Prediction of strength characteristics of cement composite using artificial neural network

Abstract

The design of concrete is still time-consuming, reliant on experience, and unpredictable. Concrete is a heterogeneous, complex substance comprised of cement, water, fine and coarse aggregate, as well as chemical and mineral admixtures. Due to the unpredictability of concrete materials such as cement, fine and coarse aggregates, the conventional concrete mixture proportion algorithms were trial and error, which wastes money, labor, and time. This might result in the erroneous computation of important variables such as the compressive and tensile strength of the concrete, necessitating the use of an Artificial Neural Network (ANN) in cement composite. ANN is inspired by the learning ability of the human brain, which is their highly interconnected and parallel nature, giving them the ability to learn from prior instances while simultaneously capturing unknown associations, making them a versatile modeling tool for concrete designs. The computed strengths would be more precise because ANN utilizes the preceding instances to produce excellent designs. This article focuses on the use of artificial neural networks as a computational tool for simulating complicated functional interactions among varied elements that impact the compressive strength of concrete as well as its behavioral characteristics. Using the capabilities of ANN, the design plot and the required corrective equations are developed to create a user-friendly approach for determining the strength of concrete, as well as its behavioral characteristics. ANN also gives a greater degree of accuracy than the currently available regression models and a reduction in the number of highly dispersed erroneous predictions. Since ANN is resilient and dependable, the results are 99.5% effective. Consequently, the review suggests that ANN may be utilized efficiently in the mix proportioning of concrete to reach the desired strength with little cost and labor waste.