Landslide susceptibility mapping is a critical task for risk management, yet many existing approaches struggle with limited accuracy and model instability. To address these challenges, this study develops a hybrid Artificial Neural Network (ANN) framework optimized with four metaheuristic algorithms (BHA, COA, MVO, and VSA). The case study is conducted in East Azerbaijan Province, Iran, a region with sufficient landslide records for robust testing. The results show that the optimized ANN models achieved strong predictive performance, with Area Under the Curve (AUC) values ​​exceeding 0.97 across training datasets. Among them, the MVO-MLP and COA-MLP models yielded the highest accuracy, highlighting the advantage of optimization in enhancing model robustness. Overall, the developed models predict landslide occurrence with more than 80% accuracy. These findings suggest that integrating optimization algorithms with neural networks provides a reliable, cost-effective approach for spatial modeling of landslide susceptibility. Furthermore, the proposed framework offers valuable insights for disaster preparedness, risk reduction, and emergency management strategies.

This study develops and compares hybrid artificial intelligence models for predicting the shaft resistance of driven piles using 65 field records. A multilayer perceptron neural network was combined with four metaheuristic optimization algorithms: whale optimization algorithm, artificial bee colony, ant lion optimization, and ant colony optimization. The input parameters included pile length, pile diameter, effective vertical stress, and undrained shear strength, while shaft friction was considered as the target output. The models were trained and tested, and their performance was evaluated using R² and RMSE. The final comparison showed that ALO–MLP provided the best overall performance, with a testing R² of 0.981 and RMSE of 5.26. WOA–MLP also showed strong and reliable prediction ability, ranking second with a testing R² of 0.981 and RMSE of 6.79. ABC–MLP produced competitive results, while ACO–MLP was less accurate and more sensitive to population size. Overall, the findings suggest that ALO–MLP and WOA–MLP can serve as practical and dependable tools for estimating pile shaft resistance in geotechnical design.

Accurate air quality predictions are vital for environmental management and public health, especially in urban areas facing rising pollution levels. This study investigates the potential of hybrid machine learning models based on Multilayer Perceptron neural networks, optimized using three nature-inspired metaheuristic algorithms: Optics-Inspired Optimization, Biogeography-Based Optimization, and the Multiverse Optimizer. The aim is to enhance the prediction of key air quality indicators, carbon monoxide, ozone, nitrogen dioxide, sulfur dioxide, and the Air Quality Index, using long-term air quality measurements collected from Shiraz Province, Iran, from April 2019 to November 2024. These data were used to train and evaluate the models. Each metaheuristic was applied to optimize the Multilayer Perceptron training across varying population sizes, and the performance of the resulting models was evaluated using the root mean squared error and the coefficient of determination on both the training and test datasets. Among the tested models, the hybrid combining the Multiverse Optimizer with the Multilayer Perceptron achieved the best generalization performance, with a test root mean squared error of 6.71 and a test coefficient of determination of 0.9903. The Biogeography-Based Optimization with Multilayer Perceptron model also yielded accurate results, confirming the strength of nature-inspired optimization in neural network training. This study provides a novel comparative framework for air quality forecasting by integrating underexplored metaheuristic algorithms with neural models. The results suggest strong potential for application in smart environmental monitoring systems. Future research could explore hybrid deep learning methods and real-time predictive models using broader environmental datasets.

Accurate prediction of shallow-foundation bearing capacity in layered cohesionless soils remains challenging because footing geometry and the strength contrast between soil layers interact nonlinearly. This study developed and compared four hybrid multilayer perceptron models optimized using biogeography-based optimization, the league championship algorithm, sunflower optimization, and whale optimization. The models used footing width, upper-layer thickness, upper-layer friction angle, and lower-layer friction angle as predictors. A complete factorial database of 2,304 limit-equilibrium analyses was generated for surface strip footings under zero surcharge and divided into 1,613 training, 346 validation, and 345 untouched test cases. Population sizes and model configurations were selected exclusively from validation performance and convergence behavior, while repeated 80/20 holdout analysis within the training set was used to assess stability. Among the finalized models, BBO-MLP provided the strongest overall agreement, achieving a test-set correlation of 0.9834, a standard-deviation ratio of 1.002, and a centered RMSE of 130.19 kPa. SHAP analysis across all 2,304 cases consistently ranked footing width as the dominant predictor, followed by lower-layer friction angle, upper-layer friction angle, and upper-layer thickness.

