Adaptive Active Noise Control: Structures, Algorithms and Convergence Analysis.

摘要

The first objective of this paper is to provide a unified framework for AANC, to make precise its similarity to and difference from well understood standard problems in recursive identification, adaptive signal processing and control. It is found that various structures for AANC are naturally classified into: 1) adaptive feedforward control with or without input contamination; 2) adaptive feedback control; 3) adaptive feedback/feedforward control with or without input contamination. The second objective is to present adaptive algorithms for each structure and analyze their convergence properties via the ordinary differential equation (ODE) method. Worth highlighting is the solution to the problem of input contamination. We show that as long as the controller is of sufficient order (usually higher than when input contamination is absent) and the transfer function from the input of the secondary source to the output of the residual microphone is of minimum phase, the presence of input contamination places no limitation on the ultimate achievable attentuation, and the algorithms designed for the case without input contamination based on the certainty equivalence principle is still convergent in the presence of input contamination. Reprints. (edc)