基于Adaboost算法的输电线路舞动预警方法

输电线路舞动是目前尚未被全面准确认识的世界性难题,已严重威胁输电系统的安全稳定运行。文章分析影响舞动的外界气象环境因素,并在此基础上提出一种基于Adaboost集成学习算法的输电线舞动预警方法。采用基于Gini指标的决策桩作为弱分类器,通过对多个弱分类器的训练及加权求和,输出舞动预测结果及其置信度,可为电网运维人员提供决策支撑。最后,使用历史数据进行验证性实验,结果证明了所提方法的有效性。

An early warning method of transmission line galloping based on Adaboost algorithm

Transmission line galloping is a worldwide problem which has not been fully understood, and it has seriously threatened the safe and stable operation of a transmission system. We investigated the factors of meteorological environment that influence galloping, and proposed an early warning method of transmission line galloping based on the Adaboost ensemble learning algorithm. In this method, the decision stump based on the Gini index is used as the weak classifier. The prediction result and its confidence are obtained by training and weighted summing of multiple weak classifiers, which are helpful information for the decision making of operators and dispatchers of power grids. The effectiveness of the proposed method is proved by the verification experiment with historical data.