Gene Expression Programming Model Research Articles

Proposal and evaluation of new models for predicting the FRP contribution to shear strength in reinforced concrete beams using gene expression programming

AbstractFiber-reinforced polymers (FRP) have been widely used in shear strengthening applications of reinforced concrete (RC) beams. The accurate prediction of the FRP contribution to the shear strength of beams is essential for reliable design. Gene expression programming (GEP) has been widely utilized because it reliably expresses complex relationships between experimental variables. In this study, three new GEP models are proposed for three different strengthening configurations of FRP such as fully-wrapping, U wrapping, and side-bonding to predict the FRP contribution to shear strength. These models are developed using the most comprehensive database containing a total of 811 strengthened beams (350 fully-wrapped, 328 U-wrapped, and 133 side-bonded. Many variables have been considered in the proposed GEP models, including those that have been experimentally effective but are often neglected in existing literature equations, such as the shear span-to-effective depth ratio $$(a/d)$$ ( a / d ) and the stirrup ratio ($${\rho }_{w}$$ ρ w ). Additionally, the reliability of existing equations in the literature and the proposed GEP models for predicting the FRP contribution to shear strength was statistically evaluated. As a result of this evaluation, the proposed GEP models for each strengthening configuration of FRP yielded the most accurate statistical results, with the lowest coefficient of variation (COV), and the highest coefficient of correlation (R).

Compressive strength of waste-derived cementitious composites using machine learning

Abstract Marble cement (MC) is a new binding material for concrete, and the strength assessment of the resulting materials is the subject of this investigation. MC was tested in combination with rice husk ash (RHA) and fly ash (FA) to uncover its full potential. Machine learning (ML) algorithms can help with the formulation of better MC-based concrete. ML models that could predict the compressive strength (CS) of MC-based concrete that contained FA and RHA were built. Gene expression programming (GEP) and multi-expression programming (MEP) were used to build these models. Additionally, models were evaluated by calculating R 2 values, carrying out statistical tests, creating Taylor’s diagram, and comparing theoretical and experimental readings. When comparing the MEP and GEP models, MEP yielded a slightly better-fitted model and better prediction performance (R 2 = 0.96, mean absolute error = 0.646, root mean square error = 0.900, and Nash–Sutcliffe efficiency = 0.960). According to the sensitivity analysis, the prediction of CS was most affected by curing age and MC content, then by FA and RHA contents. Incorporating waste materials such as marble powder, RHA, and FA into building materials can help reduce environmental impacts and encourage sustainable development.

A comprehensive analytical and computational assessment of soil water characteristics curves in Atlantic Canada: Application of a novel SelectKbestbased GEP model

For sustainable agriculture, irrigation is one of the most important management practices in a semi-arid climate. Determining the site-specific soil type-based irrigation water requirements requires the development of a soil water characteristics curve (SWCC). This study focuses on exploring the suitability of analytical SWCC models based on the observed data of SWCC developed using the pressure membrane apparatus. A novel SelectKbest gene expression programming (GEP) model is used to create the generalized machine learning SWCC model for Atlantic Canadian soils. The composite soil samples with representative soil textures of loamy sand, loam, and sandy loam are collected from the selected locations in Prince Edward Island and New Brunswick provinces of Canada. The suitability of models at three regions shows that at all sites, including PEI-1 (loam soil) 1, PEI-2 (loamy sand), and NB-3 (sandy loam) if SWCC data has four measured suction potential levels, then McBee & Bomb and Tani exponential models may be selected. Van Genuchten, double exponential, Ruso, Kosuki, and Gardner's four parameters models should be a good choice for five-section potential. Similarly, for six measured suction potential values, the Fredlund & Xing, Omuto, and Van Genuchten models having parameters should be selected for determining the optimum soil hydraulic parameters. The HydroMe program with Brooks and Corey, four, and five parameters Van Genuchten models performed well. The HydroMe program is calibrated and validated with standard RETC for selected regions of Atlantic Canada, which suggests that the HydroMe program is flexible in developing efficient agricultural water management algorithms. Various performance metrics like R2 and RMSE are employed to assess the robustness of the model. A generalized SWCC model (R2=0.976 and RMSE=0.04) is developed using a novel SelectKbest-based GEP modeling approach for Atlantic Canadian soils. These novel approaches can significantly improve site-specific agricultural water management and decision support systems.

