Algorithm-based Artificial Neural Network Research Articles

Determining post-fire residual compressive strength of reinforced concrete shear walls using the BAT algorithm

The effect of fire on the reinforced concrete structures' lateral load-bearing systems is not well established in literature and design codes. Since in case of an earthquake this influence can be crucial, in this study we aimed to predict post-fire residual compressive strength of concrete in shear walls using artificial neural network (ANN) models. The network parameters were fine-tuned using the bat optimization metaheuristic algorithm. The accuracy of the BAT-based ANN models was validated by comparing their predictions with the particle swarm optimization (PSO) algorithm-based ANNs and multiple linear regression models. The results for BAT-trained networks revealed a mean squared error (MSE) of 4.881 MPa, and prediction-target correlation of 0.987 on testing data which are more accurate than their PSO-trained counterparts.

Promising MPPT Methods Combining Metaheuristic, Fuzzy-Logic and ANN Techniques for Grid-Connected Photovoltaic.

This paper addresses the improvement of tracking of the maximum power point upon the variations of the environmental conditions and hence improving photovoltaic efficiency. Rather than the traditional methods of maximum power point tracking, artificial intelligence is utilized to design a high-performance maximum power point tracking control system. In this paper, two artificial intelligence-based maximum power point tracking systems are proposed for grid-connected photovoltaic units. The first design is based on an optimized fuzzy logic control using genetic algorithm and particle swarm optimization for the maximum power point tracking system. In turn, the second design depends on the genetic algorithm-based artificial neural network. Each of the two artificial intelligence-based systems has its privileged response according to the solar radiation and temperature levels. Then, a novel combination of the two designs is introduced to maximize the efficiency of the maximum power point tracking system. The novelty of this paper is to employ the metaheuristic optimization technique with the well-known artificial intelligence techniques to provide a better tracking system to be used to harvest the maximum possible power from photovoltaic (PV) arrays. To affirm the efficiency of the proposed tracking systems, their simulation results are compared with some conventional tracking methods from the literature under different conditions. The findings emphasize their superiority in terms of tracking speed and output DC power, which also improve photovoltaic system efficiency.

Fourier Transform and Autoregressive HRV Features in Prediction and Classification of Breast Cancer

ABSTRACT The decision making in diagnosis of breast cancer (BC) at the earliest is the necessity to decrease the mortality rate. The 5-minute electrocardiogram of 114 BC subjects and 13 age-matched healthy controls were recorded and spectral features of heart rate variability (HRV) were calculated. Fast Fourier transform (FFT) and autoregressive (AR) spectral methods were compared to analyse the frequency domain of HRV. Lavenberg–Marquardt algorithm-based artificial neural network (ANN) and support vector machine (SVM) classified all the spectral measures with maximum accuracy of 54.2% and 100%, respectively. ANOVA with Tukey’s HSD Posthoc test has also been employed to evaluate the significant variations in different parameters due to BC with respect to control subjects with the help of statistical analysis. The FFT and AR results were found almost similar. Clinicians can achieve an insight of the severity of the disease as per the findings of spectral measures as per Eastern Cooperative Oncology Group Performance Status and improvise the quality of their patients.

Levenberg–Marquardt Backpropagation for Numerical Treatment of Micropolar Flow in a Porous Channel with Mass Injection

In this research work, an effective Levenberg–Marquardt algorithm‐based artificial neural network (LMA‐BANN) model is presented to find an accurate series solution for micropolar flow in a porous channel with mass injection (MPFPCMI). The LMA is one of the fastest backpropagation methods used for solving least‐squares of nonlinear problems. We create a dataset to train, test, and validate the LMA‐BANN model regarding the solution obtained by optimal homotopy asymptotic (OHA) method. The proposed model is evaluated by conducting experiments on a dataset acquired from the OHA method. The experimental results are obtained by using mean square error (MSE) and absolute error (AE) metric functions. The learning process of the adjustable parameters is conducted with efficacy of the LMA‐BANN model. The performance of the developed LMA‐BANN for the modelled problem is confirmed by achieving the best promise numerical results of performance in the range of E‐05 to E‐08 and also assessed by error histogram plot (EHP) and regression plot (RP) measures.

