基于改进DETR的继电保护压板状态识别方法研究

打开文本图片集

中图分类号：TB9 文献标志码：A文章编号：1674-5124（2025）11-0173-08

Abstract： The current inspection work mainly relies on manual work， which has the problems of low work efficiency and inspection quality.For this reason，a voltage plate state recognition method based on improved DETR（detection transformer） is proposed.Firstly，EfcientNet isused as the backbone feature extraction network of the DETR model to enhance the image feature extraction capability and detection performance of the model. Secondly，in the neck network part， a weighted bidirectional feature pyramid module （BiFPN） is introduced to fuse the information of multiple features. At the same time，to addressthe problem of information loss due to featurefusion，a BiFormer attention mechanism is used to retain the key information and improve the detection accuracy. Then，for the hidden faults existing in the voltage plate，the electrostatic field induction technique is used to further calibratethe stateofthe voltage plate to ensure the reliabilityof the identification. Finally，the voltage plate image of the substation is used for experimental validation.Theresults show that the improved model in this paper exhibits beter performance for voltage plate state recognition compared with other current mainstream models.

Keywords： relay protection; platen state recognition; DETR; target detection; electrostatic field induction

0 引言

随着电网技术不断向智能化方向发展，以及电网相关自动化设施和配套设备日益丰富，全球范围内针对智能变电站巡检技术的研究也在不断深化。（剩余12475字）