Classification-Based Regression Models for Prediction of the Mechanical Properties of Roller-Compacted Concrete Pavement

Fracture behaviour assessment of high-performance fibre-reinforced concrete at high strain rates using interpretable modelling approaches

High-performance fibre-reinforced concrete (HPFRC), a type of cementitious composite material known for its exceptional mechanical performance, has widespread applications in structures exposed to severe dynamic loading conditions. However, understanding nonlinear HPFRC fracture behaviour, particularly under high strain rates, remains challenging given the complexities of assessment procedures and cost-intensive nature of experiments. This study presents an interpretable framework for modelling and analysing HPFRC fracture strength at high strain rates. A wide range of machine learning methods, including ensemble techniques, were employed to capture multivariate effects of eight essential input features (e.g., mortar compressive strength, fibre physical and mechanical properties, cross-sectional area, and strain rate) on fracture strength response. To assess the derived models, a novel evaluation procedure was proposed involving a data-based analysis, employing established metrics (i.e., coefficient of determination, root mean squared error, and mean absolute error via K-fold cross-validation) and a domain experts-involved evaluation utilising global sensitivity analysis to discern first-order and higher-order interactions among input factors. The proposed approach efficiently yielded both quantitative and qualitative insights into crucial input factors governing HPFRC fracture strength with limited experimental data. The obtained findings highlight the significance of multivariate effects, such as the interaction between strain rate and fibre tensile strength, and between fibre volume and fibre diameter, on fracture behaviour. The proposed interpretable framework aims to provide a powerful tool for proactive material failure analysis by understanding fracture behaviour and identifying potential weak and strong interactions among input factors of HPFRC-based samples. Moreover, the utilisation of the proposed approach enables researchers and civil engineers to efficiently focus on the most critical input parameters during the early design stage and ensuring the structural integrity and safety of HPFRC-based constructions.

Compressive strength prediction of admixed HPC concrete by hybrid deep learning approaches

The estimation of compressive strength includes time-consuming, finance-wasting, and laboring approaches to undertaking High-performance concrete (HPC) production. On the other side, a vast volume of concrete consumption in industrial construction requires an optimal mix design with different percentages to reach the highest compressive strength. The present study considered two deep learning approaches to handle compressive strength prediction. The robustness of the deep model was put high through two novel optimization algorithms as a novelty in the research world that played their precise roles in charge of model structure optimization. Also, a dataset containing cement, silica fume, fly ash, the total aggregate amount, the coarse aggregate amount, superplasticizer, water, curing time, and high-performance concrete compressive strength was used to develop models. The results indicate that the AMLP-I and GMLP-I models served the highest prediction accuracy. R2 and RMSE of AMLP-I stood at 0.9895 and 1.7341, respectively, which declared that the AMLP-I model could be presented as the robust model for estimating compressive strength. Generally, using optimization algorithms to boost the capabilities of prediction models by tuning the internal characteristics has increased the reliability of artificial intelligent approaches to substitute the more experimental practices.

Systematic Multiscale Models to Predict the Compressive Strength of Cement Paste as a Function of Microsilica and Nanosilica Contents, Water/Cement Ratio, and Curing Ages

Sustainable construction requires high-strength cement materials that additives with silica content could provide the requirements as well. In this study, the effect of the micro and nano-size of silica on the compressive strength of cement paste using different mathematical approaches is investigated. This study compares the strength of preferentially replaced cement pastes with microsilica (MS) and nanosilica (NS) incorporation by proposing several mathematical models. In this study, 205 data were extracted from the literature and analyzed. The modeling processes considered the most significant variables as input variables that influence the compression strength, such as curing time, which ranged between 3 and 90 days, the water-cement ratio, which varied between 0.4 and 0.85, and NS ranged between 0 and 15%. MS ranged between 0 and 40% based on the weight of cement. In this process, the compressive strength of cement paste modified with NS and MS was modeled using four different models, including the Linear Regression Model (LR), Nonlinear Model (NLR), Multi-Logistic Regression Model (MLR), and artificial neural network (ANN). The efficiency of the suggested models was evaluated using different statistical assessments, such as the Root Mean Squared Error (RMES), the Mean Absolute Error (MAE), Scatter Index (SI), Objective value (OBJ), and coefficient of determination (R2). The findings revealed that the ANN model conducted better performance for predicting compressive strength for cement paste than the other models based on the statistical assessment. In addition, based on the statistical assessment of the sensitivity of parameters, NS had more of an effect on the compressive strength of cement paste, with 6.3% more than MS.

Data-Driven Compressive Strength Prediction of Fly Ash Concrete Using Ensemble Learner Algorithms

Concrete is one of the most popular materials for building all types of structures, and it has a wide range of applications in the construction industry. Cement production and use have a significant environmental impact due to the emission of different gases. The use of fly ash concrete (FAC) is crucial in eliminating this defect. However, varied features of cementitious composites exist, and understanding their mechanical characteristics is critical for safety. On the other hand, for forecasting the mechanical characteristics of concrete, machine learning approaches are extensively employed algorithms. The goal of this work is to compare ensemble deep neural network models, i.e., the super learner algorithm, simple averaging, weighted averaging, integrated stacking, as well as separate stacking ensemble models, and super learner models, in order to develop an accurate approach for estimating the compressive strength of FAC and reducing the high variance of the predictive models. Separate stacking with the random forest meta-learner received the most accurate predictions (97.6%) with the highest coefficient of determination and the lowest mean square error and variance.

Predicting Compressive Strength of Concrete Containing Recycled Aggregate Using Modified ANN with Different Optimization Algorithms

Using recycled aggregate in concrete is one of the best ways to reduce construction pollution and prevent the exploitation of natural resources to provide the needed aggregate. However, recycled aggregates affect the mechanical properties of concrete, but the existing information on the subject is less than what the industry needs. Compressive strength, on the other hand, is the most important mechanical property of concrete. Therefore, having predictive models to provide the required information can be helpful to convince the industry to increase the use of recycled aggregate in concrete. In this research, three different optimization algorithms including genetic algorithm (GA), salp swarm algorithm (SSA), and grasshopper optimization algorithm (GOA) are employed to be hybridized with artificial neural network (ANN) separately to predict the compressive strength of concrete containing recycled aggregate, and a M5P tree model is used to test the efficiency of the ANNs. The results of this study show the superior efficiency of the modified ANN with SSA when compared to other models. However, the statistical indicators of the hybrid ANNs with SSA, GA, and GOA are so close to each other.