Enhancing the exploitation of natural resources for green energy: An application of LSTM-based meta-model for aluminum prices forecasting

Efficient resource allocation for electric car production can be achieved by anticipating aluminum future pricing. For the electric car sector to maintain a steady supply chain, efficient production, and long-term growth, accurate forecasts are essential for policymakers to plan resource policies. The accuracy of ensemble learning models in predicting Aluminum prices was analyzed using Gene Expression programming and long short-term memory (LSTM) method. The analysis was based on monthly frequency spot settlement price data of Aluminum, copper, silver, and crude oil prices, as well as currency inflation rates of USA, China, and Peru. The study period spanned from January 1, 1994 to December 31, 2022, and also included the gross domestic product (GDP) of USA, China, and other metal minerals. Initially, we assess the level of multicollinearity between each chosen input parameter and Aluminum price volatility using Variance inflation factor-based multicollinearity. The primary innovation of this study lies in the creation of a highly precise model that utilizes the LSTM algorithm to estimate monthly Aluminum prices. This model serves as a revolutionary forecasting method. An extensive evaluation of various ensemble learning models reveals that XGBoost is the optimal model for accurately forecasting monthly Aluminum prices. The LSTM-based recurrent neural network model exhibited the lowest error, with a Mean Relative Error (MRE) ranging from 0.001 to 0.008, RMSE ranging from 54.527 to 136.044, and R2 ranging from 0.95 to 0.983. The findings demonstrated that the LSTM-based super learner models had superior performance compared to the CatBoost, LightBoost, Random Forest, XGBoost, and AdaBoost models, as evidenced by higher values of the determination coefficient (R2), value account for (VAF), and other 10 error analysis evaluators. Therefore, the utilization of an LSTM-based super learner model can serve as a dependable approach for forecasting future monthly Aluminum prices. In addition, this study introduces a logical mathematical model utilizing gene expression programming (GEP) to anticipate future Aluminum prices, which can be utilized by other researchers. Thus, we conclude that both the GEP and ensemble learning models, particularly the LSTM-based Super learner model, are appropriate for precise metal price predictions, which can benefit policymakers and assist in resource policy planning.

Modeling and future projection of streamflow and sediment yield in a sub-basin of Euphrates River under the effect of climate change

Abstract Recognizing the differential impacts of climate change across geographical scales, this study emphasizes the importance of statistical downscaling. Using Gene Expression Programming (GEP) and Linear Genetic Programming (LGP), statistical downscaling transforms broad climate trends into region-specific insights. This allowed for detailed analyses of anticipated changes in sediment yield and discharge within a Euphrates River sub-basin in Türkiye using large-scale variables from the CanESM2 model. The dataset is divided into calibration (1970–1995) and validation (1996–2005) periods. To assess the models’ accuracy, statistical measures such as RMSE, MAE, and R were used. The analysis revealed that LGP outperformed GEP in both discharge and sediment yield during validation, with RMSE = 51.79 m3/s and 4,325.66 tons/day, MAE = 27.14 m3/s and 1,593.34 tons/day, NSE = 0.684 and 0.627, and R = 0.841 and 0.788, respectively. However, when simulating future periods based on the observed period (2006–2020), the GEP model was superior to LGP under RCP2.6, RCP4.5, and RCP 8.5 scenarios from CanESM2. In 2021–2100, models suggest a moderate decrease in discharge and sediment yield, indicating potential shifts in the basin's hydrodynamics. These changes could disrupt hydropower generation, challenge water management practices, and alter riverine ecosystems. The results necessitate a thorough assessment of potential ecological consequences.

