Conjugate Gradient Backpropagation Research Articles

New strategy to optimize in-situ fenton oxidation for TPH contaminated soil remediation via artificial neural network approach

In-situ remediation of total petroleum hydrocarbon (TPH) contaminated soils via Fenton oxidation is a promising approach. However, determining the proper injection amount of H2O2 and Fe source over the Fenton reaction in the complex geological conditions for in-situ TPH soil remediation remains a daunting challenge. Herein, we introduced a practical and novel approach using soft computational models, a multilayer perception artificial neural network (MPLNN), for predicting the TPH removal performance. In this study, we conducted 48 sets of TPH removal experiments using Fenton oxidation to determine the TPH removal performance of a wide range of different ground conditions and generated 336 data points. As a result, a negative Pearson correlation coefficient was obtained in the Fe injection mass and the natural presence of Fe mineral in the soil, indicating that the excess of Fe could significantly retarded the TPH removal performance in the Fenton reaction. In addition, the MPLNN model with 6-6-1 training using Scaled conjugate gradient backpropagation (SCG) with tangent sigmoid as the transfer function demonstrated a high accuracy for TPH removal prediction with the correlation determination of 0.974 and mean square error value of 0.0259. The optimized MPLNN model achieved less than 20% error for predicting TPH removal performance in actual TPH-contaminated soil via Fenton oxidation. Hence, the proposed MPLNN can be useful in improving the Fenton oxidation of TPH removal performance in-situ soil remediation.

Integrating Machine Learning for Improved Prediction of Temperature and Moisture in Pavement Granular Layers

Abstract Pavement temperature and moisture content within the base and subgrade layers affect the load-bearing capacity of the pavement and dominate the pavement performance in cold regions. Accurately predicting pavement temperature and moisture content can improve pavement design and management. Conventional approaches, including numerical and statistical models, have been implemented to predict pavement temperature and soil moisture content. However, they have weaknesses, such as being only suitable for warm regions or only for predicting pavement temperature within the asphalt layer. Furthermore, none of them can simultaneously predict the pavement temperature and moisture content. To address this issue, data collected from an instrumented test road in Alberta, Canada, were used to train a model to predict the daily average pavement temperature and moisture content at various depths through three parameters, namely depth, day of the year, and air temperature. The MATLAB toolbox, Neural Net Fitting, was used, and the performance of three built-in algorithms, Levenberg–Marquardt, Bayesian regularization, and Scaled conjugate gradient backpropagation, was compared. The model with Bayesian regularization showed the highest accuracy, with an R2 value of 0.99 and a root mean square error (RMSE) of 1.49°C for pavement temperature prediction, and an R2 value of 0.95 and an RMSE of 0.025 m3/m3 for moisture content prediction. The model developed in this research is the first to simultaneously estimate pavement temperature and moisture content, so its performance was separately compared with two existing models in the literature. The artificial neural network (ANN) model shows higher accuracy than the two existing models, so it was found that the ANN could be a robust method for pavement temperature and moisture content prediction at various depths.

Financial constraints prediction to lead socio-economic development: An application of neural networks to the Italian market

This study applies a neural network framework to optimize the classification of firms and to predict their difficulties in collecting external financial resources in the short term. In detail, we adopt replicated bootstrapped algorithms optimized on sensitivity and specificity as error measures and we propose a comparative analysis to identify the best-performing one. According to our results, the Conjugate gradient backpropagation with Fletcher-Reeves updates (i.e., CGF) is the best-performing algorithm, with sensitivity equal to 74.41 % and specificity equal to 70.11 %. Then, we use this algorithm and its weights to provide a classification of the Italian manufacturing industry in 2019, identifying the geographical areas in which firms under financial constraints are located, as well as the most critical industrial sectors. Based on this evidence, and considering the implementation of a cohesion policy, we highlight interventions by policy makers to support firms’ access to the capital market, fostering their investments and the consequent socio-economic development.

