The use of GA-ANNs in the modelling of compressive strength of cement mortar

In this paper, results of a project aimed at modelling the compressive strength of cement mortar under standard curing conditions are reported. Plant data were collected for 6 months for the chemical and physical properties of the cement that were used in model construction and testing. The training and testing data were separated from the complete original data set by the use of genetic algorithms (GAs). A GA-artificial neural network (ANN) model based on the training data of the cement strength was created. Testing of the model was also done within low average error levels (2.24%). The model was subjected to sensitivity analysis to predict the response of the system to different values of the factors affecting the strength. The plots obtained after sensitivity analysis indicated that increasing the amount of C₃S, SO₃ and surface area led to increased strength within the limits of the model. C₂S decreased the strength whereas C₃A decreased or increased the strength depending on the SO₃ level. Because of the limited data range used for training, the prediction results were good only within the same range. The utility of the model is in the potential ability to control processing parameters to yield the desired strength levels and in providing information regarding the most favourable experimental conditions to obtain maximum compressive strength.     

Fuzzy logic model for the prediction of cement compressive strength

A fuzzy logic prediction model for the 28-day compressive strength of cement mortar under standard curing conditions was created. Data collected from a cement plant were used in the model construction and testing. The input variables of alkali, Blaine, SO₃, and C₃S and the output variable of 28-day cement strength were fuzzified by the use of artificial neural networks (ANNs), and triangular membership functions were employed for the fuzzy subsets. The Mamdani fuzzy rules relating the input variables to the output variable were created by the ANN model and were laid out in the If-Then format. Product (prod) inference operator and the centre of gravity (COG; centroid) defuzzification methods were employed. The prediction of 50 sets of the 28-day cement strength data by the developed fuzzy model was quite satisfactory. The average percentage error levels in the fuzzy model were successfully low (2.69%). The model was compared with the ANN model for its error levels and ease of application. The results indicated that through the application of fuzzy logic algorithm, a more user friendly and more explicit model than the ANNs could be produced within successfully low error margins.     

A unified global self-consistent model of a capacitively and inductively coupled plasma etching system

Based on the concept of independent control of ion flux and ion-bombardment energy, a global selfconsistent model was proposed for etching in a high-density plasma reactor. This model takes account of the effect on the plasma behavior of separate rf chuck power in an Inductively Coupled Plasma etching system. Model predictions showed that the chuck power controls the ion bombardment energy but also slightly increases the ion density entering the sheath layer, resulting in an increase in etch rate (or etch yield) with increasing this rf chuck power. The contribution of the capacitive discharge to total ion flux in the ICP etching process is less than about 6% at rf chuck powers lower than 250 W. As a model system, etching of InN was investigated. The etch yield increased monotonically with increasing the rf chuck power, and was substantially affected by the ICP source power and pressure. The ion flux increased monotonically with increasing the source power, while the dc-bias voltage showed the reverse trend.     

Synthesis of ferromagnetic nanopowders from iron pentacarbonyl in capacitively coupled RF plasma

Ferromagnetic nanopowders (NP), synthesized with magnetically decoupled cores, may represent interesting raw materials for their further conversion into nanostructured magnetic bulk materials, with properties suitable for their utilization in low and high-frequency applications. Capacitively coupled RF discharges at a pressure of 100-250 Pa, were used for the synthesis of binary iron/carbon NP from iron pentacarbonyl (Fe(CO)₅) vapours, diluted with argon. The properties of the obtained NP products were investigated by TEM, XRD, TGA, FTIR and magnetization measurements. Diamagnetic NP, with ≈4% iron content, were obtained at lower pressures (100-130 Pa), whereas increasing the pressure to 250 Pa resulted in the synthesis of ≈40% iron content NP that demonstrated ferromagnetic properties (M_s ≈ 50 emu/gr). The first type of these NP is formed of carbon/iron rods, being 10-50 nm long and several nm wide, covered by graphitic layers and integrated into an amorphous carbon matrix. The second NP type, ferromagnetic NP, consists of isolated hexagonal shaped 20-60 nm crystals of Fe₇C₃ or Fe₂C₃ carbides, again embedded in an amorphous carbon matrix. Factors for favoring the formation of the ferromagnetic carbide NP at the higher pressure are discussed. These factors include: the lower temperatures presumed in the reaction zone, the shifting of the chemical equilibrium from non-bound Fe and its oxides towards the state of carburization, and the decreasing rate of removal of the Fe oxidation products towards the reactor walls.     

