卢翔智 吴晓强 李慧 杜锦阳 郑茵

摘 要：针对计量自动化终端在运行过程中因性能退化或故障，导致整个电力系统运行不稳定，而现有计量自动化终端自动诊断无法及时或提前预判其故障的问题，研究基于边缘计算框架，通过引入GBDT故障检测模型，提出一种基于边缘计算的计量自动化终端自我评估诊断系统，并完成了系统的硬件和软件设计。最后通过仿真实验，对GBDT故障检测模型进行了验证，以及对整个系统进行了测试。结果表明，GBDT模型在精准率、召回率、F1值上表现良好，识别的准确率高于CART模型、RF模型，达到98.82%，具有一定的优越性，可用于计量自动化终端故障的检测与识别;本系统可实现计量自动化终端的趋势监测和故障预测，相较于传统计量自动化终端诊断系统，具有更高安全性和实时性，可用于实际计量自动化终端自我评估诊断。

关键词：GBDT模型;邊缘计算;计量自动化终端;自动诊断

中图分类号：TM933.4 文献标识码：A 文章编号：1001-5922（2021）05-0088-05

Research on self Evaluation and Diagnosis technology of Measurement Automation Terminal based on edge Computing

Lu Xiangzhi， Wu Xiaoqiang，Li Hui，Du Jinyang，Zhen Yin

（Measurement center of Guangzhou Power Supply Co.， Ltd.， Guangzhou 510700， China）

Abstract：Due to the performance degradation or fault of metering automation terminal in the operation process， the whole power system is unsTab， and the existing metering automation terminal automatic diagnosis can not predict the fault in time or in advance. Based on the edge computing framework， by introducing the gbdt fault detection model， a self-evaluation diagnosis method of metering automation terminal based on edge computing is proposed The hardware and software of the system are designed. Finally， through the simulation experiment， the gbdt fault detection model is verified，and the whole system is tested. The results show that gbdt model performs well in accuracy， recall rate and F1 value，and the recognition accuracy is higher than cart model and RF model， reaching 98.82%. which has certain advantages and can be used for fault detection and recognition of metering automation terminal; this system can realize trend monitoring and fault prediction of metering automation terminal， and is more effective than traditional metering automation terminal diagnosis system It has higher security and real-time，and can be used for self-evaluation and diagnosis of actual measurement automation terminal.

Key words：Gbdt model; Edge computing; Measurement automation terminal; Automatic diagnosis

随着电力工业的快速发展，电力计量自动化终端的应用越来越广泛。电力计量自动化终端是电力系统的关键组成，其是否健康稳定运行决定了整个电力系统的运行情况。然而由于外部环境和系统干扰等因素的影响，计量自动化终端容易出现各种故障问题。为确保计量自动化终端的稳定运行，有必要对其进行故障检测。目前，主要是通过人工定期检测或校准以发现计量自动化终端的故障和潜在故障。这种方法虽然可排除一定的计量自动化终端故障，但其对故障的识别率不高，且工作量较大，工作效率也较低。……