Cascade Forward Back Propagation Network Research Articles

Compositional Modeling of the Oil Formation Volume Factor of Crude Oil Systems: Application of Intelligent Models and Equations of State.

This communication primarily concentrates on developing reliable and accurate compositional oil formation volume factor (Bo) models using several advanced and powerful machine learning (ML) models, namely, extra trees (ETs), random forest (RF), decision trees (DTs), generalized regression neural networks, and cascade-forward back-propagation network, alongside radial basis function and multilayer perceptron neural networks. Along with these models, seven equations of state (EoSs) were employed to estimate Bo. The performance of the developed ML models and employed EoSs was assessed through various statistical and graphical evaluations. Overall, the ML models could provide much more accurate predictions in comparison to EoSs. However, the results indicated that tree-based models, specifically ET models, could outperform the other models and can be reliably applied for estimating Bo. The most reliable ET model could predict Bo with a total average error of 1.17%. Lastly, the outlier detection approach verified the dataset’s consistency detecting only 17 (out of 1224) data points as outliers for the proposed Bo models.

Forecast of Bayesian-based dynamic connectedness between oil market and Islamic stock indices of Islamic oil-exporting countries: Application of the cascade-forward backpropagation network

This paper is the first attempt to forecast the time-varying total return and volatility connectedness between the oil prices and the Islamic stock indices of seven oil-exporting countries, namely Iran, Oman, Saudi Arabia, Qatar, Kuwait, Bahrain, and United Arab Emirates. The entire analysis follows two main stages. The first carries out the connectedness analysis using the Bayesian time-varying parameter vector autoregressive (BTVP-VAR) model, while the second stage applies the training process in a cascade-forward backpropagation network (CFBPN) to five groups of input (spillover TO others, spillover FROM others, both spillovers TO & FROM others, raw returns, and raw volatilities, and ALL inputs) to forecast the dynamic total connectedness of returns/volatilities. The results of the connectedness analysis show that the Islamic stock index of Iran is not connected to the oil market and the Islamic stock indices of other Islamic oil-exporting countries. The Islamic stock indices of the United Arab Emirates and Saudi Arabia have a leadership role in the network. Besides, oil demonstrates a net receiving spillover status. In addition, connectedness increases during periods of crisis and significant oil price changes. Finally, the forecast analysis shows that the transmitted return and volatility spillovers from markets (TO spillovers) are the most important factor information in predicting the total connectedness of the network. The findings showcase important implications for investors, policy makers, and future studies.

Experimental measurement and compositional modeling of bubble point pressure in crude oil systems: Soft computing approaches, correlations, and equations of state

No one can deny the ever-increasing importance of oil since it has influenced every aspect of humans' life. One of the most important pressure-volume-temperature (PVT) properties of crude oil, which is needed in a majority of production and reservoir engineering calculations, is the saturation pressure (bubble point pressure (Pb)). Having accurate knowledge about Pb is significant for both academia and industry. This communication concentrates on providing reliable experimental data from constant composition expansion (CCE) test as well as rigorous compositional models to predict saturation pressure of crude oils based on oil composition (H2S, N2, CO2, C1 to C7+), reservoir temperature, C7+ specifications (molecular weight and specific gravity). Seven advanced machine learning approaches, namely, decision trees (DTs), random forest (RF), extra trees (ETs), cascade-forward back-propagation network (CFBPN), and generalized regression neural networks (GRNN) as well as multilayer perceptron (MLP) and radial basis function (RBF) neural networks were used for modeling. The CFBPN and MLP models were trained by three different training algorithms, namely scaled conjugate gradient (SCG), Bayesian regularization (BR), and Levenberg-Marquardt (LM). The modeling was done based on a databank consisting of 206 data points (130 points were previously published in the literature plus 76 points determined experimentally in this study). The results show that the DT model could provide the most reliable prediction with an average absolute percent relative error (AAPRE) of 4.43%. The efficiency of various equations of state (EoS) and empirical correlations were checked. According to the results, Peng-Robinson (PR) and the correlation developed by Elsharkawy were the most efficient EoS and empirical correlation with AAPRE values of 8.46% and 12.05%, respectively. Then, the sensitivity analysis revealed that the Pb was extremely affected by methane and C7+ mole percent. Finally, the Leverage approach confirmed the validity of the employed data, detecting only 7 points as outliers.

