University of Twente Student Theses
Predicting Contact Wire Thickness with Machine Learning Algorithms Based on Historical Data
Mulder, Jeroen (2023) Predicting Contact Wire Thickness with Machine Learning Algorithms Based on Historical Data.
|
PDF
2MB |
| Abstract: | For the application of predictive maintenance, a new method is needed which can reliable predict the thickness of contact wires for multiple years into the future. For this, a machine learning model is created using multiple linear regression. The most important features for this model are the train passages, historical wear trend and the thickness of the current wire. A wear rate is predicted for every 10 meters which then is projected into the future to determine the remaining thickness. It is found that this approach performs better than typical machine learning methods whereby the thickness is predicted directly. The predicted thickness is compared with the real thickness for individual predictions and that of a whole wire section. For 95% of the predictions, the average thickness of the whole wire can be predicted with an accuracy ±0.12 mm for a prediction horizon of 4.6 years. The results of this study show that even for noisy data useful predictions can be produced with a novel strategy. |
| Item Type: | Essay (Master) |
| Faculty: | ET: Engineering Technology |
| Subject: | 56 civil engineering |
| Programme: | Civil Engineering and Management MSc (60026) |
| Link to this item: | https://purl.utwente.nl/essays/97128 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page