Proposed Least Square Support Vector Machine Model Research Articles

Machine-learning predictions of solubility and residual trapping indexes of carbon dioxide from global geological storage sites

Ongoing anthropogenic carbon dioxide (CO2) emissions to the atmosphere cause severe air pollution that leads to complex changes in the climate, which pose threats to human life and ecosystems more generally. Geological CO2 storage (GCS) offers a promising solution to overcome this critical environmental issue by removing some of the CO2 emissions. The performance of GCS projects depends directly on the solubility and residual trapping efficiency of CO2 in a saline aquifer. This study models the solubility trapping index (STI) and residual trapping index (RTI) of CO2 in saline aquifers by applying four robust machine learning (ML) and deep learning (DL) algorithms. Extreme learning machine (ELM), least square support vector machine (LSSVM), general regression neural network (GRNN), and convolutional neural network (CNN) are applied to 6811 compiled simulation records from published studies to provide accurate STI and RTI predictions. Employing different statistical error metrics coupled with supplementary evaluations, involving score and robustness analyses, the prediction accuracy of the models proposed is comparatively assessed. The findings of the study revealed that the LSSVM model delivers the lowest RMSE values: 0.0043 (STI) and 0.0105 (RTI) with few outlying predictions. Presenting the highest STI and RTI prediction scores the LSSVM is distinguished as the most credible model among all the four models studied. The models consider eight input variables, of which the time elapsed and injection rate displays the strongest correlations with STI and RTI, respectively. The results suggest that the proposed LSSVM model is best suited for monitoring CO2 sequestration efficiency from the data variables considered. Applying such models avoids time-consuming complex simulations and offers the potential to generate fast and reliable assessments of GCS project feasibility. Accurate modeling of CO2 storage trapping indexes guarantees successful geological CO2 storage operation, which is, in fact, the cornerstone of properly controlling and managing environmentally polluting gases.

Modeling stability conditions of methane Clathrate hydrate in ionic liquid aqueous solutions

Mixing ionic liquid (IL) and water can influence the thermodynamic characteristics such as water activity. Hence, some ILs can be utilized as inhibitors for hydrate systems. This work was intended to determine the three-phase equilibria of methane hydrate in aqueous solutions of 32 ILs using the least squares support vector machine (LSSVM), adaptive neuro-fuzzy inference system (ANFIS), and classification and regression tree (CART). The modeling studies on clathrate hydrates in aqueous solutions of ILs are also reviewed. The used databank contains anion groups such as sulphate, dicyanamide, tetrafluoroborate, and halides. The dissociation temperature of methane hydrate in aqueous solutions of ILs is modelled considering the ILs' critical pressure and critical temperature, the pressure of hydrate system, and the concentration of IL in aqueous phase as the independent parameters. All the developed models are found to well represent/predict the methane hydrate dissociation temperatures in aqueous solutions of ILs. The calculated values of average absolute relative deviation for the presented LSSVM, ANFIS, and CART models are equal to 0.08, 0.31, and 0.10, respectively. Hence, the results reveal that the introduced CART and ANFIS models cannot compete with the developed LSSVM approach in representing the dissociation temperatures of the methane + hydrate + IL + water systems studied in this research work. Using the proposed LSSVM model, all the investigated database equilibrium temperatures are within the absolute deviation percent of 0.0-1.0% except the data that show an absolute deviation percent of 1.35%. The findings of this study can help researchers and engineers to better understand the effects of ILs on methane hydrate stability conditions toward effective hydrate management.

Estimation of basis weight, ash content and moisture content in papermaking plants: A comparative study

The papermaking process is a typical nonlinear process with multiple input-output variables, so it is difficult to construct an accurate model for the process. Data-based modeling techniques may be used to establish a reliable paper plant model. In particular, the LSSVM (least-squares support vector machine) can be used to create a highperformance papermaking process model based on operation data. In this paper, we present a paper plant model that can predict three key output variables (basis weight, ash content, moisture content) with four input variables (stock flow, filler flow, speed, steam pressure) using LSSVM. The proposed LSSVM model is compared with other data-based models (the ANN (artificial neural network) model and the state-space model). The LSSVM model turned out to exhibit better estimation performance compared to others.