Smart Instrumentation Based Electromechanical Influences on Material Properties for Ultrasonically Joined Conductive Wires—A Study

This paper presents the investigations to analyze the microstructural behavior, mechanical properties of ultrasonically welded dissimilar Al–Cu wires deployed for electrical appliances and automobile parts. The study emphasizes on the metallurgical characterization to access the uniformity during the material diffusion and the corresponding changes in their mechanical properties during the high vibrational heat. The experimental trials to join the wire materials are performed based on L9 Taguchi Design of Experiments. The Levenberg-Marquardt algorithm based Artificial Neural Network Optimization is implemented to predict and validate the strength and joint resistance of the welded wires to obtain optimal parametric window. Experimental validations are carried out by performing destructive testing. Subsequently, the welded samples are also examined through microscopic studies to investigate the surface morphologies. It is observed that there is a good agreement between the predicted and experimental results yielding an optimal parametric window. The results also indicate that Ultrasonic welding of dissimilar Al–Cu joints is more feasible for their application in electrical and automobile applications.

Gaussian Process Regression Technique to Estimate the Pile Bearing Capacity

A commonly-encountered problem in foundation design is the reliable prediction of the pile bearing capacity (PBC). This study is planned to propose a feasible soft computing technique in this field i.e.; the Gaussian process regression (GPR) for the PBC estimation. The established database includes 296 number of dynamic pile load test in the field where the most influential factors on the PBC were selected as input variables. Several GPR models were designed and built. These models were assessed using three performance indices namely value account for (VAF), coefficient of determination (R2) and system error. To have a comparison, a genetic algorithm-based artificial neural network (GA-based ANN) model was also employed. It was found that the GPR-based model with VAF value of 86.41%, R2 of 0.84 and system error of 0.2006 is capable enough to predict the PBC and outperforms the GA-based ANN model. The results showed that the GPR can be utilized as a practical tool for the PBC estimation.

Experimental and theoretical evaluation of thermophysical properties for moist air within solar still by using different algorithms of artificial neural network

In this research article, an attempt has been made to predict the thermophysical properties of moist air in a solar still cavity with the help of Artificial Neural Network (ANN) modelling. Six training algorithms have been used to train, test, and validate the ANN model (viz. OSS, CGP, CGF, RP, SCG, and LM). Water and inner glass cover temperatures were selected as the input parameters whereas, the model output are: thermal conductivity, partial vapour pressure at water & glass surface, thermal conductivity, volumetric expansivity, specific heat, latent heat of vaporization, and dynamic viscosity. The findings have revealed that the proposed ANN model can be used to predict the thermophysical properties of moist air with excellent accuracy. The results of ANN model were tested against the well-established relations of Tsilingiris. Out of all the training algorithms used LM was found to be the best in all the stages of ANN modelling, as the results are well within an accuracy level of more than 95%. Hence, the developed LM algorithm-based ANN model is one of the most suitable algorithm for the prediction of the thermophysical properties of moist air.

A parametric study of 3D printed polymer gears

The selection of printing parameters for 3D printing can dramatically affect the dynamic performance of components such as polymer spur gears. In this paper, the performance of 3D printed gears has been optimised using a machine learning process. A genetic algorithm (GA)–based artificial neural network (ANN) multi-parameter regression model was created. There were four print parameters considered in 3D printing process, i.e. printing temperature, printing speed, printing bed temperature and infill percentage. The parameter setting was generated by the Sobol sequence. Moreover, sensitivity analysis was carried out in this paper, and leave-one cross validation was applied to the genetic algorithm-based ANN which showed a relatively accurate performance in predictions and performance optimisation of 3D printed gears. Wear performance of 3D printed gears increased by 3 times after optimised parameter setting was applied during their manufacture.

