南京信息工程大学学报2025，Vol.17Issue(3)：301-314,14.DOI:10.13878/j.cnki.jnuist.20240605001
基于HPO优化ECA-CNN-BiLSTM的变压器运行状态分类与识别方法
Classification and recognition of transformer operating state based on ECA-CNN-BiLSTM optimized by HPO
摘要
Abstract
The classification and accurate recognition of transformer operating states are crucial for the stable oper-ation of transformers and the safe power supply of power systems.Current research in this field faces challenges such as limited attention to transformer load data,high complexity of mechanism models,and unclear correspondences be-tween data like oil temperature and overload states.Therefore,this paper proposes an improved hybrid model that combines the Hunter-Prey Optimization(HPO)algorithm and the Efficient Channel Attention(ECA)module,ap-plied to a Convolutional Neural Network(CNN)and Bidirectional Long Short-Term Memory(BiLSTM)neural net-work,for transformer operating state classification and overload fault identification.Data from a main transformer con-taining nine features related to transformer load are selected as samples,and the load state category is determined through K-Means++clustering and analysis of the transformer's normal periodic load.The parameters of the hybrid model are optimized by HPO to improve the model's performance and generalization ability.After pre-processing and feature extraction of transformer load data,the improved model is used for accurate recognition of load stages.Experi-mental results show that the proposed method achieves a recognition accuracy of 99.24%,yielding excellent results in classification and recognition of transformer operating states.关键词
电力变压器/状态分类识别/高效通道注意力/卷积神经网络/双向长短时记忆Key words
power transformer/state classification and recognition/efficient channel attention(ECA)/convolutional neural network(CNN)/bidirectional long short-term memory(BiLSTM)分类
动力与电气工程引用本文复制引用
邹德旭,毛雅婷,权浩,周涛,彭庆军,洪志湖,代维菊,王山..基于HPO优化ECA-CNN-BiLSTM的变压器运行状态分类与识别方法[J].南京信息工程大学学报,2025,17(3):301-314,14.基金项目
国家自然科学基金(51907090) （51907090）
云南电网有限责任公司电力科学研究院项目(YNKJXM20220009) （YNKJXM20220009）
