Gaussian Process Regression for Vessel Performance Monitoring

Benjamin Pjedsted Pedersen, Jan Larsen

AbstractIt is showed how Gaussian Process Regression (GPR) can be used equally good or better than Artificial Neural Networks (ANN) for short and long term predictions of the energy consumption on a ship. Using different data sets, from five sister container ships, the method is tested with very different data quality, quantity and prediction horizon and shows that GPR is equally accurate and computational more efficient than ANN, but also offer the possible of finding the predictive variance. Furthermore, the used of so-called characteristic length-scales can be used for evaluating the importance of the input variables for the propulsion performance. A crude introduction to GPR is given together with how it has been applied for this purpose.
TypeConference paper [With referee]
Conference12th International Conference on Computer and IT Applications in the Maritime Industries (COMPIT 13)
Year2013    Month April
Address15-17 April 2013, Cortona, Italy
Electronic version(s)[pdf]
BibTeX data [bibtex]
IMM Group(s)Intelligent Signal Processing

Back  ::  IMM Publications