Exploring the Potential of Machine Learning in Stochastic Reliability Modelling for Reinforced Soil Foundations

This study introduces a novel application of gene expression programming (GEP) for the reliability analysis (RA) of reinforced soil foundations (RSFs) based on settlement criteria, addressing a critical gap in sustainable construction practices. Based on the principles of probability and statistics, the soil uncertainties were mapped using the first-order second-moment (FOSM) approach. The historical data generated via a parametric study on a validated finite element numerical model were used to train and validate the GEP models. Among the ten developed GEP frameworks, the best-performing model, abbreviated as GEP-M9 (R2 = 0.961 and RMSE = 0.049), in the testing phase was used to perform the RA of an RSF. This model’s effectiveness in RA was affirmed through a comprehensive evaluation, including parametric sensitivity analysis and validation against two independent case studies. The reliability index (β) and probability of failure (Pf) were determined across various coefficient of variation (COV) configurations, underscoring the model’s potential in civil engineering risk analysis. The newly developed GEP model has shown considerable potential for analyzing civil engineering construction risk, as shown by the experimental results of varying settlement values.

Application of group method of data handling and gene expression programming to modeling molecular diffusivity of CO2 in heavy crudes

Successful enhancement of heavy oil and bitumen production using CO2-based enhanced oil recovery (EOR) techniques requires extensive knowledge about the diffusivity of CO2 in heavy crudes. In this work, two advanced correlative approaches, namely group method of data handling (GMDH) and gene expression programming (GEP), were utilized to establish simple-to-use mathematical correlations by applying a comprehensive dataset including 260 laboratory data of CO2 diffusion coefficient (DC) in heavy crudes and taking into account all the parameters affecting the accuracy of the models. Moreover, the response surface and contour plots were analyzed to obtain the desired response values and operating conditions. The GMDH correlation represented more reliable results with better statistical criteria including average absolute percentage error (AAPRE) values of 7.47%, 7.83%, and 7.54% for training, testing, and the total sets, respectively. Also, the results showed that the increase in pressure and/or temperature caused an increment in the amount of diffusivity of CO2 in heavy crudes; and the maximum amount of CO2 DC was obtained at the upper limit of these operational parameters. At a given temperature and pressure, it appears that the highest CO2 DC in heavy crudes was obtained when the CO2 mass fraction was less than 0.02. Furthermore, the Pearson and Spearman correlation coefficients were used to investigate linear and non-linear relationships between input variables and the responses of GMDH and GEP correlations. The mass fraction of CO2, temperature, pressure, and temperature_PVT had a direct relationship, and on the other hand, the viscosity and density of crudes had a reverse relationship with CO2 DC. Finally, the reliability of the data bank of CO2 DC in heavy crudes along with the high validity of GMDH and GEP correlations was verified by the leverage technique. Notably, no out-of-leverage data were identified, with only seven and five data points considered as suspected for GMDH and GEP models, respectively.

A prediction model for plastic hinge length of rectangular RC columns using gene expression programming

In the reinforced concrete (RC) columns which are exposed extreme loads such as earthquake effects, the plastic hinge length can be defined as the length of the region where flexural moments exceed the yielding capacity, and the plastic deformations are concentrated. More accurate estimation of plastic hinge length increases the reliability of the seismic design. However, a sensitivity prediction of plastic hinge length is difficult due to a large number of model parameters. Therefore, this study aims to predict the plastic hinge length using the gene expression programming (GEP). An experimental database of 133 RC columns gathered from the literature was utilized for prediction with GEP. The results of GEP model are statistically compared with those of 13 models existing in the literature proposed by various researchers. The comparison results reveal that the proposed GEP-based formulation has the best efficiency among all models. Furthermore, a sensitivity analysis and parametric study are conducted to identify the most influential parameters affecting the GEP formulation.

