This paper is about development of an AE signal-based monitoring system for detecting defects on the bearings and rails of an LMS which is widely used in the automated production lines. It is examined whether the defects really contribute to generation of the AE signal and what frequency components of the AE signal are closely related to the defects. The cepstrum analysis is adopted to get a time interval of the purely defect-driven AE wavelets, which is compared with the theoretical interval based on the kinematic modeling of LM rail and bearing of an LMS. From the defect-driven AE wavelet picked up from a series of AE signal, the frequency features very unique to each defect of Bearing/Rail are extracted. The features are grouped three bands, that is, 100-150, 150-200, 400-450 kHz. Three-band power spectrum shows different patterns each other for normal, rail defect, small bearing defect and worse bearing defect. Through these experiments, it is verified that AE signal is useful to detect the bearing/rail defect and a couple of bands around the AE frequency features can be used as a real-time monitoring device for the status monitoring of LMS.