Artificial Neural Network Methodology Research Articles

Application of artificial neural networks in predicting the performance of ice thermal energy storage systems

Efficient prediction of thermal system performance is crucial for optimizing building energy systems. This paper introduces a predictive model to forecast Heating, Ventilation, and Air Conditioning (HVAC) systems' performance with Ice Thermal Energy Storage (ITES). The indicators encompass storage temperature, Coefficient of Performance (COP), cooling thermal load, and chiller power consumption. These forecasts play a pivotal role in cost-effective experimentation, energy optimization, reduced consumption, and minimized operational costs. The study focuses on developing an Artificial Neural Network (ANN) model using a shallow Neural Net Fitting application (nftool). This ANN model predicts necessary ITES system configurations based on nonlinear input variables. Utilizing a comprehensive dataset, incorporating dry bulb temperature, component inlet and outlet temperatures, flow rates, and control valve states, the ANN model is built and validated. Performance metrics, including Mean Squared Errors (MSE) and Regression (R), are employed for training, validation, and testing to predict individual configurations. The results reveal R values ranging from 0.94 to 0.99, with MSE mostly below 20 %. This study exemplifies machine learning's application to ITES cooling systems, serving as a benchmark for evaluating ANN methodologies' effectiveness in prediction. It sheds light on parameter variation's impact on ITES cooling systems and underscores ANN's accuracy in predicting HVAC system parameters with ITES.

Application of the Artificial Neural Network Algorithm to Predict the Realization of the Duty Tax on the Name of Motor Vehicles in Lampung Province

Regional taxes, specifically the Motor Vehicle Name Return Tax (BBNKB), provide the primary source of revenue for regions from the several forms of taxes. The BBNKB tax is crucial in funding government and regional development due to its significant annual growth, encompassing four-wheeled and two-wheeled vehicles. Furthermore, the BBNKB tax catalyzes regional economic expansion and significantly contributes to the government's income. Hence, predicting and forecasting the BBNKB Tax in Lampung Province is necessary to monitor future tax rate fluctuations. That will enable the government to devise innovative tax payment systems and establish tax revenue targets. This study utilizes the Artificial Neural Network (ANN) methodology, using many approaches for distributing training and testing data to forecast. In addition, we utilize hyper-tuning on several factors to obtain the most favourable configurations. The ideal model achieved has a training data allocation of 80% and a testing data allocation of 20%. It was trained for 50 epochs and used a batch size of 16. The model has exceptional predictability, attaining an accuracy rating of 96.51%. Additionally, it showcases a low Root Mean Square Error (RMSE) of 0.246 and a minimal Mean Absolute Percentage Error (MAPE) score of 3.48%. Therefore, it is appropriate to predict the next two-year term. As a result, the forecast for the amount of tax collected from motor vehicle name returns in Lampung has fluctuated.

RSM and ANN methodologies in modeling the enhanced biodiesel production using novel protic ionic liquid anchored on g-C3N4@Fe3O4 nanohybrid

Herin, a new nanohybrid acid catalyst was fabricated for the efficient biodiesel production. At the first, magnetic porous nanosheets of graphitic carbon nitride (g-C3N4@Fe3O4) was prepared and then functionalized with sulfonic acid. Next, the preparation of the catalyst was completed by mixing this surface modified support with n-methyl imidazolium butyl sulfonate zwitterion to achieve non-covalent immobilized acidic ionic liquid on g-C3N4@Fe3O4 support. The catalyst underwent characterization through various techniques such as 1H and 13C NMR, FTIR, SEM, TEM, TGA, EDX and BET which revealing that the magnetic support loaded acidic ionic liquids via a robust charge interaction effect enabling the one-pot production of biodiesel from low-quality oils. Furthermore, the catalyst could be simply recovered using a permanent magnet and reused multiple times without a significant decline in catalytic activity. Consequently, the solid catalyst based on ionic liquids holds promise for the sustainable and eco-friendly production of biodiesel from low-quality oils. Furthermore, Response Surface Methodology (RSM) and Artificial Neural Networks (ANN) were used to model the yield and various process parameters. The findings underscore the enhanced predictive capabilities of ANN in comparison to RSM.

