Three-layer Feedforward Backpropagation Neural Network Research Articles

Artificial neural network-based sensitivity analysis and experimental investigation of liquid–solid fluidization technique for low-grade coal upgradation

Liquid-solid fluidization technique is being applied where low-grade coal or minerals enrichment is mostly density-based. Static and dynamic behavior of particles in a fluid medium has been extensively investigated over the years because of its dynamic applications across various industries. In this work, bed characterization studies and experiments have been conducted to study coal washing ability of the liquid-solid fluidized bed separator. Results have been recorded in terms of ash rejection%, combustible recovery% and separation efficiency%. Minimum fluidization velocity and pressure drop values have been predicted using existing theoretical correlations and compared with the experimental values. A three-layered (4:5:3) feedforward back-propagation (FFBP) neural network model was developed using Levenberg-Marquardt algorithm, LOGSIG and MSE as training, transfer and performance functions respectively. Garson’s algorithm and connection weight approach have been employed for sensitivity analysis to interpret the neural network results physically. Coefficients of correlation, all R (including training, validation & testing datasets) obtained for outputs ash rejection (R = 0.9960), combustible recovery (R = 0.9952) and separation efficiency (R = 0.9944) suggest that predicted values are in agreement with the experimental values and the developed model is a good fit.

Optimization of hydrogen production via toluene steam reforming over Ni–Co supported modified-activated carbon using ANN coupled GA and RSM

Hydrogen (H2) is a clean fuel that can be produced from various resources including biomass. Optimization of H2 production from catalytic steam reforming of toluene using response surface methodology (RSM) and artificial neural network coupled genetic algorithm (ANN-GA) models has been investigated. In RSM model, the central composite design (CCD) is employed in the experimental design. The CCD conditions are temperature (500–900 °C), feed flow rate (0.006–0.034 ml/min), catalyst weight (0.1–0.5 g) and steam-to-carbon molar ratio (1–9). ANN model employs a three-layered feed-forward backpropagation neural network in conjugation with the tangent sigmoid (tansig) and linear (purelin) as the transfer functions and Levenberg-Marquardt training algorithm. Best network structure of 4-14-1 is developed and utilized in the GA optimization for determining the optimum conditions. An optimum H2 yield of 92.6% and 81.4% with 1.19% and 6.02% prediction error are obtained from ANN-GA and RSM models, respectively. The predictive capabilities of the two models are evaluated by statistical parameters, including the coefficient of determination (R2) and root mean square error (RMSE). Higher R2 and lower RSME values are reported for ANN-GA model (R2 = 0.95, RMSE = 4.09) demonstrating the superiority of ANN-GA in determining the nonlinear behavior compared to RSM model (R2 = 0.87, RMSE = 6.92). These results infer that ANN-GA is a more reliable and robust predictive steam reforming modelling tool for H2 production optimization compared to RSM model.

RETRACTED:An assessment of software defined networking approach in surveillance using sparse optimization algorithm

Abstract Unmanned aerial vehicles (UAV) can be used as basic elements of the sensor network or an upgrade of existing network that are built with static wireless sensor nodes. Wireless Networks (WN) are utilized across in surveillance in all domains like natural disasters, agriculture, water, forest, military, buildings, health monitoring, disaster relief & emergency management, area and industrial surveillance, due to its wider applicability. Software Defined Networking (SDN) provide a hopeful resolution in bendy supervision WSNs by allowing the separation of the control logic from the sensor nodes/actuators. The advantage with this SDN-based supervision in structure of WSNs is that it enables centralized control of the entire WSN making it simpler to deploy network-wide management protocols and applications on demand. Synthetic Aperture Radar (SAR) images are difficult to analyze due to the presence of speckle noise. Speckle noise must be filtered out before applying to other image processing applications. Three Layered Feed Forward Back Propagation Neural Network (TLFFBPNN) has been proposed to suppress the speckle noise. GLCM properties have been extracted and Back propagation training algorithm is used to train the neural network.

