当前位置: X-MOL 学术Resour. Conserv. Recycl. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Fuzzy-metaheuristic ensembles for predicting the compressive strength of brick aggregate concrete
Resources, Conservation and Recycling ( IF 13.2 ) Pub Date : 2021-02-23 , DOI: 10.1016/j.resconrec.2021.105443
Wafaa Mohamed Shaban , Jian Yang , Khalid Elbaz , Jianhe Xie , Lijuan Li

Several million tons of demolition and construction (D&C) wastes are being produced worldwide. Brick waste is one of the eminent D&C wastes and several models have been performed on recycling brick waste to produce environmentally friendly concrete. This study develops three fuzzy-metaheuristic ensembles, based on adaptive neurofuzzy inference system (ANFIS) with a fuzzy c-mean clustering approach to forecasting the compressive strength of brick aggregate concrete (BAC). Such models incorporate ANFIS with particle swarm optimization (PSO), genetic algorithm (GA), and firefly algorithm (FFA). For the model development, 132 datasets were used for standalone and hybrid models. All models were trained and tested with 80% and 20% of the datasets, respectively. A k-fold cross-validation method has been applied to validate the generalization accuracy of the established models. A sensitivity analysis is also used for evaluating the influence of each input variable on the proposed techniques. Among the different designed and trained models, the results revealed that the hybrid ensemble models are more successful than ANFIS based fuzzy c-mean clustering approach in forecasting the 28-days compressive strength of BAC. The developed ANFIS-PSO model yields better prediction compared with the other two hybrid GA and FFA models so the correlation coefficient (R2), a root means square error (RMSE), and the mean absolute error (MAE) were acquired as 0.913, 0.057, and 0.032, respectively. A sensitivity analysis indicated that the content of cement and the specific gravity of brick aggregates played a significant role in the model output.



中文翻译:

预测砖混混凝土抗压强度的模糊力学集成

全球正在产生数百万吨的拆除和建筑(D&C)废物。砖废料是著名的D&C废料之一,已经对回收砖废料进行了多种模型的生产,以生产出环保型混凝土。基于自适应神经模糊推理系统(ANFIS),采用模糊C均值聚类方法,预测砖混混凝土(BAC)的抗压强度,开发了三种模糊综合方法。这种模型将ANFIS与粒子群优化(PSO),遗传算法(GA)和萤火虫算法(FFA)结合在一起。对于模型开发,将132个数据集用于独立模型和混合模型。所有模型都分别使用80%和20%的数据集进行了训练和测试。已使用k折交叉验证方法来验证所建立模型的泛化精度。灵敏度分析还用于评估每个输入变量对所提出的技术的影响。在不同的设计和训练模型中,结果表明,在基于BAC的28天抗压强度预测中,混合集成模型比基于ANFIS的模糊c均值聚类方法更为成功。与其他两个GA和FFA混合模型相比,开发的ANFIS-PSO模型可提供更好的预测,因此相关系数(结果表明,在基于BAC的28天抗压强度预测中,混合集成模型比基于ANFIS的模糊c均值聚类方法更为成功。与其他两个GA和FFA混合模型相比,开发的ANFIS-PSO模型可提供更好的预测,因此相关系数(结果表明,在基于BAC的28天抗压强度预测中,混合集成模型比基于ANFIS的模糊c均值聚类方法更为成功。与其他两个GA和FFA混合模型相比,开发的ANFIS-PSO模型可提供更好的预测,因此相关系数(R 2),均方根误差(RMSE)和平均绝对误差(MAE)分别为0.913、0.057和0.032。敏感性分析表明,水泥含量和砖骨料的比重在模型输出中起重要作用。

更新日期:2021-02-23
down
wechat
bug