The advancement of novel data mining and optimization algorithms has significantly enhanced traditional engineering structural analysis models, particularly those based on swarm intelligence. This study delves into refining the neural assessment of shaft friction capacity in driven pile systems by exploring the social behavior of four hybridized algorithms: Wind-Driven Optimization (WDO), Spotted Hyena Optimization (SHO), Grasshopper Optimization Algorithm (GOA), and Moth-Flame Optimization (MFO). Four crucial influencing variables - pile length (m), diameter (cm), effective vertical stress (Sv), and undrained shear strength (Su) - are considered in constructing the requisite dataset. After applying optimized structures, each ensemble undergoes a sensitivity analysis based on its individual swarm size. The predictive precision of the models is compared using the results of two sensitivity analyses. Neural network simulations exhibit improved results with an increased number of neurons in a single hidden layer. The root mean square errors (RMSEs) for the training and test datasets, employing Multilayer Perceptron (MLP)-based solutions, are (0.05241, 0.32861, 0.06155, and 0.03874) and (0.04334, 0.18155, 0.05382, and 0.03626), respectively. In the training and testing datasets for proposed predictive models using WDO, SHO, GOA, and MFO, R-2 values of (0.996, 0.853, 0.992, and 0.997) and (0.985, 0.732, 0.997, and 0.997) were found, respectively. Notably, MFO outperforms its counterparts when integrated with MLP for predicting engineering solutions.

Soil is a critical natural resource, and accurate erosion susceptibility assessment is vital for the optimal management and development of soil resources. Erosion susceptibility assessment is necessary for long-term conservation plans, but the process can be expensive and time-consuming over large areas. It is imperative to examine the impact of water-induced erosion on cultivated lands, as it can cause significant damage. This study evaluates the effectiveness of four data-driven approaches (biogeography-based optimization, earthworm optimization algorithm, symbiotic organisms search, and whale optimization algorithm) combined with artificial neural network models for the assessment of erosion susceptibility. The examined criteria include 14 geographic and environmental criteria, and the data used in a ratio of 70 to 30 for training and testing operations. And its results were measured by AUC values. The evaluation of AUC accuracy indices revealed compelling results. Specifically, in the case of SOS-MLP, the highest AUC values were observed, reaching 0.9973 for test data and 0.9296 for train data. Conversely, for WOA-MLP, the AUC values obtained were slightly lower but still notable, registering at 0.9809 for test data and 0.959 for train data. These values were also calculated for BBO-MLP (0.999 and 0.9327) and EWA-MLP (0.9304 and 0.9296) in the training and testing phases, respectively. Results showed that all four methods could successfully evaluate erosion susceptibility according to AUC values greater than 0.92, especially the BBO-MLP with the highest AUC values. Therefore, the findings of this study have shown that the combined optimization algorithms and Machine Learning used in this research have a suitable ability to optimize the artificial neural network and are very useful for identifying areas sensitive to erosion.

Assessing the seismic susceptibility of urban highway and road networks is crucial for strengthening the most vulnerable parts of the road network in advance and effectively responding to and recovering from infrastructure system damage after a disaster. The paper suggests utilizing building information modeling (BIM) and artificial neural network (ANN) techniques to evaluate the seismic vulnerability of an urban road network. This assessment considers the spatial seismic hazard, the components' vulnerability, and the impact of structural damage on the network's functionality. The current article's primary objective was to assess road networks' susceptibility to earthquake hazards using neural networks and building information modeling (BIM) through a comprehensive and systematic approach. To determine the most precise and effective model, a comprehensive evaluation was conducted comparing BIM and ANN with the inclusion of advanced algorithms such as black hole algorithm (BHA), future search algorithm (FSA), particle swarm optimization (PSO), and wind-driven optimization (WDO). The current study's findings regarding implementing machine learning algorithms suggest that the WDO-MLP method achieved an accuracy of 0.9863. Furthermore, while evaluating the model's efficiency using the Area under the curve (AUC), the WDO-MLP algorithm demonstrated an efficiency of 0.98. The BHA-MLP, VS-MLP, and PSO-MLP algorithms demonstrated prediction accuracies of 0.9752, 0.986, and 0.9646, respectively, in assessing the vulnerability of highway construction. Thus, due to its superior accuracy, the WDO-MLP algorithm is the most precise and efficient method for predicting the vulnerability of highway construction during hazardous events. This algorithm can be precious and effective in aiding planners and policymakers in pre-crisis management decision-making.

