MECHANICAL PROPERTIES OF BLENDED CEMENT CONCREAT USING MACHINE LEARNING TECHNIQUES

Abstract

Correlations have been important since the beginning; in some circumstances, they are required because it is challenging to quantity the value directly, and in others, they are beneficial since the results of other tests may be verified by correlations. Machine learning techniques like artificial neural networks (ANN) and support vector machines (SVM) were used to create prediction models to estimate the required parameters. Compressive strength and durability of blended cement concrete have been modelled in this research. The compressive strength of blended cement concrete was anticipated given its composition and other characteristics such as time, curing, and so on in the first problem. In the second problem, the carb0nation depth of fly-ash c0ncrete has been predicted from input factors such as exposure-time, curing, relative humidity, temperature, CO2 concentration, fly-ash percentage, cement per cum and studied predictability of ensemble methods were f0und to be precise. In last problem, prediction of sulphate resistance of blended cement cօncrete c0ntaining fly-ash and silica fume was done using ANN model. The results of the performance was compared and revealed that the machine learning techniques are an effective tool for reducing uncertainty in concrete mix design projects. Soft computing may give new ideas and methodologies for reducing the risk for correlation inconsistency.