Dijagnostika kvarova u ležajevima kotrljajućih elemenata poboljšana AI - sveobuhvatan pristup analizi vibracija

This research presents a comprehensive approach for bearing fault diagnosis using artificial intelligence (AI), particularly through the application of artificial neural networks (ANNs). By integrating these networks into vibration analysis, the approach aims to meet the critical need for prompt fault detection. The methodology comprises three key steps: vibration signal acquisition, feature extraction, and fault classification. Experiments were conducted to acquire vibration signals for the test bearings on a machinery fault simulator. Six time-domain features were extracted using MATLAB, creating a comprehensive dataset for training the ANN models with three algorithms: Levenberg-Marquardt backpropagation (LMBP), scaled conjugate gradient backpropagation (SCGBP), and Bayesian regularization backpropagation (BRBP). The BRBP algorithm achieved the highest correct classification rate (97.2%), followed by LMBP (90%) and SCGBP (83.6%). To evaluate their efficacy in bearing fault classification, these three networks were simulated, revealing that BRBP could predict all four classes of bearings with zero errors.

Deriving groundwater major ions from electrical conductivity using artificial neural networks supported by analytical hydrochemical solutions

Tracing groundwater chemistry requires broad and continuous sampling campaigns to keep track of groundwater deterioration. The need is pressing in coastal aquifers due to high water demand to meet rapid population growth. This paper uses artificial intelligence (AI) to develop an algorithm capable of predicting seven major chemical ions in water (Ca2+, Mg2+, Na+, K+, HCO3−, Cl− and SO42−) from electrical conductivity (EC), a single input that can be easily acquired manually or via automated monitoring programs. The algorithm mainly relied on artificial neural network (ANN) to tackle complex and nonlinear relationships. It included a set of procedures among others, Levenberg-Marquardt back propagation multilayer perceptron (MLP), conjugate gradient back propagation with Fletcher-Reeves updates MLP, and support vector regression. The low accuracy of ANN models in predicting Ca2+ and HCO3− concentrations justified their calculation via advanced hydrochemical analytical solutions that were embedded in one AI scheme. The full scheme was trained and tested over an EC20 range of 400–3000 μS/cm, and then validated for new measurements elsewhere. It showed <15% discrepancy for the four ions that constitute the majority of groundwater chemical composition (Cl−, Na+, HCO3− and Ca2+). Higher uncertainties were associated with concentrations derived for EC values outside the training limits. The developed algorithm can save time and money in return of labor and laboratory work. It is an open source code publicly available for further improvements by training it on additional data covering a wider EC range.

Optimizing buckling load of sandwich plates with cutouts using artificial neural networks and genetic algorithms

In this research, the optimization of the geometric parameters of a 3-layer composite plate with an elliptical cutout in the center was investigated utilizing an artificial neural network and genetic algorithm for the first time to obtain the minimum ratio of linear buckling load to weight and the ratio of the diameters of the ellipse. After completing 294 simulations in Abaqus finite element software and forming an artificial neural network with two hidden layers, the function obtained in the artificial neural network was used as the objective function in the genetic algorithm, and the optimal values for the angle of the ellipse with the horizon axis, the diameter ratio of ellipses as well as the fiber angle of the middle layer of the composite were obtained. using artificial neural networks, three algorithms, Levenberg-Marquardt, Bayesian regularization; and scaled conjugate gradient backpropagation, were compared and the number of neurons in each algorithm was compared. The results showed that Levenberg-Marquardt’s algorithm is more accurate compared to other algorithms. In the end, the optimal values for the angle of the ellipse with the horizontal axis; as the fiber angle of the middle layer of the composite and, the ratio of the oval diameters as α = 12.3001 degree , θ = 5.9481 degree and D = 1.5690 .

Multi-Objective Optimization Design of Permanent Magnet Eddy Current Coupler Based on SCG-BP Neural Network Modeling and the ONDX-NSGA-II Algorithm

There is a complex coupling relationship between the structural parameters and various performance indicators of a permanent magnet eddy current coupler. In order to obtain the optimal combination of structural parameters that can improve the overall performance of the coupler, it is necessary to reasonably balance the contradiction and competition among performance indicators of the permanent magnet eddy current coupler. A multi-objective optimization method for permanent magnet eddy current couplers based on scaled conjugate gradient back propagation neural network modeling, improved opposition-based learning, and normal distribution crossover operator non-dominated sorting genetic algorithm-II is proposed. The optimization results are compared with those of the traditional non-dominated sorting genetic algorithm-II and the Pareto envelope-based selection algorithm-II, and it is verified that the proposed multi-objective optimization algorithm is accurate, reliable, and has better convergence and versatility. Compared with the original model, the output torque of the optimized coupler increased by 8.54%, and the eddy current loss and cost decreased by 3.71% and 8.74%, respectively. Finally, the correctness of the theoretical analysis was verified through 3D finite element simulation and an experimental simulation platform.