Modelling conventional and solar earth still by using the LM algorithm-based artificial neural network

ABSTRACT An experimental and theoretical study has been carried out for the performance evaluation of conventional solar still (CSS) and modified solar still integrated with earth (MSSIE). The Levenberg–Marquardt (LM) algorithm-based artificial neural network (ANN) model has been developed for the prediction of distillate output from solar stills. The solar radiation, water, glass, and atmospheric temperatures were selected as an input parameter of ANN, whereas distillate yield is the output variable. The model has been trained, tested, and validated with the experimental data obtained from CSS and MSSIE at 95% confidence level. The trained and validated model has been used for the prediction of the distillate output of CSS and MSSIE by using another set of input parameters. A maximum deviation of 3.1% and 4.6% has been observed between the obtained and experimental results for CSS and MSSIE, respectively. Results depict that the proposed ANN model is precisely accurate for the forecasting of distillate output from the distiller units.

A robust neural network model for monitoring online voltage stability

Incessant assessment of voltage stability is a lively aspect to safeguard the electrical power system (EPS) operation. The conservative methods for online assessment in terms of stability check for voltage are extremely time engrossing and also absurd for supervising any application online. In line to this, a Salp Swarm Algorithm-based artificial neural network (SSA–ANN) model is opted for online monitoring of voltage stability in this manuscript. Artificial neural network is an influential and promising predictive tool. To upsurge the efficacy and accuracy and minimalize the training time. SSA is used for tuning the metaparameters such as the activation functions and number of nodes along with the learning rate. The method anticipated for the aforesaid problem utilizes the magnitude of voltage and its corresponding phase angle which are attained from the PMU as the inputs to the neural network model and the output is the voltage stability margin index (VSMI). The efficiency of the proposed model is verified with ICA–ANN and GA–ANN underrobust test cases and compared with the same data set to attest its preeminence.

Recognition of Odia Handwritten Digits using Gradient based Feature Extraction Method and Clonal Selection Algorithm

This article aims to recognize Odia handwritten digits using gradient-based feature extraction techniques and Clonal Selection Algorithm-based (CSA) multilayer artificial neural network (MANN) classifier. For the extraction of features which contribute the most towards recognition from images, are extracted using gradient-based feature extraction techniques. Principal component analysis (PCA) is used for dimensionality reduction of extracted features. A MANN is used as a classifier for classification purposes. The weights of the MANN are adjusted using the CSA to get optimized set of weights. The proposed model is applied on Odia handwritten digits taken from the Indian Statistical Institution (ISI), Calcutta, which consists of four thousand samples. The results obtained from the experiment are compared with a genetic-based multi-layer artificial neural network (GA-MANN) model. The recognition accuracy of the CSA-MANN model is found to be 90.75%.

Metal recovery prediction of elements from anode slime

Metal recovery prediction of elements remains as one of the significant problems in the metal industry due to limited application data. In this research, quasi-Newton training algorithm-based artificial neural network (ANN) was applied for estimating the recovery amount of the elements during the anode slime emerging processes. ANN models were designed with the approximation of the inverse Hessian at each iteration by using gradient information. Temperature, leaching time, solid-to-liquid ratio and ionic liquid concentration were taken as input parameters for the recovery amount estimation. The results showed that the proposed algorithm is highly efficient in predicting metal recovery amount from anode slime. As a precious one, maximum Au recovery amount is predicted by ANN as 82.11% when the solid-to-liquid ratio is 1/25, the temperature is 45 °C, ionic liquid concentration is 40% and the leaching time is 0.5 h.

