Markov Latent Frequency Transition Analysis for Robust Bearing Diagnosis in Transient Noise Scenarios

In real-world applications, the diagnostic accuracy of rolling bearings is often compromised by environmental factors such as variable rotational speeds, transient noises, and unexpected mechanical impacts. These factors can obscure damage symptoms, making precise detection of bearing failures challenging. Conventional diagnostic methods struggle to handle these complexities and often miss failure features hidden in multifaceted random transient noise. To address these issues, this research introduces the Markov Latent Frequency Transition Peak Rate (MLFTPR) methodology. MLFTPR focuses on accurately monitoring temporal state transitions and identifying anomalous signals, aiming to mitigate transient noise at the source and improve resistance to external disturbances. This enhances precision and reliability in demodulation band selection. Additionally, an amplitude interference-reduction feature is integrated to recognize and suppress transient noise effectively. Experimental results validate the efficacy of MLFTPR, demonstrating its capability to accurately diagnose bearing defects even amidst transient disturbances.