基于深度自编码器优化的电网通信异常诊断监测技术

打开文本图片集

中图分类号：TN915.853；TP277.3 文献标志码：A 文章编号：1001-5922（2025）09-0184-04

Diagnosis and monitoring technology of power grid communication anomaly based on deep autoencoder optimization

YU Hua，JIN Xiaojing （State Grid Jiangsu Electric Power Co.，Ltd.，Lianyungang Power Supply Company， Lianyungang 22200o，Jiangsu China）

Abstract：Inviewof the nonlinear and fluctuating characteristics of power communication signals，an anomalyrecognition model for power communication signals based on the joint training of GRU deep autoencoder was proposed. The test results showed that the accuracy and AUC values of the model were 97.8% and 0.93，respectively，which were significantly higher than those of the comparison model.In the empirical analysis of the power grid communicationmonitoring model，it was found thatthe accuracyand timeliness scores of the monitoring model proposed byrelevant experts were 95.3 and93.7 points，respectively，which were significantly higher than thoseof the traditional model.Theresults indicate thatthis study provides a goodsolution for anomaly identification in power communication system，and improves the accuracy and timeliness of anomaly diagnosis.

Key words：GRU network ;deepautoencoder;support vector machine;joint training;communicationsignals;abnormal recognition

随着智能电网的迅猛发展，电网通信系统面临着越来越多的挑战[1]。（剩余4382字）