Optimizing effective perceived noise in distributed electric propulsion with neural networks and differential propeller rotation

• Grant number: 24/19943-7
• Support Opportunities: Scholarships abroad - Research Internship - Scientific Initiation
• Start date: March 31, 2025
• End date: June 29, 2025
• Field of knowledge: Engineering - Aerospace Engineering - Aerospace Propulsion
• Principal Investigator: Gabriel Pereira Gouveia da Silva
• Grantee: Gabriel Clemente Carrari
• Supervisor: Lourenco Tercio Lima Pereira
• Host Institution: Faculdade de Engenharia. Universidade Estadual Paulista (UNESP). Campus Experimental São João da Boa Vista. São João da Boa Vista , SP, Brazil
• Institution abroad: Delft University of Technology (TU Delft), Netherlands
• Associated to the scholarship: 24/07804-2 - Machine learning and vortex lattice method for surrogate modeling in multi-objective wing optimization, BP.IC

Abstract

Aeroacoustic studies are critical to developing quieter aircraft, especially for urban applications. Propeller noise is categorized into two main components, i.e. the tonal and the broadband one. While the broadband component pertains to small and low-Reynolds propellers, the tonal one, known for its prominent frequencies, dominates the emitted sound pressure. The latter is also particularly disturbing to humans and incurs penalties in noise certification metrics such as EPNdB. While little can be done to modify the noise signature of single or twin rotor aircraft, different options can be explored for the use of distributed electric propulsion. By the differential operation of the propellers, the noise signature can be significantly shifted while the overall performance is maintained. This study aims to reduce the effective perceived noise level from a distributed propeller system by optimizing individual propeller RPMs using a machine learning-based model. The approach focuses on offsetting the tonal frequencies of each propeller to disperse acoustic energy across a broader spectrum, therefore reducing acoustic annoyance. To support this study, experimental data fromwind-tunnel tests at Delft University of Technology will be used to train a neural network surrogate model. The obtained model will be leveraged to predict noise levels during optimizations of the propellers speeds to reduce the EPNdB while ensuring required thrust levels. By achieving an optimal balance between noise reduction and aerodynamic performance, this methodology seeks to advance aeroacoustic technologies and reduce the noise impact of next-generation aircraft.