Speed and Anomaly Detection using Accelerometers
AHMED, SAYED (2025)
2025
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2025060219259
https://urn.fi/URN:NBN:fi:amk-2025060219259
Tiivistelmä
This thesis focused on anomaly detection in fan systems using the accelerometer sensor from the NICLA Sense ME platform. The aim was to develop a reliable method for analyzing sensor data to identify irregularities in fan operations, such as imbalances, wear, or malfunctions. Accurate anomaly detection is crucial for predictive maintenance, which helps ensure the longevity and efficiency of mechanical systems like fans.
The methodology involved collecting accelerometer data from the NICLA Sense ME platform during both the normal fan operation and under various fault conditions. Machine learning techniques, including time-series analysis and classification algorithms, were employed to differentiate between normal and anomalous states. Feature extraction and signal processing techniques were applied to preprocess the raw sensor data, reducing noise and improving data quality to enhance model accuracy.
The study made use of the NICLA Sense ME device, a standard fan unit, and a computer equipped with the necessary software for data processing and analysis. MATLAB was used for feature extraction and to train anomaly detection models.
The results demonstrated the system's ability to detect anomalies with high accuracy, confirming the effectiveness of accelerometer-based monitoring for predictive maintenance. This approach holds significant potential for reducing operational downtime and lowering maintenance costs in industrial environments, ensuring the continued performance and reliability of fan systems.
The methodology involved collecting accelerometer data from the NICLA Sense ME platform during both the normal fan operation and under various fault conditions. Machine learning techniques, including time-series analysis and classification algorithms, were employed to differentiate between normal and anomalous states. Feature extraction and signal processing techniques were applied to preprocess the raw sensor data, reducing noise and improving data quality to enhance model accuracy.
The study made use of the NICLA Sense ME device, a standard fan unit, and a computer equipped with the necessary software for data processing and analysis. MATLAB was used for feature extraction and to train anomaly detection models.
The results demonstrated the system's ability to detect anomalies with high accuracy, confirming the effectiveness of accelerometer-based monitoring for predictive maintenance. This approach holds significant potential for reducing operational downtime and lowering maintenance costs in industrial environments, ensuring the continued performance and reliability of fan systems.