岩石卸荷本构关系的BP神经网络模型  

A rock constitutive model under unloading condition by BP neural network

作　　者：王在泉[1] 张黎明[1] 贺俊征[1] 

机构地区：[1]青岛理工大学,山东青岛266033

出　　处：《岩土力学》2004年第z1期119-121,共3页Rock and Soil Mechanics

摘　　要：卸荷试验包括应力控制和变形控制两种方法。应力控制方式的特点是试验机必须对岩样一直进行轴向压缩,变形控制方式的特点是卸荷过程中试验机轴向不再对岩样压缩做功,岩样破坏是通过自身储存的能量实现的。本文利用BP神经网络,对采用变形控制方式的粉砂岩试样峰前、峰后卸围压试验数据进行训练,并建立了岩石卸荷本构关系的BP神经网络模型。结果表明,BP神经网络模型的模拟值与试验值非常接近,模拟效果比较理想。Unloading confining pressure tests could be controlled by stress and deformation. The specimens failing in axial compression are absorbing energy. However, the specimens failing in confining pressure reduction while keeping axial deformation constantly are releasing energy through their circumference expanding in hydraulic oil. Unloading confining pressure tests at the pre-peak and post- peak of siltsand with fixed axial deformation are conducted in this paper. Based on these test data, BP neural network constitutive model is established. The results show that the BP neural network model could approach the stress-strain curve accurately.

关 键 词：卸荷 峰前、峰后卸围压 BP神经网络 

分 类 号：TU451[建筑科学—岩土工程]