Feed Forward Back Propagation Learning Research Articles

Simultaneous spectrophotometric determination of Co (II) and Co (III) in acidic medium with partial least squares regression and artificial neural networks

This study aims at the application of two chemometric techniques to visible spectra of acetic acid solutions of Co (II) and Co (III) for simultaneous determination thereof. Spectral data of 145 samples in the range of 400–700 nm were used to build the models. Partial least squares regression models were developed for which latent variables were determined using internal cross-validation with a leave-one-out strategy and 3 and 2 latent variables were selected for Co(II) and Co(III) based on root mean square error of cross-validation. For these models, root mean square errors of prediction were 1.16 and 0.536 mM and coefficients of determination were 0.975 and 0.892 for Co (II) and Co (III). As an alternate method, artificial neural networks consisting of three layers, with 10 neurons in hidden layer, were trained to model spectra and concentrations of cobalt species. Levenberg-Marquardt algorithm with feed-forward back-propagation learning resulted root mean square errors of prediction of 0.316 and 0.346 mM for Co (II) and Co (III) respectively and coefficients of determination were 0.996 and 0.988.

The effectiveness of multivariate and univariate spectrophotometric techniques for the concurrent estimation of ornidazole and ciprofloxacin HCl in tablet formulation and spiked serum: estimating greenness and whiteness profile

In this manuscript, the effectiveness of multivariate and univariate tools in conjunction with spectrophotometric techniques was evaluated for the concurrent analysis of ciprofloxacin (CI) and ornidazole (OR) in prepared mixtures, tablets, and human serum. The artificial neural network was chosen as the multivariate Technique. Bayesian regularization (trainbr) and Levenberg–Marquardt algorithms (trainlm), were constructed and trained using feed-forward back-propagation learning. The optimal logarithm was determined based on mean recovery, mean square error of prediction (MSEP), relative root mean square error of prediction (RRMSEP), and bias-corrected MSEP (BCMSEP) scores. Trainbr outperformed trainlm, yielding a mean recovery of 100.05% for CI and 99.84% for OR, making it the preferred algorithm. Fourier self-deconvolution and mean-centering transforms were chosen as the univariate Techniques. Fourier self-deconvolution was applied to the zero-order spectra of ciprofloxacin and ornidazole by electing an appropriate full width at half maximum, enhancing peak resolution at 380.1 nm and 314.2 nm for CI and OR, respectively. Mean centering transform was applied to CI and OR ratio spectra to eliminate constant signals, enabling accurate quantification of CI and OR at 272.0 nm and 306.2 nm, respectively. The introduced approaches were optimized and validated for precise CI and OR analysis, with statistical comparison against the HPLC method revealing no notable differences. The sustainability of these approaches was confirmed through the green certificate (modified eco-scale), AGP, and whiteness-evaluation tool, corroborating their ecological viability.

Artificial Neural Network Approach for Relativistic Polytropes

Over the last few decades, artificial neural networks (ANN) have played an essential role in many areas of human activity and have found application in many branches of natural sciences. ANNs have been widely used to tackle problems related to linear and nonlinear differential equations, and numerous paradigms for ANN architecture have been employed. This research proposes a computational technique based on ANN and the Taylor series to solve difficulties connected to the relativistic gas spheres' Tolman-Oppenheimer-Volkoff equations (TOV). We used ANN to study two cases related to relativistic polytropes. The first is to simulate both the Emden and the relativistic functions. The second is to predict the zeros of the Emden function and its corresponding relativistic functions. In its feed-forward back-propagation learning scheme, we used the ANN framework. The efficiency of the proposed algorithm is evaluated by running it through seven models. Comparing the analytical and the ANN solutions gives good agreement for the two cases under study.

