在变压器保护中关于励磁涌流状态的识别一直是困扰继电保护研究人员的棘手问题。本文基于人工神经网嵚纾酆峡悸潜溲蛊骼庞苛髯刺凸收献刺奶卣鳎岢霾建立了一个三层前向神经网络模型，用来实现变压器的微机嵄；ぁ＠茫牛停裕薪辛舜罅康姆抡婕扑悖扑憬峁魑祶练样本，对所建立的神经网络模型进行训练。对该模型进行故嵳献刺煅榈慕峁砻鳎⒌纳窬缒芄欢员溲蛊魉鶏发生的故障状态作出正确响应，响应时间小于１０ｍｓ。本文所嵦岢龅姆椒ㄊ抢蒙窬缡迪治⒒溲蛊鞅；ぷ爸玫幕鶏础。 The identification of magnetizing inrush conditions in transformer protection has been a thorny issue for relay protection researchers. In this paper, based on artificial neural network, we have established a three-layer forward neural network model to realize the transformer microcomputer嵄  @   牛 停 停 停 停 停 牛 牛 牛 牛    Shin Shin Shin Kang 抡 婕 Jie flutter Pu   魑  祶 training samples, the established neural network model for training. The model was so preoccupied with calcined palm trees as a result that the response time was less than 10ms. This article 嵦 岢 龅 Pepper ㄊ grab  Meng 窬  缡 Di ⒒ ⒒ 溲 蛊 Mad Marting;