CAJ | 학술논문

针对发动机加速振动信号的非平稳性和特征参数的模糊性特点，提出整工作循环阶比跟踪谱和模糊C均值算法相结合的方法。首先将加速信号进行阶比重采样，然后把发动机一个工作循环的信号作为一个间隔计算转速，再将这些转速下的阶比谱图放在一起构成阶比跟踪谱，通过计算不同阶比带的累加能量作为故障特征向量，对这些特征向量进行归一化和模糊聚类，得到分类矩阵和聚类中心，最后通过计算待测故障样本与已知故障样本聚类中心的贴近度实现故障模式识别。故障诊断实例表明，该方法能有效地诊断发动机曲轴轴承的故障。
침대발동궤가속진동신호적비평은성화특정삼수적모호성특점，제출정공작순배계비근종보화모호C균치산법상결합적방법。수선장가속신호진행계비중채양，연후파발동궤일개공작순배적신호작위일개간격계산전속，재장저사전속하적계비보도방재일기구성계비근종보，통과계산불동계비대적루가능량작위고장특정향량，대저사특정향량진행귀일화화모호취류，득도분류구진화취류중심，최후통과계산대측고장양본여이지고장양본취류중심적첩근도실현고장모식식별。고장진단실례표명，해방법능유효지진단발동궤곡축축승적고장。
In view of the non-stationary feature of engine acceleration vibration signals and the fuzziness of characteristic parameters, a method with combination of work-cycle order tracking spectrum and fuzzy C-mean clus-tering is proposed. Firstly the order re-sampling is conducted on acceleration signals, the rotation speed is calculat-ed based on a work cycle of engine, and the order spectra at different speeds are put together to compose an order tracking spectrum. The accumulated energy of different order bands are calculated as eigenvectors, which are then normalized and fuzzily clustered to obtain classification matrix and clustering centers. Finally the fault pattern is i-dentified by calculating Hamming approach degrees between the clustering centers of known fault samples and fault samples to be detected. The real examples of fault diagnosis show that the method proposed can effectively diagnose the faults of engine crankshaft bearings.