Multilayer Artificial Neural Network Research Articles

Prediction of ground subsidence due to underground mining through time using multilayer feed-forward artificial neural networks and back-propagation algorithm – case study at Mong Duong underground coal mine (Vietnam)

The paper is devoted to studying the possibility of using artificial neural networks (ANN) to estimate ground subsidence caused by underground mining. The experiments showed that the most suitable network structure is a network with three layers of perceptrons and four neurons in the hidden layer with the back propagation algorithm (BP) as a training algorithm. The subsidence observation data in the Mong Duong underground coal mine and other parameters, including: (1) the distance from the centre of the stope to the ground monitoring points; (2) the volume of mined-out space; (3) the positions of the ground points in the direction of the main cross-section of the trough; and (4) the time (presented by cycle number), were used as the input data for the ANN. The findings showed that the selected model was suitable for predicting subsidence along the main profile within the subsidence trough. The prediction accuracy depended on the number of cycles used for the network training as well as the time interval between the predicted cycle and the last cycle in the training dataset. When the number of monitoring cycles used for the network training was greater than eight, the largest values of RMS and MAE were less than 10 % compared to the actual maximum subsidence value for each cycle. If the network training was less than eight cycles, the results of prediction did not meet the accuracy requirements.

Prediction of Static Characteristic Parameters of an Insulated Gate Bipolar Transistor Using Artificial Neural Network.

Breakdown voltage (BV), on-state voltage (Von), static latch-up voltage (Vlu), static latch-up current density (Jlu), and threshold voltage (Vth), etc., are critical static characteristic parameters of an IGBT for researchers. Von and Vth can characterize the conduction capability of the device, while BV, Vlu, and Jlu can help designers analyze the safe operating area (SOA) of the device and its reliability. In this paper, we propose a multi-layer artificial neural network (ANN) framework to predict these characteristic parameters. The proposed scheme can accurately fit the relationship between structural parameters and static characteristic parameters. Given the structural parameters of the device, characteristic parameters can be generated accurately and efficiently. Compared with technology computer-aided design (TCAD) simulation, the average errors of our scheme for each characteristic parameter are within 8%, especially for BV and Vth, while the errors are controlled within 1%, and the evaluation speed is improved more than 107 times. In addition, since the prediction process is mathematically a matrix operation process, there is no convergence problem, which there is in a TCAD simulation.

Information and communication technologies for intelligent control of ground unmanned multipur-pose vehicles

At present, there is a situation when the practice of creating modern devices, units and systems of ground unmanned vehicles (GUV) has preceded the theory of information analysis and synthesis of complex systems. Some existing solutions for the information support of the GUV need to be generalized, standardized and unified, the definition of new special requirements for the creation of computer systems and networks in transport should be made. Therefore, it is necessary and actual to develop information and communication technology for intelligent control of multi-purpose GUV. Goal. The purpose of the article is to develop information and communication technology for intelligent control of multi-purpose GUV, which work in conditions of intensive loads, difficult operating conditions, increased responsibility of mechanisms, resulting in reduced human losses and energy consumption, increased machine reliability and control accuracy. Methodology. The methodology of tasks is based on the use of a synergetic approach to the creation of information and communication technology for intelligent control of multi-purpose GUV, as the development of a systematic approach taking into account the adaptation and integration of the vehicle to transport infrastructure and/or rough terrain. The principles of synergetics underlie the construction of mechatronic systems – a combination in one unit of components of different technical nature (mechanical, electrical, computer), which adaptively interact with the external environment as a single functional and structural organism. The synergetic approach deals with phenomena and processes, as a result of which the system – as a whole – may have properties that none of the parts has. Results. The concept of intelligent control of multi-purpose GUV on the basis of artificial hybrid neuro-phase regulators with the use of cloud computing services and deep learning technology is proposed, substantiated and implemented, which allows to qualitatively increase the efficiency of both one vehicle and the transport system on the base combining a synergetic approach and evolutionary methods of learning multilayer artificial neural networks by objectively forming the architecture of these networks on the basis of learning functionalities and appropriate control objectives. Originality. Use of cloud computing services based on deep learning for intelligent control technology of GUV. Practical value. The proposed information and communication technology of intelligent control of multi-purpose GUV will significantly reduce energy consumption, psychophysical stress on crew members, as well as increase the accuracy of location.

