钢筋混凝土雷达信号的BP人工神经网络识别  

Radar Signal Interpretation of Structural Concrete by BP Neural Network

作　　者：谢慧才[1] 徐茂辉 

机构地区：[1]汕头大学土木工程系,广东汕头515063 [2]深圳市建筑科学研究院,深圳518031

出　　处：《华中科技大学学报（城市科学版）》2007年第1期1-4,12,共5页Journal of Huazhong University of Science and Technology

基　　金：广东省自然科学基金资助项目(032023)

摘　　要：在混凝土雷达检测中，钢筋直径的识别是一个难点，因此探求使用人工神经网络的方法来解决。试验制作了一系列埋有钢筋的混凝土试件，这些埋置的钢筋直径在8～25mm间且埋深在15-125mm之间。这些试件经探地雷达检测后，雷达信号被提取出几个特征参数，这些参数包括雷达波的峰值、能量、波宽以及斜率，然后将这些参数作为BP神经网络识别的特征参数，并通过对神经网络进行有针对性地细分以提高网络识别精度，网络识别的结果表明BP神经网络能成功地识别钢筋的直径。Recognition of rebar diameter is still a difficulty that is not resolved by GPR survey. A new method was found to resolve this problem. Some concrete specimens were made embedding rebars with diameters of 8-25 mm and depths between 15- 125 mm. Then these specimens were tested using GPR and several characteristic parameters were recognized from radar wave directly by the method. The parameters including the peak value,wave energy,wave band width and wave slope of radar wave were chosen as the characteristic parameters of BP neural network. The BP neural network was also subdivided in order to improve the network recognize precision. The outputs of the networks show that the BP neural network is successful in identifying the size of the buried rebars.

关 键 词：探地雷达 混凝土 钢筋 BP人工神经网络 自动识别 

分 类 号：TU375[建筑科学—结构工程] TU317