Comparison of LSSVM model results with artificial neural network model for determination of the solubility of SO2 in ionic liquids

Abstract In this study, the least square support vector machine (LSSVM) as a robust approach along with genetic algorithm (GA) was utilized for prediction of SO2 solubility in ionic liquids (ILs). The proposed model used the pressure, temperature, critical temperature, boiling temperature, critical pressure, critical compressibility factor, and acentric factor as input parameters. To develop the highly accurate model, the 232 data points were randomly split into training and testing sets. This model provides an average absolute relative deviation (AARD) of 4.6%, illustrating good accuracy and validity. In addition, an artificial neural network (ANN) was developed for prediction of SO2 solubility in ionic liquids. A thorough comparison between the proposed LSSVM and ANN models was conducted through statistical and graphical tools demonstrating the superiority of the former over the latter to predict the solubility of SO2 in ILs. Afterward, a sensitivity analysis was performed on the LSSVM model, elucidating that the pressure and temperature have the greatest effects on the SO2 solubility. Furthermore, the Leverage approach identified that only 2.8% of data points are outside the applicability domain of the proposed LSSVM model, which highlights the model is statistically acceptable and can be used as a predictive approach.

Least squares support vector machine (LS-SVM)-based chiller fault diagnosis using fault indicative features

Fault detection and diagnosis (FDD) of chillers, generally the single most energy consuming piece of building equipment, is an important but hard task where many parameters are involved, and the problem is always quite non-linear. This study proposes a least squares support vector machine (LS-SVM) model optimized by cross validation to implement FDD on a 90-ton centrifugal chiller. Four component-level and three system-level faults were investigated. The effectiveness and efficiency of the eight fault-indicative features extracted from the original 64 parameters have been validated and employed in the detailed discussions. The results indicated that as compared with the other two machine learning methods, the proposed LS-SVM model with optimization showed a better FDD performance in terms of the overall correct rate for all the samples, the individual correct rate for each fault, the diagnostic efficiency, the detection rate and the false alarm rate, etc., especially when it comes to the faults of system level where the detection and diagnosis become rather more difficult due to the system-level effect (SLE)—widespread symptoms caused throughout the system. The correct rates for system-level faults of refrigerant leak/undercharge, refrigerant overcharge and excessive oil were as high as 99.59%, 99.26% and 99.38%, respectively, and the running time was only 36.7% of that of the SVM model.

Implementing PSO-LSSVM model to approximate vaporization enthalpy of pure hydrocarbons and petroleum fractions

The purpose of the present research was to estimate the heat of vaporization for petroleum fractions and pure hydrocarbons by the least-square support vector machine (LSSVM) as a function of the specific gravity, molecular weight, and boiling point temperature. Moreover, a particle swarm optimization technique was applied to determine optimal dependent parameters of LSSVM. In addition, results obtained from the proposed LSSVM model have been compared to some developed correlations by scholars. According to statistical observations, the LSSVM model has acceptable predictions by the value of mean relative deviation and R-squared (R2) equal to 0.51% and 0.9998, respectively. The present predictive manner can be used in petroleum engineering for an accurate approximation of heat of vaporization.

