基于KNN打分算法的电力计量自动化终端通信故障的检测和预警

目前,电力计量自动化系统是通过公共GPRS移动网络发送到主站的,收到数据的质量容易受到终端故障和通信质量的影响。针对这个问题,提出了基于K最近邻(KNN,K-Nearest Neighbor)算法改进的KNN打分算法,使用终端通信流量相关的数据,对终端的故障进行检测和预警,并模拟实际情况对算法进行了测试。本算法在传统KNN算法的基础上增加了打分功能,不仅可以提前对故障进行检测和预警,还能通过得分衡量判决结果的可靠性,使得相关人员可以灵活决定是否进行现场排查,节约成本。

Detection & Warning of Communication Faults with Automatic Power Measurement Terminals Based on KNN-Score Algorithm

The quality of data received at present from the master station of the automatic power measurement system via the public GPRS mobile network may be easily affected by terminal fault and communication quality.In this respect,this paper presents an improved KNN-score algorithm based on KNN (K-Nearest Neighbor)algorithm.Data related to terminal communication flow is used for the detection and warning of terminal faults,and the algorithm is tested through simulation of real environment.The additional scoring function on the basis of traditional KNN algorithm does not only allow early fault detection and warning,but also can measure the reliability of the result of judgment through the score,so that related personnel can decide whether on-site trouble shooting is required,thus reducing the cost.