综合采集混凝土振捣施工过程多源异构信息, 进而及时、客观地分析振捣质量, 对于保障高拱坝坝体混凝土施工质量至关重要。针对高拱坝混凝土振捣施工信息以“空-地”感知技术为主, 存在信息感知不全和数据质量有待提高的问题, 建立空天地一体化的混凝土振捣施工信息智能感知体系, 实现混凝土浇筑过程中多源、多维度、多模态施工信息的立体感知。在此基础上, 针对数值型、视频流以及图像型信息, 分别提出基于Kalman滤波的全球导航卫星系统(GNSS)定位信息降噪方法、基于改进Faster R-CNN的视频信息解析方法和基于DeblurGAN-v2的表面图像去模糊方法。以杨房沟水电站为例, 应用所提空天地一体化感知方法与技术, 实现混凝土振捣质量智能分析与监控。

It is essential to ensure the concrete construction quality of high arch dam via comprehensively collecting multi-source heterogeneous information in the concrete vibration construction and timely and objectively analyzing vibration quality. For incomplete information perception and undesirable data quality in current space-ground perception technology during concrete vibration construction, an intelligent sensing scheme with space-air-ground integration for concrete vibrating construction information was established to realize the stereoscopic perception of multi-source, multi-modal and multi-scale construction information in the concrete pouring process. On this basis, the Kalman-based denoising method for positioning data of Global Navigation Satellite System (GNSS), improved Faster R-CNN-based video analysis method and DebrGAN-v2-based surface image deblurring method were proposed, respectively, for numerical type, video stream, and image information. Taking Yangfanggou hydropower station as an example, the intelligent analysis and monitoring of concrete vibration quality can be realized by the proposed space-air-ground integration sensing method.