基于边缘计算的电网主设备状态实时监测方法

Real-time monitoring method of power grid main equipment status based on edge computing

针对电网主设备状态实时监测性能和异常检测准确率低的问题,构建了一个基于边缘计算的电网主设备状态实时监测系统。该系统通过传感器模块对电网主设备状态进行实时数据采集;基于即时定位与地图创建(Simultaneous Localization and Mapping,SLAM)对电网主设备进行监测;边缘节点模块利用小波去噪对采集的数据进行处理,采用局部异常因子(Local Outlier Factor,LOF)算法对数据进行异常值检测;网络节点模块利用消息队列遥测传输协议(Message Queuing Telemetry Transport,MQTT)将处理后的结果值传输到管理平台;通过管理平台模块的分析展示、告警处理和权限设定实现对电网主设备状态的实时检测和预警。实验结果表明,该系统采用LOF算法对数据进行异常值检测,其误报率为1.5%,检测率为98.5%,准确率能达到98.6%。采用MQTT协议传输数据的平均时延为31.52 ms,到报率能达到99.78%,具有较好的实用性。

Aiming at the problems of low real-time monitoring performance and anomaly detection accuracy of power grid main equipment,a real-time monitoring system of power grid main equipment based on edge computing is constructed.The system collects real-time data of the state of power grid main equipment through the sensor module.Simultaneous Localization and Mapping(SLAM)is used to monitor the main equipment of power grid.The edge node module uses wavelet denoising to denoise the collected data,and uses the Local Outlier Factor(LOF)algorithm to detect the data outliers.The network node module uses Message Queuing Telemetry Transport(MQTT)to transmit the processed result value to the management platform.Through the analysis and display,alarm processing and permission setting of the management platform module,the real-time detection and early warning of the power grid main equipment status are realized.The experimental results show that the system uses LOF algorithm to detect outliers.The false positive rate is 1.5%,the detection rate is 98.5%,and the accuracy rate can reach 98.6%.The average time delay of data transmission using MQTT protocol is 31.52 ms,and the transmission rate can reach 99.78%,which is highly practical.