High throughput screening of promoted bimetallic catalysts supported on alumina for production of light olefins via Fischer-Tropsch synthesis with artificial neural network and response surface methodology

The production of light olefins via Fischer-Tropsch Synthesis (FTS) was investigated by both experimental and modeling studies. Experimental studies were conducted by employing promoted FeCo/α-Al2O3 (FeCo) bimetallic catalysts. The Response Surface Methodology (RSM) and Artificial Neural Network (ANN) methods were used in modeling. Specifically, effects of the types and concentration of promoters on the catalytic performances of α-Al2O3 supported iron-cobalt bimetallic catalysts are investigated by using Ce, Ni, La, Ho, Ga Cu, Mn, and Zn used as promoters. A total of 49 catalysts were prepared by co-impregnation method and tested at 310 °C and 1 bar in a high-speed catalyst performance test system (HT-CPA). The FTS performances of catalysts showed that both the types and amounts of the promoters can affect light olefin production. The light olefin (C2=–C3=) production of unpromoted FeCo catalyst was about 3.87 × 10−3 (mol C/g active metal. h). The best performances were obtained from FeCo catalysts promoted with Ho, Cu, and Zn. Presence of these promoters increased olefin production up to ∼ 84 %, 76 % and 70 %, respectively.Performance parameters (MSE and R2) of RSM and ANN models revealed that both methods can be used to support experimental studies and provide a procedure for optimizing catalyst formulations yielding the desired product selectivity in FTS processes. The RSM outcomes aligned with experimental findings, suggest that that Ho, Cu, and Zn can serve as suitable promoters for the light olefin production in FTS. ANN model attained higher performance parameters (R2 = 0.95 and MSE = 2E−7) than RSM model (R2 = 0.85 and MSE = 3.4E−7). This indicates that ANN model, which designed as one hidden layer including 7 neurons, can provide better predictions and a fitting capacity. Therefore, it may be used in further optimization and sensitivity analysis approaches. Alignment degree between the experimental and modeling results demonstrates that the proposed methodology can provide a reliable and efficient way to guide experiments to optimize catalyst formulations through identification of potential promoters and their amounts to be used.

Harnessing the power of artificial neural networks methodology and multi-objective optimization for enhanced yield and bioactivity of plants polyphenolic compounds

The extraction of polyphenolic compounds from plants is crucial in the industrial production of functional nutraceuticals, but traditional methods often yield low and variable results. In this research, an innovative strategy for optimizing polyphenol extraction from two plants Cistus creticus L. and Ephedra alata subsp. alenda (Stapf) Trab., known for their rich composition in polyphenols and their bioactivities, using Ultrasound-Assisted Extraction in conjunction with artificial neural networks (ANNs) and multi-objective optimization is presented. ANNs were trained to model the intricate relationships among UAE parameters, including solvent concentration, temperature, and time, and the outcomes, encompassing polyphenol yield and bioactivity. Multi-objective optimization techniques were subsequently applied to identify extraction conditions that maximize both yield and bioactivity simultaneously. Results validate the accuracy of the ANNs model in predicting polyphenol yields and the significant enhancement in extraction efficiency and bioactivity achieved through multi-objective optimization. The extracts prepared in the optimal conditions have demonstrated superior antioxidant activities, compared to the non-optimized extracts, with the smallest values of IC50 of 242,378 µg/mL, and 146,736 µg/mL for the plants Ephedra alata subsp alenda (Stapf) Trab. and Cistus creticus L. respectively. This study introduces a promising approach for elevating the extraction of plant-derived polyphenols, augmenting their bioactivity with ANNs and multi-objective optimization. In light of the obtained results, it is recommended that further research explore the scalability and applicability of the presented innovative strategy in larger-scale industrial settings. Considering the demonstrated success in optimizing polyphenol extraction from Cistus creticus L. and Ephedra alata subsp. alenda (Stapf) Trab., extending the application of Ultrasound-Assisted Extraction, coupled with artificial neural networks (ANNs) and multi-objective optimization, to other plant species could offer valuable insights. Additionally, investigating the economic feasibility and environmental impact of implementing this strategy on an industrial scale would contribute to its practical viability.

