Session: 08-04-02 Machine Learning and Optimization for Marine Design and Operation

Paper Number: 128151

128151 - Data-Driven Multi Objective Ship Voyage Optimization for Underwater Noise Emission 

Anthropogenic noise from marine shipping and other sources poses a serious threat to marine mammals and the ocean environment. This paper aims to enhance a ship's adaptability to varying ocean conditions, with the primary objective of reducing its contribution to anthropogenic noise pollution as well as energy consumption. It presents the implementation of a synchronized adaptive control system for ships, achieved by the integration of an optimization toolbox and a machine learning model.  The environmental noise and exposure levels from the ship are predicted in real time using a machine-learning model. These data-driven models are trained using data from an acoustic solver, encompassing various oceanographic conditions. Moreover, empirical models are used to compute the ship's fuel consumption along the voyage. Using these two pieces of information, a multi-objective optimization algorithm is employed to optimize the ship's operating parameters, simultaneously minimizing both noise emissions and fuel consumption. This approach is proven to be more reliable and effective than the standard conventional slowdown-regulated ships in terms of reducing noise. Furthermore, the paper deliberates on future prospects, emphasizing the use of Computational Fluid Dynamics (CFD) for precise noise level estimation at the source level, replacing conventional empirical models. The discussion also highlights the fine-tuning of data-driven models using noise measurements.

Presenting Author: Akash Venkateshwaran The University of British Columbia

Presenting Author Biography: Akash Venkateshwaran is a MASc student in Mechanical Engineering at the University of British Columbia, currently researching ship voyage optimization and underwater acoustics at the Computational Multiphysics Laboratory under the guidance of Prof. Rajeev Jaiman.

Authors:

Akash Venkateshwaran The University of British Columbia
Indu Kant Deo The University of British Columbia
Rajeev Jaiman The University of British Columbia
Jasmin Jelovica The University of British Columbia