Acoustic boundary control for quieter aircraft.

摘要

There is a strong interest in reducing the volume of low-frequency noise in aircraft cabins. Active noise control (ANC), in which loudspeakers placed in the cabin are used to generate a sound field which will cancel these disturbances, is now a commercially available solution. A second control approach is active structural acoustic control (ASAC), which uses structural control forces to reduce sound transmitted into the cabin through the fuselage. Some of the goals of current research are to reduce the cost, weight, and bulk of these control systems, along with improving global control performance.; This thesis introduces an acoustic boundary control (ABC) concept for active noise control in aircraft. This control strategy uses distributed actuator arrays along enclosure boundaries to reduce noise transmitted into the enclosure through the boundaries and to reduce global noise levels due to other disturbances. The motivation is to provide global pressure attenuation with small, lightweight control actuators. Analytical studies are conducted of acoustic boundary in two-dimensional and three-dimensional rectangular enclosures and in a finite cylindrical enclosure. The simulations provide insight into the control mechanisms of ABC and demonstrate potential advantages of ABC over traditional ANC and ASAC implementations.; A key component of acoustic boundary control is the “smart” trim panel, a structurally modified aircraft trim panel for use as an acoustic control source. A prototype smart trim panel is built and tested. The smart trim panel is used as the control source in a real-time active noise control system in a laboratory-scale fuselage model. It is shown that the smart trim panel works as well as traditional loudspeakers for this application.; A control signal scheduling approach is proposed which allows for a reduction in the computational burden of the real-time controller used in active noise control applications. This approach uses off-line system analysis to compute a database of optimal control solutions for the operating range of the system. The off-line computations will allow for a reduced degree-of-freedom real-time controller. Both table lookup with linear interpolation and a neural network trained on the optimal control solutions are studied as methods for scheduling the optimal control signals.