Surrogate modeling with non-stationary-noise based Gaussian process regression and K-Fold ANN for systems featuring uneven sensitivity distribution

In the realm of aircraft design, there is a prevalent need for surrogate modeling techniques capable of efficiently and accurately modeling objects with high evaluation costs and uneven sensitivity distributions, such as those encountered in the nonlinear buckling of thin-walled structures and aerodynamics at transonic speeds. However, the non-stationary Gaussian Process Regression (GPR) model requires extensive hyperparameters or judicious assumptions, limiting its applications in such tasks, and the commonly used K-Fold methodology suffers from inherent instability. Addressing these issues, this paper proposes an active learning method for surrogate modeling in systems featuring uneven sensitivity distribution, integrating non-stationary-noise GPR and K-Fold Artificial Neural Network (ANN) methodology, namely NSN-GPR-KF. Instead of relying on a non-stationary kernel function, the algorithm regards noise as a non-stationary factor within a stationary GPR framework, with noise levels estimated via the K-Fold ANN methodology, allowing GPR to efficiently adapt to non-stationary systems while avoiding the aforementioned requirements for non-stationary GPR. Case validations, performed on two test functions with pronounced uneven spatial sensitivity, demonstrate that NSN-GPR-KF outperforms commonly used stationary GPR and pure K-Fold methodology. It synthesizes the exploration capabilities of GPR with the exploitation proficiency of the K-Fold methodology, achieving comparable global accuracy with reduced sample sizes—up to 26 % and 22 % less than the other two algorithms, respectively. The algorithm was successfully applied to predict the buckling load of thin-walled pipe beams, demonstrating accelerated convergence and approximately 25 % greater accuracy compared to pure K-Fold model. These results suggest that NSN-GPR-KF can serve as an efficient and precise modeling tool for engineering tasks with high evaluation costs and complex sensitivity distributions.

Effects of temperature and nanoparticle mixing ratio on the thermophysical properties of GNP–Fe2O3 hybrid nanofluids: an experimental study with RSM and ANN modeling

This study investigated the impact of temperature and nanoparticle mixing ratio on the thermophysical properties of hybrid nanofluids (HNFs) made with graphene nanoplatelets (GNP) and iron oxide nanoparticles (Fe2O3). The results showed that increased temperature led to higher thermal conductivity (TC) and electrical conductivity (EC), and lower viscosity in HNFs. Higher GNP content relative to Fe2O3 also resulted in higher TC but lower EC and viscosity. Artificial neural network (ANN) and response surface methodology (RSM) were used to model and correlate the thermophysical properties of HNFs. The ANN models showed a high degree of correlation between predicted and actual values for all three properties (TC, EC, and viscosity). The optimal number of neurons varied for each property. For TC, the model with six neurons performed the best, while for viscosity, the model with ten neurons was optimal. The best ANN model for EC contained 18 neurons. The RSM results indicated that the 2-factor interaction term was the most significant factor for optimizing TC and EC; while, the linear term was most important for optimizing viscosity. The ANN models performed better than the RSM models for all properties. The findings provide insights into factors affecting the thermophysical properties of HNFs and can inform the development of more effective heat transfer fluids for industrial applications.

Tác động của Thông tư 200 đến mô hình xếp hạng tín dụng sử dụng phương pháp trí tuệ nhân tạo

The study investigates the impact of Circular 200 on the credit rating model for small, medium, and large enterprises in Vietnam during the period 2008-2018, using Artificial Neural Network (ANN) methodology. The research compares credit rating models before and after the implementation of Circular 200 between 2008-2014 and 2015-2018, using data from 39,162 businesses in Vietnam obtained from the Orbis database. Results indicate that NITA, ROE, liquidity, and current payment ratio are important independent variables with significant differences in credit rating models before and after Circular 200. The research method proves useful for investors in analyzing investment risks for informed investment decisions. Vietnamese financial institutions can apply the model to identify specific credit rating issues for borrowers, enabling them to formulate appropriate credit policies and set different interest rates for varying risk levels. Future research directions include (1) enhancing data and variables, (2) improving data analysis methods, and (3) enhancing performance metrics.