Predicting the properties of concrete incorporating graphene nano platelets by experimental and machine learning approaches

Modern infrastructure requirements necessitate structural components with improved durability and strength properties. The incorporation of nanomaterials (NMs) into concrete emerges as a viable strategy to enhance both the durability and strength of the concrete. Nevertheless, the complexities inherent in these nanoscale cementitious composites are notably intricate. Traditional regression models face constraints in comprehensively capturing these intricate compositions. Thus, posing challenges in delivering precise and dependable estimations. Therefore, the current study utilized three machine learning (ML) methods, including artificial neural network (ANN), gene expression programming (GEP), and adaptive neuro-fuzzy inference system (ANFIS), in conjunction with experimental investigation to study the effect of the integration of graphene nanoplatelets (GNPs) on the electrical resistivity (ER) and compressive strength (CS) of concrete containing GNPs. Concrete containing GNPs demonstrated an improved fractional change in resistivity (FCR) and strength. The experimental measures depict that strength enhancement was notable at GNP concentrations of 0.05% and 0.1%, showcasing increases of 13.23% and 16.58%, respectively. Simultaneously, the highest observed FCR change reached −12.19% and −13%, respectively. The prediction efficacy of the three models proved to be outstanding in forecasting the characteristics of concrete containing GNPs. For CS, the GEP, ANN, and ANFIS models demonstrated impressive correlation coefficient (R) values of 0.974, 0.963, and 0.954, respectively. For electrical resistivity, the GEP, ANN, and ANFIS models exhibited high R-values of 0.999, 0.995, and 0.987, respectively. The comparative analysis of the models revealed that the GEP model delivered precise predictions for both ER and CS. The mean absolute error (MAE) of the GEP-CS model demonstrated a 14.51% reduction compared to the ANN-CS model and a substantial 48.15% improvement over the ANFIS-CS model. Similarly, the ANN-CS model displayed an MAE that was 38.14% lower compared to the ANFIS-CS model. Moreover, the MAE of the GEP-ER model demonstrated a 56.80% reduction compared to the ANN-CS model and a substantial 82.47% improvement over the ANFIS-CS model. The Shapley Additive explanation (SHAP) analysis provided that curing age exhibited the highest SHAP score. Thus, indicating its predominant contribution to CS prediction. In predicting ER, the graphene content exhibited the highest SHAP score, signifying its predominant contribution to ER estimation. This study highlights ML's accuracy in predicting the properties of concrete with graphene nanoplatelets, offering a fast and cost-effective alternative to time-consuming experiments.

A decision model based on gene expression programming for discretionary lane-changing move

ABSTRACT This study focuses on modeling Discretionary Lane-Changing (DLC), which accounts for the majority of lane-change moves in traffic flows. A binary decision model for lane-changing moves was improved with the method of Gene Expression Programming (GEP). The decision to prefer GEP is due to its high performance in a variety of engineering solutions in the literature. The GEP model was trained with Next Generation SIMulation (NGSIM) trajectory data gathered at the I-80 Freeway in Emeryville, California, and then tested with data gathered at the U.S. Highway 101 in LA, California. The test results indicate that the model made decisions of “change lane” with 92.98% accuracy, and “do not change lane” with 99.65% accuracy. A sensitivity analysis was also conducted to discover potential limits of the performance of the GEP model. The performance of this model was compared with other high-performance decision models developed with the NGSIM's DLC data in the literature and with TransModeler's gap acceptance model. This comparison indicates that the GEP model is the most successful decision model for discretionary lane-changing moves. The GEP model has a high potential to be applied in DLC decision support systems in (semi-) automated vehicles, as well as traffic simulation software.