Numerical modelling of gabion is challenging. Few studies analyzed and compared the fundamental mechanical properties between gabion and concrete. This paper employs discrete element methods (DEM) and introduces clusters to simulate the initiation and development of aggregate cracks. Basic element models for concrete and gabion are established, and numerical parameters are calibrated experimentally. Subsequently, this paper investigates the failure modes and mechanical properties of concrete and gabion under uniaxial compression loading. Results indicate that, for concrete, peak particle contact forces occur at the axial compression ratio (ACR) corresponding to its peak load, while force distribution in gabion is more dispersed. The force within concrete is one order of magnitude higher than that in gabion. The failure of concrete specimens involves both shear and tensile failure, with shear failure dominating initially and then transitioning to tensile failure (critical threshold is found at ACR=1.44%). In contrast, the failure of gabion is characterized by shear failure only. Furthermore, the energy required for gabion failure is significantly greater than that for concrete failure, indicating superior ductile behavior. This paper fills the research gap and serves as a reference for future numerical studies involving concrete and gabion structures.

Soil erosion, as a natural and complex geomorphological process driven by gravitational forces, wind, and water, results in the displacement and deposition of soil particles, leading to sedimentation in both natural and artificial water bodies, loss of fertile soil, and increased flood risk in vulnerable regions. This study aims to enhance the precision and robustness of soil erosion susceptibility mapping through the optimization of artificial neural networks (ANNs) using four advanced swarm intelligence algorithms: Particle Swarm Optimization (PSO), Black Hole Algorithm (BHA), Evaporation Rate Water Cycle Algorithm (ERWCA), and Vibrating Search Algorithm (VSA). The research was carried out in Kermanshah Province, Iran, integrating 14 environmental, geographical, and structural-spatial criteria that influence erosion processes. The results revealed that hybrid ANN-Swarm Intelligence models demonstrated superior predictive capability, with ERWCA-MLP and VSA-MLP achieving the best performance. The main innovation of this study lies in combining advanced metaheuristic optimization with neural network learning to enhance both the accuracy and interpretability of erosion susceptibility maps, supporting sustainable land and soil resource management.

As is known, evaluating strength degradation characteristics of rock mass of great importance in civil engineering. This research is conducted to assess the influence law of disrupt characteristics and confining pressure on the mechanical characteristics of rock mass. The results show that the strength of rock samples linearly increases with the increase of confining pressure. The strength reduction rate of rock samples decreases more obviously with the increase of fracture length and quantity, and the strength effect is most obvious when the confining pressure increases. With the increase of confining pressure, the failure mode of intact rock samples changes from tension to shear, the rock samples with different lengths/quantities of prefabricated fractures all undergo shear failure, but as the quantity of fractures increases the rock samples ultimately exhibit a "rhombus" failure mode. On this basis, a damage constitutive model was established that can reflect the characteristics of the entire deformation and failure process of fractured rock masses. The model parameters m and F0 can reflect the brittle and strength characteristics of the rock mass. The stress-strain curve and damage evolution curve of the rock mass have a good correspondence with the macroscopic failure process induced by its structural changes. The macroscopic defects formed by prefabricated fractures have a significant impact on the initial damage state of the rock mass, leading to a deterioration of its mechanical properties. All in all, the research results provide important theoretical basis for the prevention and safety evaluation of similar fractured rock engineering disasters.

The primary objective of this study is to predict the pullout capacity of belled piles by developing and evaluating various hybrid modeling techniques, including Evolution Strategy (ES), Moth Flame Optimizer (MFO), Grasshopper Optimization Algorithm (GOA), and League Championship Algorithm (LCA). Each hybrid model combines an artificial neural network (ANN) with an optimization algorithm, trained using a hybrid learning approach that incorporates back-propagation and least squares estimation, implemented in MATLAB. A total of 36 samples were used, with 25 designated for training and 11 for testing. The performance of each model was assessed using statistical metrics, namely the coefficient of determination (R2) and root mean square error (RMSE). Among the models, MFO-ANN demonstrated the highest predictive accuracy, followed by LCA-ANN, ES-ANN, and GOA-ANN, respectively. The results confirm the robustness and reliability of the ANN-based hybrid models in estimating pullout capacity.

Flooding is a devastating natural disaster that causes fatalities and property damage worldwide. Effective flood susceptibility mapping (FSM) has become crucial for mitigating flood risks, especially in urban areas. This study evaluates the performance of artificial neural network (ANN) algorithms for FSM using machine learning classification. Traditional flood prediction models face limitations due to data complexity and computational constraints. This research incorporates artificial intelligence, particularly evolutionary algorithms, to create more adaptable and robust flood prediction models. Four specific algorithms-black hole algorithm (BHA), future search algorithm (FSA), heap-based optimization (HBO), and multiverse optimization (MVO)-were tested for predicting flood occurrences in the Fars region of Iran. These evolutionary algorithms simulate natural processes like selection, mutation, and crossover to optimize flood predictions and management strategies, improving adaptability in dynamic environments. The novelty of this study lies in using evolutionary AI algorithms to not only predict floods more accurately but also optimize flood management strategies. The ANN was trained with geographical data on eight flood-impacting factors, including elevation, rainfall, slope, NDVI, aspect, geology, land use, and river data. The models were validated with historical flood damage data from the Fars area using metrics like mean square error (MSE), mean absolute error (MAE), and the receiver operating characteristic (ROC) curve. Results showed significant improvements in accuracy for BHA-MLP, FSA-MLP, MVO-MLP, and HBO-MLP, with accuracy indices and AUC values increasing. The study concludes that hybridized models offer an effective and economically viable approach for urban flood vulnerability mapping, providing valuable insights for flood preparedness and emergency response strategies.

