Abstract: In order to improve the accuracy of motor fault diagnosis, a method of multi sensor information integration is applied. After the vibration spectrum signal collected with multi sensors processed, the ant colony algorithm-neural network was used to implement local fault diagnosis and acquire each other independent evidence, then the evidence theory was employed to fuse them. Ultimately diagnosis of motor faults was realized. Experimental results show that this method is effective to improve diagnostic reliability and reduces the uncertainty in motor fault classification and recognition.

Key words: vibration and wave ; , information fusion, motor , fault diagnosis , evidence theory , ant colony algorithm -neural network

摘要： 为了提高电机故障诊断的准确性，引入一种多传感器信息融合的诊断方法。将多个传感器所采集的转子振动频谱信号处理后，利用蚁群神经网络进行故障局部诊断，以获得彼此独立的证据，再由证据理论对各证据进行融合，最终实现对电机故障的准确诊断。实验结果表明，该方法有效提高诊断的可信度，减少电机故障分类识别的不确定性。

关键词: 振动与波, 信息融合, 电机, 故障诊断, 证据理论, 蚁群神经网络