基于机器学习的电网监控通信网络运行态势感知方法

A Method for Power Grid Monitoring Communication Network Operation Situation Awareness Based on Machine Learning

针对没有删除电网监控通信网络运行数据的冗余特征,导致感知结果不理想,平均计算时间和迭代次数增加的问题,提出一种基于机器学习的电网监控通信网络运行态势感知方法。通过方差过滤器删除数据集中的冗余及无相关特征,采用决策树递归特征消除算法,过滤电网监控通信网络运行数据,进行特征提取和分类预测,结合具有人工经验的K-means聚类方法,对常态因子添加状态标签,通过误差预测值对态势因子进行误差修正,构建电网监控通信网络运行态势评估模型,实现电网监控通信网络运行态势感知。实例测试结果表明,所提方法能够有效减少平均计算时间和迭代次数,获取更加准确的感知结果。

Aiming at the problem that the redundant features of the operation data of the power grid monitoring communication network are not deleted, which leads to the unsatisfactory sensing results and the increase of the average computing time and the number of iterations, a method of power grid monitoring communication network operation situation awareness based on machine learning is proposed.The redundant and irrelevant features in the data set are deleted by variance filter, and the recursive feature elimination algorithm of decision tree is used to filter the operation data of power grid monitoring communication network.The feature extraction and classification prediction are carried out.Combined with the K-means clustering method with artificial experience, the state label is added to the normal factors, and the error of situation factors is corrected by the error prediction value, The operation situation assessment model of power grid monitoring communication network is constructed to realize the operation situation awareness of power grid monitoring communication network.The experimental results show that the proposed method can effectively reduce the average computing time and the number of iterations, and obtain more accurate sensing results.