基于XGBoost集成机器学习算法的供电台区停电敏感度仿真

Simulation of Power Outage Sensitivity in Power Station Area Based on XGBoost Integrated Machine Learning Algorithm

为了提升电力用户用电满意度,增加频繁停电检测命中率,研究基于XGBoost集成机器学习算法的供电台区停电敏感度仿真。将供电台区用户分为普通用户与重要用户,并判断是否为停电敏感度用户,采集不同类型用户历史数据,经数据预处理以及关联度分析后,将其作为输入,建立基于XGBoost集成机器学习算法的供电台区停电敏感度预测模型,通过XGBoost算法,预测供电台区停电敏感度,集成贝叶斯机器学习算法进行参数调优,获取最优分类阈值,精准预测供电台区用户停电敏感度。实验结果表明,该方法能够准确划分停电敏感用户群,有效预测供电台区不同类型用户的停电敏感度,用户覆盖率与对敏感用户预测的命中率均可达到95%以上。

In order to improve the satisfaction of power users and increase the hit rate of frequent power outage detection,the power outage sensitivity simulation of power station area based on XGBoost integrated machine learning algorithm is studied.It divides the users in the power station area into ordinary users and important users,judges whether they are power outage sensitive users,collects historical data of different types of users,takes them as input after data preprocessing and correlation analysis,and establishes a power outage sensitivity prediction model in the power station area based on XGBoost integrated machine learning algorithm.It predicts the power outage sensitivity in the power station area through XGBoost algorithm.Bayesian machine learning algorithm is integrated to optimize parameters to obtain the optimal classification threshold.It accurately predicts the power outage sensitivity of users in the power station area.The experimental results show that this method can accurately divide the power outage sensitive user groups,effectively predict the power outage sensitivity of different types of users in the power station area,and the user coverage,and the hit rate of the prediction of sensitive users can reach more than 95%.