Prediction Of Compressive Strength Research Articles

Prediction of compressive strength of two-stage (preplaced aggregate) concrete using gene expression programming and random forest

The aim of this research is to predict preplaced-aggregate concrete (PAC) compressive strength (CS) by using machine learning approaches such as gene expression programming (GEP) and random forest (RF). PAC requires injecting a portland cement-sand grout with admixtures into a mold after coarse aggregate has been deposited, making CS prediction complicated and requiring substantial study. Machine learning methods were used to cut down on the time and money needed for extensive experimental testing. The database includes 135 values for CS with eleven input variables. There is an acceptable degree of agreement between predicted and experimental values, as shown by the CS R2 values of 0.94 for GEP and 0.96 for RF. When comparing RF with GEP, RF performed better as measured by R2. The lower values displayed by the statistical error also showed that RF performed better than GEP. To compare, the GEP model's COV, MAE, RSME, and RMSLE were 0.527, 1.569, 2.706, and 0.133, whereas those for RF were 0.450, 1.648, 2.17, and 0.092. The SHAP analysis showed the effects of each input parameter, illuminating the positive effect of increasing the superplasticizer content on strength and the negative effect of raising the water-to-binder ratio. Using machine learning approaches to forecast the CS of PAC, this study has the potential to boost environmental protection and economic advantage.

Prediction of compressive strength of high-performance concrete via coupled meta-heuristic random forest regression techniques

Prediction of compressive strength of high-performance concrete via coupled meta-heuristic random forest regression techniques

Sustainable alternate binding material for concrete using waste materials: A testing and computational study for the strength evaluation

This study targets to create a firsthand marble-based binding material (MC) for assessing its strength mixed with fly ash (FA) and Rice husk ash (RHA). MC was produced by scorching waste marble powder and clay. The MC was then mixed distinctly with fluctuating quantities of RHA and FA (20, 30, and 40% by mass of MC) to find the most appropriate blend for concretes in terms of strength achievement. Different machine learning (ML) methods, such as support vector machine (SVM) and AdaBoost regressor (AR), were utilized for compressive strength (CS) prediction after the CS test and scanning electron microscopy. Moreover, the performance of ML models was evaluated by comparing their R2 values, performing statistical checks and k-fold analysis, and calculating the deviation between the experimental and predicted CS. The CS of blended marble cement concretes (MCMs) was lower at the age of 28 days. However, at later ages (90–364 days), the CS of MCMs with 30% RHA (R30) was found slightly greater than the ordinary Portland cement concrete. The SVM model was reasonably accurate, but the AR model was more precise in estimating the CS of MCMs. Sustainable development is promoted through the mitigation of ecological distress caused by the discarding of marble debris, RHA, and FA through their incorporation into construction materials.

Using the automated random forest approach for obtaining the compressive strength prediction of RCA

Using the automated random forest approach for obtaining the compressive strength prediction of RCA

Prediction of flexural and uniaxial compressive strengths of sea ice with optimized recurrent neural network

The mechanical properties of sea ice are important parameters for determining the ice load on engineering structures, such as offshore platforms, ships and ports. In the investigations of the mechanical properties of sea ice, experimental data samples are few, whereas the results predicted from an empirical formula feature a significant error and thus do not satisfy engineering requirements. Therefore, in this study, the stochastic gradient descent method combined with the Adam optimization algorithm is used to optimize a recurrent neural network (RNN), and an RNN model is established to predict the bending and compressive strengths of sea ice. This model can output the mechanical properties of sea ice using only physical parameters such as sea ice temperature, salinity, density, and loading rate. The overall average error in the prediction of the bending and compressive strengths of sea ice is less than 20%. The model is compared with an empirical formula, back-propagation neural network, and genetic-algorithm back-propagation neural network to predict the mechanical properties of sea ice. The results show that the model is optimized in terms of goodness-of-fit, mean relative error, mean absolute error, and mean square error, which implies that the optimized RNN is more suitable for predicting the bending and compressive strengths of sea ice. The model can be applied to rapidly and accurately predict the mechanical properties of sea ice using easily measured physical sea ice parameters, thus providing a reference for the anti-ice design of ocean engineering structures.