Artificial Neural Network Models for Predicting and Optimizing the Effect of Air-frying Time and Temperature on Physical, Textural, Sensory, and Nutritional Quality Parameters of Fish Ball

ABSTRACT The study was conducted to see the effect of air-frying time and temperature on physical, textural, sensory and nutritional quality parameters of air-fried fish balls and compared with the quality parameters of deep-fried fish balls. Multilayer feed-forwarded artificial neural network (ANN) models were fitted to the experimental data to predict the response variables of air-fried fish balls as a function of frying temperature and time. The model was validated using the holdback method. Based on the R2 and RMSE values, ANN model with three hidden layers was found to be best fitted model to explain the variability in the proximate composition, texture and sensory data. The desired air-frying condition obtained was temperature at 200 oC and time at 8 minutes by multi-response desirability score. The air-fried fish balls at desired condition had 81 % reduction in fat content and 17 % increase in protein content compared to deep fried sample.

Forecasting of Turkish Sovereign Sukuk Prices Using Artificial Neural Network Model

Recently, artificial neural networks have been successfully applied in many areas such as forecasting financial time series, predicting financial failure, and classification of ratings. However, it has hardly been applied in forecasting sukuk prices, which is considered the most common Islamic capital market instrument. Since sukuk is a new financial asset, there are not enough studies in this area. Therefore, this study aims to forecast the Turkish sovereign sukuk prices using with artificial neural network model and to reveal the determinants in the forecasting of sukuk prices. For this purpose, a multi-layer feed forward artificial neural network model is designed using dollar-based international sovereign sukuk price data issued by the Turkish Ministry of Treasury and Finance. The dollar index, volatility index, geopolitical risk index, Standard and Poor’s Middle East and North Africa sukuk index, and Eurobond prices constituted as input variables of the designed model and the sovereign sukuk prices formed the output. As a result, the sovereign sukuk prices were forecasted accurately at the success rate of 99.98%. The accurate forecasting of sukuk prices will play a critical role in reducing the risk perception of sukuk investors and increasing their profitability. The findings of the study are important in terms of proving that the artificial neural network model is an effective model for forecasting the sukuk prices and revealing that the dollar index, volatility index, geopolitical risk index, Standard and Poor’s MENA sukuk index, and Eurobond prices are determinants in forecasting sukuk prices.

Deep Learning for Rapid Identification of Microbes Using Metabolomics Profiles.

Rapid detection of viable microbes remains a challenge in fields such as microbial food safety. We here present the application of deep learning algorithms to the rapid detection of pathogenic and non-pathogenic microbes using metabolomics data. Microbes were incubated for 4 h in a protein-free defined medium, followed by 1D 1H nuclear magnetic resonance (NMR) spectroscopy measurements. NMR spectra were analyzed by spectral binning in an untargeted metabolomics approach. We trained multilayer (“deep”) artificial neural networks (ANN) on the data and used the resulting models to predict spectra of unknown microbes. ANN predicted unknown microbes in this laboratory setting with an average accuracy of 99.2% when using a simple feature selection method. We also describe learning behavior of the employed ANN and the optimization strategies that worked well with these networks for our datasets. Performance was compared to other current data analysis methods, and ANN consistently scored higher than random forest models and support vector machines, highlighting the potential of deep learning in metabolomics data analysis.

Artificial neural network study of the electrical conductivity of mould flux

The electrical conductivity of mould flux with chemical constitution of CaO-SiO2-Al2O3-NaO-K2O-MgO-CaF2-Cr2O3-FeO-MnO has been investigated. The assessed database contains one unitary, five binary, nine ternary, four quaternary, two quinary, two senary and one octonary subsystems. Each constitutional component is in connection to another via some direct or indirect links. A multilayer artificial neural network method was developed and implemented in the database. The work provides a method to calculate the relationships between the composition, temperature and electrical property of the mould flux within the defined parameter ranges. The results have been validated against those experimental data that are not included in the training of the neural networks.