Predictive Modeling Analysis for the Quality Indicators of Matsutake Mushrooms in Different Transport Environments.

Matsutake mushrooms, known for their high value, present challenges due to their seasonal availability, difficulties in harvesting, and short shelf life, making it crucial to extend their post-harvest preservation period. In this study, we developed three quality predictive models of Matsutake mushrooms using three different methods. The quality changes of Matsutake mushrooms were experimentally analyzed under two cases (case A: Temperature control and sealing measures; case B: Alteration of gas composition) with various parameters including the hardness, color, odor, pH, soluble solids content (SSC), and moisture content (MC) collected as indicators of quality changes throughout the storage period. Prediction models for Matsutake mushroom quality were developed using three different methods based on the collected data: multiple linear regression (MLR), support vector regression (SVR), and an artificial neural network (ANN). The comparative results reveal that the ANN outperforms MLR and SVR as the optimal model for predicting Matsutake mushroom quality indicators. To further enhance the ANN model's performance, optimization techniques such as the Levenberg-Marquardt, Bayesian regularization, and scaled conjugate gradient backpropagation algorithm techniques were employed. The optimized ANN model achieved impressive results, with an R-Square value of 0.988 and an MSE of 0.099 under case A, and an R-Square of 0.981 and an MSE of 0.164 under case B. These findings provide valuable insights for the development of new preservation methods, contributing to the assurance of a high-quality supply of Matsutake mushrooms in the market.

Using the numerical simulation and artificial neural network (ANN) to evaluate temperature distribution in pulsed laser welding of different alloys

The temperature field during laser welding process plays an important role on determining the quality and quantity of the weld bead size, microstructure characterizations and mechanical properties of the welding interface in the thermal engineering applications. In this study, using the numerical simulation, the influence of pulse duration and frequency on the temperature distribution and velocity field in distinctive laser welding of stainless steel 420 (S.S 420)/stainless steel 304 (S.S 304), and Bohler 303 (B 303)/stainless steel 304 (S.S 304) was examined. The results of numerical modeling illustrated that shear stress of Marangoni and buoyancy force are the most curtail aspects in the formation of the flow of liquid metal. A novel artificial intelligence method is proposed to optimally predict the melting ratio, and maximum temperature of the materials. To this end, a combination of ANN and Particle Swarm Optimization (PSO) algorithms are employed. The PSO algorithm is used to optimize the architecture and training algorithm of the ANN, while the ANN is employed for the regression problem. Based on the results, a three-layer feed-forward architecture with sigmoid transfer functions having 17 and 8 neurons in the hidden layers combined with the scaled conjugate gradient backpropagation training scheme is recognized by the PSO as the optimal configuration. Application of optimal ANN to the regression problem results in an acceptable level of error for the training, validation, and test datasets. Finally, the optimized ANN can be utilized to anticipate the melting ratio and thereby the resultant temperature.

A Comprehensive Pattern Recognition Neural Network for Collision Classification Using Force Sensor Signals

In this paper, force sensor signals are classified using a pattern recognition neural network (PRNN). The signals are classified to show if there is a collision or not. In our previous work, the joints positions of a 2-DOF robot were used to estimate the external force sensor signal, which was attached at the robot end-effector, and the external joint torques of this robot based on a multilayer feedforward NN (MLFFNN). In the current work, the estimated force sensor signal and the external joints’ torques from the previous work are used as the inputs to the proposed designed PRNN, and its output is whether a collision is found or not. The designed PRNN is trained using a scaled conjugate gradient backpropagation algorithm and tested and validated using different data from the training one. The results prove that the PRNN is effective in classifying the force signals. Its effectiveness for classifying the collision cases is 92.8%, and for the non-collisions cases is 99.4%. Therefore, the overall efficiency is 99.2%. The same methodology and work are repeated using a PRNN trained using another algorithm, which is the Levenberg–Marquardt (PRNN-LM). The results using this structure prove that the PRNN-LM is also effective in classifying the force signals, and its overall effectiveness is 99.3%, which is slightly higher than the first PRNN. Finally, a comparison of the effectiveness of the proposed PRNN and PRNN-LM with other previous different classifiers is included. This comparison shows the effectiveness of the proposed PRNN and PRNN-LM.