Enhancing compressive strength prediction in self-compacting concrete using machine learning and deep learning techniques with incorporation of rice husk ash and marble powder

Focusing on sustainable development, the demand for alternative materials in concrete, especially for Self-Compacting Concrete (SCC), has risen due to excessive cement usage and resulting CO2 emissions. As Compressive Strength (CS) is dominant among concrete properties, this research concentrates on developing SCC by incorporating Rice Husk Ash (RHA) and Marble Powder (MP) as cement and filler replacements, respectively, while applying Machine Learning (ML) and Deep Learning (DL) techniques to forecast the CS of RHA/MP-based SCC. The research further evaluates material characteristics, with a strong emphasis on ML and DL for CS prediction. Concrete samples with various mixed ratios were cast and examined after 91 days to collect data for model application. In the experimental technique, 133 samples were gathered, and CS was predicted using seven input factors (cement, RHA, MP, superplasticizer, coarse aggregate, fine aggregate, and water) in an 80:20 ratio. Various ML algorithms, including linear regression, ridge regression, lasso regression, K-nearest neighbors (KNN), support vector machine (SVM), decision tree (DT), random forest (RF), and boosting methods such as gradient boost (GB), XG boost (XGB), and adaptive boosting (ADB) are employed, along with the DL technique of backpropagation neural network (BPNN) with different optimizer algorithms (Adam, SGD, RMSprop) to predict CS in RHA/MP-based SCC. The predicted data is further validated using evaluation parameters such as R-squared (R2), mean squared error (MSE), normalized root mean squared error (NRMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE). Comparatively, ML ensemble algorithms and BPNN using Adam and RMSprop optimizers demonstrate high accuracy in predicting CS outcomes, indicated by their high coefficient correlation R2 values and low error values.

Machine learning approach for investigating compressive strength of self-compacting concrete containing supplementary cementitious materials and recycled aggregate

Supplementary cementitious materials (SCMs) and recycled coarse aggregate (RCA) have the potential for sustainable development and resource utilization and have been widely applied in self-compacting concrete (SCC). However, traditional experiment approaches for predicting its compressive strength suffer from inefficiency, time-consuming, and high cost. To address these issues, machine learning (ML) models are employed, evaluated, and compared in this paper. First, a database of 337 samples with 10 features is collected and established. Then, the 12 ML models, including SVM, KNN, DT, RF, ANN, GBoost, XGBoost, AdaBoost, CatBoost, LightGBM, HistGBM, and GEP are applied for modeling to predict the compressive strength. Finally, both feature importance, individual conditional expectation, and partial dependence plot are utilized to assess the importance and the effect of each and coupled parameters. The results show that the CatBoost and LightGBM models generally possess superior performance in various aspects, including higher R2 values, lower RMSE, MSE, and MAE values, as well as smaller errors. Among the Boosting models, except for the AdaBoost model, satisfactory predictive ability is observed. The ANN and RF models also demonstrate relatively good performance, and the SVM, DT, KNN, and GEP models present relatively poorer performance. Curing age and W/CMs exhibit the most pivotal features on model prediction. Furthermore, RCA replacement ratio plays a more prominent role than CA, and SCMs contribute somewhat to the compressive strength of concrete, but not as much compared to cement. Also, it is important to acknowledge the significant contribution of SP. This study offers a systematic evaluation of the prediction of compressive strength in SCC containing SCMs and RCA, making a notable contribution to both the existing literature and practical applications.

