FILTERED-REFERENCE LMS ALGORITHM WITH ESTIMATED ERROR SIGNAL FOR ACTIVE NOISE CONTROL IN EARPLUGS

Earplugs have been proven to be the only solution for personal hearing protection against excessive noise in many working environments, where other passive or active techniques are not accepted. For instance in mines or power plants, people need to move over a large area and pass by different noise sources. Large and heavy headsets annoy the users and cause skin problems due to high ambient temperature. The passive earplugs, however, do not provide sufficient reduction of the very high level noise existing in those environments. As a consequence the working time of those people is shorten to avoid damage of their hearing system. The industry is therefore greatly interested in increasing the noise reduction. This can be done by supplementing the passive noise reduction with the active noise reduction. However, contrary to classical headsets, the earplugs applied directly to the ear canal do not leave enough room for installing an error microphone. Therefore, existing applications of active earplugs are mainly concentrated on fixed-parameter feedforward control. In this paper an adaptive approach is presented. It is based on the Filtered-Reference LMS algorithm. The reference microphone is fixed directly at the back side of the earplug. However, instead of measuring the error signal, it is estimated. Since for this particular application the plant paths change little, the aim of the adaptive algorithm is to tune the control filter to the noise, which can be non-stationary, in general case. The system is verified based on the data obtained from the real industrial environment. Experiments are performed using an active personal earplug applied to the G. R. A. S. artificial ear.