The development of a ship performance model in varying operating conditions based on ANN and regression techniques

Maximizing the ship operational energy efficiency requires robust tools to monitor, estimate, and predict the ship's performance under dynamic sea environments. Knowledge of ships' fuel consumption using an appropriate prediction algorithm prior to (or during) a voyage can be a very important tool in reducing GHG emissions from international shipping. Classical methods for power estimation are approximate procedures that lack the required sensitivity to track sea environmental effects on ship performance. Meanwhile, Artificial Neural Network (ANN) as a computing system has proven its applicability in estimating its systems outputs. It also has the ability to capture, learn and adapt to the changes that may occur within the system's variables. The proposed model has employed combined ANN and Multi-Regression (MR) techniques to estimate the ship's power and fuel consumption. The proposed model has the ability to function in a real-time environment and adapt to changes that may occur to the ship environment. Additionally, the proposed model has been developed by processing intensive datasets rather than traditional Noon Reports that have been relied on in many previous studies. Finally, the developed model was utilized to predict potential fuel saving in a Just-In-Time (JIT) scenario for one of the ship's voyages.