针对旋转机械振动的监测与诊断 ,提出了一种分级自适应的多传感器数据融合方法 .每个测点首先进行传感器级的时间跨度的特征融合 ,然后再进行全局的空间跨度的融合 .融合过程基于自适应模式特征提取 ,提出的方法可以在系统状态未知的情况下 ,自适应地融合不同测点的信息 ,从而较全面、准确、及时地反映系统的振动状态 .对转子试验台上模拟故障数据的融合处理 ,说明了所提出方法的有效性 . Aiming at the monitoring and diagnosis of rotating machinery vibration, a hierarchical adaptive multisensor data fusion method is proposed, in which each sensor point first combines the features of the sensor-level time span and then the global spans. Based on the adaptive pattern feature extraction, the proposed method can adaptively fuse the information of different measuring points under the condition of unknown system status, so that the system vibration state can be reflected more comprehensively, accurately and timely. The fusion of data shows the effectiveness of the proposed method.