Prediction of solid circulation rate in an internal circulating fluidized bed: An empirical and ANN approach

Internal Circulating Fluidized beds (ICFBs) are interested in various industries because of their higher thermal efficiency and high reaction rates. However, understanding the complex flow hydrodynamics inside an ICFB is challenging. Also, the experiments needed are expensive and time-consuming. Therefore this study aims to predict the solids circulation rate in an in-house ICFB (0.3 m internal diameter x 3.0 m height) using an empirical model and an Artificial Neural Network (ANN) technique. The solid circulation rates measured at different operating and design conditions using a high-speed video camera are utilized to develop the models mentioned earlier. A dimensionless approach and nonlinear regression models are adopted to derive the empirical model. The Analysis of Variance (ANOVA) technique calculates F-number and their corresponding probabilities (P-values). The ANN model is developed with four input variables: particle size, static bed height, gap height, and gas superficial velocity. Multi-layer Perception model (MLP) with the Feedforward Back Propagation learning rule is employed to build the ANN model. A single hidden layer with nine neurons predicted a close solid circulation rate with minimal error over the empirical model compared to experimental data. Further, the empirical and ANN model's predicting capability is tested against literature data.

Multi-objective optimisation of mild steel embossed plate shear connector using artificial neural network-integrated genetic algorithm

A hybrid artificial neural network (ANN), integrated with a genetic algorithm (GA), was used for the multi-objective optimisation of mild steel embossed plate shear connector and the optimisation responses were examined by the push-out tests in accordance with BS 5400-5:2005 code. This innovative shear connection enables a greater frictional interlocking with the composite girder, allowing it to sustain a higher transverse load and gain a better bearing capacity at the same time retaining ductility. To obtain the ultimate load and relative slip data, the Taguchi experimental design was used, which took into account the three most influencing parameters: length, height, and thickness of the shear connector, which are the input elements for optimisation. The ANN-integrated GA approach is used for multi-objective optimisation, in which optimisation begins with modelling of the ANN and its output is optimised using GA. Five multi-layered (three) perceptron models that have different number of neurons in the hidden layer are developed using a feed-forward backpropagation learning algorithm. Logarithmic sigmoid (Logsig) and linear (Purelin) transfer functions from the Levenberg–Marquardt approach are adopted in this model. Using the best ANN model response values (3−11−2), multi-objective optimisation with GA was performed. The following optimum process parameters are established in this study: length, 40 mm; height, 124.97 mm; thickness of the connector, 12 mm for the maximum ultimate load of 431.991 kN and minimised relative slip of 2.056 mm, both with more than 95 % confidence level.

Statistical and intelligent reliability analysis of multi-layer ceramic capacitor for ground mobile applications using Taguchi’s approach

PurposeAlthough Multi-Layer Ceramic Capacitors (MLCC) are known for its better frequency performance and voltage handling capacity, but under various environmental conditions, its reliability becomes a challenging issue. In modern era of integration, the failure of one component can degrade or shutdown the whole electronic device. The lifetime estimation of MLCC can enhance the reuse capability and furthermore, reduces the e-waste, which is a global issue.Design/methodology/approachThe residual lifetime of MLCC is estimated using empirical method i.e. Military handbook MILHDBK2017F, statistical method i.e. regression analysis using Minitab18.1 as well as intelligent technique i.e. artificial neural networks (ANN) using MATLAB2017b. ANN Feed-Forward Back-Propagation learning with sigmoid transfer function [3–10–1–1] is considered using 73% of available data for training and 27% for testing and validation. The design of experiments is framed using Taguchi’s approach L16 orthogonal array.FindingsAfter exploring the lifetime of MLCC, using empirical, statistical and intelligent techniques, an error analysis is conducted, which shows that regression analysis has 97.05% accuracy and ANN has 94.07% accuracy.Originality/valueAn intelligent method is presented for condition monitoring and health prognostics of MLCC, which warns the user about its residual lifetime so that faulty component can be replaced in time.