A Two-stage Method for Damage Detection in Z24 Bridge Based on K-nearest Neighbor and Artificial Neural Network

In this paper, we propose an effective approach for identifying damages in the Z24 bridge, a large-scale bridge in Switzerland. The dataset of the Z24 bridge is evaluated as a benchmark, reflecting the behavior of a real structure that has been used for numerous studies. However, most of the previous studies have only addressed the issues of updating the model or estimating the location and severity of damages. The core idea behind our proposed approach is to leverage the strengths of two effective Machine Learning (ML) algorithms: K-Nearest Neighbor (KNN) and Artificial Neural Network (ANN), to assess both the location and severity of damages in the Z24 bridge. First, we employ KNN, an unsupervised learning algorithm, for pinpointing the damage location. This strategy proves highly efficient, significantly reducing computation time by circumventing the need for a loss function during KNN training. By adopting this approach, KNN effectively mitigates the risk of encountering local minima in the ANN optimization process. Subsequently, we deploy ANN to determine the damage severity. When compared to previous studies on the Z24 bridge, our proposed method (KNN-ANN) exhibits promising results. Furthermore, our results illustrate that KNN-ANN consistently outperforms traditional ANN methodologies.

Next-generation skin cancer treatment: A quality by design perspective on artificial neural network-optimized cationic ethosomes with bleomycin sulphate

Skin cancer, exacerbated by environmental factors like ozone layer depletion, poses a global health concern. Topical administration of chemotherapy, a promising approach, minimizes systemic toxicity. Bleomycin Sulphate (BLM), effective intravenously, faces challenges due to limited skin penetration. Cationic ethosomes, known for superior skin penetration, address this by enhancing electrostatic interactions with cancer cells. This ensures localized drug delivery, preventing systemic circulation. Employing thin film hydration, we prepared cationic ethosomes with BLM, optimizing through Artificial Neural Network (ANN) and Response Surface Methodology (RSM). Characterization included entrapment efficiency, vesicle size, morphology, and zeta potential. In-vitro drug release and various kinetic models were analyzed. Ex-vivo evaluations, like dermatokinetic studies and fluorescence microscopy, and in-vivo skin sensitivity testing were conducted. Results showed the optimized formulation with 27.1 % entrapment efficiency, 144.3 nm vesicle size, smooth surface morphology, and positive zeta potential. Ex-vivo assessments indicated favorable skin retention and in-vivo tests suggested suitability for topical administration, highlighting its potential in skin cancer treatment. This research offers a promising strategy for localized and effective skin cancer therapy, addressing challenges associated with systemic toxicity and limited drug penetration.

Comparative analysis of conventional optimization techniques with Artificial Neural Network (ANN) and Response Surface Methodology (RSM) models for extracting oil from Chrysophyllum albidum (C. albidum) seed for biodiesel production

Comparative analysis of conventional optimization techniques with Artificial Neural Network (ANN) and Response Surface Methodology (RSM) models for extracting oil from Chrysophyllum albidum (C. albidum) seed for biodiesel production

Simultaneous increase of parameters of an experimental absorption system: Neural network inverse optimization methodology with multi-inputs

