水下耦合机械噪声源的量化识别及试验

目前水下机械噪声源及其传递路径识别效果较难。为此，将盲源分离算法和传递路径分析方法融合和集成。视多振源信号为卷积混叠，结合LU分解，提出一种新的非正交联合块对角化方法进行耦合振动源的分离。将分离振源作为工况传递路径分析方法的输入振源，建立水下机械振动噪声源识别算法，并对潜艇舱段模型的水下振动-声辐射试验对算法进行验证。结果表明，与现存方法相比，该盲源分离算法具有易实现、收敛速度快、精度高等优点；所集成的源识别算法在水下声场预报和振源贡献量排序中的性能均优于振源耦合时的结果，与实际情况吻合好，达到了高效、准确地识别机械噪声源的目的。

Quantitative Identification and Experiments of the Underwater Coupled Mechanical Noise Source

It is difficult to efficiently identify the main vibration sources and the transmission paths of submarines using the existing methods only. In this paper, the blind source separation （BSS） method combined with transfer path analysis is put forward. First of all, to make up the imperfection of the existing BSS algorithms, a new kind of simple Jacobi-type algorithms for non-orthogonal matrix joint block diagonalization based on the LU factorization is presented, in which the signals of multi-vibration-sources are regarded as convolution overlap. After solving the uncertainty of the sequence of the separated vibration sources, an operational transfer path analysis method based on BSS is proposed, which can eliminate the impact of the amplitude uncertainty on identifying the vibration sources. Finally, the algorithm is successfully applied to identifying the underwater vibration sources of the submarine cabin model. The results demonstrate that the proposed BSS algorithm has the merits of simplicity, fast convergence and high accuracy, and is much better than the existing methods. The performance of the integrated source identification algorithm is superior in underwater sound field prediction and the sequence of noise contribution to the algorithm whose vibration sources are coupled. Therefore, the purpose to effectively and accurately identify the mechanical noise source is achieved.