A constructive algorithm to synthesize arbitrarily connected feedforward neural networks

In this work we present a constructive algorithm capable of producing arbitrarily connected feedforward neural network architectures for classification problems. Architecture and synaptic weights of the neural network should be defined by the learning procedure. The main purpose is to obtain a parsimonious neural network, in the form of a hybrid and dedicate linear/nonlinear classification model, which can guide to high levels of performance in terms of generalization. Though not being a global optimization algorithm, nor a population-based metaheuristics, the constructive approach has mechanisms to avoid premature convergence, by mixing growing and pruning processes, and also by implementing a relaxation strategy for the learning error. The synaptic weights of the neural networks produced by the constructive mechanism are adjusted by a quasi-Newton method, and the decision to grow or prune the current network is based on a mutual information criterion. A set of benchmark experiments, including artificial and real datasets, indicates that the new proposal presents a favorable performance when compared with alternative approaches in the literature, such as traditional MLP, mixture of heterogeneous experts, cascade correlation networks and an evolutionary programming system, in terms of both classification accuracy and parsimony of the obtained classifier.     

Taste perception and purchase intent of oil‐in‐water spreads: effects of oil types and salt (NaCl or KCl) concentrations

Associations of sodium intake with heart‐related problems are creating awareness towards reducing sodium. Potassium chloride (KCl), a substitute for sodium chloride (NaCl), has the disadvantage of imparting bitterness at high concentrations. We evaluated physical characteristics, taste perception and purchase intent of KCl and NaCl in oil‐in‐water spreads/emulsions composed by olive, rice bran and soya bean oils. Consumers (N = 300) evaluated saltiness/bitterness of emulsions prepared with 65% oil, and NaCl (0.5% and 1.0%) or KCl (0.75% and 1.5%). Olive oil spreads (104.07–107.43 Pa s) had higher viscosity compared to other spreads (59.16–74.96 Pa s). Type of oil had significant effects on bitterness, overall taste liking and viscosity. Taste liking decreased due to bitterness of olive oil spreads (mean drop=1.72–2.43). Purchase intent was positively associated with saltiness and pH and increased with oil claims (increase = 1.3%–22.1%) compared to sodium claims (increase = 0.0%–12.9%). These findings are useful for understanding taste perception of emulsions.     

Experimental and Calculational Studies of the Interactions of BF3 with Fluoroethers

BF₃ was co-condensed with (C₂H₅)₂O, (CF₃CH₂)₂O, and (C₂F₅)₂O in excess argon at 15 K. Infrared spectra of the matrices showed a definite Lewis acid-base interaction between BF₃ and diethyl ether; a weak but definite interaction with bis(2,2,2,-trifluorodiethyl)ether, and no observable interaction with perfluorodiethyl ether. Molecular orbital (MO) calculations complemented the experimental observations by revealing that fluorine atoms on the ethers decreased electron localization about the oxygen atom. Thus, the experimental data and MO calculations indicated a clear trend between strength of interaction with BF₃ and the degree of ether F substitution. The implications of the results for commercial perfluoro ether lubricant/metal oxide surface interactions are discussed.     

Expert system for the assessment of power transformer insulation condition based on type-2 fuzzy logic systems

An efficient expert system for the power transformer condition assessment is presented in this paper. Through the application of Duval’s triangle and the method of the gas ratios a first assessment of the transformer condition is obtained in the form of a dissolved gas analysis (DGA) diagnosis according IEC 60599. As a second step, a knowledge mining procedure is performed, by conducting surveys whose results are fed into a first Type-2 Fuzzy Logic System (T2-FLS), in order to initially evaluate the condition of the equipment taking only the results of dissolved gas analysis into account. The output of this first T2-FLS is used as the input of a second T2-FLS, which additionally weighs up the condition of the paper-oil system. The output of this last T2-FLS is given in terms of words easily understandable by the maintenance personnel. The proposed assessing methodology has been validated for several cases of transformers in service.     