Exploring the viability of AI-aided genetic algorithms in estimating the crack repair rate of self-healing concrete

Abstract As a potential replacement for traditional concrete, which has cracking and poor durability issues, self-healing concrete (SHC) has been the research subject. However, conducting lab trials can be expensive and time-consuming. Therefore, machine learning (ML)-based predictions can aid improved formulations of self-healing concrete. The aim of this work is to develop ML models that could analyze and forecast the rate of healing of the cracked area (CrA) of bacteria- and fiber-containing SHC. These models were constructed using gene expression programming (GEP) and multi-expression programming (MEP) tools. The discrepancy between expected and desired results, statistical tests, Taylor’s diagram, and R 2 values were additional metrics used to assess the constructed models. A SHapley Additive exPlanations (SHAP) approach was used to evaluate which input attributes were highly relevant. With R 2 = 0.93, MAE = 0.047, MAPE = 12.60%, and RMSE = 0.062, the GEP produced somewhat worse predictions than the MEP (R 2 = 0.93, MAE = 0.033, MAPE = 9.60%, and RMSE = 0.044). Bacteria had an indirect (negative) relationship with the CrA of SHC, while fiber had a direct (positive) association, according to the SHAP study. The SHAP study might help researchers and companies figure out how much of each raw material is needed for SHCs. Therefore, MEP and GEP models can be used to generate and test SHC compositions based on bacteria and polymeric fibers.

Online state of health estimation for lithium-ion batteries based on gene expression programming

Lithium-ion battery is a kind of energy storage devices with complex internal reaction and many factors affecting its performance. Accurate prediction of its SOH (State of Health) is of great significance to prolong its service life and improve safety performance. However, the current prediction for SOH has the difficulties of selecting health factors and using data-driven methods with opaque mechanisms. In this paper, a data-driven model will be established with the capacity change during aging of lithium-ion batteries as a health indicator to realize precise prediction of capacity degradation. Firstly, the feature parameters were extracted and analyzed from the battery cyclic aging test dataset. Two in-situ nondestructive characterization methods, incremental capacity analysis curve and differential thermal voltammetry curve, were utilized to resolve the evolution paths of the feature parameters. After that, a data-driven battery capacity degradation estimation is realized based on the GEP (Gene Expression Programming) algorithm, the performance is compared with the existing vehicle-end and cloud-end models, and a higher-accuracy SOH stacking model is developed with an increase of less than 1 ms in computational time. The results indicated that the GEP model proposed in this paper has obvious advantages in terms of physical explain-ability, computational efficiency and robustness.

An optimized prediction of FRP bars in concrete bond strength employing soft computing techniques

The precise estimation of the bonding strength between concrete and fiber-reinforced polymer (FRP) bars holds significant importance for reinforced concrete structures. This study introduces a new methodology that utilizes soft computing methods to enhance the prediction of FRP bars' bonding strength. A significant compilation of experimental bond strength tests is assembled, covering various variables. Significant variables that affect bonding strength are found in the study of this database. The prediction process is optimized using soft computing methods, particularly Gene Expression Programming (GEP) and the Multi-Objective Genetic Algorithm Evolutionary Polynomial Regression (MOGA-EPR).The proposed soft computing approaches accommodate complex relationships and optimize prediction accuracy depending on the input variables. Results demonstrate its effectiveness in predicting bond strength and comparing it with existing codes and other models from the literature. The results have shown that the MOGA-EPR and the GEP models have high R2 values between 0.91 and 0.94. The proposed new models enhance the reliability and efficiency of designing and assessing FRP-reinforced concrete.

