Search for Author, Title, Keyword
Fault Detection and Prediction for a Wood Chip Screw Conveyor
More details
Hide details
Polytechnic University of Coimbra, Portugal
RCM2+ - Research Centre for Asset Management and Systems Engineering
Submission date: 2024-01-03
Final revision date: 2024-03-17
Acceptance date: 2024-05-27
Online publication date: 2024-06-16
Publication date: 2024-06-16
Corresponding author
Mateus Mendes   

Polytechnic University of Coimbra, Portugal
Eksploatacja i Niezawodność – Maintenance and Reliability 2024;26(3):189323
  • Detection of faults in a screw conveyor, using different classifiers.
Equipment maintenance is a key aspect to maximize its availability. The present work focuses on data analysis of a screw conveyor of a biomass industry. The screw velocity and load were monitored and analysed, in order to detect and predict possible faults. A machine learning approach was used to detect anomalies, where different algorithms were tested with the data available, in order to train an anomaly classifier. The anomaly classifier was able to accurately identify most anomalies, based on historical data, temporal patterns and information of the maintenance interventions performed. The research carried out allowed to conclude that the Extra Trees Classifier algorithm achieved the best performance, among all algorithms tested, with 0.7974 F-score in the test set. The anomaly classifier has been shown to achieve remarkable accuracy in identifying anomalies. This research not only improves understanding of the performance of screw conveyors in biomass industries, but also highlights the practical utility of employing machine learning for proactive fault detection.
Journals System - logo
Scroll to top