Parameter analysis of atomized droplets sprayed evaporation in flue gas flow

Spraying flue gas desulfurization (FGD) wastewater into the flue duct is a promising technology to achieve zero emission of wastewater by evaporation in thermal power plants. To deal with the scale and corrosion on flue duct walls resulted from the incomplete droplets evaporation, a combined Eulerian-Lagrangian model was developed to reveal the thermo-fluid behavior of the FGD wastewater spray evaporation in the flue gas. The effects of several control factors under various operating conditions were conducted numerically and validated against experimental data. Because of the complex influences of numerous parameters, the Least-Square support vector machine (LSSVM) model based on numerical results is employed to predict the evaporation rate of droplets. The dominant factors, including that of droplets size distribution, flue gas velocity and flow rate, full cone angle and spray direction of nozzle, were analyzed to reveal their impacts on the droplets sprayed evaporation rate and distance, respectively. It is concluded that spaying droplets in the co-flow direction of flue gas can contribute the dispersion of droplets and enhance the relative movement of two-phase flow, achieving the maximum evaporation rate of droplets. An optimized arrangement of multiple nozzles having a small flow rate is proposed to improve the evaporation rate of droplets. The proposed LSSVM model can expeditiously predict the evaporation rate of droplets along the flue duct with high accuracy. The findings can be used to guide the design of spraying FGD wastewater treatment under practical operating conditions in power plants.

Estimating solubility of supercritical H2S in ionic liquids through a hybrid LSSVM chemical structure model

Abstract Development of a predictive tool for H2S solubility estimation can be very helpful in gas sweetening industry. Experimental databases on H2S solubility were rarely available, so as reliable predictive models. Thus, in this study the H2S solubility database was established, and then a Least-Squares Support Vector Machine (LSSVM) approach based on the established database is proposed. Group contribution method was also applied to eliminate the model's dependence on experimental data. Accordingly, our proposed LSSVM model can predict H2S solubility as a function of temperature, pressure, and 15 different chemical structures of Ionic liquids (ILs). Root Mean Square Error (RMSE) and coefficient of determination (R2) are 0.0122 and 0.9941, respectively. Moreover, comparison of our model with other existing models showed its reliability for H2S solubility in ILs. This can be very useful for engineers dealing with gas sweetening process in different applications of analysis, simulation, and designation.

Toward prediction of kinematic viscosity of biodiesel using a robust approach

ABSTRACTEspecially by using a renewable source of fuels such as biodiesel, a large number of high-quality researches have been performed on the reduction of pollution released from fossil fuels. Transesterification process is a common way for the production of biodiesel from vegetable oil, animal fat, and algae oil in the presence of alcohol and catalyst. Viscosity is one of the important physical fuel properties used in the selection of biodiesel. Experimental measurement of viscosity is a time-consuming task. Hence, in this contribution, applicability and performance of two artificial neural network-based models named least square support vector machine (LSSVM) and genetic algorithm-radial basis function (GA-RBF) for the prediction of kinematic viscosity of biodiesel were investigated. Root-mean-square error, coefficient of determination (R2), and average absolute relative deviation of each modeling were reported for each LSSVM and GA-RBF models. Modeling results show that the proposed LSSVM model is more accurate and robust than GA-RBF model.

Applying SVM framework for modeling of CO2 solubility in oil during CO2 flooding

CO2 solubility is one of the most important parameters that affects CO2 flooding, because gas dissolution into crude oil results in oil swelling, viscosity reduction, IFT reduction, oil mobilization, and oil recovery improvement. Therefore, a better understanding of CO2 solubility mechanisms and its influence on physical properties of crude oil are essential to any effective CO2 flooding process. In this study, Least-Square Support Vector Machine (LSSVM) as a newly established soft computing algorithm is applied for developing a new correlative model for CO2 solubility in both dead and live oil systems. CO2 solubility in dead oil is basically affected by the oil saturation pressure (Ps), oil specific gravity (γ), oil molecular weight (MW), and reservoir temperature (T). Moreover, the impact of bubble point pressure is considered in constructing the LSSVM model for the live oil. A number of statistical quality measures are utilized to assess and demonstrate the superior capability of the newly developed LSSVM model in comparison with the previous empirically derived correlations. The average absolute relative deviation (AARD) and coefficient of determination (R2) of 2.2783% and 0.9933 for the dead oil system, and 1.7432% and 0.9958 for the live oil system, respectively, verify the acceptable accuracy and efficient performance of the proposed LSSVM model over a wide range of dataset used in this study within the range of the used databank. However, the impact of CO2 liquefaction pressure is ignored, the LSSVM model gives the best result. In conclusion, it is worth mentioning that the proposed LSSVM model can serve as an accurate correlative tool for fast and effective estimation of CO2 solubility in both dead and live crude oils.