Artificial neural networks (ANN) have been the focus of several studies when it comes to evaluating the pile's bearing capacity. Nonetheless, the principal drawbacks of employing this method are the sluggish rate of convergence and the constraints of ANN in locating global minima. The current work aimed to build four ANN -based prediction models enhanced with methods from the black hole algorithm (BHA), league championship algorithm (LCA), shuffled complex evolution (SCE), and symbiotic organisms search (SOS) to estimate the carrying capacity of piles in cold climates. To provide the crucial dataset required to build the model, fifty-eight concrete pile experiments were conducted. The pile geometrical properties, internal friction angle phi shaft, internal friction angle phi tip, pile length, pile area, and vertical effective stress were established as the network inputs, and the BHA, LCA, SCE, and SOS -based ANN models were set up to provide the pile bearing capacity as the output. Following a sensitivity analysis to determine the optimal BHA, LCA, SCE, and SOS parameters and a train and test procedure to determine the optimal network architecture or the number of hidden nodes, the best prediction approach was selected. The outcomes show a good agreement between the measured bearing capabilities and the pile bearing capacities forecasted by SCE-MLP. The testing dataset's respective mean square error and coefficient of determination, which are 0.91846 and 391.1539, indicate that using the SCE-MLP approach as a practical, efficient, and highly reliable technique to forecast the pile's bearing capacity is advantageous.

This study evaluates a new hybrid approach for determining homes’ heating load (HL). The crow search algorithm (CSA), heap-based optimizer (HBO), seeker optimization algorithm (SOA), political optimizer (PO), and harmony search (HS) are the five components of the suggested paradigm. A nonlinear analysis of the effects of eight independent factors on the HL was conducted using the best structure identified in each model. The assessment procedure for the HS technique in this study consisted of three parts. The appropriate population size to utilize in the first phase was found to be the one that yields the best coefficient of determination (R2) value and the lowest root mean squared error (RMSE) value. For CSA-MLP, HBO-MLP, SOA-MLP, PO-MLP, and HS-MLP, respectively, the first phase yielded R2 = 0.96473, 0.95618, 0.96931, 0.97048, and 0.96702, and RMSE = 2.57119, 2.85968, 2.40125, 2.3554, and 2.46872. A battery of tests using a range of different nNew values (between 10-100) was applied to the HS-MLP with a population size of 50 in the second phase. The data indicates that the most reliable results are obtained with a nNew-value of 60. For training and testing, this value has RMSE values of 2.61518 and 2.4387 and R2 values of 0.9669 and 0.96783. In the third stage, an experiment with a population size of 50 and nNew of 60 was examined using a range of HMCR values (between 0.5-1.4). Concerning training and testing, the results indicate that the HMCR value 1.1 produces the most reliable results; its R2 values are 0.9739 and 0.97207, and its RMSE values are 2.32691 and 2.27488. Lastly, the results demonstrate that the accuracy of the HS-MLP method has been improved by the 3-phase analysis method.

The use of five optimization techniques for the prediction of a strength -based concrete mixture's best -fit model is examined in this work. Five optimization techniques are utilized for this purpose: Slime Mold Algorithm (SMA), Black Hole Algorithm (BHA), Multi -Verse Optimizer (MVO), Vortex Search (VS), and Whale Optimization Algorithm (WOA). MATLAB employs a hybrid learning strategy to train an artificial neural network that combines least square estimation with backpropagation. Thus, 72 samples are utilized as training datasets and 31 as testing datasets, totaling 103. The multi -layer perceptron (MLP) is used to analyze all data, and results are verified by comparison. For training datasets in the best -fit models of SMAMLP, BHA-MLP, MVO-MLP, VS-MLP, and WOA-MLP, the statistical indices of coefficient of determination (R2) in training phase are 0.9603, 0.9679, 0.9827, 0.9841 and 0.9770, and in testing phase are 0.9567, 0.9552, 0.9594, 0.9888 and 0.9695 respectively. In addition, the best -fit structures for training for SMA, BHA, MVO, VS, and WOA (all combined with multilayer perceptron, MLP) are achieved when the term population size was modified to 450, 500, 250, 150, and 500, respectively. Among all the suggested options, VS could offer a stronger prediction network for training MLP.

