摘要
研究目的:本文以天津市河北大街混合梁斜拉桥为工程背景,基于人工神经网络模型,提出适用于混合梁斜拉桥的分步识别方法,分别采用概率和径向基函数神经网络对子结构和钢主梁子结构局部构件进行损伤识别。此外还提出适用于钢主梁局部构件识别的动-静组合损伤指标,并建立相应的径向基函数网络模型,分别针对单损伤、双损伤和三损伤的不同损伤情况进行数值模拟。研究结论:识别结果表明:(1)本文所提出的分步识别方法具有较高的识别精度,网络识别速度快,适用于大型混合梁斜拉桥的智能诊断过程;(2)所提出的动-静组合损伤指标对混合梁斜拉桥的局部损伤识别也较为敏感;(3)单处损伤测试工况中,识别精度几乎高达100%;(4)在两处和三处损伤测试工况中,位置识别正确率分别达到82.61%和78.3%。
Research purposes： It has significant engineering value and research meaning to do research on intelligent diagnosis methods of a hybrid girder cable - stayed bridge. Taking Tianjin Hebei Street hybrid girder cable - stayed bridge as the engineering background, based on artificial neural networks, the method of hierarchical damage identification which is suitable to hybrid girder cable - stayed bridge is presented： the damaged substructure and damaged steel girder substructural components can be detected by using Probablistic Neural Network （PNN） and Radial Basis Function （RBF） Neural Network individually. Furthermore, an combined static and dynamic damage sensitive index which is suitable in the second step is presented, a RBF Neural Network model is constituted and used to simulate three damage conditions, i.e. single damage and double or three damages which occurred simultaneously. Research conclusions： The identification results show that ： （ 1 ） The proposed hierarchical damage identification method has a identified precision and efficiency, it is suitable to intelligent diagnosis process of hybrid girder cable - stayed bridges. （2） The combined static - dynamic damage identification index is also sensitive to cable -stayed hybrid girder bridges. （3） The identified precision for single damage cases is nearly to 100%. （4） For double and triple damage cases individually, the identified precision is nearly to 82.61% and 78.3%.
出处
《铁道工程学报》
EI
北大核心
2011年第12期57-63,共7页
Journal of Railway Engineering Society
关键词
斜拉桥
混合梁桥
智能诊断
损伤识别
损伤指标
径向基函数神经网络
概率神经网络
cable - stayed bridge
hybird girder
intelligent dignosis
damage identification
damage index
Radial Basis Function （RBF）
Probabilistic Neural Networks （PNN）
