Application of Data Dependent Systems to Diagnostic Vibration Analysis

Abstract

Application of Data Dependent Systems (DDS) is proposed for diagnostic monitoring of vibrations. It is exemplified by monitoring tool wear via tool holder vibrations in a turning operation. Hardware Fourier analysis shows that only a relatively less prominent set of peaks in the power spectra are affected by wear making it difficult to use them for diagnostic purposes. The DDS model, however, gives a comprehensive quantitative description of the various modes and succeeds in isolating the mode sensitive to wear. It is shown that the damping ratio of this mode provides a sensitive and accurate indicator of wear, which can be used to diagnose failure much before it occurs. Finite element analysis confirms and elucidates the physical significance of DDS results. The DDS approach is thus useful not only for on-line monitoring but also for detailed research leading to theoretical development and improved performance.