Knowledge-based Convolutional Neural Networks for Transformer Protection

作     者：Zongbo Li Zaibin Jiao Anyang He Zongbo Li;Zaibin Jiao;Anyang He

作者机构：School of Electrical EngineeringXi’an Jiaotong UniversityXi’an 710000China 

出 版 物：《CSEE Journal of Power and Energy Systems》 (中国电机工程学会电力与能源系统学报（英文）)

年 卷 期：2021年第7卷第2期

页      面：270-278页

核心收录：

学科分类：12[管理学] 080801[工学-电机与电器] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 0808[工学-电气工程] 081104[工学-模式识别与智能系统] 08[工学] 0835[工学-软件工程] 0811[工学-控制科学与工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：supported in part by the National Natural Science Foundation of China(No.51877167) 

主　　题：Convolutional neural network equivalent magnetization curve generalization ability,knowledge transformer protection 

摘      要：Deep learning based transformer protection has attracted increasing ***,its poor generalization abilities hinder the application of deep learning in the power system owing to the limited training *** order to improve its generalization abilities,this paper proposes a knowledge-based convolutional neural network(CNN)for the transformer *** general,the power experts can reliably discriminate between faulty transformers and healthy transformers only through the unsaturated parts of equivalent magnetization curve(voltage of magnetizing branch-differential current curve)but deep learning intends to focus on the combined features of saturated and unsaturated *** by the identification process of power experts,CNN adopted a specially designed loss function in this paper which is used to identify the running states of power ***,the presented Restrictive Weight Sparsity substitutes a special regularization term for the common LI *** presented Adaptive Sample Weight Adjustment endows the softmax loss of each sample with the optimizable weight the softmax loss of each sample with the optimizable weights to increase the impact of more-difficult-to-identify cases on the training *** the modified loss function,the knowledge is abstractly introduced into the training process of CNN so as to successfully imitate the identification process of power ***,the proposed knowledge-based CNN will pay more attention to the unsaturated parts of equivalent magnetization curve even if only limited samples are included in the training *** results of simulations and dynamic model experiments reveal that the knowledge-based CNN exhibits an improved generalization ability and the knowledge-based deep learning algorithm is a promising research direction.