A novel time difference of arrival localization algorithm using a neural network ensemble model

In a localization system, time difference of arrival technique is widely used to estimate the location of a mobile station. To improve the performance of mobile station location estimation, a novel algorithm-based artificial neural network ensemble and time difference of arrival information is proposed in non-line-of-sight environments. Back propagation neural network is a classic artificial neural network and may be effectively used for mathematical modeling and prediction, and an artificial neural network ensemble has better generalization ability and stability than a single network. First, the parameters, such as the weights and biases of the single neural network are optimized by the ant lion optimization method which is novel and effective. Then four types of different information from the time difference of arrival measurements are respectively used to train the individual neural network. Finally, the weighted average method is improved to combine the outputs of the different individual neural network, where weights are determined by the training errors. The estimation accuracy of the locating system is evaluated through experimental measurements. The simulation results show that the proposed algorithm is efficient in improving the generalization ability and localization precision of the neural network ensemble model.

Optimizing the Tribological Behavior of Hybrid Copper Surface Composites Using Statistical and Machine Learning Techniques

Copper-based surface composite dispersed with varying fractions of hybrid reinforcement was fabricated through friction stir processing (FSP). Hybrid reinforcement particles were prepared from aluminum nitride (AIN) and boron nitride (BN) particles of equal weight proportion. Based on design of experiments, wear characteristics of the developed copper surface composites were estimated using pin-on-disk tribometer. Experimental parameters include volumetric fraction of hybrid reinforcement particles (5, 10, and 15 vol %), load (10, 20, 30 N), sliding velocity (1, 1.5, and 2 m/s), and sliding distance (500, 1000, and 1500 m). Microstructural characterization demonstrated uniform dispersion of hybrid reinforcement particles onto the copper surface along with good bonding. Hardness of the developed surface composites increased with respect to increase in hybrid particle dispersion when compared with copper substrate while a reduction in density values was revealed. Analysis on wear rate values proved that wear rate decreased with increase in hybrid particle dispersion and increased with increase in load, sliding velocity, and distance. Analysis of variance (ANOVA) specified load as the most significant factor over wear rate values followed by volume fractions of particle dispersion, sliding velocity, and distance. Regression model constructed was found efficient in predicting wear rate values. Analysis of worn out surfaces through scanning electron microscopy (SEM) revealed the transition of severe to mild wear with respect to increase in hybrid reinforcement particle dispersion. A feed forward back propagation algorithm-based artificial neural network (ANN) model with topology 4-7-1 was developed to predict wear rate of copper surface composites based on its control factors.

A genetic algorithm-based artificial neural network model with TOPSIS approach to optimize the engine performance

ABSTRACTThis paper describes the application of an artificial neural network, genetic algorithm and analytic network process-technique for order preference by similarity to ideal solution for the selection of the optimum fuel blend. A single-cylinder, constant-speed direct-injection diesel engine with a rated output of 4.4 kW was used for exploratory analysis at different load conditions. Ten objectives – brake thermal efficiency, maximum rate of pressure rise, NOx, CO2, CO, HC, smoke, exhaust gas temperature, ignition delay and combustion delay were considered. The proposed ANN model is integrated with GA and the hybrid multi-criteria decision-making technique of ANP-TOPSIS to evaluate the optimum blend. First the ANN model was developed to predict the performance, combustion and emission parameters of the engine. A multi-layer perception network was used for non-linear mapping between input and output parameters. The performance of the ANN model is determined and shows the efficiency of the model to predict the performance, emission and combustion parameters, with a determination coefficient of 0.9627. Second, GA was used to determine the optimum load and blend based on the predicted ANN output parameter. Third, an approach based on the TOPSIS method was used for finding the best blend from the observed optimum parameters of GA.

