摘要
通过一个假想混凝土拱坝的模型试验对混凝土拱坝裂缝损伤检测的振动法进行了研究.试验模拟了一半坝厚深度、3/4坝厚深度和贯穿全部坝厚3种深度的裂缝损伤,通过振动模态测试研究了裂缝损伤对拱坝固有频率的影响,当拱坝出现一半坝厚深度裂缝时,其固有频率就出现了明显的下降,频率下降的规律与数值模拟结果基本一致;基于实测频率数据,采用统计神经网络(SNN)可以检测出一半坝厚深度和3/4坝厚深度裂缝损伤的存在;针对两种损伤指标,利用实测模态振型数据指示裂缝损伤位置的效果进行了比较,结果表明,基于模态应变能变化的损伤指标能够指示出未贯穿混凝土拱坝坝厚裂缝损伤的位置.
The vibration-based crack damage detection method for concrete arch dams was studied through the model experiment of a hypothetical dam. Three different crack depths, half, three quarters, and whole dam thickness were simulated on the physical model. The modal tests were conducted on the physical model to analyze the influence of crack damage on the natural frequencies. The measured natural frequencies were obviously reduced due to the crack in the case of half depth. The relative frequency reductions accorded with the numerical results. The crack damage with depths of half and three quarters of thickness was correctly detected by statistical neural network（SNN） based on the variation of the measured natural frequencies. The effectiveness of two damage indices to indicate crack location was discussed. The damage index can correctly indicate the crack location with depth of partial dam thickness by using the measured mode-shape data.
出处
《浙江大学学报（工学版）》
EI
CAS
CSCD
北大核心
2009年第4期738-742,共5页
Journal of Zhejiang University：Engineering Science
关键词
混凝土拱坝
裂缝损伤检测
振动法
固有频率
统计神经网络
损伤指标
concrete arch dam
crack damage detection
vibration-based method
natural frequency
statistical neural network （SNN）
damage index
