基于SVM的隧道衬砌空洞填充物雷达图像识别研究

Radar Image Recognition of Tunnel Lining Cavity Fillings Based on SVM

衬砌背后空洞及其填充物对隧道结构安全具有重要影响,开展空洞探测识别对于结构安全评估和病害处置具有重要意义。首先采用室内试验和FDTD正演模拟相结合的方法,获得了空洞内填充空气、水、干砂、湿砂条件下的雷达图谱数据,并对不同填充物波形规律进行对比分析;然后,基于支持向量机算法对波形特征进行提取和分类识别,建立了一种空洞填充物的人工智能辨识方法。研究结果表明,采用傅里叶变换前的平均值、方差、平均绝对离差和傅里叶变换后的最大幅度值max(fft(X))四个统计量作为支持向量机的识别特征,可以有效区分出衬砌背后填充物的六种类型;当采取单一倾向数据时,识别准确率较好,六种物质二分类问题准确率均可以达到90%以上。

Cavities and fillings behind the tunnel lining have an important impact on the safety of the tunnel struc⁃ture,and it is crucial to carry out cavity detection and identification to realize the assessment of the structural safety and the treatment of defects.Firstly,this paper adopts a combination method of indoor tests and FDTD forward simu⁃lations to obtain the radar mapping data under the conditions of filling the cavity with air,water,dry and wet sand,and compares and analyzes the waveform patterns of different fillings.Then,this paper extracts and classifies the waveform features based on support vector machine(SVM)algorithm and establishes an artificial intelligence recog⁃nition method for cavity fillings.The results show that the six types of fillings behind the lining can be effectively distinguished by taking four statistics(the mean,variance,mean absolute deviation before Fourier transform and maximum amplitude value max(fft(X))after Fourier transform)as the SVM identifying features;and the recognition accuracy is better when single propensity data is taken,where all the six substances′binary classification can reach an accuracy of over 90%.