Predicting formation damage of oil fields due to mineral scaling during water-flooding operations: Gradient boosting decision tree and cascade-forward back-propagation network

Water-flooding is one of the main options employed by the oil industry to meet the world's ever-increasing demand for oil, as the primary source of energy. This approach is highly prone to cause formation damage if the injected water is not compatible with the formation brine. In this study, decision tree optimized with gradient boosting (GBDT), cascade-forward back-propagation network (CFBPN), and generalized regression neural networks (GRNN) were employed, as relatively novel intelligent models, for the first time to develop accurate models to estimate the formation damage during a waterflooding operation in terms of damaged permeability. To compare the performance of these models, radial basis function (RBF) and multilayer perceptron (MLP) neural networks were also developed. The Levenberg-Marquardt algorithm (LMA), scaled conjugate gradient (SCG), and Bayesian regularization (BR) were used for training the MLP and CFBPN models. The results of this study showed the outperformance of the proposed GBDT model compared to the other developed models as well as previously proposed ones with an average absolute percent relative error (AAPRE) of 0.1465 % and correlation coefficient (R 2 ) of 0.9991 for the whole dataset. According to the results, the accuracy of the developed models could be ranked as follows: GBDT > CFBPN-LM > CFBPN-BR > RBF > MLP-LM > GRNN > MLP-BR > CFBPN-SCG > MLP-SCG. Moreover, it was shown that the GBDT mode could estimate more than 90 % of points with an absolute relative error of lower than 0.5 %. The trend analysis showed the high capability of the developed models in detecting the physical trend of the formation damage with variation of inputs. Then, the variable impact analysis was performed for this model, and the results reflect the high dependency of the model's predictions on the volume of injected water (Vinj), initial permeability (Ki), and ionic concentration of sulfate. Lastly, the Leverage approach was employed to determine suspected points as well as the applicability realm of the GBDT model. The results of the outlier detection indicated that only 4 points (0.93 % of the dataset) were detected as outliers, and the applicability realm of the proposed GBDT was verified. The findings of this communication shed light on the application of intelligent models and their power in predicting the formation damage caused during water-flooding operations before their occurrence. • Formation damage during water-flooding is modelled using GBDT, RBF, MLP, CFBPN, and GRNN. • LMA, SCG, and BR are used for optimizing MLP and CFBPN. • GBDT outperforms all models with AAPRE of 0.14 %. • The Leverage approach was used to determine suspected points and as the applicability realm of the GBDT model.

Extraction of Phenolic Compounds with Antioxidant Activity from Strawberries: Modelling with Artificial Neural Networks (ANNs).

This research study focuses on the evaluation of the total phenolic compounds (TPC) and antioxidant activity (AOA) of strawberries according to different experimental extraction conditions by applying the Artificial Neural Networks (ANNs) technique. The experimental data were applied to train ANNs using feed- and cascade-forward backpropagation models with Levenberg-Marquardt (LM) and Bayesian Regulation (BR) algorithms. Three independent variables (solvent concentration, volume/mass ratio and extraction time) were used as ANN inputs, whereas the three variables of total phenolic compounds, DPPH and ABTS antioxidant activities were considered as ANN outputs. The results demonstrate that the best cascade- and feed-forward backpropagation topologies of ANNs for the prediction of total phenolic compounds and DPPH and ABTS antioxidant activity factors were the 3-9-1, 3-4-4-1 and 3-13-10-1 structures, with the training algorithms of trainlm, trainbr, trainlm and threshold functions of tansig-purelin, tansig-tansig-tansig and purelin-tansig-tansig, respectively. The best R2 values for the predication of total phenolic compounds and DPPH and ABTS antioxidant activity factors were 0.9806 (MSE = 0.0047), 0.9651 (MSE = 0.0035) and 0.9756 (MSE = 0.00286), respectively. According to the comparison of ANNs, the results showed that the cascade-forward backpropagation network showed better performance than the feed-forward backpropagation network for predicting the TPC, and the FFBP network, in predicting the DPPH and ABTS antioxidant activity factors, had more precision than the cascade-forward backpropagation network. The ANN technique is a potential method for estimating targeted total phenolic compounds and the antioxidant activity of strawberries.