Prediction of CO2 loading capacities of aqueous solutions of absorbents using different computational schemes

In this communication, carbon dioxide solubility in aqueous solutions of various absorbents (2-amino-2-hydroxymethyl-1,3-propanediol; 2-amino-2-methyl-1-propanol; 2-amino-2-methyl-1,3-propanediol; potassium chloride; monoethanolamine; methyldiethanolamine; 2-piperidineethanol; potassium glycinate; piperazine; sodium glycinate; triethanolamine; potassium carbonate; diethanolamine) is predicted through four different methods namely Multi-layer Perceptron Artificial Neural Network (MLP-ANN), Radial Basis Function Artificial Neural Network (RBF-ANN), Least Square Support Vector Machine (LSSVM), and Adaptive Network-based Fuzzy Inference System (ANFIS). These tools predict CO2 loading capacity as a function of temperature, pressure, mass composition of solution, and average molecular weight of solution. In addition, an extrapolative ability of the proposed LSSVM model has been evaluated for different tertiary aqueous mixtures. Moreover, different statistical analyses have been conducted on the estimated and actual reported data. Among these approaches, results indicate that LSSVM model is more accurate than the other proposed ANFIS, MLP-ANN, and RBF-ANN models. Obtained values of mean squared error (MSE) and R-squared (R2) for the LSSVM, ANFIS, MLP-ANN, and RBF-ANN models are 0.001 and 0.991, 0.011 and 0.888, 0.004 and 0.964, 0.009 and 0.916 respectively. This LSSVM model can help us to have accurate prediction of carbon dioxide solubility in aqueous solutions and it can be utilized instead of thermodynamic approaches which have difficult concepts and calculations.

Effective Thermal Conductivity Modeling of Sandstones: SVM Framework Analysis

Among the most significant physical characteristics of porous media, the effective thermal conductivity (ETC) is used for estimating the thermal enhanced oil recovery process efficiency, hydrocarbon reservoir thermal design, and numerical simulation. This paper reports the implementation of an innovative least square support vector machine (LS-SVM) algorithm for the development of enhanced model capable of predicting the ETCs of dry sandstones. By means of several statistical parameters, the validity of the presented model was evaluated. The prediction of the developed model for determining the ETCs of dry sandstones was in excellent agreement with the reported data with a coefficient of determination value (\({R}^{2})\) of 0.983 and an average absolute relative deviation of 0.35 %. Results from present research show that the proposed LS-SVM model is robust, reliable, and efficient in calculating the ETCs of sandstones.

An intelligent modeling approach for prediction of thermal conductivity of CO2

In the design of a carbon dioxide capture and storage (CCS) process, the thermal conductivity of carbon dioxide is of special concern. Hence, it is quite important to search for a quick and accurate determination of thermal conductivity of CO2 for precise modeling and evaluation of such a process. To achieve this aim, a robust computing methodology, entitled least square support vector machine (LSSVM) modeling, which is coupled with an optimization approach, was used to model this transport property. The model was constructed and evaluated employing a comprehensive data bank (more than 550 data series) covering wide ranges of pressures and temperatures. Before constructing the model, outlier detection was performed on the whole data bank to diagnose and delete erroneous measurements and doubtful data from the experimental dataset. It was found that the proposed LSSVM model had a very accurate prediction of thermal conductivity of CO2 with an average absolute relative error of 0.79% and a coefficient of determination of 0.999. In addition, more than 90% of the experimental data points were estimated with an absolute relative error smaller than 2% by the developed model.

