Research purposes:Aseismic layer design is important to ensure the stability of underground engineering structures in strong earthquake area in the western China.The development and application of aseismic materials are the key to enriching the design of aseismic layer structures and ensure their performance.A comprehensive understanding of the energy consumption characteristics of rubber-sand concrete lays the foundation for its effective application in underground engineering aseismic layers.In this paper,the energy consumption characteristics of rubber-sand concrete were tested by Hopkinson pressure bar test,and four different swarm intelligence optimization algorithms were used to optimize the back-propagation neural network algorithm based on the test results,so as to build four hybrid intelligent prediction models.Research conclusions:(1)The importance of affecting the energy consumption performance of rubber-sand concrete ranges from high to low,with rubber content＞cement content＞rubber particle size.(2)The optimal population numbers for the hybrid intelligent models are 150(PSO-BPNN),75(FOA-BPNN),75(LSO-BPNN),and 80(SSA-BPNN).(3)The LSO-BPNN hybrid intelligent model has the highest prediction accuracy for the proportion of transmission energy of rubber-sand concrete,while the other models have prediction performance of PSO-BPNN,FOA-BPNN,and SSA-BPNN.(4)The proposed hybrid intelligent model can be used to develop suitable rubber-sand concrete for aseismic layer materials in underground engineering such as railway tunnelling,and provide guidance for aseismic design to ensure safe construction and stable operation for railway tunnelling.
关键词
地下工程减震层/橡胶-砂混凝土/耗能特性/智能预测模型
Key words
underground engineering aseismic layer/rubber-sand concrete/energy consumption characteristics/intelligent prediction model
维普
万方数据