To search, Click below search items. |
| All Published Papers Search Service |
|
|
|
Title |
Diagnosing Faults in BWTS based on Machine Learning |
|
Author |
Jae Kyun Kim, Jae-Hoon Kim, and Seong Dae Lee |
| Citation |
Vol. 20 No. 8 pp. 104-111 |
|
Abstract |
Due to environmental regulations on navigation ships of the IMO (International Maritime Organization), demand for eco-ships and associated equipment is soaring. Eco-ship equipment includes BWTS (ballast water treatment system) and Sox scrubbers. The BWTS is a device for purifying ballast water, which is a major cause of marine pollution. This paper proposes a fault diagnosis system based on machine learning. The proposed system is a classification model that judges the status of BWTS faults through various sensor data sent to the BWTS. The operation data of the BWTS are times series data, and normal state or diverse faults are attached to the data as class. The operation data provided for an experiment in this paper were divided into learning data and evaluation data, and were analyzed through a SVM (Support Vector Machine). The accuracy of each fault cause on the evaluation data was 86.93% on average, and the false alarm rate was 5.9%, signifying room for improvement. Improvements will be made through sufficient collection of learning data, fault data augmentation, and imbalance learning. |
|
Keywords |
Eco-Ship, BWTS, Fault-diagnosis, Big-data, Machine Learning |
|
URL |
