人工神经网络及其在耐火材料研究中的应用

介绍了人工神经网络的发展、特点、模型及其在材料研究中的应用现状;并以反应烧结原位ZrO2-SiC（p）材料中SiC生成量的拟合预报为例,讨论了其在耐火材料研究中的应用。     

Application of artificial neural networks to predict chemical desulfurization of Tabas coal

This paper presents a neural network model to predict the effects of operational parameters on the organic and inorganic sulfur removal from coal by sodium butoxide. The coal particle size, leaching temperature and time, sodium butoxide concentration and pre oxidation time by peroxyacetic acid (PAA) were used as inputs to the network. The outputs of the models were organic and inorganic sulfur reduction. Feed-forward artificial neural network with 5-7-10-1 arrangement, were capable to estimate organic and inorganic sulfur reduction, respectively. Simulated values obtained with neural network correspond closely to the experimental results. It was achieved quite satisfactory correlations of R² = 1 and 0.96 in training and testing stages for pyritic sulfur and R² = 1 and 0.97 in training and testing stages, respectively, for organic sulfur reduction prediction. The proposed neural network model accurately reproduces all the effects of operational variables and can be used in the simulation of Tabas coal desulfurization plant.     

NEURAL NETWORK MODEL FOR FLUIDISED BED DRYERS

The application of an artificial neural network (ANN) to model a continuous fluidised bed dryer is explored. The ANN predicts the moisture and temperature of the output solid. A three-layer network with sigmoid transfer function is used. The ANN learning is made by using a set of data that were obtained by simulating the operation by a classical model of dryer. The number of hidden nodes, learning coefficient, size of learning data set and number of iterations in the learning of the ANN were optimised. The optimal ANN has five input nodes and six hidden nodes. It is able to predict, with an error less than 10%, the moisture and temperature of the output dried solid in a small pilot plant that can treat up to 5 kg/h of wet alpeorujo. This is a wet solid waste that is generated in the two-phase decanters used to obtain olive oil.     

Experimental and artificial neural network modeling study on soot formation in premixed hydrocarbon flames

The formation of soot in premixed flames of methane, ethane, propane, and butane was studied at three different equivalence ratios. Soot particle sizes, number densities, and volume fractions were determined using classical light scattering measurement techniques. The experimental data revealed that the soot properties were sensitive to the fuel type and combustion parameter equivalence ratio. Increase in equivalence ratio increased the amount of soot formed for each fuel. In addition, methane flames showed larger particle diameters at higher distances above the burner surface and propane, ethane, and butane flames came after the methane flames, respectively. Three-layer, feed-forward type artificial neural networks having seven input neurons, one output neuron, and five hidden neurons for soot particle diameter predictions and seven hidden neurons for volume fraction predictions were used to model the soot properties. The network could not be trained and tested with sufficient accuracy to predict the number density due to a large data range and greater uncertainty in determination of this parameter. The number of complete data set used in the model was 156. There was a good agreement between the experimental and predicted values, and neural networks performed better when predicting output parameters (i.e. soot particle diameters and volume fractions) within the limits of the training data.     

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.     

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...     

Predicting the flux decline in milk cross-flow ceramic ultrafiltration by artificial neural networks

Fouling is complex phenomenon and an important drawback in the operation of membrane processes, thus its modeling involves scientific and commercial interest. In this research work, experimental data were collected by carrying out a sequence of cycles comprising both milk ultrafiltration through a 50 kDa tubular ceramic membrane and cleaning protocols with different agents. Then, it was developed an artificial neural network model that receives as inputs the operational cycle, the aggressivity of the cleaning and the filtration time and returns as output the permeate flux. Several training algorithms were tested and excellent fitting was obtained with the Levenberg-Marquardt one.     

PREDICTION OF EQUILIBRIUM PHASE COMPOSITIONS AND β-GLUCOSIDASE PARTITION COEFFICIENT IN AQUEOUS TWO-PHASE SYSTEMS

Artificial neural networks (ANN) and Flory-Huggins (F-H)-type models were implemented to simulate the binodal curve of an aqueous two-phase, system (ATPS) composed of poly(ethylene glycol), potassium phosphate, and water. The ANN model outperformed the F-H model in predicting the equilibrium compositions of the PEG-rich phase (average percent deviation: 10.0 versus 56.6). However, the estimation of interaction parameters was feasible only in the thermodynamic framework. Beta-glucosidase was introduced into the system under various temperature (25°–50°C) and pH conditions (6.5–8.0). The β-glucosidase partition coefficient increased with the temperature and pH over a range of 0.11–1.18. The network was better suited to predict the partitioning behavior of the enzyme because of the increased number of interaction parameters. The artificial intelligence–guided approach for isolating the enzyme has the potential to reduce costs, improve performance, and identify the most favorable purification conditions.     