Hot Deformation Characterization of Pure Aluminum Using artificial neural network (ANN) and Processing Map Considering Initial Grain Size

In this investigation, processing maps and artificial neural network (ANN) models were carried out to describe and predict the flow behavior of pure aluminum at various initial grain sizes in the hot working conditions. The elevated temperature flow behavior of AA1070 aluminum was done through isothermal hot compressive tests in a large range of initial grain size (IGS) (50–450 µm), strain rate (0.005–0.5 s−1) and temperature (623–773 K). Consequences showed that the flow stress can be remarkably influenced by the initial grain size at high temperatures. Based on the results, the ANN model trained with a feed-forward back-propagation learning algorithm which was prepared to describe the flow behavior of pure aluminum at the elevated temperatures. In which the initial grain size, strain, temperature and strain rate were taken as input data and true stress was used as target data. The results showed that the developed ANN model was a powerful method to predict the complex non-linear of the hot flow behavior of pure aluminum. The processing map was plotted and analyzed via the dynamic material model as “stable” and “unstable” regions were determined by observing the microstructure evolution. Based on this, The optimum ranges for temperature and strain rate were 623–773 K and 0.05 s−1 respectively, for fine-grained microstructure (lower than 50 µm) and were 650–720 K and 0.005–0.5 s−1 respectively, for coarse-grained microstructures (over than 50 µm).

Improving face recognition by artificial neural network using principal component analysis

The face-recognition system is among the most effective pattern recognition and image analysis techniques. This technique has met great attention from academic and industrial fields because of its extensive use in detecting the identity of individuals for monitoring systems, security and many other practical fields. In this paper, an effective method of face recognition was proposed. Ten person's faces images were selected from ORL dataset, for each person (42) image with total of (420) images as dataset. Features are extracted using principle component analysis PCA to reduce the dimensionality of the face images. Four models where created, the first one was trained using feed forward back propagation learning (FFBBL) with 40 features, the second was trained using 50 features with FFBBL, the third was trained using the same features but using Elman Neural Network. For each person (24) image used as training set for the neural networks, while the remaining images used as testing set. The results showed that the proposed method was effective and highly accurate. FFBBL give accuracy of (98.33,97.14) with (40, 50) features respectively, while Elman gives (98.33, 98.80) for with (40, 50) features respectively.

Modeling fractional polytropic gas spheres using artificial neural network

Lane–Emden differential equations describe different physical and astrophysical phenomena that include forms of stellar structure, isothermal gas spheres, gas spherical cloud thermal history, and thermionic currents. This paper presents a computational approach to solve the problems related to fractional Lane–Emden differential equations based on neural networks. Such a solution will help solve the fractional polytropic gas spheres problems which have different applications in physics, astrophysics, engineering, and several real-life issues. We used artificial neural network (ANN) framework in its feed-forward back-propagation learning scheme. The efficiency and accuracy of the presented algorithm are checked by testing it on four fractional Lane–Emden equations and compared with the exact solutions for the polytropic indices n = 0,1,5 and those of the series expansions for the polytropic index n = 3. The results we obtained prove that using the ANN method is feasible and accurate and may outperform other methods.

Polypropylene nanofibre behaviour modelling for uranium adsorption by forcespinning method: artificial neural network modelling

ABSTRACT In this study, a handmade device was developed to prepare polymeric polypropylene (PP) nanofibres by forcespinning method. The characterisations of the prepared adsorbent were investigated by Fourier transform infrared spectroscopy (FTIR) and Scanning electron microscopy (SEM). Then, the performance of the prepared nanofibres was investigated as an adsorbent for the removal of uranium (VI) ion from the aqueous solution. Experimental parameters such as contact time and pH were optimised. The maximum adsorption was obtained with the adsorbent weight of 0.03 g, contact time of 60 min, and pH of 4. This study showed that PP nanofibre is an efficient adsorbent for uranium removal from aqueous solution. Also, the uranium removal process was modelled by artificial neural network (ANN) for predicting capacity removal. In this method, feed-forward back-propagation learning with Levenberg-Marquardt (LM), adaptive learning rate back-propagation (GDX) and resilient back-propagation (RP) algorithms were used.

Using artificial neural network and multivariate calibration methods for simultaneous spectrophotometric analysis of Emtricitabine and Tenofovir alafenamide fumarate in pharmaceutical formulation of HIV drug.

