Graduation: Automation of Sound level predictions

We offer you an Ocean of Possibilities. Join our family.

This assignment is intended for a University Master student Naval Architecture / Maritime Technology / Mechanical Engineering / Applied Physics with:

• Affinity with acoustics such as spectrograms and harmonic analysis
• Familiarity with signal processing and Machine Learning frameworks
• Analytical oriented mind set
• Good programming/scripting skills in Matlab/Python
• Mastery of the English language (spoken and written)

• Academic level mentoring throughout the internship 

• Internship compensation and travel allowance in line with company policy 

• A project with direct relevance to multiple business functions and future digital products 

• Potential for internal rollout, and possibly a publication or conference contribution depending on results 

At Damen, much effort is put into designing silent and comfortable ships. Therefore, sound level predictions are made to judge the feasibility of each design. This of course asks for accurate prediction methods.

Aspects covered in these prediction methods are:

• The prediction of structure borne as well as airborne excitation levels of diesel engines, gearboxes, generator sets, propellers, etc.
• Predicting the dynamic response of the ship structure, such as the input mobility of the engine foundation and the transfer of vibrations through the ship structure.
• Prediction of attenuations due to acoustic countermeasures, such as resilient mountings, floating floors, visco-elastic damping, additional absorption, etc.
• Prediction of radiation of lining and absorption of the sound levels in the cabins.
• Prediction methods always need validation by means of measurements. For this purpose, use is made of advanced measurement equipment.

Within the RD&I department, the research team Sound & Vibrations is dedicated to the development of these tools. One of the team members of this team will act as daily supervisor.
 
The role

During Damen sea trials, extensive datasets are collected (audio, accelerations, operational parameters). Analyzing these data is labor-intensive and highly dependent on expert knowledge. By using automatization, source identification and characterization can be (partly) automated. This leads to faster interpretation of noise measurements and objective detection of critical events (cavitation, resonances, out-of-tolerance behavior). All in all, better linkage between measurements, modeling (SEA/FEM), and design parameters.

Your work as an intern is to develop the required methodology to analyze the time domain data and demonstrate the method in a tool. 

Project Objective
Develop a method and build a tool that analyzes an audio recording from a sea trial and automatically:

• Identifies acoustic sources (diesel engine, electric motor, propeller, cavitation, turbulence, pump/HVAC, etc.).
• Determines source characteristics (rotational speed/RPM, number of blades/cylinders, etc.).
• Quantifies the acoustic contribution per source (level, spectrum, and derived metrics).
• Performs source separation (acoustic representation of the contribution of each detected source).

The final result is a prototype that supports acoustic experts in analysis, troubleshooting, and design validation.