Abstract

Nowadays, because of human activities, the earth's environment is in danger. So, all industries including construction and building industries should be concerned about the environment, and their activities have to be in this direction. Using waste materials in construction is one of the solutions for decreasing environmental damage, for example; using recycled aggregate in various types of concrete including self-compacting concrete (SCC). Moreover, the usage of a non-destructive method for determining the mechanical properties of concrete can improve the environmental situation. To do so, in this study the radial basis function neural network (RBFNN) assisted by firefly algorithm (FA) is proposed for predicting the compressive strength of recycled aggregate self-compacting concrete (RASCC). In these regards, the information of 310 samples of RASCC has been collected from previous studies. In this model, the water to binder ratio, age, the ratio of recycled coarse aggregate, coarse aggregate, fine aggregate, and superplasticizer have been considered as input variables, and compressive strength as the output variable. Also, an ANN model has been utilized to conduct comparisons. The performance of the models has been evaluated based on statistical parameters, and by comparing the results of the model with experimental results. The values of statistical parameters of RBFNN model (for all data: R-values:0.97 & RMSE:3.3) show that the correlation between results of the model and experimental results are high and the error of the model’s results are acceptable. Moreover, results of the models indicate that both models have good ability and acceptable accuracy for predicting the compressive strength of RASCC.

摘要

如今，由于人类活动，地球环境处于危险之中。因此，包括建筑业和建筑业在内的所有行业都应该关注环境，他们的活动必须朝着这个方向发展。例如，在建筑中使用废料是减少环境破坏的解决方案之一；在各种类型的混凝土中使用再生骨料，包括自密实混凝土 (SCC)。此外，使用无损方法确定混凝土的力学性能可以改善环境状况。为此，在这项研究中，提出了一种由萤火虫算法（FA）辅助的径向基函数神经网络（RBFNN）来预测再生骨料自密实混凝土（RASCC）的抗压强度。在这些方面，从以前的研究中收集了 310 个 RASCC 样本的信息。在该模型中，将水胶比、年龄、再生粗骨料、粗骨料、细骨料和减水剂的比例作为输入变量，将抗压强度作为输出变量。此外，人工神经网络模型已被用于进行比较。模型的性能已根据统计参数进行评估，并将模型结果与实验结果进行比较。RBFNN模型的统计参数值（对于所有数据：R值：0.97 & RMSE：3.3）表明模型结果与实验结果之间的相关性很高，模型结果的误差是可以接受的。而且，