Using explainable machine learning to predict compressive strength of blended concrete: a data-driven metaheuristic approach

Asian Journal of Civil Engineering, Journal Year: 2023, Volume and Issue: 25(1), P. 219 - 236

Published: June 27, 2023

Language: Английский

---

Machine Learning Method to Explore the Correlation between Fly Ash Content and Chloride Resistance

Materials, Journal Year: 2024, Volume and Issue: 17(5), P. 1192 - 1192

Published: March 4, 2024

Chloride ion corrosion has been considered to be one of the main reasons for durability deterioration reinforced concrete structures in marine or chlorine-containing deicing salt environments. This paper studies relationship between amount fly ash and concrete, especially resistance chloride erosion. The heat trend map total factor correlation displayed that ranking correlations was as follows: sampling depth > cement dosage dosage. In order verify effect on resistance, three different machine learning algorithms (RF, GBR, DT) are employed predict content proportioned with varying admixture ratios, which evaluated based R2, MSE, RMSE, MAE. results predicted by RF model show threshold chlorinated environments is 30–40%. Replacing part mixture below this ash, it could change phase structure pore structure, improve permeability reduce free ions system. Machine modeling using sample data can accurately properties, effectively engineering tests. development models essential decarbonization intelligence engineering.

Language: Английский

---

Evaluation of regional green innovation performance in China using a support vector machine-based model optimized by the chaotic grey wolf algorithm

Clean Technologies and Environmental Policy, Journal Year: 2024, Volume and Issue: unknown

Published: May 16, 2024

Language: Английский

---

Study on the design method of multi-component industrial solid waste low carbon cementitious material with cement as the activator

Case Studies in Construction Materials, Journal Year: 2024, Volume and Issue: 21, P. e03478 - e03478

Published: July 2, 2024

The relationship between microstructure and mechanical properties of multi-component solid waste low-carbon cementitious materials has been widely pay attention to. However, industrial is a complex system with many variable factors, which makes it difficult to design the formulation materials. This paper pioneered application machine learning (ML) models, algorithms error rates analyze compressive flexural strength fly ash-based pastes. Coefficient determination (R2), mean squared (MSE), root square (RMSE), absolute (MAE) a20-index were used evaluate robustness. X-ray diffraction (XRD), scanning electron microscope (SEM) Brunauer-Emmett-Taylor (BET) carried out evolution evaluation results ML models exhibited that Gradient boosting regression (GBR) model had best parameters steep normal distribution fitting curve an 0.861. GBR key factors identified by Pearson's coefficient, was benefit determine Furthermore, experiments also demonstrated optimum ratio low carbon material 10 % gypsum, metakaolin, 45 ash, 15 slag 20 cement, respectively. It worth noting this kind reached 35 MPa, superior P·O 32.5 cement. phase, SEM images pore structure showed synergistic effect effectively filled voids facilitated formation variety gelatinous through gelling reactions in late stage (14–28 d). work will promote resource utilization waste, contribute reduction, can accelerate green revolution concrete.

Language: Английский

---

Using Explainable Machine Learning to Predict Compressive Strength of Blended Concrete: A Data-Driven Metaheuristic Approach

Research Square (Research Square), Journal Year: 2023, Volume and Issue: unknown

Published: June 1, 2023

Abstract In this study, we use highly developed machine learning techniques to accurately estimate the compressive strength (CS) of blended concrete, considering its composition, including cement, SCMs (ground granulated blast furnace slag (GGBFS) and fly ash (FA)), water, superplasticizer, fine/coarse aggregate, curing age. addition these, examine an array models, XGBoost, decision trees (DT), deep neural networks (DNN), linear regression (LR). Among them, XGBoost has best performance in every category. We Bayesian optimization method for hyperparameter fine-tuning improve forecast accuracy. Our in-depth examination demonstrates better predictive skills ensemble models like RF over LR, which is limited ability capture data complexity beyond relationships. With R 2 0.952, RMSE 4.88, MAE 3.24, MAPE 9.94%, performs noticeably than rivals. Using SHAP analysis, determine that age, water content cement concentration constitute main factors influencing capacity model, with contributions superplasticizer being minimal. Curing age have interesting positive association CS, but a negative link CS. These results highlight value learning, more especially effectiveness as potent device forecasting CS mixed concrete. Additionally, knowledge gained from our research provides designers researchers field concrete materials useful direction, highlighting most important strength. Future studies should work toward additional by attempting verify these across wider variety compositions test settings.

