摘要
为获得最优性能湿喷混凝土参数设计方案,提出一种基于BP-NSGAⅡ的湿喷混凝土配合比参数多目标优化模型。首先以湿喷混凝土料浆塌落度和混凝土28 d单轴抗压强度作为目标函数,以水泥掺量、水灰比、砂石比、减水剂掺量等作为决策变量建立模型基本框架;然后结合具体工程要求,设计并开展了湿喷混凝土配合比试验,得到全面可靠的湿喷配合比及抗压强度试验数据;其次将所得数据预处理后划分为训练集与测试集并作为输入变量训练BP神经网络;然后将训练好的BP神经网络作为整体集成为一个适应度函数,利用改进的非支配遗传算法(NSGAⅡ)根据目标函数对适应度函数中的决策变量求Pareto最优解。最终得到满足C25强度的最优湿喷混凝土配合比参数:水灰比为0.53,水泥掺量为440 kg/m^(3),砂石比为8∶2,减水剂掺量为0.60%;满足C30强度的最优配合比参数:水灰比为0.50,水泥掺量为460 kg/m^(3),砂石比为7∶3,减水剂掺量为0.65%;满足C35强度的最优配合比参数:水灰比为0.46,水泥掺量为480 kg/m^(3),砂石比为8∶2,减水剂掺量为0.63%。
In order to obtain the design scheme with the optimal performance of wet shotcrete parameter,a multi-objective optimization model of wet shotcrete proportioning parameters based on BP-NSGAⅡwas proposed.Firstly,the slump of wet shotcrete slurry and 28 d uniaxial compressive strength of concrete were taken as the objective function,the cement content,water-cement ratio,sand-gravel ratio and water reducer content were taken as the decision variables,and the basic framework of the model was established.Then,combined with the specific engineering requirements,the tests on wet shotcrete mix proportion were designed and carried out.The comprehensive and reliable wet shotcrete mix proportion and compressive strength test data were obtained.Secondly,the preprocessed data was divided into training set and test set,and used as input variables to train BP neural network.Then the trained BP neural network was integrated into a fitness function as a whole,and the Pareto optimal solution was obtained for the decision variables in the fitness function according to the objective function by using the improved non-dominated genetic algorithm(NSGAⅡ).Finally,the optimal mix proportion parameters of wet shotcrete meeting C25 strength were obtained as follows,the water-cement ratio of 0.53,the cement content of 440 kg/m^(3),the sand-gravel ratio of 8∶2,and the water reducer agent content of 0.60%;the optimal mix proportion parameters meeting C30 strength were the water-cement ratio of 0.50,the cement content of 460 kg/m^(3),the sand-grave ratio of 7∶3,and the water reducer agent content of 0.65%;the optimum mix proportion parameters meeting C35 strength were the water-cement ratio of 0.46,the cement content of 480 kg/m^(3),the sand-grave ratio of 8∶2,and the water reducer agent content of 0.63%.
作者
韩斌
王建栋
李少平
张金来
HAN Bin;WANG Jiandong;LI Shaoping;ZHANG Jinlai(School of Civil and Resource Engineering,University of Science and Technology Beijing,Beijing 100083,China;Key Laboratory of Ministry of Education for Efficient Mining and Safety of Metal Mines,University of Science and Technology Beijing,Beijing 100083,China)
出处
《矿业研究与开发》
CAS
北大核心
2022年第5期173-178,共6页
Mining Research and Development
基金
国家重点研发计划项目(2018YFC1900603,2018YFC0604604)。
关键词
湿喷混凝土
配合比
BP神经网络
非支配排序遗传算法
多目标决策
Wet shotcrete
Mix proportion
BP neural network
Non-dominated sorting genetic algorithm
Multi-objective decision
