基于BWO-DBSCAN和CSA-OCRKELM的变电站数据流异常检测方法  

Research on Anomaly Detection Method of Substation Data Flow Based on BWO-DBSCAN and CSA-OCRKELM

作　　者：黄欣 赵敏彤 郇嘉嘉 吴伟杰 刘嘉文 HUANG Xin;ZHAO Mintong;HUAN Jiajia;WU Weijie;LIU Jiawen(Grid Planning&Research Center of Guangdong Power Grid Co.,Ltd.,Guangzhou,Guangdong 510075,China)

机构地区：[1]广东电网有限责任公司电网规划研究中心,广东广州510075

出　　处：《广东电力》2023年第5期39-48,共10页Guangdong Electric Power

基　　金：中国南方电网有限责任公司科技项目(GDKJXM20190387)。

摘　　要：为了提升变电站数据流检测的实时性与准确性,提出一种使用白鲸优化(beluga whale optimization,BWO)算法优化基于密度的噪声应用空间聚类(density based spatial clustering of applications with noise,DBSCAN)算法,与使用圆圈搜索算法(circle search algorithm,CSA)优化单分类正则核极限学习机(one class regularized kernel extreme learning machine,OCRKELM)相结合的变电站通信网络数据流异常检测方法。首先,利用BWO-DBSCAN对正常数据流进行聚类,形成样本簇;其次,通过CSA-OCRKELM模型对异常数据流进行实时检测;最后,利用OPNET仿真软件仿真模拟变电站的通信行为并进行对比分析,验证所提方法的有效性。仿真实验结果表明所构建检测模型的检测率约为99%,较其他检测模型具有较高的性能与准确率。In order to improve the real-time performance and accuracy of the substation data flow detection,this paper proposes to use a data anomaly detection method for the substation communication network combined the beluga whale optimization(BWO)algorithm for optimizing density based spatial clustering of applications with noise(DBSCAN)and the circle search algorithm(CSA)optimizing one class regularized kernel extreme learning machine(OCRKELM).Firstly the BWO-DBSCAN is used to cluster the normal data flow to form sample clusters.Secondly,the CSA-OCRKELM model is used for real-time detection on the abnormal data flow.Finally,OPNET simulation software is used to simulate and analyze the communication behaviors of the substation,and the effectiveness of the proposed method is verified.The simulation results show that the detection rate of the constructed detection model is about 99%,and it has higher performance and accuracy than other detection models.

关 键 词：变电站数据流 白鲸优化算法 密度聚类算法 圆圈搜索算法 单分类正则核极限学习机 

分 类 号：TM764[电气工程—电力系统及自动化] TP274[自动化与计算机技术—检测技术与自动化装置]