Establishing a cost-effective sensing system and signal processing method to diagnose preload levels of ball screws

Abstract

This paper presents an embedded sensing system for precisely measuring acceleration and temperature of interest points on a ball screw structure and diagnosis of different ball-screw preloads based on processing acquired signals with further classification using the support vector machine (SVM) method. The sensing system consists of a sensing unit and a hardware signal-processing unit. The core sensors utilize a MEMS-type accelerometer and glass-type SMD PT-100, integrated into a 1cm×1cm circuit board and packaged with a metal housing sensing unit with a dimension less than 1.5cm3. The sensing unit is embedded into the screw nut of a designed preload-adjustable ball screw, installed on a computer-controlled single-axis stage for testing. Acquired signals with good noise immunity in an industrial environment are achieved through the developed hardware signal-processing unit. Measured acceleration and temperature data for different ball-screw preload levels based on time and frequency domain analysis are performed. The results demonstrate achieving diagnosis of a ball-screw preload within 20s, with a preload level classification reaching nearly 100%. The developed sensing system and analysis method can apply to monitor ball-screw health and would be very useful in industrial applications.