宽窄带混合主动噪声控制系统高效稳健算法及应用研究

Active noise control (ANC) has important applications in the field of noise suppression and can compensate the shortcomings of passive noise control (PNC). So far, ANC has achieved remarkable effect on reducing noise in duct, however, both the system performance and effectiveness have been restricted due to nonstationarity of practical environment noises, wideband and narrowband mixing characteristics, frequency mismatch, nonlinear secondary paths and many other factors. In this project, we plan to response above problems in theory and practice. Theoretically a high performance active noise control system will be studied by introducing the adaptive variable step-size algorithm, weighted average cost function, broadband and narrowband hybrid structure, double input structure immune to frequency mismatch, and adaptive network-based fuzzy inference system (ANFIS) model. The comprehensive system performance under complex environments will be achieved including fast tracking ability, reasonable convergence speed, small steady-state error, strong robustness, etc. In addition, we will conduct many simulations and experiments to optimize system model parameters, validate and improve corresponding algorithms, such as expanding current one-dimensional pipe platform to an enclosed space, formulating ANC models in large factories, aircraft cabin and other tough situations. The project's innovative achievements will greatly enhance the level of ANC theory and applications in complex environments.

主动噪声控制（ANC）在噪声抑制领域具有重要应用价值，可以弥补被动降噪的不足。目前ANC在管道噪声抑制上取得了显著效果，但实际环境噪声的非平稳性、宽带和窄带混合特征、频率失调、次级通道非线性等问题，严重影响了ANC系统性能，制约了ANC在复杂环境中的应用。对此，本项目在理论层面拟研究基于宽窄带混合结构的主动噪声控制系统，通过引入自适应变步长方法、加权平均代价函数、宽窄带混合结构、双输入抗频率失调结构和自适应网络模糊推理模型，提升系统在复杂环境噪声下追踪能力、收敛速度、稳态误差、实时性和鲁棒性等综合性能。在应用层面，将一维管道试验平台扩展到封闭空间，针对厂房、飞机客舱等特定空间进行声场建模、仿真和实验研究，优化模型参数，验证和改进算法性能。课题的创新性成果将极大地提升复杂环境ANC系统的理论和应用水平。

主动噪声控制（Active Noise Control, ANC）在噪声抑制领域具有重要应用价值，弥补了传统被动降噪的不足，完善了噪声控制体系。在宽窄带混合ANC系统研究中，对其系统性能进行了全面深入的分析，并针对系统的关键问题展开详细研究，主要包括对宽窄带混合ANC系统的频率失调（Frequency Mismatch, FM）问题、系统收敛追踪速度问题、非线性声学路径问题、参考信号为混沌噪声问题、反馈非线性预测问题。针对各问题，分别研究了误差分离技术的窄带ANC系统解决频率失调分析，总体平均经验模态分解技术的ANC系统解决宽窄带混合噪声，FLANN-FIR（Functional Link Artificial Neural Network and Finite Impulse Response）前反馈混合ANC系统解决宽窄带混合噪声和非线性路径问题，RFSLMS（Recursive Filtered-S Least Mean Square）算法的反馈非线性ANC系统解决混沌噪声问题。此外，在各系统的基础上研究了参数优化和系统复杂度问题，为进一步增强系统性能提供了可能。该项目研究增强了ANC系统对复杂环境的适应能力，在不影响系统稳态误差的前提下极大地提高了收敛速度以及系统的追踪性能。搭建管道ANC实验平台，对提出的算法和方案在该平台上进行验证。