CAJ | 학술논문

STATCOM主回路主要由电力电子器件构成。电力电子器件过载能力小，损坏速度快，故障发生前征兆较难捕捉。设计合理有效的STATCOM主回路故障诊断方法可在故障发生后及时对故障管进行定位，尽快排除故障，使装置恢复正常运行。人工神经网络技术具有很强的非线性映射特征和自学习能力，非常适用于STATCOM主回路电力电子器件的故障诊断。以三相三桥臂STATCOM为研究对象，建立其主回路IGBT或续流二极管单管开路故障模式并提取故障特征，根据神经网络非线性映射特性，建立BP神经网络来学习及存储故障特征信号和故障类型的映射关系。 MATLAB仿真表明，该神经网络能够准确地诊断和定位STATCOM主回路开关元件单管开路故障，具有很好的故障识别能力。
STATCOM주회로주요유전력전자기건구성。전력전자기건과재능력소，손배속도쾌，고장발생전정조교난포착。설계합리유효적STATCOM주회로고장진단방법가재고장발생후급시대고장관진행정위，진쾌배제고장，사장치회복정상운행。인공신경망락기술구유흔강적비선성영사특정화자학습능력，비상괄용우STATCOM주회로전력전자기건적고장진단。이삼상삼교비STATCOM위연구대상，건립기주회로IGBT혹속류이겁관단관개로고장모식병제취고장특정，근거신경망락비선성영사특성，건립BP신경망락래학습급존저고장특정신호화고장류형적영사관계。 MATLAB방진표명，해신경망락능구준학지진단화정위STATCOM주회로개관원건단관개로고장，구유흔호적고장식별능력。
The main circuit of STATCOM is composed of power electronic devices.Due to the fact that the swit-ches have a low overload capacity and can be damaged in an instant and the sign of a coming failure is difficult to catch, it’ s necessary to devise a reasonable and effective fault diagnosis method for the STATCOM main circuit, which can locate and debug the faulty switches in time, and clear the failure as soon as possible, and then the de-vice can be back to normal running.With the strong nonlinear mapping characteristics and self-taught ability, ap-plication of artificial neural network technique has been widely used in STATCOM of the fault diagnosis of the main circuit of the power electronic device.Set the three-phase bridge arm STATCOM as its research object, the main loop of the IGBT or the freewheeling diode single open circuit fault modes was established and the fault feature was extracted.Based on the nonlinear neural network mapping characteristics, BP neural network method of fault diag-nosis was used, and the structure and parameters were designed to train the network.MATLAB simulation showed that The artificial neural network can accurately diagnose and localize the STATCOM main circuit switching element single open circuit faults and has a good ability in fault identifying.