A comparative study of LightGBM, XGBoost, and GEP models in shear strength management of SFRC-SBWS

Steel fiber reinforced concrete (SFRC) was revealed to have superior mechanical properties compared to plain concrete, especially under shear loads. However, Due to a lack of prediction models for SFRC-Slender Beams without Stirrups (SFRC-SBWS) in the available literature, the application of SFRC-SBWS in reinforced concrete (RC) constructions is limited. This research aims to fill a literature gap by analyzing different Machine Learning (ML) algorithms, including Extreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), and Gene Expression Programming (GEP), to forecast the shear strength of SFRC-SBWS specimens. A database of 333 tested SFRC-SBWS specimens was collected and analyzed. The results revealed the LightGBM and XGBoost algorithms surpassed other studied algorithms with Coefficient of determination (R2) = 95.74% and 93.27%, Root Mean Square Error (RMSE) = 20.04 kN and 25.19 kN, respectively. A GEP-based closed-form model was suggested to forecast the shear strength, considering the most important factors, including beam depth effect, beam width, and longitudinal steel reinforcement ratio. The proposed GEP model has an R2 = 91.12% and RMSE = 101.44 kN, showing a slightly lower prediction accuracy. The effectiveness of the studied models was evaluated from various angles, and the results demonstrate the importance of managing new technological advancements to create reliable support methods. Engineers that use these approaches get a dependable tool for estimating shear resistance, which is crucial in guaranteeing the safety and lifespan of RC structures. The successful combination of classic engineering approaches and revolutionary ML strategies indicates the potential for improving accuracy and reducing bias in prediction-oriented investigations. As the engineering field supports technological improvements, combining robust and efficient prediction models, as described in this paper, can contribute considerably to resilient structure design and safety assurance. This leads to sustainable and robust management of structures with greater confidence in predictions.

Quantitative forecasting of bed sediment load in river engineering: an investigation into machine learning methodologies for complex phenomena

Abstract The intricate calculation of bed sediment load (BSL), which is influenced by hydraulic, hydrological, and sedimentary factors, is vital for informed decision-making in water resource management. Machine learning models, which are gaining popularity due to their accessibility and ability to reveal complex relationships, play a significant role in tackling these challenges. The efficacy of gene expression programming (GEP) models, support vector machines (SVMs), multi-layer perceptron (MLP), and multivariate adaptive regression splines (MARS) has been assessed through measured data of number 540 obtained from six rivers, namely Oak Creek, Nahal Yatir, Sagehen Creek, Elbow River, Jacoby River, and Goodwin Creek from 1954 to 1992. The assessment of model performance has been conducted utilizing root mean square error (RMSE), R2, Nash–Sutcliffe coefficient (NSE), and developed discrepancy ratio (DDR) as indices. Following data normalization within the range of 0–1, the data models underwent training and testing processes with a partition ratio of 80% for training and 20% for testing. Four dimensionless parameters, denoted as Fr = U/√gy, U/U*, Se, and ω = τU/γs√gyDs3, were employed as inputs in the models. The outcomes indicate that they exhibit superior performance compared to other methods, as evidenced by the following metrics in predicting BSL during the test stage: RMSE = 1.4088, NSE = 0.73054, R2 = 0.8729, and maximum QDDR(max) = 1.9564.

Fatigue Life Prediction Model of FRP-Concrete Interface Based on Gene Expression Programming.

Under fatigue loading, the interfacial fatigue life of fiber-reinforced polymer(FRP)-concrete is an important index for the analysis of the fatigue performance of reinforced concrete beams strengthened with FRP materials and the evaluation of the reinforcement effect. To solve the problems of the inconsistent and limited accuracy of existing fatigue life prediction models, gene expression programming (GEP) was used to study the interfacial fatigue life of FRP-concrete. Firstly, 219 sets of interfacial fatigue test data were collected, which included two kinds of reinforcement methods, namely, externally bonded (EB) reinforcement and near-surface-mounted (NSM) reinforcement; secondly, Pearson correlation analysis was used to determine the key factors affecting the fatigue life, and then GEP was used to explore the influence of different input forms on the prediction accuracy of the model. Fatigue life calculation formulas applicable to the two kinds of reinforcement methods, i.e., EB and NSM, were established, and a specific calculation formula was established. The model was subjected to parameter sensitivity analysis and variable importance analysis and was found to reflect the intrinsic relationship between the fatigue life and various factors. Finally, the GEP model was compared with the models proposed by other researchers. Five statistical indices, such as the coefficient of determination and the average absolute error, were selected to assess the model, and the results show that the GEP model has higher prediction accuracy than other models, with a coefficient of determination of 0.819, and indicators such as the average absolute error are also lower than those of the rest of the models.

