Machine learning method to prediction ship wake distribution
When sailing at harsh sea environments, a ship’s fuel consumption and air emission can be easily increased by more than 100% due to reduced propulsive efficiency at sea. The ship propulsive efficiency is often estimated/determined by analysing wake field distribution around the ship propeller. Therefore, it is essential to estimate a ship’s wake field distribution under various operational and wave environmental conditions in order to have a reliable prediction of ship energy performance at sea and develop energy efficiency measures, which can help to reduce a ship’s fuel consumption and air emissions. A ship’s wake field distribution is normally predicted by computational fluid dynamics (CFD) methods, which often require huge computational efforts and impossible to get reliable prediction for all possible ship sailing conditions. In this project, we will first develop Long-Short Term Memory (LSTM) recurrent neural network to model the waterflow in time around a ship’s boundary layer based on some experimental tests. With the inputs from the LSTM machine learning model, the time required for CFD analysis will be significantly reduced. Then it will allow us to estimate the wake distribution in different speed and wave environments. Based on the large amounts of wave distribution data generated by the machine learning improved CFD analysis, some conventional statistical learning methods will be implemented to establish relationship between wake distribution, propulsive efficiency and different operational conditions. Finally, the obtained propulsive efficiency will be integrated into a ship’s performance model to estimate the ship’s fuel consumption and air emission at actual sailing environments. A case study ship with 5 years’ full-scale measurements will be used to check the accuracy of the proposed method for the estimation of wake distribution and propulsive efficiency.
This page is only available in english
- Chalmers (Site, Sweden)
- Chalmers (Publisher, Sweden)