摘要
针对润扬长江公路大桥北汊斜拉桥的拉索损伤进行了520种工况的动力计算。定义了归一化固有频率,并且分别分析了拉索损伤位置和损伤程度对归一化固有频率的影响。通过对比研究得出以下结论:拉索损伤位置对不同阶固有频率的影响模式有本质的差异;同一拉索位置处,不同的损伤程度对各阶固有频率的影响模式相似;第4、6、7阶归一化频率对拉索损伤位置不敏感;其他各阶归一化固有频率的分布较为开展,分别包含了损伤模式的个性化信息。据此建立了描述7维损伤模式映射空间的BP神经网络,分别采用不同工况组的数据进行训练和识别,对该方法进行了验证。并且定义了识别误差评价数作为描述识别效果的定量指标。
The author calculated 520 cable-damage cases of the Runyang cable-stayed Bridge （north bridge）crossing the Yangtse River. By defining a normalized natural frequency, one may find the location and extent of cable damage with respect to the changes on the normalized nature frequency. Careful comparison reveals the following： ①There is intrinsic difference in the patterns of location versus frequency, whereas the patterns of extent versus frequency are similar;②While the fourth, sixth and seventh frequencies are not sensitive to cable damage location, the distribution patterns of the other frequencies describing the 7-dimension mapping space of damage patte based on the different damage cases validated the method, defined to quantify the identification effect. are rather spread. There upon, a BP neural network m was founded. The neural and an evaluation number training and identification of identification error was defined to quantify the identification effect.
出处
《土木工程学报》
EI
CSCD
北大核心
2006年第5期72-77,95,共7页
China Civil Engineering Journal
基金
国家自然科学基金(50378017)
关键词
斜拉桥
BP网络
损伤识别
拉索损伤
cable-stayed bridge
BP neural network
damage identification
cable damage
