混凝土冻融损伤特性的神经网络研究

Investigation of Freeze-Thaw Damage of Concrete by Artificial Neural Network

根据连续介质损伤力学理论,将混凝土的冻融损伤定义为抗压强度的劣化;研究在经历不同冻融循环次数后混凝土损伤的演化规律.首次运用RBF神经网络的理论和方法,建立了混凝土在不同冻融循环次数作用下抗压强度劣化的神经网络模型,对混凝土冻融损伤特性进行了研究.本文的工作为复杂应力状态下经受冻融循环作用的普通混凝土的损伤特性提供了一个崭新的研究途径.

Based on the theory of continuum damage mechanics, the freeze-thaw damage of concrete is defined as the reduction of compressive strength. The rule of damage evolution after freeze-thaw cycles is investigated with radial basis function （RBF） neural network. The neural network model for the compressive strength damage of concrete after different times of freeze-thaw cycles is put forward. The characteristic of freeze-thaw damage of concrete is studied. It turns out that the RBF neural network is a new approach to evaluate the freeze- thaw damage of concrete under complex stress state.