Machine learning for shaft power prediction and analysis of fouling related performance deterioration

Improving operational performance and reducing fuel consumption is increasingly important for shipping companies. Ship degrades over time due to hull propeller fouling; therefore assessing when fouling effects are significant enough warrant cleaning critical. Advancements in onboard data logging systems, combined with machine learning techniques, unlock the potential predict accurately determine clean. This study evaluates five models shaft power prediction: Multiple Linear Regression, Decision Tree (AdaBoost), K – Nearest Neighbours, Artificial Neural Network Random Forest. The importance of pre-processing highlighted, contributing creation a model lower errors than previous studies. significance environmental parameters was explored, novel integration wave statistics dataset, simulated power-speed curves created from predictions identify deterioration fouling. Forest most effective predicting power, an error 1.17%. addition ‘Days Since Clean’ ‘Significant Wave Height’ increased prediction accuracy by 0.07% 0.12% respectively. Simulated revealed 5.2% increase provides operators method conduct cleaning.