AI视频测流技术应用效果分析

Analysis of application effect of Al video flow measurement technology

为更好地验证AI视频测流技术应用效果,以北京市典型水文站作为研究试点,使用大尺度粒子图像测速算法,对研究试点测流断面流量进行识别及比较分析。AI训练识别在几个月的时间内识别效果稳步提升,最终接近实际数值,且有随时间推移识别精度增加的趋势;从水域结冰、示踪粒子受干扰、水面浪涌干扰及风力影响等方面提出减小测量误差的方法。本研究的AI视频测流技术在河流水文环境监测中有广阔的应用前景。

In order to better verify the feasibility and reliability of the application effect of Al video flow measurement technology,the large-scale particle image velocity measurement algorithm was used to iden-tify and compare the flow of the pilot flow measurement section in Beijing typical hydrological station.The recognition effect of AI training recognition improved steadily within a few months.The performance is that the recognition effect finally approached the actual value,and the recognition accuracy increased with the passage of time.The methods to reduce the measurement error are put forward from the aspects of water icing,disturbance of tracer particles,disturbance of water surge and influence of wind.The Al video flow measurement technology studied in this paper has a broad application prospect in river hydro-logical environment monitoring.