基于叶端定时的转子叶片裂纹监测与诊断

为了实现转子叶片裂纹非接触式监测,提出基于??0范数的叶端定时欠采样信号稀疏重构算法。针对叶端定时信号欠采样问题,建立欠采样信号稀疏表示模型,利用分块正交匹配追踪算法对叶端定时欠采样信号进行分段重构,进而监测在变转速非稳态工况下转子叶片动频的变化情况,据此判断转子叶片是否产生裂纹。针对应变片信号间隔采样的特点,研究基于同步提取变换的时频分析方法,以提高时频图的时频聚集性。开展旋转叶片裂纹扩展试验,同时采用叶端定时和应变片系统测量叶片振动。对比叶端定时和应变片的分析结果,二者均可在裂纹叶片断裂之前至少20min,监测到裂纹叶片和正常叶片的动频发生分离的现象,即裂纹萌生致使叶片动频发生偏移,表明所提出的叶端定时欠采样信号重构方法能够实现转子叶片裂纹的监测与诊断。

Crack Monitoring and Diagnosis of Rotor Blades Using Blade Tip Timing

A sparse reconstruction algorithm based on -norm for reconstructing undersampled signals of blade tip timing is proposed to achieve the non-contact monitoring of crack rotor blades. To deal with the undersampled blade tip timing signals, a sparse model for reconstructing undersampled signals is established. Blade tip timing signals are reconstructed by using the block orthogonal matching pursuit algorithm. The frequency shifts of rotor blades are tracked to monitor crack initiation under a variable speed. To improve the energy concentration of the time-frequency result, synchroextracting transform with high time-frequency resolution is developed to analyze the strain gauge signals using an interval sampling strategy. In a spin testing, blade vibration is measured by both blade tip timing and strain gauge systems simultaneously. According to the analysis results, both blade tip timing and strain gauge can detect the frequency shift of cracked blade at least 20 minutes before the blade fracture. It means that the crack initiation leads to the frequency shift of the blade. The results demonstrate that the proposed reconstruction method for undersampled blade tip timing signals is useful to achieve the crack monitoring and diagnosis of rotor blades.