This paper describes and ANC system which can suppresses the transverse mode wavs as well as the plane wave and, hence, be effective in reducing the noise inside the large ducts. The sound fields generated by the control source mounted on the walls of a rectangular duct are theoretically analysed and the arrangement of the control sources to radiate the specific mode waves only is introduced. Noise components propagating in the duct are measured by a set of microphone, modally decomposed and then suppressed by pressure wave of each mode which is generated by the proposed distribution of control sources. Each mode is controlled by an adaptive controller which has a finite impulse response charaterstics and uses the filtered-x LMS algorithm. The algorithm compensates for the changes in the characteristics of acoustic plants due to the environmental variations in the duct. Experimental results illustrate that the broadband(100-770Hz) noise propagating in transverse mode is attenuated by 10-15dB in sound pressure level.
In active control system of noise in ducts, it is common to locate the error microphone rather away from the control source for the purpose of preventing near-field effect. This paper reviews filtered-x LMS and filtered-u LMS algorithm and then illustrates the undesirable effect on convergence of controller. Hence, stability of active noise controllers designed using above two algorithms depends on the distance between the control sources and error microphone and adaptation speed of such a controller to the changes in the characteristics of acoustic plants due to environmental variations inside the duct must be limited significantly. In order to solve such a limitations by time-delay of acoustic error path, the new method named constrained LMS algorithm for active noise control is proposed in this paper. In presented algorithm, the coefficients of controller are adapted based on another residual noise defined so as to satisfy the time in variant constraint of controller during the adaptation process. The results of computer simulation for the simplified model of ANC system are presented in order to examine the performance of constrained LMS algorithm and show the relation among the above two algorithms and constrained LMS algorithm.
