Résumé:
The main objective of this thesis is the development of an approach for diagnosing gear faults, both in steady-state and variable operating conditions. The proposed methodology is based on two different but complementary approaches: a subjective approach based on auditory perception and an objective approach based on vibration analysis. Indeed, this analysis allows for the detection of faults and anomalies by interpreting vibrational signals, independently of rotational speed fluctuations. The perceptual analysis, on the other hand, focuses on the identification of abnormal sounds related to gear faults in both steady-state and variable operating conditions. By combining these two analytical approaches, we can obtain valuable information about the health status of rotating machinery in order to implement effective maintenance strategies to improve the reliability of these machines in various operational scenarios.
After conducting a literature review and synthesis, the research primarily focuses on a comparative study between objective and subjective methods to identify the severity of single and multiple gear faults based on measured signals in a noisy environment. This in-depth analysis helps to better understand the strengths and weaknesses of each method and provides recommendations for their application in real scenarios with noisy signals. Consequently, this contribution leads to the expansion of knowledge regarding objective and subjective approaches for gear fault analysis.
Secondly, an optimized hybrid method for detecting gear faults under non-stationary conditions has been proposed. This approach combines theImproved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (ICEEMDAN), wavelet denoising, and order analysis. The results demonstrate the effectiveness of this method in detecting various types of gear faults under different rotational speed variation modes.
Finally, perceptual tests were conducted at ENSIM, Le Mans, France, in a semi-anechoic chamber to simulate various types of gear faults. What makes these tests unique is that measurements were taken under variable-speed conditions, including phases of acceleration, steady-state, and deceleration. This approach allows for the assessment of gear fault perception under realistic operating conditions. The results of these tests provide valuable insights into the subjective detection of gear faults in dynamic environments.
