DATA CLUSTERING USING A REORGANIZING NEURAL NETWORK

A new approach, designed for clustering of arbitrary distributed patterns, is presented. This study is concerned with the use of a self-organizing neural network as a frame for data clustering. The nearest network nodes in feature space are treated as prototypes, assigned to the corresponding cluster. The rules for dead-node shifting and simple adjustment of coordinates of the active nodes are introduced. The performance of the proposed self-organizing neural network is examined on the benchmark synthetic and the real-world problem.     

Estimation of the number of trays for natural gas triethylene glycol dehydration column

The paper deals with the tray-to-tray method for determining the required number of trays in columns with intensive entrainment of liquid when the operating line of the “dry working regime” is not straight and with a small concentration of diffusing component in gas. Presented calculations show that the number of transfer units for gas and liquid should be calculated using the AIChE method accompanied with longitudinal mixing according to Gilbert properly transformed for case of diffusion through stagnant film. The suggested method can be applied for TEG dehydration absorbers-columns for dehydration of natural gas using triethylene glycol with system factor SF = 0.5.     

Rheological properties of hydroxyl‐terminated and end‐capped aliphatic hyperbranched polyesters

The rheological behavior of two series of aliphatic hyperbranched (HB) polyesters, based on 2,2‐bis(hydroxymethyl)propionic acid (bis‐MPA) and di‐trimethylol propane (Di‐TMP) as a tetrafunctional core, was studied. The effect of the size (pseudo‐generation number, from second to eight) and structure on the melt rheological properties was investigated for a series of hydroxyl‐terminated HB polyesters. In addition, the influence of the nature and degree of modification of the terminal OH groups in a series of fourth‐generation polyesters end‐capped with short and long alkyl chains and some aryl groups on the rheological properties was analyzed. The time–temperature superposition procedure was applied for the construction of master curves and for the analysis of the rheological properties of HB polyesters. The data obtained from WLF analysis of the HB polyesters showed that the values of the thermal coefficient of expansion of free volume α_f and the fractional free volume at the glass transition temperature, f_g, increase with increasing size of the HB polyesters. It was shown that the modified HB polyesters exhibited lower T_g and T_G′=G″ temperatures, above which viscous became dominant over elastic behavior. From an analysis of the master curves of the modified HB polyesters, it was observed that with increasing degree of modification, both storage and loss modules and complex dynamic viscosity and apparent energy for viscoelastic relaxation decrease, because of reduced intermolecular hydrogen interactions. They do not exhibit a plateau of rubbery behavior, which confirms that no entanglements are present and that the molar masses are below the critical molar mass. © 2014 Wiley Periodicals, Inc. J. Appl. Polym. Sci. 2015 , 132, 41479.     

Generalized sparse MRF appearance models

Image segmentation or registration approaches that rely on a local search paradigm (e.g, Active Appearance Models, Active Contours) require an initialization that provides for considerable overlap or a coarse localization of the object to be segmented or localized. In this paper we propose an approach that does not need such an initialization, but localizes anatomical structures in a global manner by formulating the localization task as the solution of a Markov Random Field (MRF).     

Investigation of ZnFe2O4 spinel ferrite nanocrystalline screen-printed thick films for application in humidity sensing

Zinc ferrite nanocrystalline powder was obtained by solid state synthesis of starting zinc oxide and hematite nanopowders. Field emission scanning electron microscopy and transmission electron microscopy, X-ray diffraction, X-ray photoelectron spectroscopy and Raman spectroscopy confirmed the formation of nanocrystalline zinc-ferrite powder with a mixed spinel structure with small amounts of remaining zinc oxide and hematite as impurities. Thick film paste was formed and screen printed on test interdigitated PdAg electrodes on alumina substrate. Formation of a porous nanocrystalline structure was confirmed by scanning electron microscopy and Hg porosimetry. Humidity sensing properties of zinc ferrite thick films were investigated by monitoring the change in impedance in the relative humidity interval 30%-90% in the frequency range 42 Hz-1 MHz at room temperature (25°C) and 50°C. At 42 Hz at both analyzed temperatures the impedance reduced ~46 times in the humidity range 30%-90%. The dominant influence of grain boundaries was confirmed by analysis of complex impedance with an equivalent circuit.     

Structure and enhanced antimicrobial activity of mechanically activated nano TiO2

Titanium dioxide is a photocatalyst, known not only for its ability to oxidize organic contaminants, but also for its antimicrobial properties. In this article, significant enhancement of the antimicrobial activity of TiO₂ (up to 32 times) was demonstrated after its activation by ball milling. The antimicrobial activity was analyzed for one fungal and 13 bacterial ATCC strains using the microdilution method and recording the minimum inhibitory concentration (MIC) values. In order to further investigate the correlation between the mechanical activation of TiO₂ and its antimicrobial activity, the structure, morphology and phase composition of the material were studied by means of Electron Microscopy, X-ray diffraction and nitrogen adsorption-desorption measurements. UV-Vis diffuse reflectance spectra were recorded and the Kubelka-Munk function was applied to convert reflectance into the equivalent band gap energy (E_g) and, consequently, to investigate changes in the E_g value. X-ray photoelectron spectroscopy was used to analyze the influence of mechanical activation on the Ti 2p and O 1s spectra. The presented results are expected to enable the development of more sustainable and effective advanced TiO₂-based materials with antimicrobial properties that could be used in numerous green technology applications.