Location accuracy improvement in bluetooth low energy based indoor positioning system for remote asset monitoring / Dasmond Roy Philips
Manufacturing industry is a fast-growing industry, not only in Malaysia but globally as well. Manufacturing processes are very crucial to this industry as it is the core business of the industry. Hence, companies should ensure smooth flow in their manufacturing processes by meeting their daily outpu...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
2024
|
| Subjects: | |
| Online Access: | http://studentsrepo.um.edu.my/15496/1/Dasmond_Roy_Philips.pdf http://studentsrepo.um.edu.my/15496/2/Dasmond_Roy_Philips.pdf http://studentsrepo.um.edu.my/15496/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Manufacturing industry is a fast-growing industry, not only in Malaysia but globally as well. Manufacturing processes are very crucial to this industry as it is the core business of the industry. Hence, companies should ensure smooth flow in their manufacturing processes by meeting their daily output targets in order to sustain in the global market. This is where asset tracking system comes in crucial to the industry. To ensure a smooth flow of manufacturing processes, all assets have to be tracked and made sure to be available at all times for use, to prevent unnecessary and unplanned delays in production. Asset tracking system is a dedicated system, deployed to monitor the movement of assets within an environment, in our case, within production floors. Commonly, location accuracy within an asset tracking system is often compromised due to many factors. This research project mainly aims to improve location accuracy of the asset tracking system through implementation of machine learning algorithms and parameters tuning. Machine learning algorithms that were involved in this research are Support Vector Regression (SVR), Decision Tree (DT) and K-Nearest Neighbor (KNN). Parameters tuning involves elevation angle, tag height, data rate and movement pace of the tags. KNN algorithm delivered lowest RMSE value of 0.631m whereas for parameters tuning, elevation angle of 55˚, tag height of 2.5m, data rate of 50Hz and slow pace combination gives lowest RMSE value of 0.219m respectively. Combining both machine learning and parameters tuning approaches, lowest RMSE value of 0.015m was achieved.
|
|---|