Fuzzy risk index for power transformer failures due to external short-circuits

A novel methodology to assess the risk of power transformer failures caused by external faults, such as short-circuit, taking the paper insulation condition into account, is presented. The risk index is obtained by contrasting the insulation paper condition with the probability that the transformer withstands the short-circuit current flowing along the winding during an external fault. In order to assess the risk, this probability and the value of the degree of polymerization of the insulating paper are regarded as inputs of a type-2 fuzzy logic system (T2-FLS), which computes the fuzzy risk level. A Monte Carlo simulation has been used to find the survival function of the currents flowing through the transformer winding during a single-phase or a three-phase short-circuit. The Roy Billinton Test System and a real power system have been used to test the results.     

Evaluation of Vapor Pressure and Ultra-High Vacuum Tribological Properties of Ionic Liquids

Ionic liquids are a class of salts that incorporate polyatomic anions and cations. These materials are typically viscous fluids at room temperature. The fluids are generally characterized as possessing negligible vapor pressures under ambient conditions. These beneficial properties have led us to study the effectiveness of ionic liquids containing both organic cations and anions for use as lubricants with space applications. This article presents vapor pressure measurements for four ionic liquids as well as friction coefficient data from a spiral orbit tribometer in the boundary lubrication regime under simulated space conditions using stainless steel tribocouples. In addition, we present the first tribological data on mixed ionic liquids and an ionic liquid additive. Postmortem infrared and Raman analyses of the balls and races indicates that the major degradation pathway for these organic ionic liquids is similar to those of other carbon-based liquid lubricants; that is, deterioration of the organic structure into amorphous graphitic carbon. The coefficients of friction and lifetimes of these lubricants are comparable to or better than those of several commonly used space oils.     

Tailor‐made Polymeric Membranes based on Segmented Block Copolymers for CO2 Separation

This paper reports the design of a tailor made polymeric membrane by using poly(ethylene oxide)–poly(butylene terephthalate) (PEO‐PBT) multi‐block copolymers. Their properties are controlled by the fraction of the PEO phase and its molecular weight. To explain the effect of structural changes in copolymer membranes, transport properties of four gases (CO₂, H₂, N₂, and CH₄) are discussed. After characterization, the two best copolymers are selected in order to prepare tailor made blends by adding poly(ethylene glycol) (PEG). The best selected copolymer that contained 55 wt. % of 4000 g mol⁻¹ PEO produced a blend with high CO₂ permeability (∼190 barrer), which is twice the permeability of the pure copolymer. At the same time, an enhancement of CO₂/H₂ selectivity is observed (∼13). These results suggest that the morphology of PEO‐PBT can be well controlled by the addition of low‐molecular‐weight PEG, and consequently the gas transport properties can be tuned.     

Electron Affinity Calculations For Thioethers

Previous work indicated that polyphenyl thioethers possessed chemical properties, related to their electron affinities, which could allow them to function as vapor phase lubricants (VPL). Indeed, preliminary tribological tests revealed that the thioethers could function as vapor phase lubricants but not over a wide temperature and Hertzian pressure range. Increasing the electron affinity of the thioethers may improve their VPL properties over this range. Adding a substituent group to the thioether will alter its electron affinity in many cases. Molecular orbital calculations were undertaken to determine the effect of five different substituent groups on the electron affinity of polyphenyl thioethers. It was found that the NO₂, F, and I groups increased the thioethers electron affinity by the greatest amount. Future work will involve the addition of these groups to the thioethers followed by tribological testing to assess their VPL properties.     

Analysis of the correlation structure of square time series

Abstract.  This paper analyses the asymptotic behaviour of the autocorrelation structure exhibited by squares of time series with a Wold expansion where the input error is a sequence of random variables with mean zero and finite kurtosis. Two important cases are discussed: (i) when the errors are independent and, (ii) when the errors are uncorrelated but their squares are correlated. Both situations are addressed when the process exhibits short or long memory. Consequences of these results on certain models widely used in many disciplines are also discussed.