Data preprocessing for WUM

This paper focuses on data preprocessing for WUM. Web page mining (WUM) applies data procedures to analyze user access of Well sites. As with any knowledge, discovery and data mining (KDD) process, WUM contains three main steps: preprocessing, knowledge extraction and results analysis. This data preprocessing try to determine the exact list of users who accessed the Web site and to reconstitute user sessions-the sequence of actions each user performed at the Web site. For privacy reasons, the preprocessing users use Web server log files from Web servers as well as the Website map and then anonymizing and joining log files are used. The data preprocessing involves data fusion, data cleaning, data structuration and data summarization. This data preprocessing not only reduces the log file size but also increases the quality of available data through the new data structures.     

A passivity-based controller for coordination of converters in a fuel cell system

The problem of converters coordination of a fuel cell system involving a hydrogen fuel cell with super-capacitors for applications with high instantaneous dynamic power is addressed in this paper. The problem is solved by using a non-linear controller based on passivity. The controller design is based on the interconnection and damping assignment approach, where the proof of the local system stability of the whole closed-loop system is shown. Simulation and experimental results on a reduced scale system prove the feasibility of the proposed approach for a real electrical vehicle.     

Pd-decorated CNT as sensitive material for applications in hydrogen isotopes sensing - Application as gas sensor

Applicability of multiwall carbon nanotubes (MWCNTs) decorated with palladium nanoparticles as sensitive layer in a resistive microsensor for identification of hydrogen isotopes, Deuterium (²H) and Protium (¹H), has been demonstrated. Palladium nanoparticles were anchored on the MWCNTs surface via a chemical process involving micellization, from a precursor chloride solution, in high ultrasonic density field. Pd-MWCNTs are quasi-aligned between the interdigitated gold electrodes of a SiO₂ substrate by drop casting and di-electrophoretic alignment in Tetrahydrofuran (THF) and Nafion solution. The morphostructural characterization of the sensitive material has been carried out through SEM, TEM and Raman spectroscopy and its gas sensing properties were evaluated using electrical measurements performed on a series of isotope concentrations (ranging from 0.1% up to 1%, and from 1% to 4%, value to which hydrogen becomes explosive) diluted in argon, to observe the evolution of the sensor sensibility. The two hydrogen isotopes have different behaviors related to the adsorption on the Pd-MWCNT, which is well observed in the resistance change. Therefore, the sensor based on Pd-MWCNTs could be a viable solution to be integrated in systems for hydrogen leakage detection.     

Photodegradation process of Eosin Y using ZnO/SnO2 nanocomposites as photocatalysts: experimental study and neural network modeling

A series of coupled ZnO/SnO₂ nanocomposites were prepared with different molar ratios (1:10, 1:2, 2:1, and 10:1), using a homogeneous co-precipitation method. The structural properties were evaluated by different techniques: XRD, UVDR, SEM, N₂ adsorption, and IR. The photocatalytic activity of the samples was tested with the main goal of Eosin Y degradation from wastewaters. The prepared nanocomposites/systems exhibit higher photocatalytic activity than a single semiconductor photocatalyst and ZnO can effectively improve the photocatalytic efficiency of SnO₂ under UV illumination. A direct neural network modeling methodology, based on feed-forward neural networks, was performed in order to evaluate the efficiency of the photodegradation process of Eosin Y, depending of the reaction conditions. The developed model considered the following parameters with significant influence on the approached process: crystallite size, surface area, absorbtion edge, TOC values, time of reaction, and catalyst concentration as inputs and the final dye concentration as output. Accurate results were obtained in the validation phase of the neural model: relative average error under 4 % and a correlation between experimental and simulation data of 0.999.     

New Bismaleimide-Silica Hybrid Materials: A Critical Assessment of Properties in Correlation with the Method of Synthesis

New hybrid materials have been prepared by sol–gel technique. They have been obtained from bismaleimide monomers either in reaction with N-(3-triethoxysilylpropyl)furan-2-carboxamide monomer, by a Diels–Alder reaction, or in reaction with (3-aminopropyl)triethoxysilane following a Michael addition reaction. The sol–gel process was conducted with or without adding different amounts of tetraethyl orthosilicate. The structures of the obtained compounds have been confirmed by proton nuclear magnetic resonance and Fourier transform infrared spectroscopy. A comparative study between Diels–Alder- and Michael addition-type products regarding their thermal and mechanical properties was also conducted for samples as obtained from synthesis. The thermoreversible character of the Diels–Alder hybrid materials has been demonstrated with the aids of differential scanning calorimetry and attenuated total reflectance Fourier transform infrared spectroscopy, the results from both methods being in good agreement with each other, and with literature data. The morphology of hybrid materials was studied by the atomic force microscopy, optical microscopy for three different stages: initial (24°C), at heating (150°C), and after cooling at 24°C, and scanning electron microscopy. All data confirmed the driving force for the dispersion of the Si-containing aggregates in the Michael addition series is the dynamic evolution of the sol–gel process, whereas the Diels–Alder series behavior is ruled by the thermoreversible character of the Diels–Alder cycloaddition.     

