基于神经网络的模糊理论在桥梁状态评估中的应用

Application of Fuzzy Theory on Condition Assessment of Bridge Based on Neural Network

探讨了模糊数学中的隶属函数在桥梁技术等级状态评估中的应用。在研究现有桥梁状态评估方法的基础上，把人工神经网络和模糊数学理论结合起来应用于大跨度预应力斜拉桥的等级状态评估，建立了基于三层神经元的模糊神经网络模型，并建立结构损伤度函数及等级隶属度模型，通过样本学习训练，获取评估专家的知识及直觉思维。最终确定桥梁所对应的技术状态等级。以检测的480组索力数据作为学习样本，另外4组作为验证样本进行了索力状态评估预测。计算结果表明，网络预测值与期望值吻合良好。

The applications of fuzzy mathematics in condition assessment of bridges were discussed. Based on the research of the present methods, neural network and fuzzy mathematics were integrated and applied in condition assossment of long span PC （ prestressed concrete） bridges. 3 - layer fuzzy BPNN （ back - propagation neural network） was proposed, DV （ damage value） function and degree of grade membership model were applied then. Authoritative knowledge was obtained through sample learning. And the technical grade of the bridge was eventually acquired. In the end, 480 sets of inspected data of cable tensions were used for training samples, while the other 4 for testing to predict the condition of cable performances. Results show that it agrees quite well with the expectation.