Dual-sensor blade tip timing with multi-source information and adaptive dictionary update

Abstract

Blade tip timing (BTT) has emerged as a critical non-contact vibration measurement technique, increasingly supplanting traditional strain gauge methods for monitoring the operational safety of high-speed rotating machinery, such as aircraft engines and gas turbines. Despite its advantages, conventional BTT approaches face several limitations: undersampling often leads to the loss of high-frequency signals; the requirement for multiple sensors and stringent layout specifications escalates system complexity and cost; and dependence on a single displacement data source restricts the extraction of comprehensive vibration features. Furthermore, traditional BTT methods lack adaptability to varying operating conditions, impeding accurate signal reconstruction. To overcome these challenges, this study proposes a novel dual-sensor BTT sparse reconstruction method. This approach leverages multi-source information, including both displacement and velocity data, to adapt to diverse operating conditions. It employs a dynamic dictionary update process based on the primary frequency range during sparse reconstruction, eliminating the need for at least four sensors to reconstruct synchronous vibration signals under constant speed conditions. This innovation reduces sensor layout constraints and enhances the robustness and accuracy of signal reconstruction. The proposed method is experimentally validated using dual capacitive and dual eddy current sensors for vibration signal identification. The results indicate that integrating multi-source information significantly improves the accuracy and reliability of vibration signal reconstruction, rendering the method highly effective for health monitoring and fault diagnosis in high-speed rotating machinery.