Prediction of compressive strength of brick columns confined with FRP, FRCM, and SRG system using GEP and ANN methods

This study assesses the strength capacity of brick columns under various confinement materials, including fiber-reinforced polymer (FRP), fiber-reinforced cementitious matrix (FRCM), and steel-reinforced grout (SRG) using gene expression programming (GEP) and artificial neural networks (ANN) models. To achieve this, a comprehensive database of masonry column test results from existing scientific literature is compiled. The models' performance is evaluated using statistical errors like the coefficient of linear correlation (R2), mean absolute error (MAE), mean square error (MSE), and root mean square error (RMSE). Additionally, sensitivity analysis is carried out to assess the significance of individual parameters in the models. The findings reveal that ANN predictions closely match empirical data, demonstrating a strong correlation coefficient of 0.95. The accuracy of the ANN approach is reasonably high, with only 26% of the predicted values deviating by more than 20% from the actual data. Based on the statistical analyses, the correlation coefficient between the actual and estimated data was 0.88, for GEP method. Also, the GEP model yields outcomes, with roughly 43% of the predicted values differing by 20–50% from the actual data. In a comparison of the two models, the ANN model outperforms the GEP model, displaying a 40% reduction in error when estimating the compressive strength of masonry columns. The data estimated by the GEP were sparser than those estimated by the ANN. Nevertheless, the GEP model still maintains an acceptable correlation coefficient and error rate, making it a viable choice for precise predictions. It offers a user-friendly formula and meets the needs of both customers and builders.

The prediction of compressive strength for recycled coarse aggregate concrete in cold region

Concrete in cold environments may suffer from frost damage, which effects the hydration reaction and early strength development of cement. This study aims to test the early strength development of recycled coarse aggregate concrete in cold region using an assisted heating method. Based on Nurse-Saul and Hansen-Pedersen maturity calculation methods, the early strength of recycled and natural coarse aggregate concrete is fitted and predicted using exponential and logarithmic type functions. The results showed that active heating and thermal insulation can ensure early concrete hydration. The temperature of recycled coarse aggregate concrete was slightly higher than that of natural concrete, while its strength was slightly lower. The early strength development of natural and recycled coarse aggregate concrete correlated more with the logarithmic type function. The prediction error of natural and recycled coarse aggregate concrete is basically within 10% for the first 14 days of compressive strength.

Prediction of compressive strength and feature importance analysis of solid waste alkali-activated cementitious materials based on machine learning

This study examines the analysis and treatment of diverse types of solid waste alkaline-activated cementitious materials utilizing machine learning techniques. To account for the variability of solid waste alkaline-activated cementitious materials, three crucial chemical components (aluminum trioxide, calcium oxide, and silicon dioxide) were chosen as representative features. A total of 274 pertinent datasets were collected and split into training and testing sets with an 80% to 20% ratio for analysis. Six machine learning models, including multilayer feedforward neural network, genetic algorithm neural network, support vector machine, random forest, radial basis function neural network, and long short-term memory network, were employed to construct predictive models. The optimal hyperparameters were identified via grid search. The performance of the models was assessed using the training and testing sets, revealing that all models exhibited strong predictive and generalization capabilities on both sets. Among the models, the support vector machine model achieved the highest performance, yielding an R2 value of 0.9054, MAE of 4.1460, and NRMSE of 0.0997. Through the utilization of SHAP (Shapley Additive Explanations) analysis, the interplay between factors and the significance of features were examined. Calcium oxide, water-to-binder ratio, silicon dioxide, modulus of water glass, and aluminum trioxide were recognized as the primary factors that influence the compressive strength. The research findings offer guidance for optimizing the performance of alkaline-activated cementitious materials made from solid waste.

