高级检索

融合图像识别与数值模拟的秦岭输电线覆冰预测

Prediction of Ice Accretion on Qinling Transmission Lines Integrating Image Recognition and Numerical Simulation

  • 摘要:
    目的 输电线覆冰是电力系统中常见的极端气象风险之一,为了提高输电线路的安全运行水平,针对输电线路覆冰预测与监测的关键技术问题。
    方法 构建了一种融合数值天气预报模型(WRF)与Makkonen物理覆冰模型的覆冰厚度预测方法。同时创新性地引入机器视觉图像识别技术,提出一种基于像素宽度测量的非接触式覆冰厚度实测方法,实现对模型结果的精确验证。
    结果 以2025年1月24日一次宝鸡市翻越秦岭区域输电线路覆冰事件为例研究表明:高低空环流形势配合下逆温层形成、暖湿气流输送与地形抬升作用的协同机制是造成此次输电线覆冰主要天气学背景;利用机器视觉图像识别技术的覆冰厚度识别算法误差≤3.08%,已达到实际输电线路监测预警精度要求,展现出良好的现场应用潜力;WRF模式能够较好地模拟出此次覆冰过程中的关键气象要素,其模拟值与观测值的相关系数高达0.9,具有较高的可靠性;基于WRF模式驱动的Makkonen模型模拟的覆冰厚度与基于机器视觉的实测值吻合良好,平均绝对误差仅为0.632 mm,能准确反映覆冰的动态增长趋势。
    结论 基于WRF模式和Makkonen模型的覆冰厚度预测方法,结合机器视觉图像识别技术,能够有效预测和监测输电线路覆冰厚度,为提高输电线路安全运行水平提供了新的技术手段。

     

    Abstract:
    Objective Ice accretion on transmission lines is one of the common extreme meteorological risks in power systems. To enhance the safe operation of transmission lines, this study addresses key technical issues in ice accretion prediction and monitoring.
    Methods A forecasting method for ice accumulation thickness was developed by integrating the numerical weather research and forecasting model (WRF) with the Makkonen physical icing model. Meanwhile, an innovative machine vision-based image recognition technique was introduced, proposing a non-contact measurement method for ice thickness based on pixel width analysis, enabling accurate validation of the model results.
    Results A case study of an ice accretion event on transmission lines in the Qinling region, occurring on January 24, 2025, in Baoji City, revealed that the main meteorological background for the ice accretion was the synergistic mechanism of the formation of an inversion layer under the influence of upper- and lower-level atmospheric circulation patterns, combined with the transport of warm and moist air and orographic lifting; the icing thickness identification algorithm utilizing machine vision-based image recognition technique achieved an error rate of≤3.08%, meeting the accuracy requirements for practical transmission line monitoring and early warning, demonstrating significant potential for field application; the WRF model effectively simulated key meteorological elements during the icing process, with a correlation coefficient of 0.9 between simulated and observed values, indicating high reliability; the ice thickness simulated by the Makkonen model driven by WRF matched well with the field measurements from machine vision, with an average absolute error of only 0.632 mm, accurately reflecting the dynamic growth trend of ice accretion.
    Conclusion The ice thickness prediction method based on the WRF model and Makkonen model, combined with machine vision-based image recognition technique, can effectively predict and monitor ice accretion thickness on transmission lines, providing a new technical approach to improve the safe operation of transmission lines.

     

/

返回文章
返回