Scholarlink

<p>This study used Multi Non-Linear Regression (MNLR) analysis and Artificial Neural Networks (ANN) to calculate the Compressive Strength (CS) of mortar comprising metakaolin (MK). The study gathered about 543 datasets with information on numerous mortar mixtures with varying amounts of MK from various sources. The inputs for both methods were the mix proportions, while the output was the CS of the mortar. For the MNLR analysis, a regression equation was established and built on the input and output variables. For the ANN method, a multilayer feed forward neural network was trained using the collected data to forecast the CS of the mortar. The forecasting of CS of the mortar containing MK, the results showed that both techniques were successful, with the ANN method surpassing MNLR analysis including machine learning and deep learning methods. A number of neurons are selected based on the empirical formulas from literature, and now-a-days, applications of the MNLR are used in various industrial sectors for prediction. This will be confirmed with the help of the coefficient of determination (R<sup>2</sup>), Mean Square Error (MSE) and Average Absolute Error (AAE). According to these results, ANN may be a useful tool for forecasting mortar strength with R<sup>2</sup> as 0.90, MSE as 0.36 and AAE as 1.77 (modelled and predicted) with k-fold cross validation as k = 5. With the help of a model developed from MNLR and ANN, CS is estimated experimentally and results are compared. Applications of these kinds of models can used to estimate the concrete properties (fresh and hardened state) with help of different kind of ingredients.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Comparison of metakaolin blended mortar compressive strength from multi non-linear regression analysis, artificial neural network and experimental work

  • 期刊论文

  • 发表日期：

    2026年1月4日

查看全文
  • P. Jagadesh,
  • J. Harshini,
  • Md Azree Othuman Mydin,
  • Khalid Ansari,
  • Anshul Nikhade,
  • Charuta Waghmare

摘要

This study used Multi Non-Linear Regression (MNLR) analysis and Artificial Neural Networks (ANN) to calculate the Compressive Strength (CS) of mortar comprising metakaolin (MK). The study gathered about 543 datasets with information on numerous mortar mixtures with varying amounts of MK from various sources. The inputs for both methods were the mix proportions, while the output was the CS of the mortar. For the MNLR analysis, a regression equation was established and built on the input and output variables. For the ANN method, a multilayer feed forward neural network was trained using the collected data to forecast the CS of the mortar. The forecasting of CS of the mortar containing MK, the results showed that both techniques were successful, with the ANN method surpassing MNLR analysis including machine learning and deep learning methods. A number of neurons are selected based on the empirical formulas from literature, and now-a-days, applications of the MNLR are used in various industrial sectors for prediction. This will be confirmed with the help of the coefficient of determination (R2), Mean Square Error (MSE) and Average Absolute Error (AAE). According to these results, ANN may be a useful tool for forecasting mortar strength with R2 as 0.90, MSE as 0.36 and AAE as 1.77 (modelled and predicted) with k-fold cross validation as k = 5. With the help of a model developed from MNLR and ANN, CS is estimated experimentally and results are compared. Applications of these kinds of models can used to estimate the concrete properties (fresh and hardened state) with help of different kind of ingredients.