The Absorption Heat Transformer (AHT) has become an alternative proposal to weaken the trend of thermal pollution because it contributes to the recovery of residual heat and the use of solar thermal energy for its activation. Recently, compact designs have been developed that share two heat exchangers in one shell to reduce heat losses and save maintenance costs. The evaluation of the performance of AHTs depends on multiple parameters, which have been optimized using models based on ideal conditions. Multivariable optimization with artificial intelligence has considered experimental data to obtain feasible conditions related to the operation of the AHT. However, these models have focused only on determining one parameter at a time, reflecting a limitation. This study aims to develop a new optimization strategy to simultaneously maximize two relevant parameters associated with the efficiency of a compact experimental AHT. The optimization methodology is applied to increment the values of the heat generated in the absorber (QAB) and the Exergy Coefficient of Performance (ECOP) using the same objective function. This objective function is resolved when reaching the desired output parameters by determining multiple optimal conditions of the AHT. Initially, the modeling of the experimental data was carried out using an artificial neural network (ANN) for the diagnosis and prediction of the QAB and the ECOP. The model was validated by comparing the experimental values of QAB and ECOP against predicted values through a linear regression model, with a satisfactory result of R2>0.98. Subsequently, the multivariable inverse artificial neural network methodology for multiple output parameters (ANNim-m) was used and coupled with the Particle Swarm Optimization algorithm (ANNim-m-PSO) to generate the new multivariate optimization strategy and improve QAB and ECOP parameters simultaneously. Finally, 4 random tests with different initial operating conditions were optimized. Based on the optimization of the 4 tests, the QAB heat load was increased by 87.7 %, 54.2 %, 30.79 %, and 22.66 % from initial experimental conditions of 2.7, 9.0, 14.23 and 23.29 kW, respectively. In the case of ECOP, elevations of 28.5 %, 23.4 %, 15.4 %, and 16.18 % were obtained for initial values of 0.3801, 0.455, 0.5180, and 0.5729, respectively. It is determined that the parameters QAB and ECOP have a high sensitivity when the inlet temperature of the absorber of the external circuit (TAB) decreases. The results reveal that it is feasible to use the ANNim-m-PSO model, because the optimization was able to significantly maximize both QAB and ECOP parameters of the 40 kW experimental AHT.

Optimization of biodiesel production parameters for hybrid oil using RSM and ANN technique and its effect on engine performance, combustion, and emission characteristics

The increased need for renewable energy sources and growing concerns about environmental pollution has driven the need for alternative fuels (particularly biodiesel) in the transportation industry. The most used technique for producing biodiesel is transesterification, which involves the oil reacting with alcohol in the presence of a catalyst. This study looks at the usage of artificial neural network (ANN) and response surface methodology (RSM) to optimize biodiesel synthesis factors for a hybrid oil feedstock (30% cottonseed and 70% spirulina microalgae oil). The parameters that were optimized were the methanol/oil ratio, reaction time, and catalyst concentration. Cottonseed oil was combined with spirulina microalgae oil to minimize its higher free fatty acid content to less than 1% and prevent soap formation during the transesterification process. The optimal biodiesel production was 94.37%, with 106.4 minutes of reaction time, 1.612% catalyst loading, and a methanol/oil ratio of 20:1. The optimized findings, when confirmed with experiments, gave a biodiesel yield of 91.03% with an error of 3.6%. To provide vital insights into the potential environmental benefits of using this alternative fuel, the study further investigates how the optimized biodiesel blend influences engine performance and emission characteristics. Testing was done at 1500 RPM and 17.5 CR for maximum loading conditions, diesel, HBD20, and HBD fuel produced 0.2746, 0.2805, and 0.288 kg/kWh of specific fuel consumption (SFC), respectively. The CO2 emissions of HBD and HBD20 were 2.49% and 8.30% lower, respectively, than pure diesel. NOx emissions increased by 14% and 18.3%. The highest cylinder pressure was 66.87 at 363 °C.

Process optimization of chemical looping combustion of solid waste/biomass using machine learning algorithm

Chemical Looping Combustion (CLC) is a carbon capture technology that uses an oxygen carrier to transfer the oxidizing agent to the fuel for combustion. This study used different machine learning algorithms, Artificial neural network and Response surface methodology to estimate the surface region process performance and optimize the process condition for the CLC of different solid fuels waste paper, plastic waste, and sugarcane bagasse blends. Based on the combustion efficiency, CO2 yield and CO2 capture efficiency responses, A high performance correlation (R2 > 0.8) was obtained for all the combustion parameters analyzed. The perturbation plot derived from the RSM analysis indicated that the most significant input parameters include the steam to fixed carbon, blend ratio and the fuel reaction temperature.The CLC process was optimized using RSM. For blends of SCB/WP, the best operating conditions were found to be 800 °C, a solid flow rate of 197.7 kg/h, an oxygen carrier to fuel ratio of 1.1, a steam to fixed carbon ratio of 2.16, and a blend ratio of 1. Similarly, for blends of SCB/PW, the optimal operating conditions were 800 °C, a solid flow rate of 199.4 kg/h, an oxygen carrier to fuel ratio of 1.3, steam to fixed carbon ratio of 2, and a blend ratio of 0.3. The optimum combustion performance was found to be 0.98, 0.78, and 0.96 for SCB/WP and 0.99, 0.62, and 0.96 for SCB/PW, respectively.