人工神经网络在锂铝硅超低膨胀透明微晶玻璃热处理研究中的应用

人工神经网络在锂铝硅超低膨胀透明微晶玻璃热处理研究中的应用陈建华，孙方明（盐城工业专科学校２２４００３）（华东理工大学２００２３７）ＡｒｔｉｆｉｃｉａｌＮｅｕｒａｌＮｅｔｗｏｒｋｓＡｐｐｌｉｅｄｔｏＳｔｕｄｙｏｆＨｅａｔＴｒｅａｔｍｅｎｔｏｆＬｉｔｈ...     

Artificial neural network based modelling of head and power characteristics of pump-mixer

Present work involves detailed experimentation on pump-mixer, and processing of experimental data to model its head and power characteristics using multi-variable least square based empirical correlations and artificial neural network (ANN). The latter modelling technique is shown to be superior. Trained and frozen ANN model has been used to generate head and power characteristics. Wherever possible, these characteristics have been qualitatively compared with the observations reported earlier. Also, an attempt has been made to physically explain these characteristics in order to demonstrate that the ANN model successfully captures the physics of the system.     

Artificial neural network simulation of rare earths solvent extraction equilibrium data

Separation of high purity rare earth elements from their mixed oxides, obtained from monazite or xenotime, requires multiple stages of separation by circuits incorporating one or more solvents. The separation factors being small, a large number of counter-current stages become necessary. Process development, analysis, optimization and control of rare earths are a complex task. Computer simulation provides useful tools in this area. Application of artificial neural networks (ANN) for simulation of equilibrium data in solvent extraction of rare earths is described in this paper. The back propagation ANN model has been used. The input neurons correspond to the system state variables such as equilibrium concentration and acidity. The partitioning of the metal ion into the two immiscible phases involved in solvent extraction is measured in terms of distribution ratio D. The model predicts the D value under varying process conditions. Comparison of ANN with conventional models shows that ANN is superior. The average absolute error for ANN model is one-fourth that of the conventional models. The approach has been used, in conjunction with a process simulation model, successfully for industrial process development involving production of high purity neodymium.     

Modeling and optimization of photocatalytic/photoassisted-electro-Fenton like degradation of phenol using a neural network coupled with genetic algorithm

Oxidation of phenol in aqueous media using supported TiO₂ nanoparticles coupled with photoelectro-Fenton-like process using Mn²⁺ cations as catalyst of electro-Fenton reaction was studied. The influence of the basic operational parameters such as initial pH of the solution, applied current, initial Mn²⁺ concentration, initial phenol concentration and kind of ultraviolet (UV) light on the oxidizing efficiency of phenol was studied. An artificial neural network (ANN) model was coupled with genetic algorithm to predict phenol oxidizing efficiency and to find an optimal condition for maximum phenol removal. The findings indicated that ANN provided reasonable predictive performance (R² = 0.949).     

