基于神经网络方法对混凝土损伤特性研究

首先对混凝土立方体试件施加三轴受压荷载历史，卸载后测量其各个方向单轴抗压和劈拉强度，根据连续介质损伤力学理论，将混凝土的损佃分别定义成抗压强和抗拉强度的劣化；研究在经历不同荷载历史后混凝土损伤的演化规律，首次运用神经网络的理论和方法，分别建立了在不同荷载史作用下抗拉强度和抗压强劣化的神经网络模型，对混凝土经历荷载历史后的强度劣化特征，即损伤特征性进行了研究，从中可以看出，神经网络方法是研究混凝土损伤的一个崭新途径。

Investigation of damage of concrete by artificial neural networks approach

The effects of triaxial compression loading history on the uniaxial compression strength and the splitting tensile strength are studied. Based on the continuum damage mechanics, the damage of concrete is defined as the reduction of strengths. The rule of damage evolution under the condition of different confinement pressures and triaxial equi compression is investigated using artificial neural network approach (ANN). It is concluded that ANN is absolutely a new method to evaluate the damage of concrete.