Artificial neural network and response surface methodology for modeling reverse osmosis process in wastewater treatment

The reverse osmosis (RO) process is widely utilized to reduce highly hazardous chemicals in wastewater, resulting in decreased electrical conductivity and enhanced usability. Reverse osmosis is regarded as an efficient desalination technique, and a comprehensive understanding of the mathematical modeling correlation in the RO system can contribute to the development of advanced RO systems. This research is concerned with the modeling and optimization of the RO process by leveraging machine learning techniques, such as Artificial Neural Networks (ANN) and Response Surface Methodology (RSM). Specifically, an ANN and RSM employing a central composite design (CCD) were performed to analyze the influence of key parameters, including flow rate (30–70 m3/hr), initial conductivity (2000–4000 µs/cm), feed pressure (13–17 bar), and solution temperature (11–39 °C), on the reduction of total dissolved solids (TDS) represent by conductivity in wastewater treatment. The outcomes derived from the RSM-CCD analysis demonstrated that the optimal conditions for achieving the lowest conductivity of 35 ± 10 µs/cm included a solution temperature of 31.6 °C, feed pressure of 16 bar, flow rate of 40 m3/hr, and an initial conductivity of 3500 µs/cm. Five ANN models have been suggested to evaluate the plant’s performance. Model-5 with two hidden layer, eleven hidden layer nodes (20 and 30 nodes) for first and second layers respectively. Moreover, ANN exhibited excellent performance, with a low (MSE) of < 0.0003 and a high (R2) of more than 0.99. These findings highlight the valuable utilization of RSM and ANN methodologies in the modeling and optimization procedures of the RO process.

Cross-Correlation among Seismic Events, Rainfalls, and Carbon Dioxide Anomalies in Spring Water: Insights from Geochemical Monitoring in Northern Tuscany, Italy

Variations in the CO2 dissolved in water springs have long been observed near the epicenters of moderate and strong earthquakes. In a recent work focused on data collected during the 2017–2021 period from a monitoring site in the Northern Apennines, Italy, we noticed a significant correlation between CO2 anomalies and moderate-to-weak seismic activity. Here, we extended this analysis by focusing on data collected from the same site during a different period (2010–2013) and by integrating the CENSUS method with an artificial neural network (ANN) in the already-tested protocol. As in our previous work, a fit of the computed residual CO2 distributions allowed us to evidence statistically relevant CO2 anomalies. Thus, we extended a test of the linear dependence of these anomalies to seismic events over a longer period by means of binary correlations. This new analysis also included strong seismic events. Depending on the method applied, we observed different time lags. Specifically, using the CENSUS methodology, we detected a CO2 anomaly one day ahead of the earthquake and another anomaly eleven days ahead. However, no anomaly was observed with the ANN methodology. We also investigated possible correlations between CO2 concentrations and rain events and between rain events and earthquakes, highlighting the occurrence of a CO2 anomaly one day after a rain event of at least 10 mm and no linear dependence of seismic and rain events. Similar to our previous work, we achieved a probability gain of around 4, which is the probably of earthquake increases after CO2 anomaly observations.

Modified couple stress and artificial intelligence examination of nonlinear buckling in porous variable thickness cylinder micro sport structures

This investigation focuses on the nonlinear behavior of porosity-dependent functionally graded (FG) truncated conical small-scale structures. The modified coupled stress theory, as well as the energy method, are applied to generate the nonlinear partial differential equations (PDEs) related to buckling analysis of simply supported nonuniform micro-cylindrical structures. The material dispersion is gradually changed along the length of the structures between the Nickel and concrete, while the porosity voids are scattered in the radial direction, and the external radius of the structure decreases along the length direction via nonlinear mathematic equations applicable in sports structures. The PDEs are numerically solved via the generalized differential quadrature method (GDQM) coupled with the numerical iterative technique. In this particular context, the aim is to predict nonlinear results using a newly developed methodology that employs artificial neural networks (ANNs). The predictions generated by this approach will be compared against previously obtained data and validated to ensure their accuracy and reliability. The ANN methodology is expected to provide a more robust and comprehensive framework for predicting nonlinear results, which would be helpful in a variety of settings, from scientific research to engineering applications.