WITHDRAWN: Prediction of epileptic seizures using Grey Wolf Optimized Model Driven mathematical approach

Abstract Epilepsy is the most dangerous neurological disorder which is characterized by unprovoked electrical events in the brain, called seizures. Epileptic patients are affected neurologically, cognitively, psychologically and societal behaviourally. It is crucial to understand the temporal dynamics of epileptic EEGs involving the inception, propagation and termination of seizures. In this work, we construct a data analytics approach in conjunction with a neural based mathematical model. First the modelling of seizure EEGs are obtained using an Enhanced Wendling (EW) model. Next a data driven, Grey Wolf Optimized Model Driven (GWO-MD) approach is modelled for tracking the seizure events. Finally, a Closed Loop Control (CLC) framework which has a Fractional Order Sliding Mode Controller (FOSMC) combined with a Genetic Algorithm based Cascade Forward Back Propagation network (GA-CFBP) is constructed for controlling the seizure EEGs. The proposed model-driven mathematical approach is validated using two distinct databases such as traditional CHB-MIT and a private real time SRM database. Empirical results were obtained from 16 patients 69 EEG recordings exhibit that the proposed approach accomplishes close to the real time recordings. Tracking of the seizure states, predicting and controlling results of the proposed model is exceptionally better than the state-of-the-art approaches.

Moment invariants for cancer classification based on electron–ion interaction pseudo potentials (EIIP)

Cancer is considered one of the most dangerous diseases leading to imminent death during recent decades. It spreads quickly at a frightening pace to all parts of the body and early detection of such diseases may help reduce the risks. Many methods have been used to classify and predict the outcome of cancer, such as image processing techniques, artificial intelligence, and nanotechnology. Out of these methods, artificial intelligence techniques have played an essential role and have provided satisfying results from different investigations in various fields. Considering that cancer is a genetic disease, electron–ion interaction pseudo potential can be used to convert DNA sequences from string shape into number values so that genomic signal processing can be applied for the feature extraction step. In this paper, 51 healthy and 51 cancer genes from multiple cells obtained from the National Center for Biotechnology Information Genbank were used for analysis. Then, 7 invariant moments were extracted into 2 types of learning machine methods: supervised and unsupervised learning. Finally, their results were compared using Receiver Operating Characteristic parameters. From these results, the best accuracy was obtained from the trainable cascade-forward backpropagation network. The calculated parameters were 0.3333, 0.3158, 0.6842, 0.6667, and 67.5% for the Kohonen network, 0.1053, 0.1429, 0.8571, 0.8947, and 87.5% for the feed-forward network, and 0.0000, 0.0909, 0.9091, 1.0000, and 95% for the cascade-forward network for FPR, FNR, TPR, TNR, and accuracy respectively.

Prediction of Partial Miscibility of Binary Mixtures Consisting of Ionic Liquids

Accurate knowledge and prediction of the liquid–liquid transition phase consisting of ionic liquids and organic solvents are key points in different industrial processes. In this way, the effects of 8 ionic liquids and 38 organic solvents through 58 binary mixtures on the miscibility gap and upper critical solution (or consolute) temperature are examined. Various approaches including 1) feedforward and cascade forward, which are trained with a backpropagation algorithm (that is, feedforward backpropagation network and cascade-forward backpropagation network) and 2) radial basis neural networks, an exact radial basis network, and generalized regression neural networks based on kernel regression, are used for the predictions and evaluations. Four independent variables (that is, molecular weight and density of ionic liquid and the organic solvent) are selected to differentiate among ionic liquids and organic solvents. In addition, the mole fraction of ionic liquid is the other input variable for the networks to predict the liquid–liquid phase transition. The obtained results reveal that the kernel regression-based networks are able to perform better as compared to two other considered networks trained with a backpropagation algorithm. The overall results indicated that the performance of the radial basis network leads to the lowest error, corresponding to the highest capability in predicting the partial miscibility condition of studied mixtures.

