Flow direction algorithm and Biogeography-based optimization via fuzzy neural network approach for the prediction of high-performance concrete strength property

Abstract

Concrete is one of the most useful materials in the construction industry. Conventional concrete comprises additives such as cement, water, and aggregates. This concrete cannot be used for very important and sensitive structures. For this reason, high-performance concrete (HPC) has been used to achieve the desired and more suitable compressive strength by employing some additives. The additional variable is cement, fly ash, blast furnace slag, superplasticizer, fine aggregate, and coarse aggregate. On the other hand, to obtain a mixture of these materials, laboratory work is not economical and saves time. Therefore, soft-based modeling is the order of the day to solve this problem. The adaptive network fuzzy inference system model is one of the ways to achieve compressive strength close to the laboratory model, which is a smart modeling move. This model has to be optimized to get better and more satisfying results, which is done by two optimizers, Biogeography-Based Optimization (BBO) and Flow Direction Algorithm (FDA), which have bright created and powerful for better performance. Furthermore, in the outputs of these two models, BBO-ANFIS and FDA-ANFIS, certain errors and desired percentages are used to select the most suitable and ideal model for the desired output, i.e., the compressibility of concrete in high-performance concrete. In the relevant modeling, the number of evaluators in the ANFISBBO combined model is R2 = 0.8926, RMSE = 5.0406, MAE = 3.7145, A20-index = 0.8382 and U95 = 13.881, and in ANFISFDA, R2 = 0.912, RMSE = 4.7294, MAE = 3.5367, A20-index = 0.8414, and U95 = 13.054 is obtained. According to the obtained numbers, it is clear that the ANFISFDA combined model has been able to get better results than the BBO-ANFIS model.