The goal of this work was to develop adaptive and robust non-adaptive control
algorithms for active noise cancellation in a duct and to experimentally test their performante in the cancellation of noise.
To model the propagation of Sound in the duct, System identification methods
in the frequency domain were used. The high Order of the model and the desire for active noise cancellation in a wide frequency band give rise to numerical Problems. An advanced identification procedure is developed here to solve this Problem. The model obtained is of Order 44 and Shows a good fit with the measured data in the frequency range from 20 Hz up to 1000 Hz. The H, Synthesis method is used to design the robust non-adaptive controllers.
Both feedback controls, and feedforward controls, and the combination of
feedforward and feedback controls (2-DOF) have been synthesized and tested. Due to the time delay of the secondary path, the Performance of the feedback control is limited. In the combination of feedforward and feedback control, the feedforward Controller plays a dominant role. The Performance of the feedforward Controller is very dependent on model errors. Nevertheless, the combination of feedforward and feedback control Shows a better Performance. For adaptive control, several known methods, such as FXLMS and FULMS, are introduced. The most significant advantages of adaptive control are its feedforward structure and its ability to adapt to the changes of the plant. The major Problem of the adaptive control is the trade-off between stability and convergence time. Since the existing algorithms do not take into account the model error of the secondary path, the adaptive System is unstable if the gain in the adaptation loop is large. A new, robust, and stable FULMS algorithm has been derived and implemented. The experimental results show that the new algorithm is both stable and rapidly tonverging.