Neural Network Parameter Identification Based Constitutive Modeling of Superelastic Shape Memory Alloys

AbstractSuperelastic shape memory alloys (SMAs) are two‐phase polycrystal materials with hysteretic energy dissipation and deformation recovery capabilities. This feature opens numerous application possibilities, particularly using SMA wires in structural vibration control. However, vibrations, such as wind and earthquake induced, may occur in high rate regimes with alternating amplitudes. Experiments show that the thermodynamically coupled SMA behavior is highly sensitive to strain rate and amplitude. Accordingly, the prediction accuracy of the constitutive models depends notably on their thermodynamic parameters, such as latent heat, heat transfer coefficient and specific heat. The identification of these parameters requires extensive experiments. To circumvent this challenge, this study proposes for SMA wires a machine learning based parameter identification procedure using artificial neural networks (ANNs). In this approach, a recently improved uniaxial macroscopic constitutive SMA model is utilized as a forward physics based model to compute all possible responses in an expected parameter space. With the generated data, a multilayer ANN is trained as a data based inverse model. The Latin hypercube sampling and the backpropagation learning algorithms are applied. After training, the data based model is able to identify suitable thermodynamic parameters from SMA stress responses. With the identified parameters, the constitutive model replicates the strain rate and amplitude dependent SMA response effects. The experimental results are matched by the model with high accuracy.

Enhanced Mineral Quantification and Uncertainty Analysis from Downhole Spectroscopy Logs Using Variational Autoencoders

This paper describes an innovative machine-learning application, based on variational autoencoder frameworks, to quantify the concentrations and associated uncertainties of common minerals in sedimentary formations using the measurement of atomic element concentrations from geochemical spectroscopy logs as inputs. The algorithm comprises an input(s), encoder, decoder, output(s), and a novel cost function to optimize the model coefficients during training. The input to the algorithm is a set of dry-weight concentrations of atomic elements with their associated uncertainty. The first output is a set of dry-weight fractions of 14 minerals, and the second output is a set of reconstructed dry-weight concentrations of the original elements. Both sets of outputs include estimates of uncertainty on their predictions. The encoder and decoder are multilayer feed-forward artificial neural networks (ANN), with their coefficients (weights) optimized during calibration (training). The cost function simultaneously minimizes error (accuracy metric) and variance (precision or robustness metric) on the mineral and reconstructed elemental outputs. Training of the weights is done using a set of several-thousand core samples with independent, high-fidelity elemental and mineral (quartz, potassium-feldspar, plagioclase-feldspar, illite, smectite, kaolinite, chlorite, mica, calcite, dolomite, ankerite, siderite, pyrite, and anhydrite) data. The algorithm provides notable advantages over existing methods to estimate formation lithology or mineralogy relying on simple linear, empirical, or nearest-neighbor functions. The ANN numerically capture the multidimensional and nonlinear geochemical relationship (mapping) between elements and minerals that is insufficiently described by prior methods. Training is iterative via backpropagation and samples from Gaussian distributions on each of the elemental inputs, rather than single values, for every sample at each iteration (epoch). These Gaussian distributions are chosen to specifically represent the unique statistical uncertainty of the dry-weight elements in the logging measurements. Sampling from Gaussian distributions during training reduces the potential for overfitting, provides robustness for log interpretations, and further enables a calibrated estimate of uncertainty on the mineral and reconstructed elemental outputs, all of which are lacking in prior methods. The framework of the algorithm is purposefully generalizable so that it can be adapted across geochemical spectroscopy tools. The algorithm reasonably approximates a “global-average” model that requires neither different calibrations nor expert parameterization or intervention for interpreting common oilfield sedimentary formations, although the framework is again purposefully generalizable so it can be optimized for local environments where desirable. The paper showcases a field application of the method for estimating mineral type and abundance in oilfield formations from wellbore-logging measurements.

