高压电器2025，Vol.61Issue(5)：93-102,10.DOI:10.13296/j.1001-1609.hva.2025.05.010
基于改进PKCNN的电网基建档案绿色数字化管理预警方法
Green Digital Management and Early Warning Method for Power Grid Infrastructure Files Based on Improved PKCNN
摘要
Abstract
In the process of power grid construction,power grid infrastructure files are important process data.How-ever,its complicated quantity and types bring a great challenges to the green digital management and early warning of the files.To address this challenge,a kind of electronic management and early warning method for electric grid in-frastructure files based on the improved fusion of prior knowledge convolutional neural network(PKCNN)is pro-posed in this paper.Firstly,the PKCNN convolutional layer is used to extract features from the input electric grid in-frastructure files,and prior knowledge is introduced to assist in the training of the network parameters.Then,a nonlin-ear SVM optimized by the firefly algorithm(FA)is used to improve the classification function in the PKCNN to en-hance the accuracy of the electronic management and early warning of electric grid infrastructure files.Finally,the electronic management and early warning method of electric grid infrastructure file based on improved PKCNN mod-el is set up for simulation and verification.The results show that the PKCNN network has stronger feature learning ability and faster convergence speed than traditional CNN,and the use of SVM to improve the classification function in PKCNN can significantly improve the recognition accuracy of PKCNN.Compared to the traditional CNN and CNN-SVM model-based warning methods,the method proposed in this paper has higher accuracy and stronger generaliza-tion in the identification and early warning of electric grid infrastructure file quality management.关键词
先验知识/卷积神经网络/支持向量机/电网基建档案/管理预警Key words
prior knowledge/convolutional neural network/support vector machine/electric grid infrastructure files/management and early warning引用本文复制引用
陈然,周蠡,蔡杰,贺兰菲,郑希,何峰,许小薇,王振..基于改进PKCNN的电网基建档案绿色数字化管理预警方法[J].高压电器,2025,61(5):93-102,10.基金项目
国家自然科学基金资助项目(52107108).Project Supported by National Natural Science Foundation of China(52107108). （52107108）
