Ball Screw Health Monitoring With Inertial Sensors

Abstract

In industrial applications, the mechanical wear on ball screw components can lead to a loss of positioning accuracy that reduces the operational reliability and reproducibility of production systems. Existing monitoring solutions are impractical for real industrial settings or are unable to provide quantifiable estimates of the magnitude of degradation. To address this, the proposed method strategically applies a two-phase data collection with inertial sensors to perform both health monitoring and fault magnitude estimation. The first, online phase offers a practical, non-intrusive means of monitoring the ball screw degradation during normal production operations. As deemed necessary by the first phase, the second, offline phase is implemented outside the production routine to physically quantify the detected fault. The combined methods offer a balanced approach that provides detailed information while still considering the requirements of a production environment. To validate the performance of this proposed strategy, a run-to-failure experiment was performed on a linear axis testbed. Validation results indicate that the method is a pragmatic and promising approach for incipient fault detection and absolute backlash error measurement in a linear axis.