Vapor liquid equilibrium prediction of carbon dioxide and hydrocarbon systems using LSSVM algorithm

Many supercritical processes, like monomer separation depends crucially on VLE data. The need of simple, robust and general method, which can overcome deficiencies of EOSs, especially in critical regions, is obvious. In this study, a mathematical algorithm based on Least-Squares Support Vector Machine (LSSVM) has been developed for simulating 425 VLE data of seven CO2/hydrocarbon binary mixtures in supercritical or near critical conditions. The target value, bubble point/dew point pressure, is considered as a function of reduced temperature, hydrocarbon mole fraction and the hydrocarbons acentric factor and critical pressure. The proposed LSSVM model with its magnificent R2 of 0.9932 and AARD% of 3.61 is proving able to predict VLE data of CO2/hydrocarbon binary mixture in a very precise manner. In addition, comparison of LSSVM with EOSs indicates its supremacy over conventional methods. A sensitivity analysis, with three different methods, was performed on the independent variables in an effort to determine the relative importance of each one. At the end with the aid of Leverage statistical algorithm, the statistical validity of the model was guaranteed and proved that the majority of the data points are in the applicability domain of the proposed LSSVM.

Prediction of phase equilibrium of CO2/cyclic compound binary mixtures using a rigorous modeling approach

Vapor liquid equilibrium (VLE) data has significant role in designing processes which include vapor and liquid in equilibrium. Since it is impractical to measure equilibrium data at any desired temperature and pressure, particularly near critical region, thermodynamic models based on equation of state (EOS) are usually used for VLE estimating. In recent years due to the development of numerical tools like artificial intelligence methods, VLE prediction has been find new alternatives.In the present study a novel method called Least-Squares Support Vector Machine (LSSVM) used for predicting bubble/dew point pressures of binary mixtures containing carbon dioxide (CO2)+cyclic compounds as function of reduced temperature of the system, critical pressure, acentric factor of the cyclic compound, and CO2 composition. A 333 binary equilibrium data points of CO2 and six cyclic compounds within temperature and pressure ranges of 308.15–473.15K and 0.5–27.71MPa were used to develop the model. Results show that the proposed model is able to predict VLE data for binary systems containing supercritical or near-critical CO2/cyclic compounds with an acceptable average absolute relative deviation percent (AARD%) of 3.9381% and the coefficient of determination (R2) value of 0.9980. For detection of the probable doubtful experimental data, and applicability of the model, the Leverage statistical approach performed on the data sets. This algorithm showed that the proposed LSSVM model is statistically valid for VLE prediction and the whole phase equilibrium data points are in applicability domain of the model.

Modelling of a new solar air heater through least-squares support vector machines

This paper reports on a modelling study of new solar air heater (SAH) system efficiency by using least-squares support vector machine (LS-SVM) method. In this study, a device for inserting an absorbing plate made of aluminium cans into the double-pass channel in a flat-plate SAH. A SAH system is a multi-variable system that is hard to model by conventional methods. As regards the LS-SVM, it has a superior capability for generalization, and this capability is independent on the dimensionality of the input data. In this study, a LS-SVM based method was intended to adopt SAH system for efficient modelling. For modelling, different mass flow rates in flow duct and collector types are used and then for obtaining the optimum LS-SVM parameters, such as regularization parameter, and optimum kernel function and parameters, several tests have been carried out. The performance of the proposed methodology was evaluated by using several statistical validation parameters. It is found that root mean squared error (RMSE) value is 0.0024, the coefficient of multiple determinations ( R 2) value is 0.9997 and coefficient of variation (cov) value is 2.1194 for the proposed radial basis function (RBF)-kernel LS-SVM method at 0.03 kg/s air mass flow rate. It is found that RMSE value is 0.0135, R 2 value is 0.9991 and cov value is 2.9868 for the proposed RBF-kernel LS-SVM method at 0.05 kg/s air mass flow rate. Comparison between predicted and experimental results indicates that the proposed LS-SVM model can be used for estimating the efficiency of SAHs with reasonable accuracy.