Exploring the Influencing Factors of Concrete Compressive Strength Based on Machine Learning

There are some factors that interact to influence the compressive strength of concrete. Traditional prediction methods rely on extensive experimental testing to optimize mix proportions. Machine learning, however, offers the capability to handle complex datasets, enhancing prediction accuracy while reducing experimental costs. This study primarily employs machine learning models to examine the effects of various material contents on the compressive strength of concrete. The results indicate that curing time has the most significant impact on strength, exhibiting an approximately logarithmic relationship, with an importance value of 0.3. Cement, superplasticizer, and fly ash each display an initial increase followed by a decrease in strength, with peak values, and corresponding importance values of 0.15, 0.11, and 0.06, respectively. Water content demonstrates a relationship with strength characterized by an initial increase, followed by a decrease and subsequent rise, featuring both a peak and a trough, with an importance value of 0.12. The trend for fine and coarse aggregate in relation to strength was first decreasing and then increasing, with notable fluctuations and troughs, with importance values of 0.1 and 0.07, respectively. Slag presents a gradual increase in strength, exhibiting the lowest influence, with an importance value below 0.05. This study provides a useful reference for strength prediction and mix design optimization.