Inclusive Multiple AI Models for Predicting Cement-Mortar Compressive Strength

The estimation of the ۲۸-day compressive strength of cement mortar, based on the chemical composition of cement, is a vital aspect of quality control in cement production plants. Accurate prediction of this property not only ensures the reliability and durability of concrete structures but also reduces the need for time-consuming and costly experimental testing. In this study, artificial intelligence (AI) techniques were employed to model and predict compressive strength using a dataset of ۸۰ samples obtained from Type II and Type V cements. Several AI approaches were evaluated, among which Wavelet analysis and Support Vector Machine (SVM) demonstrated superior predictive performance. To enhance the accuracy of predictions, a two-layer hybrid modeling strategy was adopted. Within this framework, SVM achieved the highest statistical efficiency. Results indicated that the correlation coefficient improved from ۰.۷۴ with a ۷% error to ۰.۸۴ with a ۴% error. Comparative analysis further confirmed that the two-layer neural network outperformed linear regression and semi-empirical models, highlighting the effectiveness of AI in cement quality control.

کلیدواژه ها:

Compressive Strength ، Linear Regression ، Blaine ، Alite ، Artificial Intelligence