关键词: 强夯施工监测, 装备研发, 摄影测量, 卷积神经网络, 模式识别

Abstract: Tamping pit positioning, tamping count and tamping settlement are three basic indicators for monitoring the quality of dynamic compaction construction. How to measure tamping settlement is usually a challenge to traditional monitoring techniques, and the dilemma between its efficiency and accuracy are difficult to reconcile. This study developes a full set of multi-perception integrated intelligent monitoring instrumentation for monitoring dynamic compaction construction quality, a supporting intelligent monitoring software, and a monitoring system with an active measurement target based on photogrammetry. A real time kinematic Global Navigation Satellite System (GNSS-RTK) and a magnetic azimuth sensor are integrated to realize the coordinated positioning of the ramming machine and tamping pit, and tamping count is recorded using machine vision and temporal pattern recognition. We have also developed an information cloud platform for releasing monitoring data collected using the instrumentation. All the above systems and instruments working together can realize a real time measuring system for tamping pit positioning, tamping count monitoring, and tamping settlement calculation. Field experiments of dynamic compaction construction prove that the accuracy and efficiency of our monitoring instrumentation and system meets the engineering requirements. The results help develop new intelligent construction monitoring instruments and promote the monitoring level.

Key words: dynamic compaction construction monitoring, instrument development, photogrammetry, convolutional neural network (CNN), pattern recognition