Prediction of compressive strength of glass fiber-reinforced self-compacting concrete interpretable by machine learning algorithms

Prediction of compressive strength of glass fiber-reinforced self-compacting concrete interpretable by machine learning algorithms

Investigating compressive strength of concrete containing steel fiber by data-driven approach

The main objective of this paper is to use the data-driven approach to predict and study the factors affecting the compressive strength of steel fiber concrete. Therefore, six machine learning (ML) models were evaluated against a database of 166 samples and ten input variables, including Cement content, Water content, Silica fume content, Steel fiber content, Coarse aggregate content, Sand content, Superplasticizer content, Fiber diameter, Fiber length, Fly ash content. SixMLmodels, including Artificial Neural Network (ANN), K-Nearest Neighbor (KNN), Categorical Boosting (CatB), Random Forest (RF), Gradient Boosting (GB), and Extreme Gradient Boosting (XGB), are evaluated against performance metrics and validated by 1000 Monte Carlo simulations. The compressive strength prediction performance by the ML models is arranged in descending order as follows XGB>GB>CatB>RF >ANN>KNN. The two best models can predict the compressive strength of steel fiber concrete to be GB with a coefficient of determination (R2) of 0.9874 and a root mean square error (RMSE) of 2.5763 MPa and XGB with an R2 of 0.9926 and an RMSE of 1.9814 MPa for the testing dataset. The influence of the ten variables can be arranged in descending order as follows: Cement content > Water content > Silica fume content > Steel fiber content >Coarse aggregate content > Sand content > Superplasticizer content > Fiber diameter > Fiber length > Fly ash content. Among them, Cement content, Silica fume content, and Steel fiber content have a positive effect on improving the compressive strength of concrete. The steel fiber content used should be less than 1.5% of concrete volume to improve the efficiency of steel fiber in concrete. Meanwhile, Fiber diameter and Fiber length have a minimal influence on the compressive strength of steel fiber concrete.

Prediction of compressive strength in additively fabricated part using long short term memory based neural network

With ability to produce complex geometries and accurate parts, additive manufacturing (AM) could be applied in various sectors including aerospace and biomedical, though the frequent occurrence of defects and unavailability of feasible monitoring and control system prohibits their effective utilization. Technological developments and use of artificial intelligence, machine learning models could be a perfect solution to the problems. In this study, a predictive model is developed using deep learning (DL) algorithm and investigated the compressive strength of fused deposition modelling (FDM) printed part. An improved recurrent neural network (RNN) model called long short-term memory (LSTM) is used for predicting the compressive strength of part produced by FDM process. Model developed with LSTM utilizes layer-wise relevance propagation (LRP) algorithm to establish relation between the FDM part and LSTM layers. Predictive model developed in this study uses the back propagation algorithm to establish the relation between the compressive strength of part and input process variables. A total of 81 poly lactic acid samples were printed and tested, out of which 57 samples were used for training the model and 16 data were used for validating the model and remaining data were used for testing the model. Effects of process variables including infill density, print orientation, raster orientation and layer thickness have been analyzed. The values obtained from the predictive model was in good agreement with experimental results along 1% error. LSTM based DL model predicted the compressive strength closer to the measured values. From the results obtained, the developed model holds good for predicting the compressive strength of FDM printed part.

Prediction of Compressive Strength of Fly Ash-Based Geopolymer Concrete Using Supervised Machine Learning Methods

Prediction of Compressive Strength of Fly Ash-Based Geopolymer Concrete Using Supervised Machine Learning Methods

Prediction of uniaxial compressive strength of the Kızılkaya ignimbrite with variable properties using MRA and ANN, Cappadocia (Turkey)

