Combined feedback and feedforward minimum-variance control of acoustic noise

Active noise control often aims at reducing the sound pressure level of the residual sound being the effect of the primary noise and secondary sound interference. It can be achieved by minimising the mean-square acoustic pressure of the residual sound. In control system terminology it corresponds to minimisation of mean-square value of the signal measured by the error sensor(s), which can be considered as the output disturbance of the acousto-electric plant. If a reference signal correlated with the disturbance is available in advance a feedforward structure is preferred. Otherwise, feedback structure is an interim solution. In the previous research of the author a system was designed where one of the control strategies supervised the other. In this paper the feedback controller with additional information about the disturbance is designed by solving an optimisation problem. Thus, the two structures are combined to operate commonly. In most publications on active control the time domain, frequency domain, or transform domain with inner-outer factorization approaches are applied. In this paper the optimisation is performed in the transform domain using polynomial operations including Diophantine decompositions. The cost function is extended by a control weighting term to allow for stable operation in case of a non-minimum phase plant. Performance of the combined system is analysed and it is confronted with performance of the individually operating feedback system. Theoretical analysis is supported by simulation experiments based on data acquired from a real-world active headrest system.