A quantitative assessment of speckle noise reduction in SAR images using TLFFBP neural network

Synthetic aperture radar (SAR) images are difficult to analyze due to the presence of speckle noise. Speckle noise must be filtered out before applying to other image processing applications. Three-layered feed forward back propagation neural network (TLFFBPNN) has been proposed to suppress the speckle noise. Gray-level co-occurrence matrix properties have been extracted, and back propagation training algorithm is used to train the neural network. The performance metrics such as peak signal to noise (PSNR), structural similarity index matrix (SSIM), edge preservation index (EPI), equivalent number of looks (ENL), and speckle suppression index (SSI) have been evaluated to find the efficiency of TLFFBPNN and compared with four recently developed de-speckling techniques. The exploratory outcomes show that the TLFFBPNN method has better de-speckling execution with great edge preservation. The comparative outcome reveals that the proposed TLFFBPNN de-speckled method outperformed in terms of PSNR of 0.98%, SSIM of 1.0%, SSI of 2.0%, EPI of 0.84%, and ENL of 0.5% when compared with the Wiener Filter Sparse Optimization in Contourlet transform domain de-speckling method.

Investigation the insights between health expenditures and air quality

In this study, models have been developed for predicting health expenditures of Turkey associated with greenhouse gas (GHG) emission levels using 27-year dataset between the years 1990 and 2016. The annual GHG emissions data consisting of carbon dioxide, methane, nitrous oxide, and fluorinated gases have been used as inputs. In order to increase the accuracy and reliability, three different models namely, the Bayesian optimisation-based support vector regression (BO-SVR), three-layered feed-forward back-propagation neural network (BPNN), and multivariate linear regression (MLR) models were employed. The coefficient determination (R2) for the BO-SVR, BPNN and MLR models were determined as 0.9893, 0.9796, and 0.9766 in the training phase and 0.9795, 0.9629, and 0.9529 in the testing phase, respectively. The results showed that the BO-SVR model is found to be superior for the estimation of Turkey's health expenditures.

Evaluation of construction projects based on the safe work behavior of co-employees through a neural network model

The safety climate of an organization influences employees’ safe work behavior. It is essential for project management to be aware of any unsafe work behavior of their workers on a project and to take any necessary remedial measures to reduce the likelihood of accidents. This study aims to predict and evaluate the work behavior of employees on construction projects using the constructs of the safety climate. Because of the prevailing nonlinear and complex interrelationships among these determinants, an artificial neural network (ANN) is employed to develop the model. Ten important safety climate features are used as inputs, and co-employees’ safe work behavior is taken as the output. A total of 240 responses from different construction projects across India were collected through a questionnaire in a two-stage process to train, test, and validate the model. A three-layer feedforward backpropagation neural network architecture 10-17-1 was found to be a suitable model. Through this model, outlier projects have been determined based on the project efficiency score and the Anderson-Darling (AD) statistics. The study advocates the practice of safe work behavior as reported by co-employees, rather than workers’ own reported safe work behavior, as the output variable. The significant constructs of safe work behavior are presented based on the sensitivity analysis. This model will be helpful to evaluate, predict, and monitor the safety performance of construction projects.

Supervised artificial neural network-based method for conversion of solar radiation data (case study: Algeria)

In this study, a backpropagation artificial neural network (BP-ANN) model is used as an alternative approach to predict solar radiation on tilted surfaces (SRT) using a number of variables involved in physical process. These variables are namely the latitude of the site, mean temperature and relative humidity, Linke turbidity factor and Angstrom coefficient, extraterrestrial solar radiation, solar radiation data measured on horizontal surfaces (SRH), and solar zenith angle. Experimental solar radiation data from 13 stations spread all over Algeria around the year (2004) were used for training/validation and testing the artificial neural networks (ANNs), and one station was used to make the interpolation of the designed ANN. The ANN model was trained, validated, and tested using 60, 20, and 20 % of all data, respectively. The configuration 8-35-1 (8 inputs, 35 hidden, and 1 output neurons) presented an excellent agreement between the prediction and the experimental data during the test stage with determination coefficient of 0.99 and root meat squared error of 5.75 Wh/m2, considering a three-layer feedforward backpropagation neural network with Levenberg–Marquardt training algorithm, a hyperbolic tangent sigmoid and linear transfer function at the hidden and the output layer, respectively. This novel model could be used by researchers or scientists to design high-efficiency solar devices that are usually tilted at an optimum angle to increase the solar incident on the surface.

