基于线性相位的风机齿轮箱无键相同步平均分析

Linear phase estimation based wind turbine gearbox time synchronous averaging without keyphasor

风电机组转速信号多由发电机编码器提供,由于SCADA与CMS存在时间差和采样率差异,难以实现高速轴键相信号与齿轮箱振动信号的同步采集。基于时域同步平均的阶次分析技术较难在风机齿轮箱故障诊断领域得到大范围应用,提出了以线性相位估计为基础的、可适应转速波动为10%的风机齿轮箱无键相同步平均分析方法。选择输出级啮合频带中信噪比最高频段,利用傅里叶FIR"理想"滤波技术,实现严格线性相位保持的窄带滤波;通过希尔伯特变换提取带通信号复相位,并进一步通过线性插值合成过零点序列;以此为基础,完成基于软件的等角度重采样,进而实现无键相的时域同步平均分析。在某型号机组连续10 s的现场实测数据中,有效地验证了所提方法正确性和工程实用性。

Aiming at the problem of traditional Time Synchronous Averaging analysis（TSA） can＇t be used when there is hard to acquire one pulse per revolution signal synchronous with acceleration signal after wind turbine has been moved out factory, a linear phase based, with speed fluctuation less than 10% wind turbine gearbox keyphasor-less TSA algorithm has been raised. FFT / IFFT based ideal filtering algorithm has been used first to choose a pass-band which has the highest signal to noise ratio among several gear meshing harmonics; then phase extracted in complex domain after Hilbert transform calculation of the band-pass signal; based on linear interpolation,zero crossing index is acquired last and angular resampling can be carried out. Real case study results of some wind turbine under constant rotation speed and vary speed show that： algorithm raised in this paper has advantage of little prior knowledge needed, but high-precision when compared with actual keyphasor analysis result, thus providing a reliable keyphasor-less TSA method for wind turbine gearbox in actual use.