有源降噪通路辨识的人因参数几何模型  

Geometric model of human factor parameters for channel identification for active noise control

作　　者：王少博 胥健 高艳 李冲 林秦豪 高大 张劲 靳伯骜 WANG Shaobo;XU Jian;GAO Yan;LI Chong;LIN Qinhao;GAO Da;ZHANG Jin;†JIN Boao(Research Center for Humanoid Sensing,Zhejiang Laboratory,Hangzhou 311121;Key Laboratory of Noise and Vibration Research,Institute of Acoustics,Chinese Academy of Sciences,Beijing 100190)

机构地区：[1]之江实验室类人感知研究中心,杭州311121 [2]中国科学院声学研究所,北京100190

出　　处：《声学学报》2024年第2期226-237,共12页Acta Acustica

基　　金：国家自然科学基金项目(12304524);之江实验室青年基金(K2023MG0AA05)资助。

摘　　要：研究了一种基于人因参数几何模型的声学通路辨识策略,可用于有源降噪技术中的通路辨识环节,以降低因人体姿态改变而导致的降噪性能退化。该策略利用深度相机输出的点云数据提取人因参数并构建几何模型,通过仿真和人工神经网络预测声学通路的传递函数矩阵。结果表明,人因参数几何模型的降噪性能与点云模型接近,适用于参数化扫描仿真。进一步通过实验验证了仿真的准确性以及所提策略的可行性。实验结果还表明,人体姿态改变会导致声学通路传递函数发生复杂且显著的变化,说明实际应用应考虑人体姿态改变所带来的影响。This paper introduces an approach to acoustic channel identification rooted in a geometric model that accounts for human factors.This method holds promise for applications in active noise control technology,specifically addressing the challenge of maintaining optimal noise reduction performance despite the changes in human posture.The approach involves extracting human factor parameters from the point cloud data acquired through a depth camera,which are then used to construct a precise geometric model.This model is utilized to predict the transfer function matrix of the acoustic channel through simulation and artificial neural networks.The results demonstrate that the geometric model incorporating human factor parameters yields noise reduction performance comparable to that of the traditional point cloud models,making it an attractive option for parametric scanning simulations.Furthermore,experimental verification also validates the accuracy of the simulated results and the feasibility of the proposed approach.The findings show the significant impact of changing human posture on the complexity and variability of the acoustic channel,emphasizing the importance of the posture effects in real-world applications.

关 键 词：有源降噪 人因学 传递函数矩阵 信号通路辨识 

分 类 号：TB535[理学—物理]