Insights into the prediction of the liquid density of refrigerant systems by artificial intelligent approaches

This study presents a novel model for accurately estimating the densities of 48 refrigerant systems, categorized into five groups: Hydrofluoroethers (HFEs), Hydrochlorofluorocarbons (HCFCs), Perfluoroalkylalkanes (PFAAs), Hydrofluorocarbons (HFCs), and Perfluoroalkanes (PFAs). Input variables, including pressure, temperature, molecular weight, and structural groups, were systematically considered. The study explores the efficacy of both the multilayer perceptron artificial neural network (MLP-ANN) and adaptive neuro-fuzzy inference system (ANFIS) methodologies in constructing a precise model. Utilizing a comprehensive dataset of 3825 liquid density measurements and outlier analysis, the models achieved R2 and MSE values of 0.975 & 0.5575 and 0.967 & 0.7337 for MLP-ANN and ANFIS, respectively, highlighting their remarkable predictive performance. In conclusion, the ANFIS model is proposed as an effective tool for estimating refrigerant system densities, particularly advantageous in scenarios where experimental measurements are resource-intensive or sophisticated analysis is required.

Prediction of the Bond Strength of Externally Bonded FRP Sheets Applied to Concrete via Grooves Technique Using Artificial Neural Networks

The present study aims to fill a gap in the literature on the estimation of the bond strength of fiber reinforced polymer sheets bonded to concrete, via the externally bonded reinforcement on grooves (EBROG) technique, employing the curve-fitting on existing datasets in the literature and the methodology of Artificial Neural Networks (ANNs). Therefore, a dataset of 39 experimental results derived from EBROG technique is collected from the literature. A mathematical equation for the bond strength of FRP sheets applied on concrete via the EBROG technique was suggested using curve-fitting and general regression. The proposed mathematical equation is compared and validated with experimental results. The developed ANN model was constructed after testing diverse hidden layers and neurons to find the optimal predictions. The validation of the model is carried out using the experimental results and a statistical analysis is applied to assess the proposed mathematical equation and the proposed ANN model. Furthermore, a parametric study using the ANN model was also performed to investigate the influence of various factors on the bond strength of FRP sheets bonded to concrete. The parametric study proves that the bond strength increases with increasing the tensile stiffness per width, the FRP sheet width, and the concrete compressive strength; however, the effect of the Groove’s width and depth is found to be not monotonous.

Prediction of purified water quality in industrial hydrocarbon wastewater treatment using an artificial neural network and response surface methodology

This study addresses the critical issue of determining the environmental impact of wastewater discharged from a petroleum complex's wastewater treatment plant (WWTP). The emphasis is on assessing the quality of purified water and its potential environmental consequences. Using the water quality index (WQI) as a comprehensive metric for water quality assessment, the study investigates the effects of key parameters such as pH, temperature, phosphates, and hydrocarbons derived from an oil refining complex WWTP. The aim is to contribute to water resource management and environmental monitoring. Machine learning models were trained on Principal Component Analysis (PCA) data, representing 70 % of the total data. The results show that water quality indices vary significantly across seasons, with maximum values for WQI in autumn (40.66), winter (32.96), spring (37.92), and summer (49.34). The ANN and RSM-BBD models effectively forecast experimental values with high predictive accuracy indicated by determination coefficients (R2). For WQI in autumn, winter, spring, and summer, R2 values for ANN were 0.9897, 0.9834, 0.9909, and 0.9956, respectively, and 0.9279, 0.9378, 0.9802, and 0.98866 for RSM-BBD. While the ANN model outperformed the RSM-BBD model, both demonstrated efficacy in predicting water quality indices. The study adds to our understanding of the ecological implications of discharged water from petroleum complexes by providing valuable insights into seasonal variations in water quality factors.