Constrained optimization of combustion in a simulated coal-fired boiler using artificial neural network model and information analysis

Combustion in a boiler is too complex to be analytically described with mathematical models. To meet the needs of operation optimization, on-site experiments guided by the statistical optimization methods are often necessary to achieve the optimum operating conditions. This study proposes a new constrained optimization procedure using artificial neural networks as models for target processes. Information analysis based on random search, fuzzy c-mean clustering, and minimization of information free energy is performed iteratively in the procedure to suggest the location of future experiments, which can greatly reduce the number of experiments needed. The effectiveness of the proposed procedure in searching optima is demonstrated by three case studies: (1) a bench-mark problem, namely minimization of the modified Himmelblau function under a circle constraint; (2) both minimization of NO_x and CO emissions and maximization of thermal efficiency for a simulated combustion process of a boiler; (3) maximization of thermal efficiency within NO_x and CO emission limits for the same combustion process. The simulated combustion process is based on a commercial software package CHEMKIN, where 78 chemical species and 467 chemical reactions related to the combustion mechanism are incorporated and a plug-flow model and a load-correlated temperature distribution for the combustion tunnel of a boiler are used.     

Artificial neural network modelling of supercritical fluid CO2 extraction of polyunsaturated fatty acids from common carp (Cyprinus carpio L.) viscera

Common carp viscera, obtained from Tikveš Lake in Macedonia, was investigated as a possible source of polyunsaturated (PUFA) fatty acids. Supercritical fluid CO₂ extraction (SFE-CO₂) was employed for extraction of investigated bioactive components. The GC-FID analysis on the total extract obtained by supercritical fluid CO₂ extraction confirmed the assumption of presence of these bioactive components. A three layer artificial neural network was created for prediction and modelling of the extraction yield of polyunsaturated fatty acids from lyophilized viscera matrixes. Operating values of pressure, temperature, mass flow of CO₂ and extraction time were defined as input vectors to the ANN where PUFA extraction yield was considered as an output vector. Created ANN model provided adequate fitting of experimental data, with a correlation coefficient of 0.9968 for the entire data set. RSM-3D method was employed for mathematical modelling of the ANN output values as a function of operating variables and their interactions.     

The modelling of the toughening of epoxy polymers via silica nanoparticles: The effects of volume fraction and particle size

Silica nanoparticles possessing three different diameters (23, 74 and 170 nm) were used to modify a piperidine-cured epoxy polymer. Fracture tests were performed and values of the toughness increased steadily as the concentration of silica nanoparticles was increased. However, no significant effects of particle size were found on the measured value of toughness. The toughening mechanisms were identified as (i) the formation of localised shear-band yielding in the epoxy matrix polymer which is initiated by the silica nanoparticles, and (ii) debonding of the silica nanoparticles followed by plastic void growth of the epoxy matrix polymer. These mechanisms, and hence the toughness of the epoxy polymers containing the silica nanoparticles, were modelled using the Hsieh et al. approach (Polymer 51, 2010, 6284–6294). However, it is noteworthy that previous modelling work has required the volume fraction of debonded silica particles to be measured from the fracture surfaces but in the present paper a new and more fundamental approach has been proposed. Here finite-element modelling has demonstrated that once one silica nanoparticle debonds then its nearest neighbours are shielded from the applied stress field, and hence may not debond. Statistical analysis showed that, for a good, i.e. random, dispersion of nanoparticles, each nanoparticle has six nearest neighbours, so only one in seven particles would be predicted to debond. This approach therefore predicts that only 14.3% of the nanoparticles present will debond, and this value is in excellent agreement with the value of 10–15% of those nanoparticles present debonding which was recorded via direct observations of the fracture surfaces. Further, this value of about 15% of silica nanoparticles particles present debonding has also been noted in other published studies, but has never been previously explained. The predictions from the modelling studies of the toughness of the various epoxy polymers containing the silica nanoparticles were compared with the measured fracture energies and the agreement was found to be good.     