Cappadocia in Turkey is an important region with emphasis on ethnic and historic structures and geological features. Because of volcanic eruptions, ash flow materials deposited at different levels. These different colored ignimbrite levels have a wide range of geomechanical properties. In the region, the Kızılkaya ignimbrite covers a very large area and is frequently used as building dimension stone in for the restoration of old historical buildings such as churches, caravansaries and mosques as well as new buildings. It contains tuffaceous matrix with some minerals and rock fragments. The rock has variable character. In this study, the relationship between petrographic and physico-mechanical properties of the Kızılkaya ignimbrite was investigated. Rate of change between uniaxial compressive strength and the other index properties of the rock was statistically examined utilizing multiple regression analysis (MRA) and Artificial Neural Network (ANN) methods. The study has shown that uniaxial compressive strength (UCS) of the ignimbrite can be successfully estimated by the use of dry unit weight, effective porosity, sonic velocity, and grain matrix ratio (GMR) parameters. A new UCS prediction equation is proposed in this paper for the ignimbrite with variable character. Since properties of the ignimbrite at different distances from the eruption area changes significantly, the approach presented in this study can be considered for the ignimbrites especially covering a large area with different welding degrees and grains.

Hybrid random aggregation model and Bayesian optimization‐based convolutional neural network for estimating the concrete compressive strength

AbstractNumerous experimental studies have shown the type and gradation of coarse aggregates effect on the mechanical properties of concrete. The type and gradation of coarse aggregates have not been taken into account in the available machine learning prediction models. In this study, a two‐dimensional concrete microscopic image was generated by using a random aggregate model (RAM), and the coarse aggregate and other concrete ingredients were represented innovatively using polygons and trichromatic chromaticity values in the RAM images. The RAM image set was created by applying this method to represent 1110 sets of different concrete mixes. Then based on the Bayesian optimization algorithm and the image set, a compressive strength prediction model considering the effect of coarse aggregate types and gradations was developed utilizing a convolutional neural network (CNN) model. Meanwhile, an artificial neural network (ANN) compressive strength prediction model was developed using 1110 sets of mix ratio data. The results show that the proposed RAM image generation method has the capability to represent different concrete mix ratios collected in this study. The prediction performance of the CNN compressive strength model considering aggregate types and gradations is better than that of the ANN model. The method can provide a new perspective for predicting other concrete mechanical properties and technically support performance‐based intelligent concrete mix design.

Development of alkali activated paver blocks for medium traffic conditions using industrial wastes and prediction of compressive strength using random forest algorithm

Geopolymer is an environment friendly construction material that could be synthesized using either the natural source or the industrial byproducts such as flyash and GGBS. The characteristics of the Geopolymer rely on the proportion of the flyash and GGBS and the concentration of the activator solution used. In this research work, the effect of partial replacement of flyash with GGBS in proportions such as 0, 10, 20, 30 and 40% is investigated. Also Molarity of NaOH are tested from 8 to 14 M and both the parameters are optimized. In this optimized Geopolymer concrete, the utilization of iron slag as a partial substitute for river sand in various proportions such as 10, 15, 20, 25, 30 35, 40 and 45% are investigated. The optimized Geopolymer concrete with iron slag is investigated for its performance as a paver block with incorporation of banana fiber in proportions such as 0, 0.5, 1 and 1.5 and is compared with conventional cement concrete paver block. The results show that there is a significant enhancement in the properties of Geopolymer concrete with the different levels of optimization and the utilization of natural banana fiber. The developed sustainable paver block was found to with stand medium traffic conditions as per IS 15658:2006. Further this study employed random forest (RF) algorithm for the prediction of compressive strength of geopolymer concrete specimens for the variable parameters such as molarity of alkaline solution, Flyash/GGBS ratio and partial replacement of river sand with iron slag. The performance evaluation parameters represented high accuracy of developed RF model. This research work unleashes a heft potential of Geopolymer concrete to develop economical eco-friendly sustainable paver blocks to the society through mitigation of environmental strain on the ecosystem.

