旋转机械瞬态声场重建与特征提取方法研究

Noise source or vibration source identification of operating rotating machinery using acoustic principles and techniques, is very important for the safe operation, noise and vibration control , and the comfort quality of the equipment. Taking car engines, motorcycle engines, motors, and other rotating machinery as the main research objects, this project intends to perform in-depth study of rotating machinery transient sound field reconstruction under non-steady operating conditions and feature extraction methods. Firstly, facing the difficulty that the existing conventional sound field reconstruction techniques are not suitable for the transient sound field, time-frequency analysis methods will be combined with sound field reconstruction methods (NAH, beam forming, wave superposition method, etc.) to build an effective solution for transient sound field reconstruction; Secondly, as the radiated sound field of rotating machinery is closely related to the rotating speed, speed tracking and order analysis will be combined with the transient sound field reconstruction methods, and the transient sound field of the rotating equipment at different orders can be reconstructed efficiently; Finally, based on the sound reconstruction results, feature extraction methods will be studied to identify the noise sources, then the machinery operating status can be known and the causes of the abnormal noises can be identified. The rotating machinery transient sound field reconstruction methods and acoustic feature extraction methods for rotating machinery based on this project will provide a reliable support for further mechanical failure diagnosis, noise control, and so on.

利用声学原理和技术来分析旋转机械设备运行中形成的噪声源（或振动源），对于保障设备的安全可靠运行、实施噪声与振动控制、提高舒适性品质等均起着非常重要的作用。本项目拟以汽车、摩托车发动机、电机等旋转机械设备为主要研究对象，深入研究变转速工况下设备的瞬态声场重建与特征提取方法。首先，针对常规声场重建技术不适用于瞬态声场的难题，本项目将时频分析方法与声场重建算法（如NAH、波束形成、波叠加等）相结合，形成一套有效的瞬态声场重建算法理论体系；其次，针对旋转机械辐射声场与转速密切相关的特点，将瞬态声场重建与转速跟踪、阶次分析相结合，高效重建设备不同阶次下随转速变化的瞬态空间声场；最后，根据声场重建结果，利用特征提取技术识别旋转机械运行状态下的噪声源，并找到产生异常噪声的原因。本项目形成的旋转机械瞬态声场重建与声学特征提取方法，将为进一步开展旋转机械故障诊断、噪声控制提供有力支持与补充。

本项目旨在利用声学原理与技术来研究旋转机械非稳态声场特性，对于保障设备的安全可靠运行、实施噪声与振动控制、提高舒适性品质等均起着非常重要的作用。课题从近场声全息和远场波束形成两个方面对非稳态声源的识别做了尝试，研究了插值时域等效源法的一些影响因素，并给出了参考数据，考虑到近场声全息在应用层面还存在诸多问题，本项目着重对波束形成非稳态声源识别做了深入研究，经过详细的综述总结，从旋转机械的阶次特性入手提出了两个理论方法：第一、利用阶次追踪和重采样技术，将旋转机械的时域非稳态信号变换成角域平稳信号，再利用窄带波束形成理论进行声源定位。第二、提取非平稳声场中的某个目标阶次，滤掉其他阶次以及噪音成分的干扰，再用宽带波束形成理论进行声源定位，该方法通过仿真与电机试验验证了其可行性，并进行了参数讨论。最后，在声学理论基础上，进一步研究了信号的特征提取技术与故障诊断技术，包含早期故障特征提取的信号稀疏表征理论及隐马尔可夫模型故障诊断技术。