基于PSO-BP和GA-BP神经网络再生砖骨料混凝土强度模型的对比研究  

A Comparative Study on Compressive Strength Model of Recycled Brick Aggregate Concrete Based on PSO-BP and GA-BP Neural Networks

作　　者：黄炜[1,2,3] 周烺 葛培 杨涛 HUANG Wei;ZHOU Lang;GE Pei;YANG Tao(State Key Laboratory of Green Building in Western China,Xi’an University of Architecture&Technology,Xi’an 710055,China;School of Civil Engineering,Xi’an University of Architecture&Technology,Xi’an 710055,China;Key Laboratory of Structural Engineering and Earthquake Resistance of Ministry of Education,Xi’an University of Architecture&Technology,Xi’an 710055,China)

机构地区：[1]西安建筑科技大学西部绿色建筑国家重点实验室,西安710055 [2]西安建筑科技大学土木工程学院,西安710055 [3]西安建筑科技大学结构工程与抗震教育部重点实验室,西安710055

出　　处：《材料导报》2021年第15期15026-15030,共5页Materials Reports

基　　金：国家自然科学基金(51978566);陕西省重点研发计划项目-重点产业创新链项目(2020ZDLNY06-04)。

摘　　要：采用两种混合算法人工神经网络模型(PSO-BP和GA-BP)预测具有不同砖骨料替代率的再生砖骨料混凝土(RBAC)的抗压强度。以RBAC的水泥质量、水灰比、碎瓷砖(CT 0—5,CT 5—32.5)替代率、碎砖(CB 0—5,CB 5—32.5)替代率及天然骨料(NA 0—5,NA 5—32.5)替代率等八个参数作为混合神经网络模型的输入参数,28 d立方体抗压强度作为输出参数。使用均方根误差(RMSE)、相关系数(R)和平均误差率对两种模型进行验证和对比分析。结果表明,PSO-BP模型与GA-BP模型都能实现高精度的预测,具有强大的泛化能力,总体而言,PSO-BP模型稍好于GA-BP模型,且都优于BP模型。同时,这也证明提出的混合算法神经网络有助于寻找最佳的RBAC配合比设计,提高实验效率。Two hybrid algorithm artificial neural network models(PSO-BP and GA-BP)are used to predict the compressive strength of recycled brick aggregate concrete(RBAC)with different brick aggregate replacement rates.The cement quality,water-cement ratio,replacement rate of broken ceramic tile(CT 0—5,CT 5—32.5),broken brick(CB 0—5,CB 5—32.5)and natural aggregate(NA 0—5,NA 5—32.5)of RBAC were used as input parameters of the hybrid neural network model,and the 28 day cube compressive strength was taken as output parameter.Root mean square error(RMSE),correlation coefficient(R)and average error rate were used to verify and compare the two models.The results show that both the PSO-BP model and the GA-BP model can achieve high-precision prediction and have strong generalization capabilities,In ge-neral the PSO-BP model is slightly better than the GA-BP,and both are better than BP model.At the same time,it also proves that the proposed hybrid algorithm neural network is helpful to find the best RBAC mix ratio design and improve the experimental efficiency.

关 键 词：人工神经网络 粒子群算法 遗传算法 再生砖骨料 抗压强度 混凝土 

分 类 号：TU528[建筑科学—建筑技术科学] TP312[自动化与计算机技术—计算机软件与理论]