The 8th International Conference on System Reliability and Safety
Sicily, Italy - November 20-22, 2024

Marco Cocconcelli, University of Modena and Reggio Emilia, Italy

 

Marco Cocconcelli received the M.S. degree in mechanical engineering and the Ph.D. degree in applied mechanics from the University of Bologna, Italy, in 2003 and 2007, respectively. In 2007, he joined the University of Modena and Reggio Emilia, Reggio Emilia, Italy, where he is currently an Associate Professor in applied mechanics with the Department of Sciences and Methods for Engineering. His research interests include diagnostics of bearings and gears,monitoring of machinery in stationary and non-stationary conditions, history of mechanisms and machine science. He is a member of the Permanent Commission for the History of Mechanism and Machine Science and of the Technical Committee for Reliability of the International Federation for the Promotion of Mechanism and Machine Science (IFToMM).

Speech Title: Fault Detection Algorithms for Rolling Bearings: The Role of Noise Modeling

Abstract: The condition monitoring of rolling bearings is gaining significant attention because production slowdowns are largely caused by damage to these components. In the past decade, numerous algorithms for fault detection have been introduced in the technical literature. Typically, their performance is evaluated using both synthetic and experimental data. However, the signal models used in computer simulations are often simplistic and fail to accurately predict real-world performance. In particular, the paradigm of white gaussian noise, despite reasonable from a theoretical point of view, it does not correspond to experimental observations. Real machine noise can best be represented as a wideband component, potentially overlapped with narrowband components. In this talk, the wideband component is modeled in three different ways: as additive white Gaussian noise, additive white noise from an alpha-stable distribution, and additive noise derived from an autoregressive process. Narrowband components are modeled using sequences of Gaussian pulses. The performance of three well-known fault detection algorithms is compared based on their ability to identify theoretical cyclic frequencies associated with damage. In these scenarios, the behavior of fault detectors significantly diverges from the predictions made by traditional wideband noise models, such as additive white Gaussian noise.