was optimized to maximize lipopeptide biosurfactants concentration using a genetic algorithm (GA). The ranges and levels of these critical process parameters were determined through single‐factor‐at‐a‐time experimental strategy. Improved ANN‐GA modeling and optimization were performed using MATLAB v.7.6 and the experimental design was obtained using Design Expert v.7.0. The ANN model was developed using the advanced neural network architecture called resilient back‐propagation algorithm. CONCLUSION: Process optimization for maximum production of marine microbial surfactant involving ANN‐GA aided experimental modeling and optimization was successfully carried out as the predicted optimal conditions were well validated by performing actual fermentation experiments. Approximately 52% enhancement in biosurfactant concentration was achieved using the above‐mentioned optimization strategy. © 2012 Society of Chemical Industry