基于多源信息融合的架空输电线路覆冰预测模型研究

Icing Prediction Model Analysis for Overhead Transmission Lines Based on Multi-Source Information Fusion

输电线路严重覆冰可能会导致输电线路的机械和电气性能急剧下降,威胁电力系统安全稳定运行。线路覆冰预测技术是电网防冰、抗冰领域难点之一。本文以电网输电线路自然覆冰监测大数据为基础,进行数据异常处理、缺失值填补等预处理,提出一种基于覆冰拉力浮动区间的区间准确率评测方法;研究基于新型深度学习的数据驱动输电线路覆冰预测技术,构建了融合历史监测拉力、微气象数据及未来天气预报的拉力时序、一阶差分拉力时序覆冰预测模型,实现覆冰监测终端未来24小时的逐小时拉力准确预测,预知输电线路是否覆冰以及覆冰严重程度,有助于防冰、融冰决策,保证电力系统稳定安全运行。

Severe icing on transmission lines can cause a significant decrease in their mechanical and electrical performance,posing a threat to the safe and stable operation of the power system.Line icing prediction technology is one of the challenging areas in the field of anti-icing and ice-resistant power grids.Based on the big data of natural icing monitoring on transmission lines,this paper conducts data preprocessing,including data anomaly handling and missing value imputation.It proposes an interval accuracy evaluation method based on the floating range of icing tension.Furthermore,it explores a data-driven transmission line icing prediction technology based on novel deep learning approaches.The proposed model integrates historical monitoring tension data,micro-weather data,and future weather forecasts to construct tension time series and first-order difference tension time series icing prediction models.This enables accurate hourly tension prediction for the next 24 hours at the icing monitoring terminal,providing information on whether the transmission line is iced and the severity of icing.This information facilitates anti-icing and de-icing decision-making,ensuring the stable and safe operation of the power system.