Abstract: According to the rules and characteristics of the test data of traffic speed deflectometer (TSD), a method based on support vector machine (SVM) model was proposed to detect the void at the bottom of cement pavement slab. The system composition and measuring principle of TSD was introduced. The contrast test of TSD and heavy weight deflectometer (HWD) was carried out. The applicability of the void determination method based on deflection ratio to the detection results of TSD was analyzed, and the classification recognition model of the void degree at the bottom of the pavement slab was established based on the SVM model. The results show that: the identification results of weak sections of the pavement by TSD and HWD show high consistency; the constructed SVM void degree classification model has good comprehensive performance, and both two and three classification have high prediction accuracy.

Key words: road and airport engineering, concrete pavement, void detection, traffic speed deflectometer (TSD), support vector machine (SVM), road detection

摘要： 针对高速弯沉仪检测数据的规律和特点，提出了基于支持向量机模型的水泥道面板底脱空检测方法。介绍了高速激光弯沉仪的系统组成及测量原理；对高速弯沉仪和重型落锤式弯沉仪进行了对比试验研究，分析了基于弯沉比的脱空判定方法对高速弯沉仪检测结果的适用性，基于支持向量机模型建立了道面板底脱空程度分类识别模型。结果表明：高速弯沉仪与重型落锤式弯沉仪对道面薄弱区段的识别结果显示较高的一致性；所建立的支持向量机脱空程度分类模型具有较好的综合性能，二分类和三分类均有较高的预测精度。

关键词: 道路与机场工程, 水泥道面, 脱空检测, 高速激光弯沉仪, 支持向量机, 道路检测