Structure and properties of polymer–wood composites based on an aliphatic copolyamide and secondary polyethylenes

Composite materials based on an aliphatic copolyamide and a secondary polyethylene, as thermoplastic matrices, and wood chips as filler were obtained and characterized. The influence of different factors (polymeric matrix type, fractional composition and geometrical characteristics of wood filler, processing parameters and ratio polymer/wood) on the properties of polymer–wood composites (PWCs) was studied. It was demonstrated that the packing factor F has an essential influence on the properties of PWCs: increasing F values determines an improvement in mechanical properties of these materials. Mechanical properties, thermal behaviour, morphological and diffusion characteristics of the analyzed composites were evaluated through specific methods and reported herein. Morphological and DSC data confirmed the presence of strong interface interactions between polymer and wood. The diffusion characteristics of PWCs showed that the diffusion coefficient D essentially depends on filler content in composites. © 2006 Wiley Periodicals, Inc. J Appl Polym Sci 101: 1700–1710, 2006     

Advanced data preprocessing for intersites Web usage mining

Web usage mining applies data mining procedures to analyze user access of Web sites. As with any KDD (knowledge discovery and data mining) process, WUM contains three main steps: preprocessing, knowledge extraction, and results analysis. We focus on data preprocessing, a fastidious, complex process. Analysts aim to determine the exact list of users who accessed the Web site and to reconstitute user sessions-the sequence of actions each user performed on the Web site. Intersites WUM deals with Web server logs from several Web sites, generally belonging to the same organization. Thus, analysts must reassemble the users' path through all the different Web servers that they visited. Our solution is to join all the log files and reconstitute the visit. Classical data preprocessing involves three steps: data fusion, data cleaning, and data structuration. Our solution for WUM adds what we call advanced data preprocessing. This consists of a data summarization step, which will allow the analyst to select only the information of interest. We've successfully tested our solution in an experiment with log files from INRIA Web sites.     

Magnetic trap effects on nanowire��s dynamics within micro-capillary vessels

The concept of using magnetic nanowires (NWs) as carriers in magnetic drug targeting was largely developed in the last decade. Magnetically controlled and manipulated nanoparticles are gaining more and more ground in various biomedical applications: drug delivery, cancer therapy by magnetic fluid hyperthermia, radiotherapy. In this study, a 2-D mathematical model was developed and implemented to examine the capture of magnetic drug carrier particles within a magnetic trap represented by high magnetic field gradients. The transient dynamics of magnetic NWs under the effect of the blood velocity (v _B) varied from 0.05 up to 1.0 cm/s, magnetic field strength and their initial position within the fluid channel have been analyzed using COMSOL Multiphysics module. The parameters related to the NW with strong influence on the magnetic trap output were systematically modified: magnetic material, cylindrical particle geometry characterized by diameter and length, its initial position related to both fluid flow direction and distance from the channel center. Considering the influence of both hydrodynamic, magnetic forces and torques the magnetic trap’s efficiency was evaluated and phase diagrams were determined for different parameters involved. A considerable number of simulations were made to obtain statistically significant results. These results are commented and discussed in detail.     

Laser Direct Writing via Two-Photon Polymerization of 3D Hierarchical Structures with Cells-Antiadhesive Properties

We report the design and fabrication by laser direct writing via two photons polymerization of innovative hierarchical structures with cell-repellency capability. The structures were designed in the shape of “mushrooms”, consisting of an underside (mushroom’s leg) acting as a support structure and a top side (mushroom’s hat) decorated with micro- and nanostructures. A ripple-like pattern was created on top of the mushrooms, over length scales ranging from several µm (microstructured mushroom-like pillars, MMP) to tens of nm (nanostructured mushroom-like pillars, NMP). The MMP and NMP structures were hydrophobic, with contact angles of (127 ± 2)° and (128 ± 4)°, respectively, whereas flat polymer surfaces were hydrophilic, with a contact angle of (43 ± 1)°. The cell attachment on NMP structures was reduced by 55% as compared to the controls, whereas for the MMP, a reduction of only 21% was observed. Moreover, the MMP structures preserved the native spindle-like with phyllopodia cellular shape, whereas the cells from NMP structures showed a round shape and absence of phyllopodia. Overall, the NMP structures were more effective in impeding the cellular attachment and affected the cell shape to a greater extent than the MMP structures. The influence of the wettability on cell adhesion and shape was less important, the cellular behavior being mainly governed by structures’ topography.     

Controlling morphology,electrical, and mechanical properties of polymer blends by heterogeneous distribution of carbon nanotubes

Electrical and mechanical properties of some new composites made of copolyamide/polypropylene/carbon nanotubes (CPA/PP/CNTs) in different ratios, in correlation with their morphology, were investigated. The polymer blends based on polypropylene and low melting point copolyamide filled with CNTs were prepared by two methods. In the first procedure, CNTs were orderly distributed into the heterogeneous polymer matrix where filler occupied only one polymer component creating a highly branched conductive phase that provides a low value of the percolation threshold. In second case, CNTs were randomly distributed in the heterogeneous polymer matrix created by the blend CPA/PP yielding in a higher value of percolation threshold. Four factors influence on the distribution of filler, namely: thermodynamics, kinetics, chemical factor, and processing. Both electrical and mechanical properties are defined by spatial distribution of the conductive nanofiller in the polymer blend. POLYM. COMPOS., 37:2467–2477, 2016. © 2015 Society of Plastics Engineers