Maximum likelihood least squares identification method for active noise control systems with autoregressive moving average noise

Maximum likelihood methods are significant for parameter estimation and system modeling. This paper derives a maximum likelihood principle based least squares identification algorithm for online secondary path modeling in feed-forward active noise control systems with autoregressive moving average noise. This derivation proves that minimizing the cost function of least squares is equivalent to the maximum of likelihood function. Proposed method requires tuning of only one parameter in comparison with other recognized methods. Simulation tests show that proposed algorithm has better estimation accuracy and noise reduction capability than the current state-of-the-art methods in the presence and absence of disturbance at the error microphone.