摘要
在分析多种疲劳损伤累积准则的基础之上,针对混凝土材料疲劳损伤发展受多种因素影响,损伤累积过程表现出非线性,难以用较准确的数学函数模型对其进行描述和分析的问题,建立了基于神经网络的预应力混凝土梁疲劳挠度计算的仿真模型,提出了3种仿真模拟方法,并在结合预应力混凝土梁疲劳实验结果分析的基础上,分析了各方法的精度、优势和适用范围.结果表明,3种仿真模型误差范围在5%~15%之间,可用于预应力混凝土桥梁结构抗疲劳设计及疲劳耐久性评估.建议根据不同情况需求选取合理高效的仿真模拟方法.
It is difficult to describe fatigue cumulative damage of concrete via a clear mathematical function, as cumulative damage of concrete is affected by several factors and it performs nonlinear properties. In order to solve this problem, through analyzing several fatigue damage cumulating theories, the simulation calculation model of prestressed concrete beam based on artificial neural network theory is proposed in this paper. Three simulation methods are built, and their precisions, advantages and application conditions are analyzed by fatigue experiment results of prestressed concrete beam. The results indicate that the errors of the three methods are 5 %-15 %. The precision is acceptable in fatigue design or durability evaluation of prestressed concrete beam. It is suggested to apply appropriate model and method in different conditions.
出处
《北京交通大学学报》
CAS
CSCD
北大核心
2013年第4期89-93,共5页
JOURNAL OF BEIJING JIAOTONG UNIVERSITY
基金
国家自然科学基金资助项目(51178042)
中央高校基本科研业务费专项资金资助(2011YJS255)
关键词
预应力混凝土
疲劳
挠度计算
人工神经网络
prestressed concrete
fatigue
deflection calculation
artificial neutral network