Artificial Neural Network Model to Predict Reservoir Storage

The rapid growth in population increases water demand thus resulting in scarcity of water which is due to improper management rather than lack of resources. Reservoir is the most important source for surface water. So, reservoir storage plays a crucial role in efficient reservoir management. Artificial neural network (ANN) is capable of simulating reservoir storage capacity. So, in present work cascade forward back propagation and feed forward back propagation network artificial neural network (ANN) models with five different architecture in each by changing the number of hidden layer neurons were developed to predict Harangi reservoir storage, Karnataka, India. The first two years (2010–2012) data on daily basis was used for supervised training and remaining (2013–2014) data on daily basis was used in prediction. The predictive accuracy using the statistical parameters like correlation coefficient (R) and mean absolute percentage error (MAPE) were found within well acceptable limit. Results shows that, compared to the cascade forward back propagation network, feed forward back propagation network architecture of 6-5-1 ANN model with correlation coefficient (R) of 0.9882 and absolute mean percentage error (MAPE) of 5.225%. is best suited for present study.

A Study on the Application of Artificial Neural Networks on Green Self Consolidating Concrete (SCC) under Hot Weather

Self-consolidating concrete (SCC) has recently drawn attention to the construction industry in hot weather countries, due to its high fresh and mechanical properties. The slump flow is routinely used for quality control of SCC. Experiments were conducted by the current authors to investigate the effects of hot weather conditions on the slump flow of SCC. Self-consolidating concrete mixtures were prepared with different dosages of fly ash and superplasticizer and under different ambient temperatures. The results showed that the slump flow of SCC is sensitive to changes in ambient temperature, fly ash dosage, and superplasticizer dosage. In this paper, several artificial neural networks (ANNs) were employed to predict the slump flow of self-consolidating concrete under hot weather. Some of the data used to construct the ANNs models in this paper were collected from the experimental study conducted by the current authors, and other data were gathered from literature. Various parameters including ambient temperature and mixing time were used as inputs during the construction of ANN models. The developed ANN models employed two neural networks: the Feed-Forward Back Propagation (FFBP) and the Cascade Forward Back Propagation (CFBP). Both FFBP and CFBP showed good predictability to the slump flow of SCC mixtures. However, the FFBP network showed a slight better performance than CFBP, where it better predicted the slump flow of SCC than the CFBP network under hot weather. The results in this paper indicate that the ANNs can be employed to help the concrete industry in hot weather to predict the quality of fresh self-consolidating concrete mixes without the need to go through long trial and error testing program.Keywords: Self-consolidating concrete; Neural networks; Hot weather, Feed-forward back-propagation, Cascade-forward back propagation.

Recurrent Neural Networks Based Differential Protection of Power Transformers

Power transformers are important electrical equipments that need fast protection, because of their essential role in power system operation and their expensive cost. The most common technique used to protect the transformer is the differential relay, but it doesn't provide discrimination between internal fault and inrush currents. This paper presents an algorithm based on recurrent neural network (RNN) as a differential protection for three phase two windings transformer. The algorithm uses both the primary and secondary currents and second order harmonics of currents to discriminate between internal fault and inrush currents. A comparison among the performance of three neural networks based classifiers is presented. These networks are: FFBPNN (feed forward back propagation), cascade-forward back propagation network (CFBPNN), and proposed recurrent network (RNN). The transformer fault conditions are simulated using PSCAD/EMTDC in order to obtain the primary and secondary current signals. These current signals are used to train and test the neural networks which implemented by Matlab/Simulink. The test results prove that the RNN is stable and give good behaviors for different fault conditions. It is more reliable for recognition of transformer inrush and internal fault currents.

Intelligent Landslide System Based on Discriminant Analysis and Cascade-Forward Back-Propagation Network

Inspired from the great potential of the discriminant analysis (DA) as a feature selection tool for pattern recognition, this paper proposes an intelligent prediction of landslides based on selected optimum factors. The proposed prediction framework consists of three parts: the collection and extraction of landslide factors, determination of the important factors using DA, and the artificial intelligence for the classification and prediction of landslide hazard mapping for Penang Island. Twenty-one factors are extracted and collected for the first part. The DA is introduced as a feature selection tool for the second part while the cascade-forward back-propagation network (CFBPN) is proposed to predict the locations where landslides are prone to occur. The CFBPN is compared with multilayered perceptron network and Elman back-propagation network. The performances were verified using the classification and prediction accuracy. Results obtained proved that the proposed DA is an effective feature selection technique. Based on the results, the CFBPN produces good performance after the factor selection process with an accuracy of 89.28 % for the classification and 92.58 % for the prediction, as compared with the accuracy before the factors selection process, 88.13 % for the classification and 91.55 % for the prediction.