Spectrophotometric analysis method based on artificial neural network (ANN), partial least squares regression (PLS) and principal component regression (PCR) models was proposed for the simultaneous determination of Emtricitabine (ETB) and Tenofovir alafenamide fumarate (TAF) in human immunodeficiency virus (HIV) drug. An artificial neural network consisting of two, five, and seven layers with 2,3,5,7, and 9 neurons was trained by applying a feed forward back-propagation learning. In this method, Levenberg-Marquardt (LM) and gradient descent with momentum and adaptive learning rate back propagation (GDX) algorithms were used. Statistical parameters indicated that the ability of LM was better than GDX algorithm. Also, root mean square error (RMSE) and recovery (%) of the PLS and PCR methods showed that PLS has worked better than PCR. The proposed models were compared to the high- performance liquid chromatography (HPLC) as a reference method. Furthermore, the obtained results of the one-way analysis of variance (ANOVA) test at the 95% confidence level represented that there was no significant difference between the proposed and reference methods.

Forecasting Gross Domestic Product per Capita Using Artificial Neural Networks with Non-Economical Parameters

Gross Domestic Product per capita is one of the most important indicators of social welfare. All countries try to increase their Gross Domestic Product per capita to contribute to their population’s happiness and well-being, as well as strengthen their nation’s standing in international relations. Economic growth is affected by economic parameters such as trade, import, and export. However, Gross Domestic Product may also be affected by non-economic factors. Therefore, for a country to increase its Gross Domestic Product per capita, it is important to employ the correct strategy. The aim of this study is to investigate the predictability of Gross Domestic Product per capita based on non-economic data by using artificial neural network with feed forward back-propagation learning algorithm. For this purpose, neural network models have been developed with different architectures. Education level, number of published academic paper per capita, number of researchers per employed, percentage of Research and Development expenditure in the Gross Domestic Product and number of patents per capita are used as input data in the models. The input data has been collected from variety of resources such as Organisation for Economic Co-operation and Development.A comparison between the model results and actual data give a high correlation coefficient (R2=0.96) and show that the model is able to predict the Gross Domestic Product per capita from non-economic parameters.

ANN-based optimization of a parabolic trough solar thermal power plant

Design and optimization of a solar power plant are very complex and require many calculations, data and time. From this point of view, artificial neural network (ANN) models are desired options to determine techno-economic performances of this kind of plants. Therefore, the objective of this study is to investigate the feed-forward back-propagation learning algorithm with three different variants; Levenberge Marguardt (LM), Scaled Conjugate Gradient (SCG), and Pola-Ribiere Conjugate Gradient (CGP), used in the ANN to find the best approach for prediction and techno-economic optimization of parabolic trough solar thermal power plant (PTSTPP) integrated with fuel backup system and thermal energy storage. The obtained statistical parameters showed that LM algorithm with 38 neurons in a single hidden layer looks as the best ANN model to predict the annual power generation (PGnet) and levelized cost of electricity (LCOE) of the presented PTSTPP. Moreover, the obtained weights from this topology were used in the LCOE analysis for determining the optimum system design. It is therefore available to get a minimum LCOE of 8.88Cent/kWh from the new optimized plant when the plant characteristics are; 34°C and 850W/m2 for both design ambient temperature and solar radiation, 23m for row spacing, 1.7 for solar multiple, and 2.5h number of hours for the storage system.

A comparative study on the phenomenological and artificial neural network models to predict hot deformation behavior of AlCuMgPb alloy

The high-temperature deformation behavior of AlCuMgPb alloy was investigated by the hot compression tests over a wide range of deformation temperature (623–773 K) and strain rate (0.005–0.5 s−1). Based on the experimental results, the phenomenological models consist of the Johnson-Cook, Arrhenius-type and Strain-compensation Arrhenius-type constitutive equations and an artificial neural network (ANN) model with a feed forward back propagation learning algorithm were developed for the prediction of the hot deformation behavior of the AlCuMgPb alloy. And then a comparative predictability of the phenomenological models and the trained ANN model were further evaluated in terms of the correlation coefficient(R), average absolute relative error (AARE), root mean square error (RMSE) and relative error. The results showed that the Arrhenius-type constitutive equation could predict the flow stress accurately except under the strain rate of 0.005 s−1. The Strain-compensated Arrhenius-type constitutive equation could represent the elevated temperature flow behavior more accurately than the other investigated phenomenological models consist of Johnson-Cook and Arrhenius-type constitutive equations in the entire processing domain. The results indicated that the trained ANN model is more efficient and accurate in predicting the hot deformation behavior in AlCuMgPb alloy than the investigated phenomenological constitutive equations and offers no physical insight.

