Shipping Radiated Noise Estimation Using AI.


This research note highlights commercial ships are the major source of underwater radiated noise, which is generated because of interaction between hull, water and propeller cavitation that lies in the low-frequency range. It states that various models such as D-Ross, RANDI, Wales-Heitmeyer, SONIC, and Wittekind are being used to estimate shipping radiated noises. However, except for the Wittekind model, all other models are limited in the frequency range. But the Wittekind model also has some drawbacks which guides us to implement Artificial Intelligence (AI) to take only the easily obtainable parameters and give the output of Wittekind thereby reducing the time complexity and hence making it deployable.

Key highlights
  • There are various sources of noises and vibrations in ships, some of which are:
    • The prime movers – typically diesel engines
    • Shaft-line dynamics
    • Propeller radiated pressures and bearing forces
    • Air conditioning systems
    • Maneuvering devices such as transverse propulsion units
    • Cargo handling and mooring machinery
    • Vortex shedding mechanisms
    • Intakes and exhausts
    • Slamming phenomena
  • There are broadly two categories under which models to estimate the shipping noises fall.
    • First are computations which are based on numerical analysis
    • Second is empirical which is based on a statistical analysis of noise data
  • All computational models are highly accurate compared to empirical models but they have a major drawback, which is, they require a lot of computational power and take time to give results.
  • Artificial intelligence is starting to rise in the shipping industry and a lot of recent research is being done to incorporate AI into various domains of maritime.
  • Automatic Identification System (AIS) data is the primary source of information for all empirical models, therefore any problem with AIS data will lead to the wrong results or no results at all.

    Common problems associated with AIS are:

    • Commercial vessels below 300 GT cannot be fitted with AIS
    • It can be switched-off intentionally to avoid giving any information
    • Its data transmission is error-prone
    • Data is manually entered so, at times, it can be manipulated to give wrong information
Key recommendations
  • There is a need to modify the empirical models like D. Ross, using data of ships that are currently in service
  • Various limitations of AIS can be fixed with the help of AI
  • AIS can be used to detect if the switching on/off of AIS is intentional or not
  • It can detect if there is any error in AIS data or if there’s any human intervention to manipulate the data
  • It can be used to reduce the time complexity of empirical noise models

“Commercial ships are the major source of underwater radiated noise and this has severely impacted marine animals especially baleen whales. Although AI has been used in many domains of maritime yet there hasn’t been any significant work done towards the use of AI for the empirical noise models.”

Aviral Tyagi, Shridhar Prabhuraman (MRC) and Dr (Cdr) Arnab Das (MRC)