基于神经元网络的碾压混凝土材料剪切本构模型研究

Shear Constitutive Model of RCC Based on Neural Networks

碾压混凝土的分层形式对其力学性能有较大的影响．为了全面探讨碾压混凝土材料力学性能及为碾压混凝土结构数值分析提供材料参数，研究和建立了基于损伤的碾压混凝土材料剪切本构模型．由于现场原位抗剪（断）试验条件的限制，试验无法量测出试件的真实应变值．在考虑碾压层面点的强度、应变局部化效应的基础上，基于岩滩水电站现场碾压抗剪（断）试验结果，采用损伤模型的本构形式进行了碾压混凝土剪切本构研究，并建立了基于神经元网络的碾压混凝土剪切本构模型．

This paper deals with the shear constitutive model induced from the sandwith structure- The damage curve of RCC under shear stress is nSn shape which is similar to the curves of uniaxial tension and compression of normal concrete. And its kinetic law is adopted with a supper-exponential expression, which can effectively describe the shape. By using the kinetic law, the damage constitutive equation is obtained based on the strain equivalent assumption. Combined the constitutive equation with the shear strength model mentioned above, a damage shear constitutive relationship of RCC based on the neural networks and field shear tests in Longtan Hydropower Station has been developed.