A robust FOD acoustic detection method for rocket tank final assembly process

Abstract

The presence of Foreign Object Debris (FOD) during the assembly of rocket tanks poses a significant risk to the success of rocket launches. The current manual listening method which relies on the human ear for detecting FOD in rocket tanks is limited by its low level of automation and poor robustness. To overcome these limitations, this study presents a Robust Foreign Object Debris Detection (RFODD) method that utilizes an Adaptive Wiener Filter (EDAWF) based on energy variation and an Adaptive Forgetting Factor Gaussian Mixture Model (AFFGMM). EDAWF improves the input signal quality, while AFFGMM addresses the challenge of updating parameters under time-varying signals. FOD detection is then performed through a multi-segment counting method and a criterion based on the anomaly percentage index. The proposed RFODD method was evaluated through an experimental study on a rocket tank and was found to offer a more robust and automated solution for FOD detection with engineering applications.