Modified self-organising maps with a new topology and initialisation algorithm

Mapping quality of the self-organising maps (SOMs) is sensitive to the map topology and initialisation of neurons. In this article, in order to improve the convergence of the SOM, an algorithm based on split and merge of clusters to initialise neurons is introduced. The initialisation algorithm speeds up the learning process in large high-dimensional data sets. We also develop a topology based on this initialisation to optimise the vector quantisation error and topology preservation of the SOMs. Such an approach allows to find more accurate data visualisation and consequently clustering problem. The numerical results on eight small-to-large real-world data sets are reported to demonstrate the performance of the proposed algorithm in the sense of vector quantisation, topology preservation and CPU time requirement.     

A novel ACO–GA hybrid algorithm for feature selection in protein function prediction

Protein function prediction is an important problem in functional genomics. Typically, protein sequences are represented by feature vectors. A major problem of protein datasets that increase the complexity of classification models is their large number of features. Feature selection (FS) techniques are used to deal with this high dimensional space of features. In this paper, we propose a novel feature selection algorithm that combines genetic algorithms (GA) and ant colony optimization (ACO) for faster and better search capability. The hybrid algorithm makes use of advantages of both ACO and GA methods. Proposed algorithm is easily implemented and because of use of a simple classifier in that, its computational complexity is very low. The performance of proposed algorithm is compared to the performance of two prominent population-based algorithms, ACO and genetic algorithms. Experimentation is carried out using two challenging biological datasets, involving the hierarchical functional classification of GPCRs and enzymes. The criteria used for comparison are maximizing predictive accuracy, and finding the smallest subset of features. The results of experiments indicate the superiority of proposed algorithm.     

Development of a 2-D 2-group neutron noise simulator for hexagonal geometries

In this paper, the development of a neutron noise simulator for hexagonal-structured reactor cores using both the forward and the adjoint methods is reported. The spatial discretisation of both 2-D 2-group static and dynamic equations is based on a developed box-scheme finite difference method for hexagonal mesh boxes. Using the power iteration method for the static calculations, the 2-group neutron flux and its adjoint with the corresponding eigenvalues are obtained by the developed static simulator. The results are then benchmarked against the well-known CITATION computer code. The dynamic calculations are performed in the frequency domain which leads to discarding of the time discretisation. Then, the developed 2-D 2-group neutron noise simulator calculates both the discretised forward and the adjoint reactor transfer function between a point source and its induced neutron noise, by assuming the neutron noise source as an “absorber of variable strength” type. The neutron noise induced by a “vibrating absorber” type of noise source may also be modeled using the calculated transfer function. The viability of the simulator is verified for different benchmark cases.     

Load Tests on Prestressed Precast Concrete and Timber Piles

This paper presents the results of six pile-load tests performed on timber and prestressed concrete piles. Five axial compressive pile-load tests and one lateral-load test are presented. Comparisons are made between the results of these tests and predictions made using available geotechnical computer software. These comparisons provide an insight into using the Brinch-Hansen method of pile-load test evaluation along with wave equation and other computer analyses to evaluate pile capacities from load test data. This paper demonstrates how wave equation analyses can be used to anticipate and mitigate excess driving stresses during the installation of prestressed concrete piles. This paper also demonstrates that the use of well-established analytical methods can adequately predict the vertical and lateral behavior of timber and prestressed concrete piles. However, the bending rigidity of concrete piles under lateral loading should be carefully selected to obtain a good prediction to the measured data.     

Clarifying the effects of comonomer distribution between short and long chains on physical and mechanical properties of ethylenic copolymers

In this paper, two Zigler-Natta catalysts (ZNCs) were used to synthesize a commercially available linear low-density polyethylene (LLDPE), widely used in the packaging industry, on an industrial scale. The catalysts differ only in their ability to distribute comonomers between short and long chains. Both catalysts were fully characterized in the first section, and two similar ethylene/1-butene copolymers were made using them. Afterward, the produced copolymers were fully characterized using different techniques; namely, differential scanning calorimetry (DSC), successive self-nucleation and annealing (SSA), oxygen induction time (OIT), melt flow index (MFI), rheometric mechanical spectroscopy (RMS), and a wide range of mechanical experiments. It was revealed that while the presence of comonomers in short chains can reduce their resistance against oxidation (by more than 30%) and can cause a dramatic change in friction coefficients (by more than 20%), some of the other main mechanical properties of the made copolymers were independent of comonomer distribution between long and short chains. In addition, it was shown that ethylenic copolymers' strain hardening modulus (SHM) takes advantage of the homogenous distribution.     

The Effects of SiO2/Al2O3 and H2O/Al2O3 Molar Ratios on SAPO-34 Catalyst in the Methanol to Olefin Process

Silicon - Informed through synthesis and characterization of NH 3 TPD, BET techniques, the current study investigated the right contribution of silicon and water content on a 9 sample SAPO-34...     

Simple Method for Prediction of the Standard Gibbs Free Energy of Formation of Energetic Compounds

A reliable simple method for prediction of the standard Gibbs energy of formation (Δ_fG^θ) of energetic compounds containing nitroaromatic, acyclic, and cyclic nitramine, nitrate ester, and nitroaliphatic compounds is introduced herein. The method is based on the contribution of elemental composition (Δ_fG_elem^θ) and the correcting function for the presence of additive and non‐additive molecular fragments (Δ_fG_corr^θ). In presence of some molecular moieties, Δ_fG_corr^θ may increase or decrease the value of Δ_fG_elem^θ, depending on the intermolecular interactions. The experimental root‐mean‐square error (RMSE) of the novel correlation (22.7 kJ mol⁻¹) is quite good. For some energetic compounds, where the computed values of two complex models of the quantitative structure‐property relationship (QSPR) theory were available, the experimental RMSE developed by the new method is smaller than the values obtained by QSPR method.     

A Flat Polymeric Membrane Sensor for Carbon Dioxide/Nitrogen Gas Mixture

A gas sensor was developed to measure the concentration of binary gas mixtures. This sensor works based on the permeability change of different gas mixtures across the polymeric membranes. Although high values of permeability and selectivity are needed for an ideal separation, the performance of this sensor mainly depends on the permeability factor. Polysulfone and silicone rubber were applied as the membrane base and coat, respectively. Moreover, in contrast to existing polymeric sensors that use hollow fibers, the present sensor is made of flat membranes. This new design is cheaper, smaller, and easier to use in comparison to the hollow fiber polymeric sensors. In order to test the sensor applicability, nitrogen and carbon dioxide were used as model gases. The effect of pressure on the response time and sensor accuracy was studied for the aforementioned gases. The response time (T_95%) of this low price sensor was 50?s, and the tolerance of measuring concentration was approximately 1.4% at 2?bar feed pressure. Also, increasing the feed pressure can improve the response time or accuracy of the sensor.     

The full cost of high-speed rail: an engineering approach

This paper examines the full costs, defined as the sum of private and social costs, of a high-speed rail system proposed for a corridor connecting Los Angeles and San Francisco in California. The full costs include infrastructure, fleet capital and operating expenses, the time users spend on the system, and the social costs of externalities, such as noise, pollution, and accidents. Comparing these full costs to those of other competing modes contributes to the evaluation of the feasibility of high-speed rail in the corridor. The paper concludes that high-speed rail is significantly more costly than expanding existing air service, and marginally more expensive than auto travel. This suggests that high-speed rail is better positioned to serve shorter distance markets where it competes with auto travel than longer distance markets where it substitutes for air. Received: May 1996 / Accepted in revised form: December 1996