分数延时时域平均方法在风电齿轮箱诊断中的应用  

Application of Fractional Delay Filtering Based on Time Domain Averaging Technique in Wind Turbine Gearbox Diagnosis

作　　者：邹今春[1,2] 朱增兵[1] 瞿雷[2] 沈玉娣[2] 

机构地区：[1]中国电建集团西北勘测设计研究院有限公司,陕西西安710065 [2]西安交通大学机械工程学院,陕西西安710049

出　　处：《水力发电》2015年第4期69-73,共5页Water Power

摘　　要：针对变速工况下风电齿轮箱频谱结构复杂,传统频谱分析难以对故障进行有效识别和定位的问题,将分数延时滤波技术与时域平均方法相结合,采用一种基于啮合频率预估与重估的方法进行转频跟踪实现信号分段截取,提出一种线性时变延时滤波算法实现在定采样频率、变回转速度条件下对信号采样点数进行统一,实现了无时标变转速时域平均。风电齿轮箱现场故障案例验证了方法的有效性。The failures in wind turbine gearbox under time-varying speed conditions can＇t be easily detected or diagnosed with conventional frequency spectrum analysis as the complexity and strong noisy background of spectrum. An integrated time domain averaging （ITDA） method combining fractional delay filtering and time domain averaging is proposed. To overcome the obstacles due to tacho sensor inaccessibility in speed dynamic tracking, a technique based on EDFT and meshing frequency locating is employed. By fore-estimating and updating, the variation of fundamental frequency is closely tracked. A new structure for time-varying delay implementation is formulated on the basis of Lagrange FIR band limited filters which facilitates an efficient algorithm for fluctuating period normalizing. The on-site diagnosis of a wind turbine gearbox has verified the effectiveness of proposed method.

关 键 词：分数延时滤波 时域平均 风电齿轮箱 变工况 

分 类 号：TM614[电气工程—电力系统及自动化]