Computer simulation of gas generation and transport in landfills. V: Use of artificial neural network and the genetic algorithm for short- and long-term forecasting and planning

In the first four parts of this series a three-dimensional model was developed for transport and reaction of gaseous mixtures in a landfill. An optimization technique was also utilized in order to determine a landfill's spatial distributions of the permeability, porosity, the tortuosity factors, and the total gas generation potential of the wastes, given a limited amount of experimental data. In the present paper we develop an artificial neural network (ANN) in order to make accurate short-term predictions for several important quantities in a large landfill in southern California, including the temperature, and the CH₄, CO₂, and O₂ concentration profiles. The ANN that is developed utilizes a back-propagation algorithm. The results indicate that the ANN can be successfully trained by the experimental data, and provide accurate predictions for the important quantities in the sector of the landfill where the data had been collected. Thus, an ANN may be used by landfills' operators for short-term plannings. Moreover, we showed that a novel combination of the three-dimensional model of gas generation, flow, and transport in landfills developed in Parts I, II, and IV, the optimization technique described in Part III, and the ANN developed in the present paper is a powerful approach for developing an accurate model of a landfill for long-term predictions and planning.     

基于人工神经网络的咖啡因溶解度的建模

利用激光法测定的0-50℃温度范围内咖啡因在水和乙醇中的溶解度数据,经过比较选择了较好的人工神经网络模型—2-2-1向后传播(BP)人工神经网络模型,利用训练后的2-2-1BP人工神经网络模型对非线性的咖啡因溶解度数据进行了再现回归、内插和外推,在水中溶解度数据的回归、内插的误差均在0．07％以内,外推的误差在0．2％以内;在乙醇中的溶解度数据的回归误差在0,07％以内,内插和外推误差均在4％以内,效果十分令人满意。     

Dynamic modelling of a carbon-in-leach process with the regression network

The regression network provides a connectionist framework in which both parametric and non-parametric modelling can be implemented. It is shown how mechanistic knowledge can be built directly within the connectionist structure that results in a semi-empirical network model. In doing so the inherent freedom of a specific model is restricted so that the generalisation performance of such a model improves accordingly. It is described how a semi-empirical regression network kinetic model is developed for the dynamic modelling of the carbon-in-leach (CIL) process for gold recovery. By providing for mechanistic knowledge in the connectionist structure and catering for poorly understood aspects of the process by use of non-parametric regions within the structure of the semi-empirical regression network, the regression network kinetic model displayed significant superiority in generalisation properties over other non-parametric regression models if evaluated during dynamic simulation runs.     

RP-HPLC optimization of econea by using artificial neural networks and its antifouling performance on the Turkish coastline

Coverage of artificial surfaces within seawater by fouling organisms is defined as biofouling. Although biofouling is a natural process, it has some disadvantages for shipping industry such as increased fuel consumption, and CO₂ emission. Therefore, the ships' hull must be covered by antifouling (AF) or fouling release type coatings to overcome biofouling. In general, the so-called self-polishing AF paints contain biocides for preventing fouling organisms. Their concentrations and release rates from AF coatings are of great importance and they definitely affect both quality and cost of the coating. In the present study, we aimed at applying a new robust method. In this method, we used a model biocide, i.e., econea, to obtain its RP-HPLC optimization through artificial neural networks (ANN) and to see its antifouling performance. Column temperature, mobile phase ratio, flow rate, concentration and wavelength as input parameters and retention time as an output parameter were used in the ANN modeling. In conclusion, the R&D groups in AF paint industry may use RP-HPLC method supported with ANN modeling in further studies.     

BP神经网络方法预测聚乙二醇的羧基化率

以聚乙二醇和高锰酸钾作为原料,用反应前后溶液电导率的变化值来表征研究聚乙二醇的羧基化率,电导率的变化值越大说明聚乙二醇的羧基化率越高。并对反应所需的反应条件（如pH值、温度和配比等）作了简单的探讨。研究这些反应条件的改变对聚乙二醇的羧基化率的影响。并以matlab语言编写BP神经网络,通过实验数据对网络进行训练,然后以训练好的网络对聚乙二醇的羧基化率进行预测,结果表明,我们所建立的两层BP神经网络对不同反应条件下聚乙二醇羧基化率的仿真结果和实验数据吻合程度最低的都达到了94％,最高的能达到99.9％左右。说明BP神经网络应用于化学反应过程的预测是切实可行的。     

Modeling Oil Content of Sesame (Sesamum indicum L.) Using Artificial Neural Network and Multiple Linear Regression Approaches

