Artificial intelligence can design more sustainable concrete mixtures to reduce costs and CO2 emissions. In this paper, a concrete mix ratio design method based on deep learning algorithm and meta-heuristic algorithm is proposed. The main parameters that control and affect the strength of the concrete were selected as input variables, including Cement, Fly ash, Blast furnace slag, Water, Superplasticizer, Coarse aggregate, and Fine aggregate of concrete per unit volume. The compressive strength of concrete as output variables, a long short-term memory network model was developed to predict and explain the compressive strength characteristics of concrete. Sensitivity analysis showed that cementitious material and concrete age are the most important factors affecting the compressive strength of concrete. This is in good agreement with the relevant theoretical research. Taking the trained prediction model as the objective function of compressive strength of concrete mix ratio design, combined with the concrete material cost and carbon emission from cradle to gate, the multi-objective particle swarm optimization algorithm is used to optimize the proportion of each component of concrete to find the optimal mix ratio under specific conditions. The results show that the long short-term memory network can predict the compressive strength of concrete with high accuracy, and replace the traditional explicit mathematical expression to simulate the complex nonlinear relationship of multiple parameters. The established algorithm can calculate the Pareto optimal solution set according to the objective function, and obtain the optimal mix ratio with the least material cost and carbon footprint while meeting the compressive strength required by the project. The results of this work provide an intelligent and efficient design method for determining concrete mix ratio design, which can provide a reference for practical engineering applications to a certain extent.
Read full abstract