一种无干扰全流程水深测试方法及其在溃坝水流试验研究中的应用
A Non-intrusive Measurement of Full-field Water Depth and Its Application in Experimental Investigations of Dam-break Flows
作者：刘鑫(四川大学 水力学与山区河流开发保护国家重点实验室, 四川 成都 610065)；王波(四川大学 水力学与山区河流开发保护国家重点实验室, 四川 成都 610065)；张建民(四川大学 水力学与山区河流开发保护国家重点实验室, 四川 成都 610065)；陈云良(四川大学 水力学与山区河流开发保护国家重点实验室, 四川 成都 610065)；伍超(四川大学 水力学与山区河流开发保护国家重点实验室, 四川 成都 610065)；刘文军(四川大学 水力学与山区河流开发保护国家重点实验室, 四川 成都 610065)；宋家俊(中国葛洲坝集团海外投资有限公司, 北京 100025)
Author：LIU Xin(State Key Lab. of Hydraulics and Mountain River Eng., Sichuan Univ., Chengdu 610065, China)；WANG Bo(State Key Lab. of Hydraulics and Mountain River Eng., Sichuan Univ., Chengdu 610065, China)；ZHANG Jianmin(State Key Lab. of Hydraulics and Mountain River Eng., Sichuan Univ., Chengdu 610065, China)；CHEN Yunliang(State Key Lab. of Hydraulics and Mountain River Eng., Sichuan Univ., Chengdu 610065, China)；WU Chao(State Key Lab. of Hydraulics and Mountain River Eng., Sichuan Univ., Chengdu 610065, China)；LIU Wenjun(State Key Lab. of Hydraulics and Mountain River Eng., Sichuan Univ., Chengdu 610065, China)；SONG Jiajun(China Gezhouba Group Overseas Investment Co., Ltd., Beijing 100025, China)
收稿日期：2018-06-19 年卷（期）页码：2019,51(3):52-58
期刊名称：工程科学与技术
Journal Name：Advanced Engineering Sciences
关键字：水深测量;无干扰;全流程;图像处理;溃坝水流
Key words：water depth measurement;non-intrusive;full-field;image process;dam-break flows
基金项目：国家自然科学基金项目（51879179）；四川省科研计划项目（2019JDTD0007）
中文摘要
基于MATLAB图像处理和计算机视觉模块函数库，开发了一种无干扰全流程水深测试方法：首先利用Camera Calibration工具箱标定摄像机以得到标定代码；其次，将记录水流运动的视频文件逐帧保存，利用标定代码消除每帧图片中由于广角镜头产生的桶形畸变，并对图片灰度化和二值化，同时加入中值滤波消除噪声；最后，读取二值图黑色像素点，将其转换为实际水深，从而获取其时空分布特性。运用该方法对强非恒定流-溃坝水流进行了试验测量，研究工况为上游水深500 mm，水深比（初始时刻下游与上游水深之比）为0～0.9。观测发现，溃坝水流可大致分为3种演进模式：简单波（水深比为0）、间断波（水深比为0~0.4）、起伏波（水深比为0.4～1.0）。分别选取3种典型工况（水深比为0、0.3、0.6）数据与波高仪测量结果进行了对比，结果表明，除上下游静水区二者数据吻合良好外，下游水面光滑平顺的渐变流区、激波前锋与下游出现强间断的激波区及水面起伏明显的起伏波区，波高仪较图像技术都有不同程度的壅高，水流演进越剧烈，壅高越明显。本文建立的测量方法无需在水流中安放测量仪器，具有无侵入性，对流场无干扰，不仅能够获取全流程水深数据、重现水流完整演进过程，还能很好地捕捉到诸如水流跃起、翻滚、波动、空气卷吸等流态细节。另外，试验中保持稳定光照及染色浓度，有助于降低图片噪点、改善图片质量，从而进一步提高测量结果的精度。
英文摘要
Based on MATLAB image processing and computer vision module, this study developed a non-intrusive method to measure full-field water depth. First, the Camera Calibration toolbox was used to calibrate the camera to obtain the calibration code; Second, the experimental video which recorded water flow movement was saved frame by frame and the barrel distortion of each image caused by wide angle lens was eliminated by calibration code. Then the images were grayed and binarized, and the median filter was added to eliminate image noise; Finally, the number of black pixels in binary images was counted and converted to actual water depth to obtain its spatial and temporal distribution characteristics. This method was applied to experiment of intensively unsteady flow-dam-break flows, in which the upstream water depth was 500 mm and the depth ratio (ratio of initial downstream to upstream water depth) was from 0 to 0.9. The observation showed that dam-break flows could be roughly divided into three modes:simple wave (depth ratio was 0), discontinuity wave (depth ratio was from 0 to 0.4) and undulation wave (depth ratio was from 0.4 to 1.0). Three typical conditions (depth ratio was 0, 0.3, 0.6) were chosen to compare with the results of wave probe. The results showed that wave probe had larger values of water level when compared with image processing data in the gradually varying flow area with smooth water surface, the shock wave area with discontinuous water flowing between downstream and front of wave, and the rolling wave area with obvious fluctuations on water surface, except for the upstream and downstream areas where both results agreed well with each other. The more intensive the water flow was, the larger the differences were. There was no need to install intrusive measuring instruments in water channel by using this method, it could not only obtain full-field water depth and water evolution process completely, but also capture water details such as leap, tumbling, fluctuations and air entrainment. In addition, the stable lighting and dyeing concentration in the experiment could help reduce the image noise and improve the image quality, thus further improve the accuracy of measurement results.
【关闭】
