| dc.contributor.advisor | Skeie, Nils-Olav | |
| dc.contributor.advisor | Viumdal, Håkon | |
| dc.contributor.author | Jaganjac, Amar | |
| dc.date.accessioned | 2023-01-03T17:41:21Z | |
| dc.date.available | 2023-01-03T17:41:21Z | |
| dc.date.issued | 2022 | |
| dc.identifier | no.usn:wiseflow:6583421:50226191 | |
| dc.identifier.uri | https://hdl.handle.net/11250/3040758 | |
| dc.description | Full text not available | |
| dc.description.abstract | Yara Porsgrunn are trying to determine the outflow of dust at their NPK prill facilities and if possible reduce it. This data needs also to be reported to the Norwegian authorities. Process data as well as dust measuring data are made available for this.
The objective of this thesis is to investigate the possibility of using the process data in a Neural Network with the sensor data as output to try to forecast the values of the sensor data. Data washing was applied to the process data. Then, two types of Neural Networks were applied to this data to see if it was possible to forecast.
The results indicate that there is a correlation between the process data and how much dust is created from this NPK prill production, using the NEO Scattering sensor as an output. Before such machine learning algorithms can be implemented as a digital twin more investigation should be performed on this model. Such information could prove to be beneficial to find other factors that can be applied to the model. | |
| dc.description.abstract | Yara Porsgrunn are trying to determine the outflow of dust at their NPK prill facilities and if possible reduce it. This data needs also to be reported to the Norwegian authorities. Process data as well as dust measuring data are made available for this.
The objective of this thesis is to investigate the possibility of using the process data in a Neural Network with the sensor data as output to try to forecast the values of the sensor data. Data washing was applied to the process data. Then, two types of Neural Networks were applied to this data to see if it was possible to forecast.
The results indicate that there is a correlation between the process data and how much dust is created from this NPK prill production, using the NEO Scattering sensor as an output. Before such machine learning algorithms can be implemented as a digital twin more investigation should be performed on this model. Such information could prove to be beneficial to find other factors that can be applied to the model. | |
| dc.language | eng | |
| dc.publisher | University of South-Eastern Norway | |
| dc.title | Machine learning models for evaluation of online dust measurements in NPK prill stack | |
| dc.type | Master thesis | |