Neural Network Approach for Safety Climate Prediction

AbstractSafety climate is a snapshot of the safety culture of any organization and it is dynamic in nature. It reflects the employees’ perceptions and attitude towards the existing safety practices in construction projects. The main goal of this paper is to evaluate and differentiate construction projects based on their safety climate and to determine the significant constructs of a safety climate. This paper develops a model to predict safety climate in a construction project based on an artificial neural network (ANN). The important constructs for a safety climate have been determined through literature review. The constructs of safety climate are used as inputs and the safety climate of the project is used as an output for the ANN algorithm. This study collected a total of 250 responses through a questionnaire survey across the country. A three-layer feed-forward back-propagation neural network (10-18-1) has been utilized for the analysis. The developed model predicts the safety climate of a constructi...

Neural Network Model for the Prediction of Safe Work Behavior in Construction Projects

AbstractA model has been developed employing an artificial neural network (ANN) to predict the safe work behavior of employees using 10 safety climate constructs determined through literature review. The model utilizes safety climate constructs (determinants) as inputs and safe work behavior as an output. Two hundred twenty-two responses from several construction projects across India were collected through a questionnaire survey. A three-layer feed-forward back-propagation neural network (10-11-1) was appropriate in building this model which has been trained, validated, and tested with sufficient data sets. The model predicts the safe work behavior of employees reasonably well. In addition, a sensitivity analysis was carried out to study the impact of each construct on the safe work behavior of employees. As a result, safety climate constructs like supervisory environment, work pressure, employees’ involvement, personal appreciation of risk, and supportive environment were significantly associated with t...

Density prediction of liquid alkali metals and their mixtures using an artificial neural network method over the whole liquid range

In this study, the application of artificial neural network (ANN) method in predicting the density of alkali metals and their mixtures is investigated. A total number of 595 different data points of these compounds were used to train, validate and test the model. A typical three-layer feedforward backpropagation neural network has been trained by the Levenberg Marquardt algorithm. The tansig-tansig transfer functions with 15 neurons in the hidden layer makes the least error, so a network with (8-15-1) structure was used to design the ANN model. The average relative deviations for train, validation, and test sets are 0.1029, 0.1396, and 0.1002, respectively. A comparison between our results and those obtained from some previous works shows that this work, as an excellent alternative, can provide a simple procedure to predict the density of these compounds in a better accord with experimental data up to high temperature, high pressure (HTHP) conditions.

An intelligent approach to evaluate drilling performance

In this paper, an attempt has been made to predict the rate of penetration (ROP) of rocks by incorporating thrust, revolutions per minute (rpm), flushing media and compressive strength of rocks using artificial neural network (ANN) technique. A three-layer feed-forward back-propagation neural network with 4-7-1 architecture was trained using 472 experimental data sets of sandstone, limestone, rock phosphate, dolomite, marble and quartz-chlorite-schist rocks. A total of 146 new data sets were used for the testing and comparison of the ROP by ANN. Multivariate regression analysis (MVRA) has also been done with same data sets of ANN. ANN and MVRA results were compared based on coefficient of determination (CoD) and mean absolute error (MAE) between experimental and predicted values of ROP. The coefficient of determination by ANN was 0.985, while coefficient of determination was 0.629 for rate of penetration. The mean absolute error (MAE) for rate of penetration by ANN was 0.3547, whereas MAE by MVRA was 1.7499.

Application of soft computing to predict blast-induced ground vibration

In this study, an attempt has been made to evaluate and predict the blast-induced ground vibration by incorporating explosive charge per delay and distance from the blast face to the monitoring point using artificial neural network (ANN) technique. A three-layer feed-forward back-propagation neural network with 2-5-1 architecture was trained and tested using 130 experimental and monitored blast records from the surface coal mines of Singareni Collieries Company Limited, Kothagudem, Andhra Pradesh, India. Twenty new blast data sets were used for the validation and comparison of the peak particle velocity (PPV) by ANN and conventional vibration predictors. Results were compared based on coefficient of determination and mean absolute error between monitored and predicted values of PPV.