Prediction of uniaxial compressive strength of rock based on lithology using stacking models

Uniaxial compressive strength (UCS) of rock is an essential parameter in geotechnical engineering. Point load strength (PLS), P-wave velocity, and Schmidt hammer rebound number (SH) are more easily obtained than UCS and are extensively applied for the indirect estimation of UCS. This study collected 1080 datasets consisting of SH, P-wave velocity, PLS, and UCS. All datasets were integrated into three categories (sedimentary, igneous, and metamorphic rocks) according to lithology. Stacking models combined with tree-based models and linear regression were developed based on the datasets of three rock types. Model evaluation showed that the stacking model combined with random forest and linear regression was the optimal model for three rock types. UCS of metamorphic rocks was less predictable than that of sedimentary and igneous rocks. Nonetheless, the proposed stacking models can improve the predictive performance for UCS of metamorphic rocks. The developed predictive models can be applied to quickly predict UCS at engineering sites, which benefits the rapid and intelligent classification of rock masses. Moreover, the importance of SH, P-wave velocity, and PLS were analyzed for the estimation of UCS. SH was a reliable indicator for UCS evaluation across various rock types. P-wave velocity was a valid parameter for evaluating the UCS of igneous rocks, but it was not reliable for assessing the UCS of metamorphic rocks.

Forecasting the strength of micro/nano silica in cementitious matrix by machine learning approaches

This work provides an intelligent machine learning-based technique for anticipating cement paste compressive strength incorporating nanocomposite. In this approach, artificial neural network (ANN), gene expression programming (GEP), random forest (RF), and ensemble approaches are used to combine multiple weak learners capable of identifying the mapping between input and output data into a robust learner. The accuracy of the strong learner as a whole will be increased by incorporating the weak learner with the least prediction error with enhanced correlation. The components of the cement paste mixture (such as water-to-cement ratio, nano-silica (%), micro-silica (%), curing time (days), and the compressive strength value are used as input data and output data. Thus, 205 datasets of cement compressive strength tests are collected to train and assess the learners. The reliability of the models was evaluated using cross-validation with K-folding and statistical error analysis (i.e., MAE, MAE, and RMSE). Therefore, interpretable methods using AdaBoost provide a significant correlation with a lower error rate of R2 = 0.96. The two that perform the best are ANN with ensemble Adaboost and RF with an R-square of 0.97. Additionally, the cross-validation findings demonstrate that the answer was correct and had minimal mistakes. The ensemble model, therefore, displays a strong connection. In addition, new expressions for the predictions of the 28-day compressive strength of nanocomposite containing silica content are developed based on the database. Sensitivity analysis reveals that curing time and silica content has a major influence on prediction. In addition, a model capable of predicting the values of the desired outputs when the necessary input parameters are provided is trained to generate the graphical user interface (GUI). It simplifies the procedure and provides a useful instrument for employing the model's civil engineering abilities.

Intelligent data-driven compressive strength prediction and optimization of reactive powder concrete using multiple ensemble-based machine learning approach

In recent years reactive powder concrete (RPC), also known as ultrahigh-performance concrete, emerged as one of the most efficient building materials due to its ultrahigh compressive strength (usually greater than 150 MPa), ductility, and durability. The existing literature presents a wealth of experimental data exploring the intricate interactions among the various components within RPC mixtures, emphasizing the requirement for a potent and efficient model to optimize, simulate, and enrich the understanding of RPC mixture characteristics. This study presents a notable attribute utilizing a multilayer stacked model that harnesses the potential to harmoniously seamlessly integrate and synergize various machine learning (ML) algorithms. An extensive dataset from reliable resources in the literature was used to develop a new ML model for predicting the compressive strength of RPC. It was found that there is a strong positive correlation between the RPC’s compressive strength and the dosage of silica fume, fiber dosage, and age of concrete. In contrast, water content and dosage of GGBFS were observed to have a low correlation with the compressive strength of the RPC. The most significant features were age, fiber dosage, silica fume content, cement content, HRWR dosage, fine aggregate content, and water content. Further, in this study, predictive efficacy identifies the stacking algorithm as optimal, followed by XGBoost, RF|ETR, and KNN. Notably, R2 strengths vary: stacking achieves 0.96, XGB sustains 0.954, RF and ETR hold 0.95, while KNN reaches 0.77.