摘要
针对传统图像跟踪算法在捕捉推移质多颗粒动态运动过程中精度不足的问题,提出了一种基于卡尔曼滤波的运动轨迹跟踪算法。开展了5组不同水流强度的推移质颗粒运动概化水槽试验,结果表明:在试验水流强度范围内(0.012~0.02),改进算法运动轨迹识别率可达100%,在运动颗粒坐标识别的相对误差较传统法下降70%,离散系数较传统法下降66%;推移质颗粒的瞬时速度总体呈现出“加速-减速-加速”交替周期性趋势,并伴有显著的局部波动;随着水流强度的增大,推移质颗粒的运动轨迹逐渐由曲折变得平滑,颗粒的流向脉动强度显著增大,横向脉动强度则增幅不明显。
To address the accuracy limitations of the traditional image tracking algorithms in capturing the dynamic motion of multiple bedload particles,a motion trajectory tracking algorithm based on Kalman filter has been proposed.Experimental results from five flume tests under varying flow intensities(0.012 to 0.02)show that the improved algorithm improves the trajectory recognition rate up to 100%,the relative error and the coefficient of variation in particle coordinate identification can be decreased by 70%and 66%respectively,compared to traditional method.Bed load particle′s instantaneous velocity exhibits an"acceleration-deceleration-acceleration"periodic pattern with notable local fluctuations.As flow intensity increases,particle trajectories become to smoothness from winding,the streamwise fluctuation intensity significantly rises,while transverse fluctuation intensity shows minor increase.
作者
刘鑫
邓敬宏
肖毅
LIU Xin;DENG Jing-hong;XIAO Yi(School of River&Ocean Engineering,Chongqing Jiaotong University,Chongqing 400074,China;SPIC Chongqing Shizitan Power Generation Company Limited,Chongqing 401220,China;National Engineering Research Center for Inland Waterway Regulation,Chongqing Jiaotong University,Chongqing 400074,China)
出处
《泥沙研究》
北大核心
2025年第5期42-49,共8页
Journal of Sediment Research
基金
国家自然科学基金项目(52179059)。
关键词
推移质
卡尔曼滤波
颗粒识别
运动轨迹
水流强度
脉动强度
bed load
Kalman filter
particle identification
particle trajectory tracking
flow intensity
fluctuation intensity
