基于灰关联信息融合的转子系统早期故障发现研究

Investigation of early faults detection of a complex rotor system based on grey correlative information fusion

设置振动信号的FFT谱序列为特征量,在稳态运行下的参考特征量与异常运行状态下的待辨识特征量间建立了通道对应的灰关联空间并将两者进行灰关联分析.对安装有多传感器的转子系统在灰关联空间内设置融合策略进行转子系统的早期故障发现研究.为消除速度波动造成的关联失真影响,对FFT谱特征量进行了基于工频的归一化处理.实验分析表明,谱特征量在关联空间内的灰关联分析及基于关联得到的关联度评价了系统的异常分布状态,关联度量化了该分布的程度,灰关联度最小的信源是系统内的故障发生部位.

Setting the FFT spectra of journal vibration signals of a complex rotor-bearing system as char-acteristic quatities, the unipath grey correlative space between the reference characteristic quantity in steady-state operation and the chacteristic quantity to be identified in abnormal operation was established, and both the characteristic quantities were analyzed by means of grey correlation. The fusion strategy was set in the grey correlative space of a rotor system with multi-transdusers to conduct the investigation of early fault detection. In order to eliminate the correlation distortion due to speed oscillation, the characteristic quantity of FFT spectra was normalized. It is shown by experimental analysis that the abnormal distribu-tion state can be assessed with the correlative fusion measure which is obtained by means of grey correlative analysis in the correlative space as well as on the basis of the correlation when the spectra is taken as the characteristic quantity. The correlativity has quantized the degree of this distribution, so that the signal source with minimum grey correlativity would be the location of fault generation in the system.