AbstractAn Artificial Neural Network Research Articles

Evaluation of adsorption potential of illite/quartz/kaolinite/montmorillonite for Cr(VI), Ni(II), and Cu(II) ions and modeling of experimental results by artificial neural networks

Abstract An artificial neural network (ANN) was used to predict the removal efficiency of Cr(VI), Ni(II), and Cu(II) ions on riverbed sand containing illite/quartz/kaolinite/montmorillonite (IQKM) clay minerals. The effect of operational parameters such as initial metal ion concentration (10–100 mg/L), initial pH (2–10), adsorbent dosage (0.025–0.15 g/L), contact time (15–90 min), agitation speed (100–800 rpm), and temperature (303–323 K) is studied to optimize the conditions for greatest removal of metal ions. Employment of equilibrium isotherm models for the description of adsorption capacities for IQKM explored better efficiency of the Langmuir model for the best representation of experimental data with the highest adsorption capacity of 8.802, 7.5125, 6.608 mg/g for Cr(VI), Ni(II), and Cu(II) ions in the solution. The kinetics of the proposed adsorption processes efficiently followed pseudo-second-order and intraparticle diffusion kinetic models. .

Application of artificial neural networks for predicting the isotopic composition of high burn-up solid plutonium sample using the 90–105 keV gamma-spectrum region

Abstract An artificial neural network (ANN) algorithm was developed to predict isotopic composition of five Pu isotopes (238Pu, 239Pu, 240Pu, 241Pu, and 242Pu) of high burn-up Pu samples. The study was carried out using the most complex but informative gamma energy region of Pu gamma spectra, 90–106 keV. This region has remained futile, due to the overlapping nature of the gamma emission lines and X-rays emitted by U, Pu, and Np. A backpropagation neural network algorithm based ANN with error minimization using the steepest gradient method was built with the help of normalized gamma spectra for ∼800 samples. The paper discusses the optimization of hidden neuron number and the layer design for best prediction. With the exception of 242Pu, the prediction accuracy and precision of the proposed technique was found to be ∼3% for all other isotopes of Pu.

Modeling and optimization of a continuous electrocoagulation process using an artificial intelligence approach

Abstract An artificial neural network (ANN) with the topology 8-94-85-2 (input – hidden layer 1 - hidden layer 2 - output) was used to model the operation of the continuous electrocoagulation (CEC) process for the removal of fluoride from water. After the ANN training, the sum of the squared errors (MSE) and the determination coefficient (R2) of the testing set model predictions were 0.0088 and 0.999, respectively, showing a good generalization and the model's predictive capacity. The optimization of the process cost using the genetic algorithm (GA) showed that the optimal conditions are highly dependent on the feed concentration and the fluoride removal requirements. For 5 L of water containing 10 mg/L of fluoride, the optimal conditions to reduce the fluoride concentration below the permissible limit (1.5 mg/L) are 88.3 mA of current intensity, a flow rate of 73.6 mL/min, and the use of a series monopolar (SM) electrode configuration, corresponding to a fluoride removal of 85% and an operating cost of 0.05 €/L.

A classification of meteor radio echoes based on artificial neural network

Abstract An artificial neural network is described for classification of meteor trails into the distinct overdense, intermediate and underdense trail categories. The neural network was trained and on model data obtained using the “KAMET” program and tested on real data. The best result of classification success rate of 95% without according to the heights of the formation of meteor trails. Results of classification with according to the heights of the formation of meteor trails are 82% - 91%.

Prediction of Drape Coefficient by Artificial Neural Network

Abstract An artificial neural network (ANN) model was developed to predict the drape coefficient (DC). Hanging weight, Sample diameter and the bending rigidities in warp, weft and skew directions are selected as inputs of the ANN model. The ANN developed is a multilayer perceptron using a back-propagation algorithm with one hidden layer. The drape coefficient is measured by a Cusick drape meter. Bending rigidities in different directions were calculated according to the Cantilever method. The DC obtained results show a good correlation between the experimental and the estimated ANN values. The results prove a significant relationship between the ANN inputs and the drape coefficient. The algorithm developed can easily predict the drape coefficient of fabrics at different diameters.

Modeling of a UASB Reactor by NARX Networks for Biogas Production

Abstract An Artificial Neural Network model of a UASB Reactor has been developed. The reactor treats bagasse wash water (containing organics), generated after washing of stored bagasse prior to its use in paper manufacture. In the process, biogas, a renewable source of energy is produced. As the UASB reactors (2×5,000 m3 volume) operate mostly with feed having varying characteristics, therefore a special type of dynamic networks, called NARX networks have been used to model it for predicting biogas production rate. The input to the model is influent flow rate, inlet and outlet COD. Model is based upon 576 days plant data. NARX model architecture consists of input, output, and 2 hidden layers each having 10 neurons and utilizes 4 days delay. The developed ANN model represents the dynamic behavior of UASB reactor and recursively predicts and forecasts the biogas production rate with acceptable deviation with respect to actual production rate.