Prediction and controlling of flyrock in blasting operation using artificial neural network

Flyrock is one of the most hazardous events in blasting operation of surface mines. There are several empirical methods to predict flyrock. Low performance of such models is due to complexity of flyrock analysis. Existence of various effective parameters and their unknown relationships are the main reasons for inaccuracy of the empirical models. Presently, application of new approaches such as artificial intelligence is highly recommended. In this paper, an attempt has been made to predict and control flyrock in blasting operation of Sangan iron mine, Iran incorporating rock properties and blast design parameters using artificial neural network (ANN) method. A three-layer feedforward back-propagation neural network having 13 hidden neurons with nine input parameters and one output parameter were trained using 192 experimental blast datasets. It was also observed that in ascending order, blastability index, charge per delay, hole diameter, stemming length, powder factor are the most effective parameters on the flyrock. Reducing charge per delay caused significant reduction in the flyrock from 165 to 25 m in the Sangan iron mine.

Prediction of surface roughness profiles for milled surfaces using an artificial neural network and fractal geometry approach

Abstract Artificial neural networks (ANNs) models were developed for the analysis and prediction of the relationship between the cutting conditions and the corresponding fractal parameters of machined surfaces in face milling operation. These models can help manufacturers to determine the appropriate cutting conditions, in order to achieve specific surface roughness profile geometry, and hence achieve the desired tribological performance (e.g. friction and wear) between the contacting surfaces. The input parameters of the “ANNs” models are the cutting parameters: rotational speed, feed, depth of cut, pre-tool flank wear and vibration level. The output parameters of the model are the corresponding calculated fractal parameters: fractal dimension “D” and vertical scaling parameter “G”. The model consists of three-layered feed-forward back-propagation neural network. ANNs models were utilized successfully for modeling and predicting the fractal parameters “D” and “G” in face milling operations. Moreover, W–M fractal function was integrated with the developed ANNs models in order to generate an artificially fractal predicted profiles at different cutting conditions. The predicted profiles were found statistically similar to the actual measured profiles of test specimens.

Evaluation of blast-induced ground vibration predictors

The present paper mainly deals with the prediction of blast-induced ground vibration level at a Magnesite Mine in tecto-dynamically vulnerable hilly terrain in Himalayan region in India. The ground vibration was monitored to calculate the safe charge of explosive to avoid the continuous complaints from the nearby villagers. The safe charge of explosive and peak particle velocity (PPV) were recorded for 75 blast events (150 blast data sets) at various distances. These data sets were used and analysed by the widely used vibration predictors. From the four predictors, vibration levels were calculated and compared with new monitored 20 blast data sets. Again, the same data sets were used to validate and test the three-layer feed-forward back-propagation neural network to predict the PPV values. The same 20 data sets were used to compare the results by the artificial neural network (ANN). Among all the predictors, a very poor correlation was found, whereas ANN provides very near prediction with high degree of correlation.

Multi-step predictive control with TDBP method for pneumatic position servo system

This paper presents a new multi-step predictive controller based on neural networks and researches the adaptability of the predictive controller for a pneumatic position servo system which has some typical characteristics of non-linearity and time-varying. A diagonal recurrent neural network (DRNN) is used to predict the system output of the multi-step ahead directly. According to the intrinsic defects of a back-propagation (BP) algorithm that cannot update network weights incrementally, a new hybrid learning algorithm combining the temporal differences (TD) method with the BP algorithm to train the DRNN is put forward. A three-layer feedforward BP neural network is used as a non-linear rolling optimal controller to realize the optimization of control input of the next step according to a single-value predictive control algorithm to simplify computation. Simulation and experimental results indicate that the proposed predictive controller is suitable for real-time control of a pneumatic position servo system because of its characteristics of a simple algorithm, fast calculation of the control input and good tracking effects.

