Soutenance de thèse : Kayacan KESTEL

« Development of signal processing techniques for vibration-based condition monitoring of industrial rotating machines »

Doctorant : Kayacan KESTEL

Laboratoire INSA : LVA

École doctorale : ED162 : MEGA de Lyon (Mécanique, Energétique, Génie civil, Acoustique)

This dissertation introduces novel signal processing techniques and enhances existing ones for vibration-based condition monitoring of complex industrial rotating machines. The vibration-based condition monitoring field literature provides many solutions to reveal faulty patterns from a vibration signal. However, their performance on the vibration signals of modern complex machines has not been tested or is generally limited. Furthermore, they may often lack a robust version ready to be applied to real measurements. This study addresses these limitations with several proposals. Key research areas include the development of a stable blind filtering method to optimize Finite Impulse Response (FIR) filters for efficient fault detection and testing the proposed method on vibration signals obtained from a wind turbine gearbox. Additionally, the thesis advances the use of Generalized Likelihood Ratio test (GLRT)-based indicators for more stable and effective monitoring and proposes new statistical indicators with defined thresholds for improved reliability. Furthermore, it explores the impact of operational and environmental conditions on the vibratory behavior of wind turbine drivetrains, aiming to enhance the understanding and assessment of gearbox health through dynamic behavior analysis under varying conditions. These contributions offer substantial advancements in the practical application of signal processing in condition monitoring.