Control signal scheduling for active noise control systems: time domain study

Abstract

This paper presents a time domain study of control actuator scheduling for active structural-acoustic control systems. It is a follow-up study of a previous work on control signal scheduling for reducing the adaptive controller dimensionality in the frequency domain. For a system whose response to acoustic disturbance is well characterized by a small set of measurable state variables, it may be possible to avoid the computational expense associated with the real-time adaptation of a large multichannel controller. Instead, the optimal solution of the controller may be identified off-line as a function of the state variables. The scheduler would identify the best solution given the measured system state, leaving only a small set of magnitudes and phases to be updated by a real-time adaptive algorithm. This work demonstrates how to implement a neural network based actuator scheduling procedure in the time domain. Numerical simulations show good performance of the proposed control strategy. Further experimental studies are needed to validate the theoretical development in real time.