Porosity- and reliability-based evaluation of concrete-face rock dam compaction quality

Abstract Compaction quality evaluation of concrete-faced rockfill dams (CFRDs) is a complex nonlinear process. Conventional evaluation methods depend on random spot tests, and porosity is considered as the main evaluation index. However, reliability of results is neglected in existing studies. Considering compaction parameters, material properties and their variability characteristics, an evaluation approach of CFRD compaction quality based on porosity and reliability is proposed. Reliability analysis is introduced to measure the variability in highly variable factors and calculate the porosity- and reliability-based index. Porosities of the entire work area are predicted using an elitist genetic algorithm-based artificial neural network, which are consistent with the measured values with an error

Neurogenetic modeling of energy demand in the United Arab Emirates, Saudi Arabia, and Qatar

Socio‐economic variables including gross domestic product, population, and energy and electricity production are used in modeling and forecasting national energy demands of the United Arab Emirates, Saudi Arabia, and Qatar. The proposed model features: (i) the nonlinear component of energy demand (removal of linear trend), (ii) application of double exponential smoothing method for input data projection, and (iii) genetic algorithm‐based artificial neural network (ANN) models. The proposed neuro‐genetic model performed very well for the three selected countries. The coefficient of determination and Willmott's index of agreement for the training and testing dataset are quite high whereas the mean absolute error, mean absolute percentage error and root mean squared error are quite low. The acceptable agreements between the observed energy consumption and the model predictions revealed its viability for the study of energy demand in the three selected member states of the energy exporting regional alliance Gulf Cooperation Council (GCC). © 2017 American Institute of Chemical Engineers Environ Prog, 36: 1208–1216, 2017

Modeling the cooling performance of vortex tube using a genetic algorithm-based artificial neural network

In this study, artificial neural networks (ANNs) have been used to model the effects of four important parameters consist of the ratio of the length to diameter(L/D), the ratio of the cold outlet diameter to the tube diameter(d/D), inlet pressure(P), and cold mass fraction (Y) on the cooling performance of counter flow vortex tube. In this approach, experimental data have been used to train and validate the neural network model with MATLAB software. Also, genetic algorithm (GA) has been used to find the optimal network architecture. In this model, temperature drop at the cold outlet has been considered as the cooling performance of the vortex tube. Based on experimental data, cooling performance of the vortex tube has been predicted by four inlet parameters (L/D, d/D, P, Y). The results of this study indicate that the genetic algorithm-based artificial neural network model is capable of predicting the cooling performance of vortex tube in a wide operating range and with satisfactory precision.

Design flood estimation in ungauged catchments using genetic algorithm-based artificial neural network (GAANN) technique for Australia

This paper focuses on the development and testing of the genetic algorithm (GA)-based regional flood frequency analysis (RFFA) models for eastern parts of Australia. The GA-based techniques do not impose a fixed model structure on the data and can better deal with nonlinearity of the input and output relationship. These nonlinear techniques have been applied successfully in many hydrologic problems; however, there have been only limited applications of these techniques in RFFA problems, particularly in Australia. A data set comprising of 452 stations is used to test the GA for artificial neural networks (ANN) optimization known as GAANN. The results from GAANN were compared with the results from back-propagation for ANN optimization known as BPANN. An independent testing shows that both the GAANN and BPANN methods are quite successful in RFFA and can be used as alternative methods to check the validity of the traditional linear models such as quantile regression technique.

Biometric gait identification based on a multilayer perceptron

In this study, we propose a novel approach for biometric gait identification. We designed a multilayered back-propagation algorithm-based artificial neural network for gait pattern classification and we compared the results obtained with those produced using the k-means and k-nearest neighbor algorithms. A novel aspect of our feature extraction procedure was the use of a kernel-based principal components analysis because the captured real-time data exhibited significant nonlinearity. The gait data were classified into four classes: normal, crouch-2, crouch-3, and crouch-4. The proposed method achieved gait identification with very good activity recognition accuracy (ARA). The experimental results demonstrated that the proposed methodology could recognize different activities accurately in outdoor and indoor environments, while maintaining a high ARA. The identification of disordered or abnormal gait patterns was the fundamental aim of this study. Thus, we propose a method for the early detection of abnormal gait patterns, which can provide warnings about the potential development of diseases related to human walking. Furthermore, this gait-based biometric identification method can be utilized in the detection of gender, age, race, and for authentication purposes.