Local scour depth at piles group exposed to regular waves: On the assessment of expressions based on classification concepts and evolutionary algorithms

An accurate estimation of local scour depth around piles group is inevitably essential to provide stability of marine structures. Over the past decades, many investigations have been made to understand the scouring process at piles group exposed to waves for field and experimental scales. This study aims to predict the wave-induced local scour depth by various robust Data Driven Models (DDMs) and Machine Learning Models (MLMs) developed by classification and evolutionary concepts: Model Tree (MT), Evolutionary Polynomial Regression (EPR), Multivariate Adaptive Regression Spline (MARS), and Gene-Expression Programming (GEP). From relevant literature, 125 data series were employed to provide empirical relationships for scour depth predictions. The raw variables were related to bed sediment, pile configuration, pile geometry, approaching flow, and wave characteristics. Non-dimensional parameters have been obtained through the Buckingham theorem in order to control local scour depth. In this way, the ratio of spacing between piles to pile diameter (G/D), sediment number (Ns), pile arrangement number (ratio of the number of piles parallel to the flow; m; to the number of piles normal to the flow; n), Shields parameter (θ), and Keulegan-Carpenter (KC) were considered. From the training and testing stages of Artificial Intelligence Models (AIMs), it was found that regression equation given by MARS model provided better performance (e.g., Correlation Coefficient [R] = 0.9297, Root Mean Square Error [RMSE] = 0.3489, and Scatter Index [SI] = 0.2765) than other AI models’ efficiency. Additionally, the performance of AI models was assessed for various ranges of dimensionless parameters (i.e., G/D, Ns, and θ) versus KC variation. MARS model had the best performance for G/D = 0–1 and KC < 10 (R = 0.9917 and RMSE = 0.2198) and 0.4≤θ < 0.5 and 15 < KC < 25 whereas EPR model had promising efficiency in the its highest level for Ns = 1–3 and 10 < KC ≤ 15 (R = 0.9941 and RMSE = 0.0771) than other AI models. For the sensitivity analysis, the mutation rate and number of chromosomes are examined for the GEP model, the K-fold number and number of parameters for the MARS model, and the number of algebraic terms and number of chromosomes for the EPR model. Furthermore, ranking analysis of AI models indicated that MARS model had the best performance (Ranking Performance Index [RPI] = 0.7348) and followed by GEP (0.3412), M5MT (0.2999), and EPR (0.2848).

Evaluation of water quality indexes with novel machine learning and SHapley Additive ExPlanation (SHAP) approaches

Water quality indexes (WQI) are pivotal in assessing aquatic systems. Conventional modeling approaches rely on extensive datasets with numerous unspecified inputs, leading to time-consuming WQI assessment procedures. Numerous studies have used machine learning (ML) methods for WQI analysis but often lack model interpretability. To address this issue, this study developed five interpretable predictive models, including two gene expression programming (GEP) models, two deep neural networks (DNN) models, and one optimizable Gaussian process regressor (OGPR) model for estimating electrical conductivity (EC) and total dissolved solids (TDS). For the model development, a total of 372 records on a monthly basis were collected in the Upper Indus River at two outlet stations. The efficacy and accuracy of the models were assessed using various statistical measures, such as correlation (R), mean square error (MAE), root mean square error (RMSE), and 5-fold cross-validation. The DNN2 model demonstrated outstanding performance compared to the other five models, exhibiting R-values closer to 1.0 for both EC and TDS. However, the genetic programming-based models, GEP1 and GEP2, exhibited comparatively lower accuracy in predicting the water quality indexes. The SHapely Additive exPlanation (SHAP) analysis revealed that bicarbonate, calcium, and sulphate jointly contribute approximately 78 % to EC, while the combined presence of sodium, bicarbonate, calcium, and magnesium accounts for around 87 % of TDS in water. Notably, the influence of pH and chloride was minimal on both water quality indexes. In conclusion, the study highlights the cost-effective and practical potential of predictive models for EC and TDS in assessing and monitoring river water quality.

