A Machine Learning Model For Estimating Shipping Noise In India Ocean Region.


This article discusses the importance of managing underwater radiated noise (URN) generated by commercial ships, which can have negative impacts on marine species that rely on sound for navigation, communication, and survival. The article explains that URN management is crucial for ensuring acoustic stealth in naval platforms, complying with regulatory norms, and minimizing degradation of the marine acoustic habitat. The three main aspects of URN management are source-path-receiver modelling, URN measurement and analysis, and prediction or estimation of URN using existing models or AI or ML based models. The article delves into the different sources of ship noise, such as propeller noise, machinery noise, and hydrodynamic noise, and how these noises are generated.

The article also discusses measurement and analysis techniques for determining URN and the challenges associated with each method. Finally, the article highlights the importance of ML-based approaches for estimating shipping noise and the need for effective and efficient hardware and software to measure and analyse URN.

Key highlights
  • Commercial ships are a major source of underwater radiated noise (URN) due to the interaction between the hull and water and propeller cavitation.
  • URN management is important for ship design and manufacturing, acoustic stealth for naval platforms, and the preservation of the acoustic vision of underwater species.
  • The International Whaling Commission (IWC), International Union for Conservation of Nature (IUCN), and International Maritime Organization (IMO) are establishing and monitoring rules and regulations related to URN management.
  • Mathematical models for estimating shipping radiated noise have some drawbacks, and an ML-based approach is needed.
  • Ship noise sources include propeller noise, machinery noise, and hydrodynamic noise.
  • Measurement systems for URN include permanently installed ranging facilities, bottom moored hydrophones, surface supported hydrophones, and near shore measurement.
Key Challenges
  • AIS data availability and accuracy
  • Challenges with Source-Path-Receiver Models, including unique ship machinery configurations, underwater channel fluctuations, and receiver-related issues
  • Data availability for Machine Learning models and monitoring model complexity
  • Interpretability of Machine Learning results.
Major Opportunities
  • Development of more advanced and efficient URN measurement systems for accurate data collection
  • Use of AI/ML-based approaches for better prediction and estimation of URN
  • Implementation of regulations and guidelines to control URN and reduce its impact on marine life
  • Integration of URN management into ship design and manufacturing for more efficient operation and maintenance
  • Improvement of underwater noise reduction technologies to minimize URN emissions from ships.

“There are many areas that require further work to be done in this field including noise model, problems with AIS data, implementing the effective AI or ML models for easier and faster solution.”

Ashutosh Khandal, IIT Delhi; Dr. (Cdr.) Arnab Das, MRC , Pune; Sridhar Prabhuraman, MRC