Optimization of neural network parameters in improvement of particulate matter concentration prediction of open-pit mining

The prediction of particulate matter (PM) concentration around open-pit mining is crucial for its control. To achieve this, machine learning (ML) techniques have been attempted in PM estimation without reaching their full potential. In the current study, an artificial neural network (ANN) is employed for the prediction of PM2.5 concentration (PM < 2.5 μm), PM10 concentration (PM < 10 μm), and total suspended particulate (TSP). The influence of training algorithms and epochs on the modeling accuracy of ANN is investigated. Moreover, the convergence of ANN modeling is studied using Monte Carlo simulations. Compared with the quasi-Newton method (QNM) and conjugate gradient backpropagation (CGB), the one-step secant algorithm (OSSM) is the most suitable training algorithm for PM concentration prediction. Using the 0.1% bound as a convergence criterion, 500 Monte Carlo simulations are enough to get the converged results. The ANN-OSSM achieves high accuracy in PM concentration prediction. The R values on the testing set of PM2.5, PM10, and TSP are 0.887, 0.943, 0.886, respectively. Compared with the results in the literature, accuracy improvement is achieved using ANN-OSSM, with R increases from 0.86 to 0.92 (TSP dataset), from 0.91 to 0.94 (PM2.5 dataset), and from 0.84 to 0.89 (PM10 dataset). The present ML model is higher in accuracy and more straightforward in methodology compared with the hybridized model in the literature, which will promote its industrial application in the near future.

Experimental and numerical analysis of shape memory sinusoidal lattice structure: Optimization through fusing an artificial neural network to a genetic algorithm

Shape memory polymers are smart materials that can “remember” their original shape and return to it when exposed to certain stimuli, such as heat, light, or a change in pH. This research investigates the compressive nonlinear response of a sinusoidal shape memory unit cell made of PLA. Different architectures with diverse thicknesses are optimized through a finite element analysis, and the values of absorbed energy, dissipated energy, and peak force are extracted. The obtained data is subsequently used to train a judiciously structured artificial neural network. Three algorithms, including Levenberg-Marquardt, Bayesian regularization, and scaled conjugate gradient backpropagation, are utilized for training the neural network. The function obtained is employed to formulate the objective function to be minimized- (or maximized), and the optimal geometric dimensions are extracted. In the experimental part, the sinusoidal cell is made using a 3D printer and PLA material, and a quasi-static test is performed. After each quasi-static test, the structure is exposed to heat, and thanks to its shape memory properties, it returns to its original shape. This cycle is repeated up to four times. Testing results showed that the proposed sinusoidal structure possessed an attractive mechanical behaviour relevant to the specific geometric parameters and loading conditions. It was observed that the amount of absorbed energy and the maximum forces decreased in each cycle compared to the previous cycle. The experimental results demonstrate that the absorbed energy by the sinusoidal cell decreased by 4.7% in the second loading, 12.98% in the third loading, and 33.7% in the fourth loading. The results reveal that the reduction in the peak force of structure is more pronounced compared to the decrease in absorbed energy under cyclic loading–unloading. Based on the obtained results, it can be concluded that the proposed structure can be used as a suitable energy absorber with the ability to recover the original shape and maintain the mechanical properties up to several cycles of loading.

