Analysis of compressive strength of nanostructure pyrolytic carbon enhanced nanocomposite mortar and forecasting using machine learning models

Matéria (Rio de Janeiro), Год журнала: 2025, Номер 30

Опубликована: Янв. 1, 2025

ABSTRACT Utilization of Nano-structure pyrolytic carbon (NSPC) particles holds significant potential in developing nanocomposites. Consequently, compressive strength is a crucial characteristic which stipulates the efficiency NSPC cementitious composites. Nevertheless, predicting this nanocomposite challenge due to distorted responses and complex structures. The main novelty research predict developed nanocomposite. Therefore, machine learning (ML) model first-time proposed for mortar incorporated with various dosages particles. In addition, bound water determined understand hydration process. This work highlights comprehensive comparison six ML algorithms, such as linear regression, random forest extra trees, gradient boost regressor, extreme boost, LightGBM, prediction accuracy Furthermore, it evaluated through multiple statistical error analysis. Seventeen parameters were considered input variables mortar. According coefficient determination analysis, regressor attained highest R2 value 0.87, while trees achieved values 0.86 0.85, respectively. low mean absolute 3.229 was earned boost. Overall, reliable performed better mapping interplay between strength.

Язык: Английский

---

Impact of Rice Husk Ash Properties on Concrete Strength: Experimental and Machine Learning Study

Advances in Civil Engineering, Год журнала: 2025, Номер 2025(1)

Опубликована: Янв. 1, 2025

This study explores the substitution of cement with rice husk ash in concrete, aiming to reduce pollution and energy consumption caused by production. Through a series experiments, including slump test, water absorption compressive strength dynamic compression effects calcination temperature average particle size on concrete performance were evaluated. The findings indicate that use lowers while increasing concrete’s capacity. With higher temperatures, both strengths initially improve, followed decline. Smaller particles lead better these tests. ideal parameters identified 650°C 5 µm. A machine learning approach, utilizing eXtreme Gradient Boosting (XGB) model, was employed predict strength, predictions cross‐referenced experimental data. Shapley analysis applied assess influence individual variables outcomes. Results confirmed accuracy XGB an R 2 value 0.9435 mean absolute error (MAE) 2.29 MPa, showing more significant impact than temperature.

Язык: Английский

---

Splitting tensile strength prediction of Metakaolin concrete using machine learning techniques

Scientific Reports, Год журнала: 2023, Номер 13(1)

Опубликована: Ноя. 16, 2023

Splitting tensile strength (STS) is an important mechanical property of concrete. Modeling and predicting the STS concrete containing Metakaolin method for analyzing properties. In this paper, four machine learning models, namely, Artificial Neural Network (ANN), support vector regression (SVR), random forest (RF), Gradient Boosting Decision Tree (GBDT) were employed to predict STS. The comprehensive comparison predictive performance was conducted using evaluation metrics. results indicate that, compared other GBDT model exhibits best test with R2 0.967, surpassing values ANN at 0.949, SVR 0.963, RF 0.947. error metrics are also smallest among MSE = 0.041, RMSE 0.204, MAE 0.146, MAPE 4.856%. This can serve as a prediction tool in Metakaolin, assisting or partially replacing laboratory compression tests, thereby saving costs time. Moreover, feature importance input variables investigated.

Язык: Английский

---

Machine Learning-Based Approach to Wind Turbine Wake Prediction under Yawed Conditions

Journal of Marine Science and Engineering, Год журнала: 2023, Номер 11(11), С. 2111 - 2111

Опубликована: Ноя. 4, 2023

As wind energy continues to be a crucial part of sustainable power generation, the need for precise and efficient modeling turbines, especially under yawed conditions, becomes increasingly significant. Addressing this, current study introduces machine learning-based symbolic regression approach elucidating wake dynamics. Utilizing WindSE’s actuator line method (ALM) Large Eddy Simulation (LES), we model an NREL 5-MW turbine yaw conditions ranging from no 40 degrees. Leveraging hold-out validation strategy, achieves robust hyper-parameter optimization, resulting in high predictive accuracy. While demonstrates remarkable precision predicting deflection velocity deficit at both center hub height, it shows slight deviation low downstream distances, which is less critical our focus on large farm design. Nonetheless, sets stage advancements academic research practical applications sector by providing accurate computationally tool optimization. This establishes new standard, filling significant gap literature application models prediction.

Язык: Английский

---

Assessing the compressive strength of eco-friendly concrete made with rice husk ash: A hybrid artificial intelligence-aided technique

Structures, Год журнала: 2024, Номер 68, С. 107050 - 107050

Опубликована: Авг. 15, 2024

Язык: Английский

---