The ball screw system is one of the most critical components in advanced manufacturing and is expected to perform with high accuracy. However, any potential failures or degradation of mechanical parts in the system would affect its efficiency and position precision; even cause severe machining errors or breakdown. This paper mainly focuses on the fault diagnosis of ball screw system components. In order to classify multiple failure modes, one full size ball screw testing machine is set up to replicate different health conditions including four failure modes — lubrication starvation, preload loss, ball nut wear, and re-circulation system failure. In this paper, the first two failure modes are introduced. Time domain and frequency domain features have been extracted from the vibration and temperature signals. A classification modeling method is used to establish a ball screw system health map. The direction of the threads on the screw shaft causes different vibration patterns when ball nut travels forward and backward. Thus, failure signatures from both traveling directions are investigated in the paper. Based on the developed health map for the ball-screw, the health values can be calculated to quantify the failure severity. Furthermore, from the perspectives of accuracy and online application efficiency, three health assessment methods, Self-Organizing Map - Minimum Quantization Error (SOM-MQE), Mahalanobis distance (MD) and Gaussian Mixture Model (GMM) are compared in the study.

This content is only available via PDF.
You do not currently have access to this content.