Reference evapotranspiration modeling using radial basis function neural network in different agro-climatic zones of Chhattisgarh

Precise estimation of evapotranspiration (ET) is extremely essential for efficient utilization of available water resources. Among the empirical models, FAO-Penman-Monteith equation (FAO-PM) is considered as standard method to determine reference evapotranspiration (ET ). In developing countries like India, application of FAO-PM equation for ET estimation has certain limitations due to unavailability of specific data requirements. Several empirical models such as Hargreaves, Turc, Blaney-Criddle etc., arealso considered for ET estimation. However, ET estimates obtain with these models are not comparable with benchmark FAO-PM ET . To address this issue, potential of radial basis function neural network (RBFNN) is investigated to estimate FAO-PM ET . Result obtained with proposed RBFNN models are compared with equivalent multi-layer artificial neural network (MLANN) and empirical approach of Hargreaves, Turc and Blaney-Criddle. Lower RMSE values obtained with RBFNN and MLANN models is an indication of improved performance over empirical models. Similarly, higher R2 and Efficiency Factor obtained with RBFNN and MLANN models also approves the superiority of machine learning techniques over empirical models. Among the two machine learning techniques, RBFNN models performed better as compared to MLANN. In a nut shell, proposed RBFNN models can simulate FAO-PM ET even with limited meteorological parameters and consistence degree of accuracy level.

Flutter speed prediction by using deep learning

Deep learning technology has been widely used in various field in recent years. This study intends to use deep learning algorithms to analyze the aeroelastic phenomenon and compare the differences between Deep Neural Network (DNN) and Long Short-term Memory (LSTM) applied on the flutter speed prediction. In this present work, DNN and LSTM are used to address complex aeroelastic systems by superimposing multi-layer Artificial Neural Network. Under such an architecture, the neurons in neural network can extract features from various flight data. Instead of time-consuming high-fidelity computational fluid dynamics (CFD) method, this study uses the K method to build the aeroelastic flutter speed big data for different flight conditions. The flutter speeds for various flight conditions are predicted by the deep learning methods and verified by the K method. The detailed physical meaning of aerodynamics and aeroelasticity of the prediction results are studied. The LSTM model has a cyclic architecture, which enables it to store information and update it with the latest information at the same time. Although the training of the model is more time-consuming than DNN, this method can increase the memory space. The results of this work show that the LSTM model established in this study can provide more accurate flutter speed prediction than the DNN algorithm.

Simultaneous optimization of thermal and electrical conductivity of high density polyethylene-carbon particle composites by artificial neural networks and multi-objective genetic algorithm

Prediction and achievement of, simultaneously, the highest thermal and lowest electrical conductivity in polymer composites prior to their manufacturing is of great interest to avoid excessive experimentation and to achieve the best material performance for diverse applications. However, traditional modeling and optimization methods are not effective due to the non-linear complex behavior of both properties in polymer composites. In this study, two artificial neural networks (ANN) were developed with the aim of approximate the thermal conductivity and the electrical conductivity of high density polyethylene (HDPE)-carbon particle composites, based on data obtained experimentally. Composites were prepared by twin-screw extrusion using four different types of carbon particles at different concentrations. The obtained ANN models were used as objective functions in a multi-objective genetic algorithm (GA) to optimize the design parameters of the composites to maximize their thermal conductivity and minimize their electrical conductivity. ANN models showed a good correlation between simulated and experimental data, evidenced by correlation factors, R, above 0.97. Multi-layer perceptron ANN with three neurons in a single hidden layer and trained by the Levenberg-Marquardt algorithm exhibited the best predictive performance in both models. As a result of the multi-objective optimization process by GA, a set of Pareto optimal solutions for maximizing thermal conductivity and minimizing electrical conductivity was obtained. Conformity tests were performed to validate the optimization capability of the GA method. The optimization and modeling procedure developed can be applied to other properties of polymer composites.

A Hybrid Approach to Classify Power Quality Problems in Distribution Systems

Electrical power systems are expected to transmit continuously nominal rated sinusoidal voltage and current to consumers. However, the widespread use of power electronics has brought power quality problems. This study performs classification of power quality disturbances using an artificial neural network (ANN). The most appropriate ANN structure was determined using the Box-Behnken experimental design method. Nine types of disturbance (no fault, voltage sag, voltage, swell, flicker, harmonics, transient, DC component, electromagnetic interference, instant interruption) were investigated in computer simulations. The feature vectors used in the identification of the different types of disturbances were produced using the discrete wavelet transform and principal component analysis. Our results show that the optimized feed forward multilayer ANN structure successfully distinguishes power quality disturbances in simulation data and was also able to identify these disturbances in real time data from substations.

