Comparative Evaluation of Structural Surface Intensity to Statistical Features for Gearbox Failures

Abstract : The key to an effective Condition-Based Maintenance (CBM) program lies in the ability to extract machinery health information through diagnostic and prognostic indicators. In an effort to develop such indicators, the CBM department at the Penn State Applied Research Laboratory (ARL) has evaluated the use of structural surface intensity (SSI) for diagnostics and prognostics. In order to characterize structural surface intensity's effectiveness as a machinery diagnostic indictor, transitional fault data for three failure modes of an industrial grade gearbox was generated and SSI parameters are extracted and compared to the more widely used statistical-based features. The comparisons were focused on early detection capability and the relative change of the indicators subsequent to fault initiation. Results of such comparisons are provided for the three test runs. The comparisons show that in certain cases, SSI, as a diagnostic indicator, may provide an earlier detection capability and result in higher decision confidence than those obtained using the "traditional" statistical-based features.