Application of Remote Error Sensor Techniques for Active Noise Control in Cabins, at the Ears of a Machine Operator

1. Context

Active Noise Control (ANC) relies on generating waves in phase opposition to undesirable noise to attenuate it. This technique is commonly used in closed systems (small volumes compared to the wavelengths to be treated), such as noise-canceling headphones. This requires the user to wear a device to protect them from noise. The recent development of surface actuators has made this technology promising for improving acoustic comfort in other spaces, particularly in vehicle and machine cabins. However, its application in such environments (modal effects) presents significant technical challenges. Noise sources can be multiple, and the interaction between waves can be complex to understand, even when limited to a certain volume around the operator's head.
An innovative approach in this field is the use of virtual microphones, which predict the sound field at specific locations from remote measurements. This aims to create noise attenuation zones without the need to install multiple sensors. For example, Convolutional Neural Networks (CNN) [1,2] can be used to analyze acoustic data and predict the sound field at targeted positions. Other methods [3], such as the additional filter method and the remote microphone method, involve using inverse optimization algorithms. These algorithms adjust acoustic models based on measured data to estimate the sound field at the virtual microphone location, enabling targeted noise attenuation.
The implementation of virtual microphone techniques combined with the use of acoustic antennas offers an interesting opportunity to reduce noise around the heads of machine operators. This would not only improve driving comfort but also ensure better intelligibility of useful acoustic signals, such as safety alerts, radio communications, or sounds associated with machine performance.

2. Thesis Objectives

The main objective of this thesis is to design and implement an active noise control system using virtual microphone techniques and/or acoustic antennas to reduce noise around the head and ears of a machine operator, specifically at the entrance of the auditory canal. This system aims to create a localized quiet zone, thereby improving the operator's acoustic comfort while avoiding the clutter associated with installing physical sensors too close to their head.
Traditionally, virtual sensors are positioned at a relatively short distance from the target area (a few centimeters). The challenge of this thesis is to test solutions that increase this distance, extending it to several tens of centimeters, to reduce clutter around the operator's head and thus offer them greater freedom of movement while maintaining system effectiveness.

3. Proposed Approach and Description of Planned Work

After a literature review phase complementing the one proposed in the appendix, the thesis will be conducted in two main phases: a numerical phase and an experimental phase.
**Numerical Phase:**
- **Modeling and Simulation:** Development of numerical models to simulate the sound field in a machine cabin. Use of acoustic simulation techniques to evaluate the effectiveness of virtual microphones and acoustic antennas.
- **Control Algorithms:** Design and implementation of adaptive active noise control algorithms based on virtual microphone techniques. Exploration of the use of neural networks to improve the accuracy of sound predictions.
**Experimental Phase:**
- **Prototyping:** Construction of prototypes of acoustic antennas and virtual microphone systems. Integration of these prototypes into a machine cabin for laboratory and, if possible, real-world testing.
- **Testing and Validation:** Conducting experimental tests to evaluate the performance of the active noise control system. Analysis of results to validate numerical models and control algorithms.

4. Organizational Framework of the Thesis

Supervision: This thesis work will be carried out within the MEGA doctoral school and CETIM, the technical center for the mechanical industries.
Funding Duration: 3 years.
Location: Fluid Mechanics and Acoustics Laboratory (LMFA UMR 5509). Occasional travel (at least 1 to 2 weeks every 6 months) is planned to CETIM in Senlis or Beauvais.

5. Desired Profile

Candidates should ideally have the following skills:
- Holder of an Engineering or Master's degree in Mechanics, Physics, Electronics, or Signal Processing
- Knowledge of Vibroacoustics, Acoustics, and Signal Processing
- Experience in numerical simulation and programming in Matlab/Simulink/Python (C/C++ would be a plus)
- Autonomy, curiosity, initiative, and ability to draw analogies
- English (ability to write and present in English with prior preparation)

6. Contact

Interested candidates are