ANN-based evaluation of wind power generation: A case study in Kutahya, Turkey

Wind energy is one of the most significant and rapidly developing renewable energy sources in the world and it provides a clean energy resource, which is a promising alternative in the short term in Turkey. The wind energy potential in various parts of Turkey is becoming economical due to reductions in wind turbine costs, and in fossil fuel atmospheric pollution. This paper is to present, in brief, wind potential in Turkey and to perform an investigation on the wind energy potential of the Kutahya region. A wind measurement station was established at Dumlupinar University Main Campus in order to figure out the wind energy potential in the province. This study analyses the electricity generation capacity of the Kutahya region, Turkey, which uses the wind power system. In the study, the wind data collected from wind measurement stations between July 2001 and June 2004 (36 months) were evaluated to determine the energy potential of the region. Using this energy potential value, the power generation capacity of Kutahya was investigated for 17 different wind turbines. In this analysis, an ANN-based model and Weibull and Rayleigh distribution models were used to determine the power generation. In the ANN model, different feed-forward back propagation learning algorithms, namely Pola-Ribiere Conjugate Gradient, Levenberg–Marquardt and Scaled Conjugate Gradient were applied. The best appropriate model was determined as Levenberg–Marquardt with 15 neurons in a single hidden layer. Using the best ANN topology, it was determined that all the turbines were profitable except turbine type 1. The system with the turbine type 3 was decisively the most profitable case as determined at the end of the study according to Net Present Value concept.

Artificial Neural Network Approach to Predict the Abrasive Wear of AA2024-B<sub>4</sub>C Composites

A neural network (ANN) model was developed to predict the abrasive wear behavior of AA2024 aluminum alloy matrix composites reinforced with B4C particles. Al2024-B4C powder mixtures with various reinforcement volume fractions (3-10%) and particle sizes (29µm and 71 µm) were prepared and Al2024-B4C composites were produced by stir-casting technique. The model was based on three layer neural network with feed forward back propagation learning algorithm. A sigmoid transfer function was developed and found to be suitable for analyzing the abrasive wear behavior of composites with the least error. The training data are collected by the experimental setup in the laboratory. The trained model was used to study the effect of ceramic particle size and volume fraction on the abrasive wear of Al2024-B4C composites. By comparing the predicted values with the experimental data, it was demonstrated that the well-trained feed forward back propagation ANN model is a powerful tool for prediction of abrasive wear behavior of Al2024-B4C composites.

Modelling of As3+ adsorption from aqueous solution using Azadirachta indica by artificial neural network

An intensive study has been made on the removal efficiency of As3+ from aqueous solution by zinc oxide nanoparticle entrenched on activated silica (ZnO-NPs-AS) using aqueous leaf extract of Azadirachta indica. ZnO-NPs-AS is characterized using SEM-EDX, FT-IR and XRD. The effect of various parameters such as initial concentration of As3+, adsorbent dosage, contact time, pH and agitation is studied systematically. The maximum adsorption of As3+ is found to be 98.31% at pH 5, equilibrium time of 50 min using adsorbent of 3 g/L and initial concentration of 0.06 mg/L at agitation speed of 250 rpm. Adsorption parameters for the Langmuir, Freundlich, Tempkin and BET isotherms were determined. The equilibrium data were best described by Langmuir isotherm model and fits quite well with the experimental data with good correlation coefficient of 0.974. The results of intraparticle diffusion model suggested that intraparticle diffusion was not the rate-controlling process. From the values, it is accomplished that the maximum adsorption corresponds to a saturated monolayer of As3+ molecules on the adsorbent surface with constant energy. The data were analysed using kinetics models akin to pseudo-first and second order. All the findings presented in this study suggested following pseudo-second-order equation for the adsorption of As3+ on to Zno-NPs-AS. The data collected from laboratory-scale experimental set up are used to train a feed forward back propagation learning algorithm having 5-20-1 three-layered architecture. The model uses tangent sigmoid transfer function at input to hidden layer whereas a linear purelin function is used at output layer. The data are divided into training (70%), testing (15%), and validation (15%) sets. The network is found to be working satisfactorily as absolute mean square percentage error of 0.0014 is obtained during training phase. Comparison between the model results and experimental data gives a high degree of correlation (R2 = 0.986) indicating that the matlab nntool 2010a neural network model is able to predict the sorption efficiency with reasonable accuracy.

