由表观损伤特征反演钢筋混凝土梁承载率神经网络模型

钢筋混凝土梁表观损伤特征与受弯承载率间存在着一定的对应关系,但这种关系极为复杂,很难用显式的数学函数表达.基于神经网络方法研究了表观损伤特征与受弯承载率之间的非线性映射关系,以裂缝宽度、裂缝相对高度、挠度、混凝土强度、配筋率、钢筋屈服强度、钢筋直径、钢筋粘结特性系数、保护层厚度为输入,以受弯承载率为输出,建立了由表观损伤特征反演钢筋混凝土梁受弯承载率的神经网络模型.网络仿真结果与试验值吻合较好,为检测评定在役钢筋混凝土梁的受弯承载率,进而评定其安全性提供了一种新方法和新思路.

Neural network model of inversing load-bearing rate of reinforced concrete beams using apparent damage characteristics

The relationship between apparent damage characteristics and load-bearing rate in bending is very complex for reinforced concrete beams. It is very difficult to be described by explicit mathematical function. Artificial neural networks technology is applied to build the complex non-linear relationship between apparent damage characteristics and load-bearing rate in bending. By using crack width, relative crack height, deflection, RC strength, ratio of reinforcement, reinforcement yield strength, reinforcement diameter, reinforcement adhesive coefficient, and thickness of cover as inputs, load-bearing ratio in bending as output, a BP neural network model was built to inverse the load-bearing rate of reinforced concrete beams by means of apparent damage characteristics. Simulation results agree with test results well. It is proven that the artificial neural network approach is a new applicable method to evaluate the load-bearing rate in bending and safety of reinforced concrete beams.