Language: Английский

---

The Role of Xanthan Gum in Predicting Durability Properties of Self-Compacting Concrete (SCC) in Mix Designs

Buildings, Journal Year: 2023, Volume and Issue: 13(10), P. 2605 - 2605

Published: Oct. 16, 2023

This study comprehensively investigates the rheological properties of self-compacting concrete (SCC) and their impact on critical parameters, including migration coefficient, penetration depth chlorine ions, specific electrical resistance, compressive strength. A total 43 mix designs were meticulously examined to explore relationships between these properties. Quantitative analysis employed a backpropagation neural network model with single hidden layer accurately predict resistant durable characteristics concrete. The optimal number neurons in was determined using fitting component selection method, implemented MATLAB software(2021b). Additionally, qualitative conducted sensitivity expert opinions determine priority research additives. main contributions this paper lie exploration SCC properties, utilization for accurate prediction, prioritization additives through analysis. demonstrated exceptional performance predicting test results, achieving high accuracy rate 14 parameters such as depth, strength, resistance. Sensitivity revealed that xanthan gum emerged most influential additive, accounting 43% observed effects, followed by nanomaterials at 35% micro-silica 21%.

Language: Английский

---

Evaluation and Prediction of Compressive Strength of Self-compacting Concrete Containing Ultrafine Ground Granulated Blast Furnace Slag Using Random Forest Algorithm

International Journal of Pavement Research and Technology, Journal Year: 2024, Volume and Issue: unknown

Published: Nov. 25, 2024

Language: Английский

---

New insights on rheology, durability and mechanical and thermal properties of polyester and steel fiber-reinforced self-compacting concretes

European Journal of Environmental and Civil engineering, Journal Year: 2024, Volume and Issue: unknown, P. 1 - 24

Published: Dec. 18, 2024

Self-compacting concrete improves fresh-state fluidity while maintaining mechanical properties, and presents an increasing research interest in fiber incorporation. However, the effects of fibers on rheological behavior durability remain insufficiently studied existing literature. This study provides new insights effect polyester steel rheological, mechanical, durability, microstructural, thermal properties SCC. Nine different mixtures were studied: one reference SCC (without fibers), four incorporating fibers, other fibers. The percentages for each type 0.25%, 0.5%, 0.75%, 1%. results showed that whatever their type, adding reduces workability improving compressive strength Incorporating 1% increased flexural by 97%, whereas had no significant effect. In terms porosity but reduced its sorptivity. For instance, 9.5% whilst reducing sorptivity 23% compared to one. Polyester also improved conductivity, inverse An proportionality between plastic viscosity was identified, highlighting importance influencing transport hardened Based these findings, it is recommended careful attention should be taken change both when a given percentage into

Language: Английский

---

The Role of Additives in Estimating Service Life of Self-Compacting Concrete Mix Design Using Fib Modeling

SSRN Electronic Journal, Journal Year: 2022, Volume and Issue: unknown

Published: Jan. 1, 2022

Concrete strength is considered to be a significant criterion in the construction and operation of reinforced concrete structures. Reinforced structures exposed corrosive environmental effects are prone short lifetime due corrosion rebar presence chloride Regarding resolve or mitigating abovementioned shortcomings, current study conducted investigate impact Xanthan Gum (polysaccharide)on durability properties associated with an increased half-life self-compacting concrete. Durability mechanical have been evaluated at different concentrations supplementary material (in 0.2 0.25% by weight), microsilica 5, 7, 10% nanosilica 2, 2 4% weight). In addition rheological concrete, other factors including permeability coefficient, depth ions infiltration, electrical conductivity, pressure resistance taken into account. To do so, service 43 mixes estimated using FIB modeling. this study, soft computing methodologies (specific Artificial Neural Network (ANN)) utilized facilitate computation burden resulting from complexity model number variables. observed results confirmed high accuracy low error ANN modeling terms compatibility, nonlinearity, proper generalizability, capability anticipate resistance, as well prediction coefficient infiltration. Also, sensitivity analysis was performed super decision software determine ranking priority additives. Our final approve that additives can enhance life. Moreover, experiment sensitive additives, change additive’s weight ratio may lead alteration rank.

Language: Английский

---