基于小波-DHNN识别变压器励磁涌流  

Recognition of Transformer Inrush Current Based on Wavelet-Discrete Hopfield Neural Network

作　　者：公茂法[1] 李岚冰[1] 于晓春[1] 王志文[1] 安彬[1] 

机构地区：[1]山东科技大学信息与电气工程学院,山东青岛266590

出　　处：《电测与仪表》2014年第2期19-22,71,共5页Electrical Measurement & Instrumentation

基　　金：山东省自然科学基金(ZR2012EEM021);山东科技大学研究生科技创新基金(YC130328);山东省大学生学术课题(自然科学类)基金(13CZR001)

摘　　要：变压器主要采用纵联差动保护,如何防止因涌流造成的误动已成为关键性问题。对于该问题,提出一种基于小波-DHNN识别励磁涌流的新的研究方案。利用小波变换对采样信号进行分析,得出励磁涌流的小波系数较内部故障电流有非常明显的差异,并且畸变特点伴随整个衰减过程。分析后的信号通过离散型Hopfield网络测试与识别,从而区分励磁涌流和内部故障电流。通过PSCAD和MATLAB仿真软件进行建模仿真,结果表明,该方法能可靠的识别励磁涌流和内部故障电流,并且准确率高达100%。Transformers mainly take use of the longitudinal differential protection. It becomes a key issue to prevent malfunction caused by inrush current. For solve this problem, a new method is put forward to identify inrush current based on Wavelet-Discrete Hopfield neural network. Through the analysis of sampling signals by wavelet transform, it is found that the wavelet coefficients of the inrush current has a significant difference with the internal fault current, and the entire decay process is accompanied by distortion. Then the signals are tested and identified through Discrete Hopfield neural network, so as to distinguish excitation inrush current and internal fault current. PSCAD and MATLAB are used to accomplish the simulation models. The simulation results show that this method can distinguish inrush current and internal fault current reliably, and the accuracy rate is up to 100%.

关 键 词：小波变换 离散型Hopfield神经网络 PSCAD 励磁涌流 变压器 

分 类 号：TM406[电气工程—电器]