中文摘要：

主要研究了磁浮列车悬浮系统加速度传感器的故障诊断问题。由于悬浮系统模型难以精确建立，因此采用了基于信号比较的故障诊断方法。提出将冗余的间隙传感器微分信号与加速度积分信号进行比较，采用贝叶斯决策理论为比较结果设计阈值以判断是否发生故障。针对微分信号难于提取的问题，提出采用非线性离散跟踪微分器来提取间隙微分信号，以降低由微分导致的噪声，并与传统线性微分器的效果进行了比较。最后提出了在加速度计故障之后采用间隙微分信号进行反馈的容错方案并进行了实验验证。

英文摘要：

Accelerometer fault diagnosis for maglev suspension system is studied. The fault diagnosis method based on signal comparison is adopted to overcome the difficulty of setting up a precise model. The differential signal of re- dundant gap sensors is compared with the integral signal of accelerometer. The result is used to set a threshold value for indicating faults based on Bayesian statistics theory. Considering the difficulty of getting differential signal, non- linear discrete tracking differentiator （TD） is adopted to improve the quality of differential signal, and its perform- ance is compared with that of conventional differentiator. At last, a fault tolerant strategy is proposed and proved through experiment, which replaces the integral signal of the accelerometer with the differential signal of the gap sensor when the accelerometer goes wrong.