A groundwater reservoir is either a solitary aquifer or a network of interconnected aquifers. A particular aquifer’s groundwater purity evaluation could be time-consuming and costly. This study quantified the properties of Na

Estimating and predicting groundwater quality characteristics so that managers may make management decisions is one of the critical goals of water resource planners and managers. The complexity of groundwater networks makes it difficult to predict either the time or the location of groundwater. Many models have been created in this area, offering better management to preserve water quality. Most of these models call for input parameters that are either seldom accessible or expensively and laboriously measured. A better option among them is the Artificial Neural Network (ANN) Model, which draws inspiration from the human brain. This study uses Na + , Mg2 + , Ca2 + , Na

In developing countries, evaluating irrigation water quality using conventional methods can be costly and time-consuming. To overcome these challenges, this study explores the potential of utilizing physical parameters and artificial intelligence (AI) models for predicting and evaluating the quality indicators of irrigation water in aquifer systems. To achieve this goal, novel hybrid methods, namely the Whale Optimization Algorithm (WOA) and Wind-Driven Optimization (WDO), are employed in conjunction with Artificial Neural Network (ANN) models. The specific objective of this study is to forecast the Sodium Adsorption Ratio (SAR) by considering independent variables such as Na+, Mg2+, Ca2+, Na percent, K+, SO42−, Cl−, pH, and HCO3−. A dataset of 540 samples from the Shiraz plain, collected over a statistical period of 16 years (2002–2018), is used to estimate the groundwater quality variables. A pre-processing technique is applied in the AI approach to enhance the model's efficiency. The results indicate that the WDO-ANN model exhibits higher accuracy (R2 = 0.9983 and RMSE = 0.10618) than the WOA-ANN model (R2 = 0.9957 and RMSE = 0.16957). The optimization of computational parameters and comparison of AI model structures demonstrate that the WDO-ANN model outperforms the WOA-ANN model in predictive ability. In general, using AI models as a tool for low-cost and timely prediction of underground water quality using physical parameters as input variables has a high potential.

This research, with a descriptive-analytical approach and in a pragmatic way, examines the evolution of housing architecture after COVID-19. Thirty experts in architecture and urban planning made up its statistical population. A combination of artificial intelligence algorithms, tree data mining techniques, and the FTOPSIS multi-criteria decision analysis method have been applied to the information analysis process. A panel of experts was formed to gather information, and the criteria of housing design that had evolved in the post-corona period were identified by designing a questionnaire on the Likert scale (i.e., a psychometric scale named after its inventor)—the primary influential factors in the pattern of housing architecture after the Corona period were determined. The findings showed a significant correlation between housing architecture in the post-corona period and people's health patterns. Results showed that the factors that have the most significant impact on the evolution of housing in the region under investigation include the new function of housing (0.152), changing behavioral interactions (0.152), design criteria (0.145) and the passive ventilation factor with a score of 0.113. Other results showed that the physical distance as well as quarantine and isolation components obtained a score of 0.359 and 0.328, respectively, and were identified as the most critical developments in housing architecture. Finally, data mining findings showed that the spatial quality variable in the first phase had the most significant influence on the growth and fortification of the evolution of housing components at the study area level and the promotion of the health of people in the community in the post-pandemic period.

Energy-related CO2 emissions are one of the biggest concerns facing urban design today, increasing rapidly as cities grow. This study uses as inputs the GDP of the G8 nations (from 1990 to 2016) depending on the utilization of various energy sources, including coal, oil, natural gas, and renewable energy. Multilayer perceptrons (MLP) are combined with various nature-inspired optimization algorithms, such as Heap-Based Optimizer (HBO), Teaching-Learning-Based Optimization (TLBO), Whale Optimization Algorithm (WOA), Vortex Search algorithm (VS), and Earthworm Optimization Algorithm (EWA), to create a dependable predictive network that takes the complexity of the problem into account. Our key contributions lie in developing and comprehensively evaluating these hybrid models assessing their efficacy in capturing the intricate dynamics of carbon emissions. The study found that TLBO and VS outperform other algorithms in CO2 emission computation accuracy. TLBO has a higher training MSE (3.6778) and lower testing MSE (4.4673), suggesting larger squared errors on training data and lower testing MSE, suggesting less overfitting due to better generalization to the testing set.