中文摘要：

机电设备运行状态的监测对保障系统稳定可靠运行、预防重大事故发生有重要意义。针对传统诊断方法由于故障信息不足导致的诊断精确度差，提出了一种基于主特征模式识别的故障诊断方法。基于多源特征信息融合，研究了基于多传感器系统的特征融合故障诊断模型，讨论了反映系统运动状态特征的指标体系及故障诊断算法。文中以滚动轴承系统故障诊断为例，首先计算了各传感器获取信号的时域特征参数，然后，借助主特征模式对特征信息进行融合与降维处理，实验测试数据显示出与传统诊断方法相比较该算法有更好的故障诊断性能。研究结果表明了该方法在重型机电设备故障诊断中应用的可行性与合理性。

英文摘要：

In electromechanical engineering, it has very important significance to monitor the running state of device to ensure the system is under a stable and reliable condition and to prevent the occurrence of significant fault. Since the conventional diagnosis method has poor accuracy due to the shortage of fault information, a sort of diagnosis algorithm of system fault is proposed based on main feature information pattern in this paper. The algorithm is based on the information fusion of multi-source feature. The fault diagnosis model is comprehensively studied based on feature fusion of multi-sensors system and the indices system of state feature reflected the system motion and the algorithm of fault diagnosis are fully discussed. In this paper, the fault diagnosis of antifriction bearing system is taken as an example. The feature parameters of time domain in signal gained from each sensor in this system are firstly computed, after that, the fusion and dimension reduction processing for feature information is made by means of main feature pattern, and then the test of antifriction bearing system is made. Compared with the conventional diagnosis method, the experimental data showed that the present diagnosis algorithm could obtain better performance of fault diagnosis. Therefore, the present algorithm is feasible and capable to be applied in the fault diagnosis for heavy electromechanical device.