Flamenco experience in Córdoba, Spain: estimating tourist profiles by multilayer perceptron's artificial neural networks

PurposeThis research aims to demonstrate the possible link between the sociodemographic profile of visitors motivated by the visit to flamenco shows and the city of Córdoba (Spain), and the preferences and sensations regarding these experiences.Design/methodology/approachThe methodology used (multilayer perceptron) is based on the development of an artificial neural network.FindingsThe results show that the variables age and educational level are determining factors in the profile of the visitor. Also, as the level of income increases, so does the interest in flamenco, a fact that can be useful to determine the target audience for this type of shows.Originality/valueFlamenco is an art that originated in the Andalusian region that arouses the interest of the visitor due to its music, way of singing and dance. Flamenco is a popular art that excites and awakens the senses of those who attend this dance, song and guitar show. Its recognition as Intangible Heritage of Humanity by United Nations Educational, Scientific and Cultural Organization (UNESCO) since 2010, makes it a tourist product that motivates visitors to travel to the city of Córdoba (Spain), being also one of flamenco's places of origin. Córdoba has this art in its tourist offer so that the identity of the city has two aspects: patrimonial and immaterial, among the flamenco highlighted.

Photo-Fenton-Like Treatment of Municipal Wastewater

In this work, the photochemical treatment of a real municipal wastewater using a persulfate-driven photo-Fenton-like process was studied. The wastewater treatment efficiency was evaluated in terms of total carbon (TC), total organic carbon (TOC) and total nitrogen (TN) removal. Response surface methodology (RSM) in conjunction Box-Behnken design (BBD) and multilayer artificial neural network (ANN) have been utilized for the optimization of the treatment process. The effects of four independent factors such as reaction time, pH, K2S2O8 concentration and K2S2O8/Fe2+ molar ratio on the TC, TOC and TN removal have been investigated. The process significant factors have been determined implementing Analysis of Variance (ANOVA). Both RSM and ANN accurately found the optimum conditions for the maximum removal of TOC (100% and 98.7%, theoretically), which resulted in complete mineralization of TOC at the reaction time of 106.06 min, pH of 7.7, persulfate concentration of 30 mM and K2S2O8/Fe2+ molar ratio of 7.5 for RSM and at the reaction time of 104.93 min, pH of 7.7, persulfate concentration of 30 mM and K2S2O8/Fe2+ molar ratio of 9.57 for ANN. On the contrary, the attempts to find the optimal conditions for the maximum TC and TN removal using statistical, and neural network models were not successful.

Forecasting the CO2 Emissions at the Global Level: A Multilayer Artificial Neural Network Modelling

Better accuracy in short-term forecasting is required for intermediate planning for the national target to reduce CO2 emissions. High stake climate change conventions need accurate predictions of the future emission growth path of the participating countries to make informed decisions. The current study forecasts the CO2 emissions of the 17 key emitting countries. Unlike previous studies where linear statistical modeling is used to forecast the emissions, we develop a multilayer artificial neural network model to forecast the emissions. This model is a dynamic nonlinear model that helps to obtain optimal weights for the predictors with a high level of prediction accuracy. The model uses the gross domestic product (GDP), urban population ratio, and trade openness, as predictors for CO2 emissions. We observe an average of 96% prediction accuracy among the 17 countries which is much higher than the accuracy of the previous models. Using the optimal weights and available input data the forecasting of CO2 emissions is undertaken. The results show that high emitting countries, such as China, India, Iran, Indonesia, and Saudi Arabia are expected to increase their emissions in the near future. Currently, low emitting countries, such as Brazil, South Africa, Turkey, and South Korea will also tread on a high emission growth path. On the other hand, the USA, Japan, UK, France, Italy, Australia, and Canada will continuously reduce their emissions. These findings will help the countries to engage in climate mitigation and adaptation negotiations.

Development and validation of a predictive model for feeding intolerance in intensive care unit patients with sepsis.