Sesame (Sesamum indicum L.) is an important ancient oilseed crop with high oil content (OC) and quality. The direct selection to improve OC of sesame (OCS) due to low heritability leads to a low profit. The OCS modeling and indirect selection through high‐heritable characters associated with OCS using advanced modeling techniques is a beneficial approach to overcome this limitation that allows breeder to get a better idea of the plant properties that should be monitored during breeding experiments. This study, carried out in 2013 and 2014, compared the potential of artificial neural network (ANN) and multilinear regression (MLR) to predict OCS in the Imamzadeh Jafar plain of Gachsaran, Iran. Principal component analysis (PCA) and stepwise regression (SWR) were used to evaluate 18 input variables. Based on PCA and SWR, the 6 traits of number of capsules per plant (NCP), number of days from flowering to maturity (NDFM), plant height (PH), thousand seed weight, capsule length, and seed yield were chosen as input variables. The network with the sigmoid axon transfer function and 2 hidden layers was selected as the final ANN model. Results showed that the ANN predicted the OCS with more accuracy and efficacy (R² = 0.861, root mean square error [RMSE] = 0.563, and mean absolute error [MAE] = 0.432) compared with the MLR model (R² = 0.672, RMSE = 0.742, and MAE = 0.552). These results showed the potential of the ANN as a promising tool to predict OCS with good performance. Based on sensitivity tests, NCP followed by NDFM and PH, respectively, were the most influential factors in predicting OCS in both models. It seems that a breeding program to select or create long sesame genotypes with a long period from flowering to maturity can be a good approach to address OCS in the future.     

Comparison of Estimation Capabilities of the Artificial Neural Network with the Wavelet Neural Network in Lipase-Catalyzed Synthesis of Triethanolamine-Based Esterquats Cationic Surfactant

In this study, estimation capabilities of the artificial neural network (ANN) and the wavelet neural network (WNN) based on genetic algorithm were investigated in a synthesis process. An enzymatic reaction catalyzed by Novozym 435 was selected as the model synthesis process. The conversion of enzymatic reaction was investigated as a response of five independent variables; enzyme amount, reaction time, reaction temperature, substrates molar ratio and agitation speed in conjunction with an experimental design. After training of the artificial neurons in ANN and WNN, using the data of 30 experimental points, the products were used for estimation of the response of the 18 experimental points. Estimated responses were compared with the experimentally determined responses and prediction capabilities of ANN and WNN were determined. Performance assessment indicated that the WNN model possessed superior predictive ability than the ANN model, since a very close agreement between the experimental and the predicted values was obtained.     

Neural network modeling and analysis of gel casting preparation of porous Si3N4 ceramics

Based on orthogonal experimental results of porous Si₃N₄ ceramics by gel casting preparation, a three-layer back propagation artificial neural network (BP ANN) was developed for predicting the performances of porous Si₃N₄ ceramics. The results indicated that BP ANN was a very useful and accurate tool for the prediction and optimization of porous Si₃N₄ ceramics performances. By using the developed ANN model, the influences of the compositions on performances of porous Si₃N₄ ceramics were investigated, and some important conclusions were drawn as follows: for the flexural strength of Si₃N₄ ceramics, solid loading has an optimum value where can achieve a maximum value, and the optimum solid loading decreases with the increase of monomer content; the porosity of sintering body monotonically decreases with the increase of solid loading, and it increases with monomer content; the porosity of sintering body monotonically increases with the increase of the ratio of crosslinking agent to monomer.     

利用等离子反应器从SiF_4与CH_4制备HF和SiC的探索研究

研究了一个新型反应,即SiF_4气体与CH_4气体反应生成无水HF气体和SiC固体。该反应在最大能量为20eV的等离子反应器的固定床中进行,反应一段时间后,收集反应器壁上固体,通过分析固体中的碳含量来判断四氟化硅和甲烷的反应情况。结果表明:四氟化硅和甲烷可以发生反应,但是速度十分缓慢。