An intelligent pattern recognition model for assessment of terrorists’ activities in Nigeria

Terrorism and its brutal tendencies constitute a major setback to the development process of the Nigerian economy leading to severe loss of lives, destruction of properties, and a decline of interest in investment by both local and foreign investors. Many models for assessment of terrorist’s activities lack the ability of learning from previous patterns in order to guide pre-emptive actions against future occurrences, and there are no established regional pattern of weaponry, types of attack, as well as types of victims of terrorists’ operations. This study seeks to build a robust intelligent model for recognizing several terrorists’ patterns in each of the six geo-political zones of Nigeria. A data set of 5,503 instances of terrorists’ activities in Nigeria was obtained and a pattern recognition model was built using Artificial Neural Networks (ANN) with 70%, 15%, and 15% data splits for training, validation, and testing respectively. A 10-10-6 ANN architecture was designed and trained using the scaled conjugate gradient backpropagation algorithm. The training was carried out using Matlab’s neural network pattern recognition toolkit. In order to numerically represent categorical data, a sort-order scheme was developed by the authors and utilized. The results showed average percentage scores for accuracy, precision, recall and F1-score as 99.89, 99.96, 100 and 99.98 respectively. This showed acceptable performance. The developed model is therefore considered a robust one for recognition of terrorists’ patterns in Nigeria. This would assist security agencies to deal with terrorists’ incidences with high intelligent information and advanced preparation for prevention and control. The developed model is highly recommended for use by the security agencies in the country.

Detection of Malicious Network Activity by Artificial Neural Network

This paper presents a deep learning approach to detect malicious communication in a computer network. The intercepted communication is transformed into behavioral feature vectors that are reduced (using principal component analysis and stepwise selection methods) and normalized to create training and test sets. A feed-forward artificial neural network is then used as a classifier to determine the type of malicious communication. Three training algorithms were used to train the neural network: the Levenberg-Marquardt algorithm, Bayesian regularization, and the scaled conjugate gradient backpropagation algorithm. The proposed artificial neural network topology after reducing the size of the training and test sets achieves a correct classification probability of 81.5 % for each type of malicious communication and of 99.6 % (and better) for normal communication.

GFDM TRANSCEIVER BASED ON ANN CHANNEL ESTIMATION

Generalized frequency division multiplexing (GFDM) is one of the candidate waveforms beyond 5G for wireless communication. Channel estimation is challenging in wireless transmission because of inherent interference. It is based on a pilot signal that uses Least Square (LS) and Normalized Least Mean Square (NLMS) to perform the estimation. This paper used the Artificial Neural Network (ANN) as a channel estimation method, considered a novel estimation process for the GFDM transceiver. The channel estimation based on ANN depends on the data set generated from LS estimation, which considers the proposed method is optimized to LS based on the backpropagation neural network. The ANN algorithm considered a fitness function to estimate the channel. Levenberg-Marquardt backpropagation (LM), Bayesian Regularization backpropagation (BR), and Scaled Conjugate Gradient backpropagation (SCG) training methods training by using the Matlab toolbox used to perform the estimation. BR training method gives the best performance than LS and other training methods at 22 neurons in hidden layers, which at 20dB give BER = 0.0369 that enhanced over LS by 0.08.

A machine learning approach to electron orbit control at the 1.5 GeV synchrotron light source DELTA

Machine learning (ML) methods have found their application in a wide range of particle accelerator control tasks. Among other possible use cases, neural networks (NNs) can also be utilized for automated beam position control (orbit correction). ML studies on this topic, which were initially based on simulations, were successfully transferred to real accelerator operation at the 1.5–GeV electron storage ring of the DELTA accelerator facility. For this purpose, classical fully connected multi-layer feed-forward NNs were trained by supervised learning on measured orbit data to apply local and global beam position corrections. The supervised NN training was carried out with various conjugate gradient backpropagation learning algorithms. Afterwards, the ML-based orbit correction performance was compared with a conventional, numerical-based computing method. Here, the ML-based approach showed a competitive orbit correction quality in a fewer number of correction steps.

