Bioautomation

Volume 14, Issue 2
Editorial
In Memoriam Prof. Hinko Hinov
Anniversary - the 60th Birthday of Prof. Diana Karklina
New Books
Bioprocess systems
Optimal Feeding Trajectories Design for E. coli Fed-batch Fermentations	89-98
Olympia Roeva, Stoyan Tzonkov, Bernd Hitzmann [ +/- abstract ][ full text ]
In this paper optimal control algorithms for two E. coli fed-batch fermentations are developed. Fed-batch fermentation processes of E. coli strain MC4110 and E. coli strain BL21(DE3)pPhyt109 are considered. Simple material balance models are used to describe the E. coli fermentation processes. The optimal feed rate control of a primary metabolite process is studied and a biomass production is used as an example. The optimization of the considered fed-batch fermentation processes is done using the calculus of variations to determine the optimal feed rate profiles. The problem is formulated as a free final time problem where the control objective is to maximize biomass at the end of the process. The obtained optimal feed rate profiles consist of sequences of maximum and minimum feed rates. The resulting profiles are used for optimization of the E. coli fed-batch fermentations. Presented simulations show a good efficiency of the developed optimal feed rate profiles.
Software Sensors Design for a Class of Aerobic Fermentation Processes	99-118
Trayana Patarinska, Vasil Trenev, Silvia Popova [ +/- abstract ][ full text ]
The problem of on-line state and parameter estimation (software sensors design) of a class of aerobic fermentation processes for metabolite product formation is considered. The class is characterized by: two limiting substrates one of which, growth factor, is practically depleted during the biomass growth where the product formation is negligible; corresponding general reaction scheme – a qualitative description of the main metabolic reactions between the main components in the liquid phase (biomass, substrates, product and dissolved oxygen concentrations). Two separate sensors – state and parameter estimators – are designed. The state estimator is developed based on knowledge of only one on-line measurable variable, the dissolved oxygen, and the yield factors assumed as constant coefficients. Parameter estimator of the specific reaction rates is developed under the assumption that all the process variables are known on-line by measurements or estimates. The yield factors are estimated also as non-stationary parameters, thus creating a basis for comparison with the specified constant values used for the state estimator design. As a case study industrial Lysine fermentation in fed-batch mode of operation is considered. Simulation investigations under different operating conditions are done in order to highlight the performances of the proposed sensors.
An Approach for Identifying of Fusarium Infected Maize Grains by Spectral Analysis in the Visible and Near Infrared Region, SIMCA Models, Parametric and Neural Classifiers	119-128
Tsvetelina Draganova, Plamen Daskalov, Rusin Tsonev [ +/- abstract ][ full text ]
An approach for identifying of Fusarium infected single maize grains based on diffuse reflectance in visible and near infrared region is proposed in the paper. Spectral characteristics were collected in the range 400-2500 nm in steps of 2 nm. Soft independent modeling of class analogy (SIMCA) is used for data processing. Maize grains classification is based on SIMCA classifier and Probabilistic neural network (PNN). Recognition accuracy which is achieved for both classes of grains is respectively 99.89% for healthy, and 93.7% for infected.
Biomedical systems
Simple Approach for Tremor Suppression in Electrocardiograms	129-138
Ivan Dotsinsky, Georgy Mihov [ +/- abstract ][ full text ]
Electrocardiogram recordings are very often contaminated by high-frequency noise usually power-line interference and EMG disturbances (tremor). Filtering out the tremor remains a priori partially successful since it has a relatively wide spectrum, which overlaps the useful ECG frequency band by aperiodic noise. The proposed simple approach for tremor suppression uses heuristic relations between the ECG signal parts and parameters of the applied moving averaging. The results obtained are assessed and compared to tremor suppression obtained by moving averaging with constant sample numbers throughout the signal.
An Object Detection Method Based on the Separability Measure using an Optimization Approach	139-146
Edward Y. H. Choy, Wai Tak Hung [ +/- abstract ][ full text ]
The detection of an object in an image can be set up as a maximization problem using separability measure as the objective function. The parameters of this objective function are the parameters used to define two regions in a mask. This mask has the same dimension as the image being considered. The two regions in the mask correspond to two regions in the image under investigation. The pixel value information in these two regions in the image would be used for the calculation of the separability measure. An optimization method would then be used to solve this maximization problem. This study demonstrated that the proposed method could detect the rectangle in the test image successfully. It showed that object detection and detailed segmentation could be carried out at the same time if the geometry of an object could be described mathematically.
Bioinformatics
An Improvement of the Grid-based Hydrophobic-hydrophilic Model	147-156
Stefka Fidanova [ +/- abstract ][ full text ]
Proteins are complex macromolecules that perform vital function in all living beings. They are composed of a chain of amino acids. The biological function of a protein is determined by the way it is folded into a specific 3D structure, known as native conformation. The high resolution 3D structure of a protein is the key to the understanding and manipulating of its biochemical and cellular functions. Protein structure could be calculated from knowledge of its sequence and our understanding of the sequence-structure realizations. Various methods have been applied to solve protein folding problem. In this paper the protein is represented like a sequence over 3 letter alphabet according the specific functions of amino acids. After that the folding problem is defined like optimization problem. Our protein model is multifunctional: it can be used to predict the 3D structure of the protein from its amino acid sequence; the model can predict the changes in the protein folding when several amino acids are mutated; by it can be constructed a protein with needed 3D folding.