摘要
采用人工神经网络技术对混凝土损伤过程中所伴生的声发射信号进行识别,可实现对混凝土损伤程度的识别.首先建立人工神经网络模型,并在标准工况下采集混凝土损伤声发射信号;然后根据加载曲线将采集到的声发射信号分为3类(分别对应混凝土的3个损伤阶段:轻度损伤阶段、中度损伤阶段和严重损伤阶段),并将这3类信号作为标准工况数据输入到神经网络学习模块中进行训练,得到混凝土损伤程度识别系统;最后将相同工况下所采集的混凝土声发射信号输入到系统中,即可识别混凝土的损伤程度.实测结果表明,识别准确率可达90%以上.
Identification of damage degree of concrete could be realized by artificial neural network technology combined with acoustic emission（AE） signals associated with concrete damage process. First, neural network model was established, and AE signals accompanied by concrete damage were collected under standard operating conditions. AE signals were divided into three kinds according to the loading curves of three different damage stages of concrete（slight damage stage, moderate damage stage and serious damage stage）. The three kinds of AE signals were inputted into the neural network for training as the standard condition data. Concrete damage degree recognition system was established. Finally, AE signals collected by the same condition to be identified were inputted into the recognition system, identification of damage degree of concrete can be realized according to the AE signals. The experimental results show that the accuracy reaches over 90%.
出处
《建筑材料学报》
EI
CAS
CSCD
北大核心
2014年第4期672-676,共5页
Journal of Building Materials
基金
国家自然科学基金资助项目(51009058)
中国博士后科学基金面上项目(2011M501160)
关键词
混凝土
声发射
神经网络
模式识别
损伤程度
concrete
acoustic emission （AE）
neural network
pattern recognition
damage degree