CFBP Network–A Technique for Crack Detection

This paper describes the Cascade Forward Back Propagation (CFBP) network for crack detection in Euler Bernoulli beam like structure through the knowledge of changes in the natural frequencies and their measurements. A comparative study has been made between theoretical, experimental as well as the CFBP network analysis. In the theoretical analysis Finite element method has been used for the calculation of natural frequencies with the help of various shape functions as well as the stiffness matrix. The results of natural frequencies, using theoretical analysis, that has been trained to the CFBP network for further analysis. In this network there are three inputs and two outputs. The first three relative natural frequencies are the inputs and the relative crack depth and relative crack location are the outputs. After obtaining the results from the network analysis, it has been validated through experimental analysis. It is found that the results are in close agreement with each other.

Artificial neural network approach for prediction of stress–strain curve of near β titanium alloy

In the present study, artificial neural network (ANN) approach was used to predict the stress–strain curve of near beta titanium alloy as a function of volume fractions of α and β. This approach is to develop the best possible combination or neural network (NN) to predict the stress–strain curve. In order to achieve this, three different NN architectures (feed-forward back-propagation network, cascade-forward back-propagation network, and layer recurrent network), three different transfer functions (purelin, Log-Sigmoid, and Tan-Sigmoid), number of hidden layers (1 and 2), number of neurons in the hidden layer(s), and different training algorithms were employed. ANN training modules, the load in terms of strain, and volume fraction of α are the inputs and the stress as an output. ANN system was trained using the prepared training set (α, 16 % α, 40 % α, and β stress–strain curves). After training process, test data were used to check system accuracy. It is observed that feed-forward back-propagation network is the fastest, and Log-Sigmoid transfer function is giving the best results. Finally, layer recurrent NN with a single hidden layer consists of 11 neurons, and Log-Sigmoid transfer function using trainlm as training algorithm is giving good result, and average relative error is 1.27 ± 1.45 %. In two hidden layers, layer recurrent NN consists of 7 neurons in each hidden layer with trainrp as the training algorithm having the transfer function of Log-Sigmoid which gives better results. As a result, the NN is founded successful for the prediction of stress–strain curve of near β titanium alloy.

Trainable cascade-forward back-propagation network modeling of spearmint oil extraction in a packed bed using SC-CO2

Supercritical extraction (SE) is a separation technique utilizes near or above critical properties of the solvents. In this technique, modeling of yield and solubility of materials are crucial points in supercritical fluid extraction processes. Generally, mathematical modeling of the supercritical oil extraction is a very difficult task since a highly nonlinear relation exists between process variables and solubility. Considering these facts, in the present study, a trainable cascade-forward back-propagation network (CFBPN) was proposed to correlate the yield of spearmint oil extracted by supercritical carbon dioxide. The results revealed the applicability of the proposed model to correlate the yield of spearmint oil extraction with an acceptable level of accuracy. Finally, the obtained results were compared to mathematical models namely Goodarznia & Eikani and Kim & Hong. The comparison between the results of proposed network and mathematical models demonstrated a better predictive capability of the proposed network.

Artificial neural network to predict the effect of thermomechanical treatments on bake hardenability of low carbon steels

Artificial neural network (ANN) is a powerful tool in optimizing many industrial processes. In the present study, an ANN was developed to model and predict the bake hardenability and final yield stress of low carbon steels. Five parameters affecting the bake hardening were considered as inputs, including the annealing/reheating temperature, cooling rate, initial yield stress, work hardening and carbon level. The network was then trained so that to predict the bake hardening amounts and the final yield stress as outputs. A Multilayer cascade-forward back-propagation network is developed and trained using experimental work. Two low carbon steels, St12 and St14, were investigated. The effects of annealing/reheating temperature (500–1000 °C) and subsequent cooling rate (0.5, 5 and 500 °C/s) on the bake hardenability of steels were modeled by ANN as well. In terms of cooling rate, two different behaviors were observed. The bake hardenability of St12 was increased from 13 ± 2, for reheating to 500 °C, to 88 ± 2 MPa for the reheating temperature of 1000 °C. As for St14, these values were respectively 8 ± 2 and 67 ± 2 MPa. The predicted values are in very good agreement with the measured ones indicating that the developed model is very accurate and has the great ability for predicting the bake hardenability.