Modelling the correlation between cutting and process parameters in high-speed machining of Inconel 718 alloy using an artificial neural network

An artificial neural network (ANN) model was developed for the analysis and prediction of the relationship between cutting and process parameters during high-speed turning of nickel-based, Inconel 718, alloy. The input parameters of the ANN model are the cutting parameters: speed, feed rate, depth of cut, cutting time, and coolant pressure. The output parameters of the model are seven process parameters measured during the machining trials, namely tangential force (cutting force, F z ), axial force (feed force, F x ), spindle motor power consumption, machined surface roughness, average flank wear (VB), maximum flank wear (VB max) and nose wear (VC). The model consists of a three-layered feedforward backpropagation neural network. The network is trained with pairs of inputs/outputs datasets generated when machining Inconel 718 alloy with triple (TiCN/Al 2O 3/TiN) PVD-coated carbide (K 10) inserts with ISO designation CNMG 120412. A very good performance of the neural network, in terms of agreement with experimental data, was achieved. The model can be used for the analysis and prediction of the complex relationship between cutting conditions and the process parameters in metal-cutting operations and for the optimisation of the cutting process for efficient and economic production.

Modified geometry of ring-wedge detector for sampling Fourier transform of fingerprints for classification using neural networks

Sampling of the Fourier transforms of fingerprints is studied with neural networks to detect regions useful for their classification. Ring-wedge detector (RWD) is modified and simulated to sample such regions. The output of the detector is propagated through a three-layer feedforward-backpropagation neural network for checking the classification performance. Modified detector's performance is also compared with that of RWD. It has been found that fingerprints scanned at 500 dpi resolution and cropped to a size of 200×200 contain useful information for their classification in a band of width 20 pixels with inner radius approx. 60 pixels.

Rolling learning-prediction of product formation in bioprocesses

A rolling learning-prediction approach based on neural networks is proposed with the aim of on-line prediction of the product formation. Commercial-scale penicillin cultivations were taken as an example to test the product predictor. Raw data are pretreated in such a way that each input vector of the neural network consists of a series of time-discretised values on a specified transient of process variables. The output vector is composed of the amount of product at the next one and two prediction steps. The process variables involved in the predictor include carbon dioxide and product formation as well as oxygen, precursor and substrate consumption. Accumulated rather than instant values of these variables were used. A simple three-layer feedforward backpropagation neural network with a tangent sigmoidal transfer function in the hidden nodes and a linear one in the output nodes was used as the main frame of the product predictor. The proposed prediction procedure is called rolling learning-prediction because the training database is updated after each sampling interval and the learning-prediction is repeated thereafter. The robustness of the predictor was illustrated by its adaptive ability to widely scattered data sets and extra added noises. The testing results indicated that a prediction accuracy of 2–5% could be generally expected in the later phase of cultivation and reliable prediction time spans may take more than 10% of the cultivation period for penicillin production. An intrinsic problem of using neural networks—occasional trap of the network in bad local minima—is automatically detected and remedied. In addition, it was illustrated by example that the prediction error signal may be potentially used to detect extraordinary charges caused, for example, by contamination. Problems associated with the industrial application of the predictor are discussed.

Artificial neural networks in computer-assisted classification of heart sounds in patients with porcine bioprosthetic valves

The paper describes the design, training and testing of a three-layer feed-forward back-propagation neural network for the classification of bioprosthetic valve closure sounds. Forty-seven patients with a porcine bioprosthetic valve inserted in the aortic position were involved in the study. Twenty-four of them had a normal bioprosthetic valve, and the other 23 had a degenerated valve. Five features extracted from the Fourier spectra and 12 linear predictive coding (LPC) coefficients of the sounds were used separately as the input of two neural-network classifiers. The performance of the classifiers was tested using the leave-one-out method. Results show that correct classifications were 85 per cent using the spectral features, and 89 per cent using the LPC coefficients. The study confirms the potential of artificial networks for the classification of bioprosthetic valve closure sounds. Clinical use of this method, however, still requires further investigation.