SBIR Phase I: A Neural Network Approach for Active Noise Control in a Three-Dimensional Enclosure

摘要：Sofge小型企业创新研究第一阶段项目将开发和测试一种在三维封闭环境中主动控制室内噪声的新方法。拟议研究的新重点是一个数字自适应系统，该系统利用来自三维外壳的分析和直接测量数据，通过使用位于外壳灵活边界上的智能材料来进行主动噪声控制。该系统旨在跟踪由于运动、开孔变化等引起的声学信息变化，并进行在线最小化，以实现理想的声学性能。与大多数假设可以使用驻波来模拟箱体内声场的方法不同，该方法旨在适应箱体内的漫射声场。此外，不假设噪声为谐波，并考虑了非周期声场。本文提出的基于神经网络的滤波器可以利用分析和实验数据来分配初始权值，这些初始权值将适用于（局部）在线最小化箱体内的声压。降低飞机噪音是业界关注的重要商业问题。在这项工作中开发的技术将广泛适用于商业直升机旋翼机机舱的降噪，并可能适用于商业航空公司的驾驶舱，以及工厂设置和其他振动临界制造环境。＊＊＊

*** ABSTRACT 9660605 Sofge This Small Business Innovation Research Phase I project will develop and test a new approach to active interior noise control in a three-dimensional enclosure. The new thrust in the proposed research is a digital adaptive system that makes use of both analyses and directly measured data from a three-dimensional enclosure to card out active noise control by using smart materials located on flexible boundaries of the enclosure. The system is intended to track the changes in the acoustic information due to movement, changes in open apertures, etc. and carry out on-line minimization to realize the desired acoustic performance. Unlike most methods that assume that the acoustic field inside the enclosure can be modeled by using standing waves, the proposed method is designed to accommodate diffuse sound fields in the enclosure. Furthermore, the noise is not assumed to be harmonic and nonperiodic sound fields are to be considered. The neural network based filter proposed herein can make use of analytical and experimental data to assign initial weights that are to be adapted to carry out (locally) on-line minimization of sound pressure in the enclosure. Noise reduction in aircraft is a subject of serious commercial concern in the industry. The techniques developed in this effort will be widely applicable to noise reduction in rotorcraft cabins of commercial helicopters, and may be adapted for use in commercial airline cockpits, as well as in factory settings and other vibration critical manufacturing environments. ***