Background:Feeding intolerance in patients with sepsis is associated with a lower enteral nutrition (EN) intake and worse clinical outcomes. The aim of this study was to develop and validate a predictive model for enteral feeding intolerance in the intensive care unit patients with sepsis.Methods:In this dual-center, retrospective, case-control study, a total of 195 intensive care unit patients with sepsis were enrolled from June 2018 to June 2020. Data of 124 patients for 27 clinical indicators from one hospital were used to train the model, and data from 71 patients from another hospital were used to assess the external predictive performance. The predictive models included logistic regression, naive Bayesian, random forest, gradient boosting tree, and deep learning (multilayer artificial neural network) models.Results:Eighty-six (44.1%) patients were diagnosed with enteral feeding intolerance. The deep learning model achieved the best performance, with areas under the receiver operating characteristic curve of 0.82 (95% confidence interval = 0.74–0.90) and 0.79 (95% confidence interval = 0.68–0.89) in the training and external sets, respectively. The deep learning model showed good calibration; based on the decision curve analysis, the model's clinical benefit was considered useful. Lower respiratory tract infection was the most important contributing factor, followed by peptide EN and shock.Conclusions:The new prediction model based on deep learning can effectively predict enteral feeding intolerance in intensive care unit patients with sepsis. Simple clinical information such as infection site, nutrient type, and septic shock can be useful in stratifying a septic patient's risk of EN intolerance.

Machine Learning with Explainable Artificial Intelligence Vision for Characterization of Solution Conductivity Using Optical Emission Spectroscopy of Plasma in Aqueous Solution

AbstractThis study presents an explainable artificial intelligence (XAI) vision for optical emission spectroscopy (OES) of plasma in aqueous solution. We aim to characterize the plasma and OES with XAI. Trained with 18000 spectra, a multilayer artificial neural network (ANN) model accurately predicted the solution conductivity. Local interpretable model‐agnostics explanations (LIME), an XAI method, interpreted the model through perturbing spectral features and fitting the feature contribution with a linear model. LIME showed that OH, Hγ, and Hβ emission lines were critical to the model, differing from the lines typically selected by humans. The results demonstrated that machine captured the spectral features neglected by humans. We believe using XAI for plasma OES analysis impacts the fields of plasma and analytical chemistry.

Behavior factor prediction equations for reinforced concrete frames under critical mainshock-aftershock sequences using artificial neural networks

ABSTRACT This paper proposes the ductility demands control of reinforced concrete frames under criticalsuccessive earthquakes using evaluation of behavior factors (R factors). The influence of RC frameperiods, PGA and magnitude of mainshocks and aftershocks is also taken into account by 10 training ideal artificial neural network (ANN) and proposing the empirical equations. Firstly, 2D RC frames are implemented in Opensees and then evaluated under as-recorded critical single and successive scenario. R factors are calculated and compared for single and successive cases. It is found that the sequences of critical records decrease R factors and capacity of RC frames about 18% and 30%, respectively. Despite what is necessitated in the seismic design codes, proposing a constant value as R factor for whole RC structure especially under single scenarios cannot lead to proper design of structures. Hence, the idealized multilayer ANNs employed to generate the empirical charts for evaluation of R factors.

Predicting the Degree of Dissolved Oxygen Using Three Types of Multi-Layer Perceptron-Based Artificial Neural Networks

Predicting the level of dissolved oxygen (DO) is an important issue ensuring the sustainability of the inhabitants of a river. A prediction model can predict the DO level using a historical dataset with regard to water temperature, pH, and specific conductance for a given river. The model can be built using sophisticated computational procedures such as multi-layer perceptron-based artificial neural networks. Different types of networks can be constructed for this purpose. In this study, the authors constructed three networks, namely, multi-verse optimizer (MVO), black hole algorithm (BHA), and shuffled complex evolution (SCE). The networks were trained using the datasets collected from the Klamath River Station, Oregon, USA, for the period 2015–2018. We found that the trained networks could predict the DO level of 2019. We also found that both BHA- and SCE-based networks could predict the level of DO using a relatively simple configuration compared to that of MVO. From the viewpoints of absolute errors and Pearson’s correlation coefficient, MVO- and SCE-based networks performed better than BHA-based networks. In synopsis, the authors recommend MVO- and MLP-based artificial neural networks for predicting the DO level of a river.