Concrete Intelligence: Predicting Test Properties with Artificial Intelligence

Abstract

Highly flowable, segregation-resistant, self-compacting concrete (SCC) ensures excellent structural performance in cramped areas and efficiently facilitates correct filling. The highly flowable concrete, known as self-consolidating concrete (SCC), forms without mechanical vibration. This paper analyzes the methods used on SCC mixed datasets gathered from the scientific community, utilizing artificial neural networks (ANN). Artificial neural networks act similarly to a fully developed human brain, storing and retrieving data to solve complex issues and learn through experience. In addition to employing soft computing to process data, it operates in a symbolic way of intelligent computation. It has numerous advantages, has begun to be used in civil engineering, and is quickly becoming a popular research area. The approach uses slump flow diameter, 28-day compressive strength, and ingredients as inputs to the ANN to maximize prediction accuracy for SCC features such as V-funnel and L-Box. SCC mixtures generated compressive strengths ranging from 14 to 86 MPa. L-Box values range from 0.8 to one, whereas V-Funnel times vary from 3 to 15. The accuracy of the anticipated ingredients is further guaranteed by the consistency of the training data.