Virtual sensing in the reverberant field based on the harmonic signal from the emitting source.

To perform active noise control it is necessary to establish sensors locations, knowing that the error sensors should be placed near to areas that we want to reduce the noise. Its quite inconvenient the need of positioning the error sensor in the center of a room, or in walking areas, or near of the observer's ears, so, It has been investigated a new method, that does a virtual detection from the physical sensors. In this paper we investigate the remote responses obtained in a reverberant room, from a generated harmonic signal. The acoustic emission is characterized from one vibrational sensor and it is used artificial neural networks to estimate a virtual detection response. Was compared and investigated: the neural network architecture, the usage of reference signals, influences due to environment changes, operating condition changes and noises from other sources. Frequency responses in magnitude and phase were also evaluated and compared. It has been investigated ways to accomplish the system identification, and the evaluation of the answers accuracy when there are changes in the dynamics of propagation.