A neuro-fuzzy decision support model for therapy of heart failure

Most medical decision support systems focus on diagnosis with little emphasis on therapy to the effect that though an accurate diagnosis is undertaken patients still have problems of drug misuse as a result of inaccurate therapy. The purpose of this paper is to design an assistive model for therapy of heart failure using artificial neural networks (ANN). Artificial neural networks have been found to be a very veritable tool in learning from existing datasets and based on the results, can perform accurate prediction on the data it has not encountered before through generalisation. It was based on this that 134 datasets on heart failure were collected from three hospitals and trained in a feed forward back propagation learning neural networks. This was further refined through the fuzzy system and some decision support filters. Results obtained from the neuro-fuzzy system indicate that the model has the ability to refine and enhance the physician’s ability to prescribe an appropriate therapy based on the diagnosis. This study is one of the few attempts at utilising soft computing technology in the diagnosis and therapy of cardiovascular diseases. The authors had previously developed neuro-fuzzy models for diagnosis of heart failure.

CONDITION MONITORING OF OFF-HIGHWAY TRUCK TIRES AT SUNGUN COPPER MINE USING NEURAL NETWORKS / MONITOROWANIE STANU TECHNICZNEGO OPON W CIĘŻKICH POJAZDACH TERENOWYCH EKSPLOATOWANYCH W KOPALNI MIEDZI SUNGUN, PRZY UŻYCIU SIECI NEURONOWYCH

Abstract Maintenance cost of the equipment is one of the most important portions of the operating expenditures in mines; therefore, any change in the equipment productivity can lead to major changes in the unit cost of the production. This clearly shows the importance and necessity of using novel maintenance methods instead of traditional approaches, in order to reach the minimum sudden occurrence of the equipment failure. For instance, the tires are costly components in maintenance which should be regularly inspected and replaced among different axles. The paper investigates the current condition of equipment tires at Sungun Copper Mine and uses neural networks to estimate the wear of the tires. The Input parameters of the network composed of initial tread depth, time of inspection and consumed tread depth by the time of inspection. The output of the network is considered as the residual service time ratio of the tires. The network trained by the feed-forward back propagation learning algorithm. Results revealed a good coincidence between the real and estimated values as 96.6% of correlation coefficient. Hence, better decisions could be made about the tires to reduce the sudden failures and equipment breakdowns.

A New Approach for Estimating Free Point in Fishing of Stuck Pipe using Artificial Neural Network

Stuck pipe is a common problem in drilling industry. It accounts for the major rig time losses each year in the petroleum industry. In cases where common solutions such as pulling up and pushing down, rotating, jarring, and changing flow rate don’t work, then backing off is the last resort. To have a successful back off operation, estimating location of the free point is vital. In this study, an attempt is made to estimate the free point in stuck pipe cases using the drilling data and artificial neural network approach. For this purpose, drilling data such as mud properties, pipe rotation, rate of penetration, and some other parameters are required. In this study, artificial neural networks (ANNs) model using field data from more than 40 wells was employed, and results were compared to field results. ANN model was constructed with a supervised learning algorithm and feed forward back propagation learning rule is used for training the network. The statistical error analysis results obtained by the models and acceptable values for correlation coefficient indicate that ANN model is successful in free point prediction.