A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Condition monitoring and predictive maintenance of ball bearings
This thesis explores the possibility of monitoring ball bearing health based on information from an accelerometer and explicitly targets the health issue of corrosion. The bearings are not affected by radial force, minimal axial force, and are commonly found in computer fans or anemometers. This is accomplished using an Artificial Neural Network approach, namely LSTM-classification on vibration data from an accelerometer. Data is generated by attaching an accelerometer to an anemometer at the optimal location of the bearings. A hairdryer generates the wind, and a Raspberry Pi collects the data and performs the classification. The bearings are corroded using an acid approach to 18-predefined levels. The network predicts the classes with an accuracy of 94% on previously unseen test data. This project highlights the possibilities of automated classification of corrosion levels with minimum need for manual engineering or domain knowledge.
Condition monitoring and predictive maintenance of ball bearings
This thesis explores the possibility of monitoring ball bearing health based on information from an accelerometer and explicitly targets the health issue of corrosion. The bearings are not affected by radial force, minimal axial force, and are commonly found in computer fans or anemometers. This is accomplished using an Artificial Neural Network approach, namely LSTM-classification on vibration data from an accelerometer. Data is generated by attaching an accelerometer to an anemometer at the optimal location of the bearings. A hairdryer generates the wind, and a Raspberry Pi collects the data and performs the classification. The bearings are corroded using an acid approach to 18-predefined levels. The network predicts the classes with an accuracy of 94% on previously unseen test data. This project highlights the possibilities of automated classification of corrosion levels with minimum need for manual engineering or domain knowledge.
Condition monitoring and predictive maintenance of ball bearings
2022-08-01
Theses
Electronic Resource
English
DDC:
624
Autonomous IoT for Condition Monitoring, Assessment and Predictive Maintenance
| Springer Verlag | 2021
| Engineering Index Backfile | 1901
| British Library Online Contents | 1993