Coherence-based performance analysis on noise reduction in multichannel active noise control systems

Abstract

Active noise control (ANC) over an extended spatial region using multiple microphones and multiple loudspeakers has become an important problem. The maximum noise reduction (NR) potential over the control area is a critical evaluation variable as it indicates the fundamental limitation of a given ANC system. In this paper, a method to mathematically formulate the NR potential for any given multichannel ANC systems is developed. First, the residual error in the multichannel feedforward ANC system is formulated, and then the multiple-input-multiple-output problem is decomposed into the parallel-channel problem. The total energy of the residual error is further decomposed into three different terms representing (i) the signal coherence between the reference signals and error signals, (ii) the filter, and (iii) the system null space. The experimental results validate the proposed evaluation method and illustrate the effectiveness on the maximum NR performance evaluation for given systems. Using the proposed analyzing method, more insight into the contribution of each component to the NR potential can be achieved.

Differential Hebbian learning with time-continuous signals for active noise reduction

“…Traditional noise cancelling headsets employ a real error microphone, but also a simulated error microphones can be used, which, in turn, tunes the LMS filter [18]. Recently, the concept of an acoustical error has been translated to a setup with multiple error microphones and speakers to eliminate noise in a seminar room, which is still based on the LMS algorithm but combined with Eigen analysis to tackle the multiple cross correlations between different error microphone signals and speakers outputs [22]. However, at the heart of both digital or acoustic cancellation is the delta rule [17], which changes the filter coefficients by correlating the error signal with the reference noise.…”

Differential Hebbian learning with time-continuous signals for active noise reduction

“…Traditional noise cancelling headsets employ a real error microphone, but also a simulated error microphones can be used, which, in turn, tunes the LMS filter [ 18 ]. Recently, the concept of an acoustical error has been translated to a setup with multiple error microphones and speakers to eliminate noise in a seminar room, which is still based on the LMS algorithm but combined with Eigen analysis to tackle the multiple cross correlations between different error microphone signals and speakers outputs [ 22 ].…”

A Decoupling Algorithm Based on PID Neural Network for Multi-Channel Active Noise Control of Nonstationary Noise

“…The development of an adaptive intelligent algorithm is a new trend in active control. The combination of neural network algorithm, 20,21 fuzzy algorithm 22 and adaptive algorithm 23 opens up a new path for active control of vibration and noise. Kurczyk 22 proposed a comprehensive method using fuzzy control to replace the inverse modeling of magnetorheological dampers, which expanded the application conditions of the control system.…”