A Combined Genetic Algorithm-Artificial Neural Network Optimization Method for Mix Design of Self Consolidating Concrete

The use of intelligent optimization and modeling methods is rapidly increasing in many fields including concrete technology. In recent years, concrete mix design has been studied using intelligent models in which the artificial neural networks are among the most popular and widely utilized method. However, this modeling depends on an optimization process, and the structured model should be tuned by implementing optimization techniques. Additionally, finding the most appropriate neural network structure for solving the concrete mix design problem was proven to be an important challenge in the state-of-the art. Therefore, this paper introduces a novel strategy in which an evolutionary algorithm and a structure of artificial neural network were fused to find the best network for modeling the compressive strength of Self Consolidating Concrete (SCC) and to extract the most optimal mix design. The novel strategy is tested using 169 data-sets with each set containing 11 concrete constituent properties. The proposed GA-ANN-GA strategy not only finds the best model but also presents the most optimal mix design of concrete to mitigate the challenges reported in recent studies. 