摘要
对发电机故障放电信号特征进行自动识别时,实现方法要求具有较高的可靠性与快速性。为达到这些要求,首先提出一种高可靠的脉冲提取方法,能够更为准确完整地提取脉冲信号,通过仿真方式对该方法进行了研究,并使用现场实测信号验证了该方法的优越性。在该方法基础上,提出了脉冲分类参数的快速计算方法,在进行脉冲提取的同时完成脉冲分类参数的计算,使用实测信号验证该方法的有效性。在检测脉冲的存在并对脉冲进行高可靠提取的同时,获取脉冲信号的分类特征参数,而且计算速度快,将其称为脉冲特征快速自动识别技术。该技术已成功运用在多台水轮发电机上,在实际应用中取得了良好的效果。
The method to automatically recognize the signal feature of fault discharge in generator should possess both reliability and rapidity. To meet the demand, firstly a high reliable pulse extraction method is proposed to extract pulse signals, which are submerged in noise, more accurately and completely, and the simulative research on such a method is performed and its superiority is validated by on-site measured signals. Based on this method, a fast calculation approach for sorting parameters of pulses, which can calculate sorting parameters of pulses during the extraction of pulses, is proposed, and this method is verified by actually measured signals. Since the characteristic parameters for the pulse signal classification can be fast attained by the proposed method and in the meantime whether the pulse exists or not is detected and the pulse is extracted rapidly and reliably, the proposed method can be called as the pulse feature fast recognition technique （PFFR）. This method has been successfully applied to multi hydropower generators and good results have been achieved.
出处
《电网技术》
EI
CSCD
北大核心
2015年第2期543-549,共7页
Power System Technology
关键词
发电机
故障放电
实时在线
特征自动识别
状态监测
generator
fault discharge
real-time online
automatic feature recognition
condition monitoring
作者简介
孙文星（1987），男，博士研究生，研究方向为高压电气设备状态监测与故障诊断，E-mail：swx_hust@ymail．com；
李朝晖（1963），男，通讯作者，博士，教授，博士生导师，研究方向为发电生产过程控制、电站集成自动化、计算机仿真与自动测试；
程时杰（1945），男，教授，博士生导师，中国科学院院士，IEEEF ellow，研究方向为人工智能在电力系统中的应用、电力系统运行与控制、超导电力等。
