Key points are not available for this paper at this time.
The virtual sensing (VS) technique enables an active noise control (ANC) system to estimate the virtual error signal for control using remote monitoring microphones. However, instances where noise characteristics and primary paths exhibit variations lead to a noticeable decline in performance for the conventional VS technique. To address this challenge, we propose the cognitive VS technique in this paper. Its objective is to enhance VS performance by providing a more precise estimate of the error signal based on environmental cognition. Differing from the previous selective VS technique, the cognitive VS technique connects both the reference and monitoring microphones to a lightweight classifier. Hence, the cognitive VS technique has the capability to dynamically adjust the VS filter in accordance with the noise and environmental conditions identified by the classifier. Simulation results demonstrate that the cognitive VS technique surpasses conventional and selective VS techniques in terms of adaptivity and generalisation when noise characteristics change and primary paths are time-varying.
Xie et al. (Mon,) studied this question.