首页|基于AI图像识别的床面形态特征参数和迁移速度分析

基于AI图像识别的床面形态特征参数和迁移速度分析

扫码查看
床面形态特征和迁移速度是影响水流阻力和推移质泥沙输移的关键因素,对于分析河床演变趋势、河底生境系统和人类活动的影响具有重要意义.本文利用AI视觉大模型技术,从试验水槽侧壁获得了长时间、高时空分辨率的床面形态动态发展数据.采用改进的床面形态量化方法和迁移速度计算方法,提取了 9种工况下的床形特征参数和迁移速度.结果表明:在动态平衡状态下,床面形态特征参数和迁移速度仍存在显著的随机波动,其中迁移速度的波动性和变异性较高,床面形态的平均迁移速度随泥沙运动强度和推移质运动强度的增加呈幂函数型增长,相对波高、陡度与泥沙运动强度符合抛物线关系;约95%的背流面角度集中在10°~30°,与陡度呈线性关系;床面形态和迁移速度均呈现明显的右偏分布且具有拖尾特征,其中波高、波长符合Birnbaum-Saunders分布,其他床面形态特征和迁移速度遵循伽马分布.
Analysis of bed morphological feature parameters and migration velocity based on AI image recognition
The characteristic parameters and migration velocity of bedforms are key factors influencing hydraulic re-sistance and the mechanism of sediment transport,which are of significant importance for analyzing riverbed evolu-tion trends,benthic habitat systems,and the impact of human activities.This paper utilizes AI Visional Founda-tion Model to obtain long-duration,high spatiotemporal resolution data on the dynamic development of bedforms from the sidewalls of flume.By adopting improved methods for quantifying bedform morphology and calculating mi-gration velocity,the bedform characteristic parameters and migration speeds under nine different conditions were extracted.The results indicate that under dynamic equilibrium states,significant random fluctuations still exist in bedform characteristic parameters and migration velocities,with migration speed showing higher volatility and varia-bility.The average migration velocity of the bedform increases exponentially with the intensification of sediment transport and bedload movement,while relative wave height and steepness follow a parabolic relationship with the intensity of sediment movement.Approximately 95%of the leeside face angles are concentrated between 10° and 30°,showing a linear relationship with steepness.Both bedform morphology and migration velocity exhibit a pro-nounced right-skewed distribution with tailing characteristics.Among them,wave height and wavelength fit the Birnbaum-Saunders distribution most closely,while other bedform morphology characteristics and migration veloci-ties follow a Gamma distribution.

bedformAI image recognitionmigration velocity of bedformsintensity of bedload movement

张凌峰、刘春晶、曹文洪、江肖鹏、张宇

展开 >

中国水利水电科学研究院流域水循环模拟与调控国家重点实验室,北京 100048

中国水利水电科学研究院水利部泥沙科学与北方河流治理重点实验室,北京 100048

床面形态 AI图像识别 沙波迁移速度 泥沙运动强度

2024

水利学报
中国水利学会

水利学报

CSTPCD北大核心
影响因子:1.778
ISSN:0559-9350
年,卷(期):2024.55(11)