Vibration analyses of railway systems using proposed neural predictors

Due to travelling on railway systems; there are many gaps and problems in cross areas. Therefore; it is necessary and very important to establish intelligent crossing systems in such areas. On the other hand, it is not possible for trains to stop or brake immediately against an obstacle due to their high speed and inertia. For this reason, it is necessary to work on the safety/warning of the other main factors and necessities (pedestrians and vehicles) in level crossings. This experimental investigation is carried out by using an experimental real-time train and crossing systems. The main vibration parameters are analysed by using neural networks. First, the dynamics of the train-rail system related to level crossings are examined, and the vibrations created by the train on rails are measured at different speeds. Then three types of proposed neural networks predictors, Levenberg-Marquardt backpropagation (LMBP), scaled conjugate gradient backpropagation (SCGB) and BFGS quasi-Newton backpropagation (BFGS) are used to predict the vibration of the train-rail system. From the results, it is seen that the proposed LMBP is more suitable for analysing and predicting the vibration of the train-rail system. It is clear that the speeds of the trains approaching the level crossing can be estimated from the vibration of the trains on the rails.

Evaluation of the Performance of ANN Algorithms with the Bidirectional Functionally Graded Circular Plate Problem

Functionally graded materials (FGMs) are materials composed of metals and ceramics in which the distribution of material components varies according to a particular volumetric function. FGMs are often used in high-temperature applications. In our study, models were created in the Artificial Neural Network depending on the equivalent stress levels in the compositional gradient exponent, which is the most important parameter in determining the thermo-mechanical behavior of circular plates functionally staggered in two directions, and the performances of these models were evaluated. These models were obtained with four different training algorithms: Levenberg-Marquardt, Backpropagation Algorithm, Resilient Propagation Algorithm, Conjugate Gradient Backpropagation with Powell-Beale Restarts To train the ANN, equivalent stress levels were obtained by performing numerical analyzes at different compositional gradient upper values. The data sets were created by considering the largest value of the equivalent stress levels, the smallest value of the largest value, the largest value of the smallest value, and the smallest value of the smallest value. In this study, training stages and performance values were examined and interpreted with 4 training algorithms in detail.

FORECASTING NIGERIAN ECONOMIC GROWTH BASED CORPORATE INCOME TAX

Developing countries like Nigeria have been hit hard by recent economic crises, and uncertainty is one of the characteristics that all prices and revenues share. A forecast of Nigerian economic growth was attempted using an artificial neural network (ANN) and statistical model to analyse the contribution of non-oil income tax generation to Nigerian economic development. While the primary objective is to develop and implement the proposed models capable of simulating a real non-oil income tax and evaluate their performances. The dataset (Corporate Income Tax) from 2015–2020 was obtained from the National Bureau of Statistics of Nigeria (NBSN). Three training algorithms for ANN were adopted, such as conjugate gradient back-propagation with Fletcher-Reeves restarts, Bayesian regularisation, and gradient descent with an adaptive learning rate, whereas in the statistical part, multiple linear regressions were applied. Comparing all the models revealed that the Bayesian regularisation produced more accurate results than the other models.

Machine Learning Model for Estimation of Local Scour Depth around Cylindrical Bridge Piers

Scour around bridge piers is a well-known threat to bridge stability worldwide. It can cause losses in lives and the economy, especially during floods. Therefore, an artificial intelligence approach called artificial neural network (ANN) was used to predict the scour depth around bridge piers. The ANN model was trained with laboratory data, including pier width, flow velocity, particle diameter, sediment critical velocity, flow depth, and scour depth. The data was divided into 70% for training, 15 for validation, and 15% for testing. Besides, the ANN model was trained using various training algrthins and a single hidden layer with 20 neurons in the hidden layer. The results showed that the ANN model with Bayesian regularization backpropagation training algorithm provides a better predicted scour depth with a correlation coefficient (R) equal to 0. 9692 and 0.926 for training and test stages, respectively. Besides, it showed a low mean squared error (MSE), which was 0.0034 for training and 0.0066 for the test. These results were slightly better than the ANN with Levenberg-Marquardt backpropagation with R training equals 0.9552 (MSE training = 0.0047), and R test equals 0.838 (MSE test = 0.007).On the other hand, the ANN model with a scaled conjugate gradient backpropagation training algorithm showed worse predictions (R training = 0.7407 and R test = 0.6409). Besides, the ANN model shows better outcomes than the linear regression model. Finally, the sensitivity analysis has shown that the pier width is the most crucial parameter for estimating scour depth using the ANN model.