YOLO-v7 Improved With Adan Optimizer: Realizing Orphaned Fiber Bragg Grating to Sense Superimposed Personalized Dynamic Strain
In recent years, condition monitoring has become the leading method for diagnosing the health of motor-based machines, with motor vibration or strain being a critical parameter for assessing machine health and detecting abnormal vibrations in multiple running motors posing a significant challenge. T...
Gespeichert in:
Veröffentlicht in: | IEEE sensors journal 2024-01, Vol.24 (23), p.39923-39933 |
---|---|
Hauptverfasser: | , , , , , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext bestellen |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | In recent years, condition monitoring has become the leading method for diagnosing the health of motor-based machines, with motor vibration or strain being a critical parameter for assessing machine health and detecting abnormal vibrations in multiple running motors posing a significant challenge. This article presents a novel approach to condition monitoring of motors that ensures the prevention of motor wear, reduces high maintenance costs, and increases the durability of machines. The method uses a single fiber Bragg grating (FBG) sensor to sense the vibrations of three electrical motors, and you only look once version 7 (YOLO-v7) as the signal detection algorithm. The Adan optimization technique is used to enhance the YOLO-v7 performance. A maximum of eight possible cases of normal and abnormal dynamic strain can be generated by these three motors operating simultaneously. YOLO-v7 ensures and evaluates the normal and abnormal signals of each motor. The detection result demonstrates the model accuracy of 98.4%. The model performance shows that strains from the motor machine are accurately detected, indicating anomalies. Thus, our proposed experimental setup is flexible, cost-effective, robust, less computation, fast, and improves the sensing quality in anomalies in machine conditioning monitoring. |
---|---|
ISSN: | 1530-437X 1558-1748 |
DOI: | 10.1109/JSEN.2024.3474747 |