Methods For Sensor Layout Planning And Weak Signal Detection With Applications To Machine And Manufacturing Process Fault Diagnosis

Sensing and signal processing are the key techniques for condition monitoring and fault diagnosis. In this thesis, several approaches for sensor layout planning and weak signal detection are addressed with applications to machine and manufacturing process fault diagnosis. The main contents of the thesis are summarized as follows.1. Several diagnostic problems arising from manufacturing processes, i.e., diagnosis of fixture fault, diagnosis of geometric imperfection and diagnosis of beam structure deformation, are modeled as a unified linear parameter estimation problem. Case of diagnosis of fixture fault is mainly discussed in this paper. A method of model-based fault diagnosis, based on the least squares estimation method is developed. This method not only can detect and classify multiple faults, but also can be applied to the situation of mean fault and variance fault coexisting. It starts with the variance detection, if no variance fault is detected in the i thfault, and then continues to do the mean fault detection. Thus, on the one hand, the repeating detection of the same i thfault is avoided; on the other hand, the possibility of deriving the wrong diagnostic conclusion is avoided if only detect one of the two types of faults while ignore the other one existing.