Estimation of land subsidence using coupled particle swarm optimization and genetic algorithm: The case of Damghan aquifer

Abstract Land subsidence, which is mainly caused by over-extraction of groundwater, is one of the most important problems in arid and semi-arid regions. In the present study, seven factors affecting the land subsidence, i.e., types of subsoil, land use, pumping, recharge, thickness of the plain aquifer, distance to the fault, and groundwater depletion were considered as input data for the ALPRIFT framework and intelligence models to map both Subsidence Vulnerability Index (SVI) and prediction of land subsidence, respectively. The hybrid of particle swarm optimization (PSO) and genetic algorithm (GA) (Hybrid PSO-GA) was then used to optimize the weights of the input layers and the estimation of the land subsidence. The capability of the PSO-GA at predictions of land subsidence compared with the typical GA model, and Gene Expression Programming (GEP). The statistical indices R2, RMSE, and MAE were used to assess the accuracy and reliability of the applied models. The results showed that the Hybrid PSO-GA model had R2, RMSE, and MAE equal to 0.91, 1.11 (cm), and 0.94 (cm), respectively. In comparison with the GA, and GEP models, the Hybrid PSO-GA model improved the prediction of land subsidence and reduced RMSE by 24.30 and 16.80%, respectively.

Assessment of the split tensile strength of fiber reinforced recycled aggregate concrete using interpretable approaches with graphical user interface

Incorporating recycled concrete aggregates into concrete represents a sustainable approach to mitigate the extraction of natural mineral resources and alleviate the adverse environmental effects associated with the concrete industry. Nevertheless, it encounters challenges due to the vulnerability of the hardened mortar adhering to natural aggregates. Thus, leading to an increased susceptibility to cracking and reduced strength. Therefore, this study utilized machine learning (ML) methods such as gene expression programming (GEP), random forest regression (RFR), artificial neural networks (ANN), ANN-Bagging, and ANN-Boosting models to forecast the split tensile strength (STS) of fiber-reinforced recycled aggregate concrete (FRRAC). To develop the models, 257 data points were gathered from experimental studies, encompassing ten influential input variables and considering the STS as the output variable. The accuracy of the models was assessed using statistical metrics. The five developed prediction models showcased excellent and reliable performance, with an impressive correlation coefficient (R) exceeding 0.80 during training and reaching 0.92 in testing. However, the GEP model outperformed the remaining four models, with an excellent R-value of 0.980 in training and 0.983 in testing. In addition, the standalone ANN model demonstrated performance on par with the ensemble ANN-Boosting model. In contrast, the ANN-Bagging ensemble model outperformed the individual ANN model, showcasing superior prediction accuracy with impressive R-values of 0.952 and 0.975 for training and testing, respectively. Moreover, the feature analysis based on SHapley Additive exPlanations (SHAP) revealed that fiber content, recycled aggregate density, cement, and water have the highest contribution in estimating the STS of FRRAC. Furthermore, a user-friendly graphical interface (GUI) has been developed to simplify the application of machine learning models in predicting the strength properties of concrete. In conclusion, the findings can encourage the utilization of FRRAC in both structural and non-structural reinforced concrete elements. Therefore, promoting the recycling and reuse of construction waste into building materials and thereby reducing environmental impact. Furthermore, incorporating recycled concrete aggregate in construction projects has the potential to decrease construction material costs, as it reduces reliance on natural resources such as natural coarse aggregate.