Predicting compressive strength of different geopolymers by artificial neural networks

In the present study, six different models based on artificial neural networks have been developed to predict the compressive strength of different types of geopolymers. The differences between the models were in the number of neurons in hidden layers and in the method of finalizing the models. Seven independent input parameters that cover the curing time, Ca(OH)₂ content, the amount of superplasticizer, NaOH concentration, mold type, geopolymer type and H₂O/Na₂O molar ratio were considered. For each set of these input variables, the compressive strength of geopolymers was obtained. A total number of 399 input-target pairs were collected from the literature, randomly divided into 279, 60 and 60 data and were trained, validated and tested, respectively. The best performance model was obtained through a network with two hidden layers and absolute fraction of variance of 0.9916, the absolute percentage error of 2.2102 and the root mean square error of 1.4867 in training phase. Additionally, the entire trained, validated and tested network showed a strong potential for predicting the compressive strength of geopolymers with a reasonable performance in the considered range.     

Nonlinear system identification and model reduction using artificial neural networks

We present a technique for nonlinear system identification and model reduction using artificial neural networks (ANNs). The ANN is used to model plant input–output data, with the states of the model being represented by the outputs of an intermediate hidden layer of the ANN. Model reduction is achieved by applying a singular value decomposition (SVD)-based technique to the weight matrices of the ANN. The sequence of state values is used to convert the model to a form that is useful for state and parameter estimation. Examples of chemical systems (batch and continuous reactors and distillation columns) are presented to demonstrate the performance of the ANN-based system identification and model reduction technique.     

The use of artificial neural networks in electrostatic force microscopy

ABSTRACT: The use of electrostatic force microscopy (EFM) to characterize and manipulate surfaces at the nanoscale usually faces the problem of dealing with systems where several parameters are not known. Artificial neural networks (ANNs) have demonstrated to be a very useful tool to tackle this type of problems. Here, we show that the use of ANNs allows us to quantitatively estimate magnitudes such as the dielectric constant of thin films. To improve thin film dielectric constant estimations in EFM, we first increase the accuracy of numerical simulations by replacing the standard minimization technique by a method based on ANN learning algorithms. Second, we use the improved numerical results to build a complete training set for a new ANN. The results obtained by the ANN suggest that accurate values for the thin film dielectric constant can only be estimated if the thin film thickness and sample dielectric constant are known.PACS: 07.79.Lh; 07.05.Mh; 61.46.Fg.     

Neural network modelling of the inductively coupled RF plasma synthesis of silicon nanoparticles

The synthesis of silicon nanopowders by an inductively coupled plasma (ICP) process is investigated. The specific surface area (SSA) of the elaborated particles is determined by nitrogen absorption (BET) as a function of the quenching gas flow rate and the precursor feeding rate. Nanopowders with specific surface areas varying from 69 to 194 m² g^− 1, corresponding to equivalent particle sizes of 37 and 13 nm respectively, could be produced. The correlation between these two input parameters and the output SSA has been numerically modelled with linear regression and artificial neural networks approaches. It has been demonstrated that with the available data set, a regression model with quadratic regressors and a neural network modelling give a similar response.     

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.     

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.     

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.     

Friction and wear behaviour of CMCs based on Portland clinker against steel countermaterial

Abstract

The wear behaviour of ceramic materials against steel has been studied with respect to the viability of using clinker as an inex pensive component. Friction and wear behaviour of composites based on Portland clinker reinforced with 3, 6, or 9 wt-% of three different oxides (alumina, magnesia, silica) was evaluated against a steel countermaterial (910 HV) using a pin on disk test. The composites were prepared by dry mixing and compacting at 180 MPa using cold isostatic pressing; sintering was carried out at 1400°C in air. All samples were polished to 0.8 μm. Friction coefficients and wear rates were determined and the materials characterised by optical and scanning electron microscopy.