Robust Virtual Sensing of the Exterior Noise Radiation from a Complex Structure in Different Acoustic Environments
* Presenting author
By using a numerical model of a vibro-acoustic system in combination with a small set of measurements in a Kalman filter, it is possible to estimate the sound pressure at locations where no microphones are present. This can be achieved with increased accuracy as compared to using only the numerical model due to the inclusion of the expected process and measurement noise. This procedure is also known as virtual sensing. In this paper, a model-based virtual sensor is built for a complex, deep drawn structure that radiates sound into the free field. A finite element model is used that is extended with infinite element to approximate the Sommerfeld radiation condition. Since the original model is too large to be used in a Kalman filter, it is reduced by a stable Krylov based model order reduction technique. The performance of this virtual sensor is tested in a semi-anechoic chamber, and additionally in several acoustical environments that do not conform to the Sommerfeld condition in order to validate whether the Kalman filter can handle the additional uncertainties. The obtained results show that the filter performs well, which means that the required modeling effort can be reduced significantly by assuming free-field conditions.