A switching strategy of the frequency-domain adaptive algorithm for active noise control

Area of Science:

• Signal Processing
• Acoustics
• Control Systems

Background:

• The normalized frequency domain block least mean square (NFBLMS) algorithm offers high convergence speed for active noise control.
• However, NFBLMS can exhibit biased steady-state solutions due to secondary path influences and limited filter length.
• The modified frequency domain block least mean square (MFBLMS) algorithm provides optimal steady-state performance but typically has slower convergence.

Purpose of the Study:

• To develop a novel mixed algorithm that leverages the strengths of both NFBLMS and MFBLMS algorithms.
• To achieve both high convergence speed and optimal steady-state performance in active noise control systems.
• To introduce an adaptive switching strategy for seamless integration of the two algorithms.

Main Methods:

• A mixed algorithm combining NFBLMS and MFBLMS is proposed.
• An intelligent switching strategy is developed to transition between algorithms based on convergence and environmental changes.
• Simulations utilized measured acoustic transfer functions to validate the approach.

Main Results:

• The proposed mixed algorithm demonstrates both rapid convergence and accurate steady-state performance.
• The switching strategy effectively manages transitions between the NFBLMS and MFBLMS components.
• Validated effectiveness using real-world acoustic transfer function data.

Conclusions:

• The mixed NFBLMS-MFBLMS algorithm offers a superior solution for active noise control compared to individual algorithms.
• The adaptive switching mechanism ensures optimal performance across different operational phases and environmental conditions.
• This approach effectively addresses the trade-offs between convergence speed and steady-state accuracy in adaptive filtering.