增强盲源分离及潜艇振动定量溯源研究

Submarines are powerful weapons for national defense and maritime interest protection. Vibration monitoring and control have great engineering values for improving the stealth performance and war power of submarines. Due to the complex structures, many vibration sources, and intensive space, it is difficult to quantitatively trace vibration sources, which is the bottleneck that restricts the stealth performance of submarines. Aiming at quantitative vibration and noise source tracing for submarines, this project studies on the mechanisms of vibration generation, transmission, and coupling of submarines, and builds an accurate vibration model by combining linear composition, nonlinear modulation, and convolutive coupling; proposes an enhancement mechanism of blind source separation via multi-separation and clustering optimization, which reduces the effects of modeling errors and parameter errors on the accuracy of blind source separation, and enhances the stability and reliability in engineering applications; studies on the relative energy contributions of vibration sources in iterative calculation of the enhanced blind source separation, and proposes a novel quantitative vibration source tracing method for submarines; carries out multi-source vibration experiments on an vibration test-bed of submarine shell structures, and validates the correctness of vibration source mixing model, the effectiveness of the enhanced blind source separation, and the accuracy of quantitative vibration source tracing method. The enhanced blind source separation method and vibration source tracing technique proposed in this project have important academic significance and engineering application values as they can provide scientific evidences for vibration monitoring and control of the submarines on active service, and vibration reduction design of the late-model submarines.

潜艇是捍卫国家安全和海洋权益的国防利器，振动噪声监测和控制对于提高潜艇隐身性能和作战能力具有重要工程价值。潜艇结构复杂、振源众多、空间密集，使得振动定量溯源困难，成为制约潜艇隐身性能的瓶颈。本项目以潜艇振动噪声定量溯源为目标，研究潜艇振动发生、传递和耦合机理，建立线性叠加、非线性调制和卷积耦合的振源混合模型；提出多次分离聚类优选的盲源分离增强机制，降低模型误差及盲源分离参数对计算精度的影响，提高盲源分离工程应用的稳定性和可靠性；研究增强盲源分离迭代计算中振源的相对能量贡献，提出潜艇振动定量溯源的新方法；在潜艇壳体结构振动实验台上开展多源振动实验，验证振源混合模型的正确性、增强盲源分离方法的有效性和定量溯源算法的准确性。本项目提出的增强盲源分离方法和潜艇振动定量溯源技术，可为现役潜艇振动噪声的监测和控制，以及新型潜艇的减振降噪设计提供科学依据，具有重要的学术意义和工程应用价值。

潜艇是捍卫国家安全和海洋权益的国防利器，振动噪声监测和控制对于提高潜艇隐身性能和作战能力具有重要工程价值。针对潜艇壳体结构振动特性，研究了基于运行工况传递路径分析的壳体结构振动传递特性，剖析了信源串扰、传递率精度低、以及路径误判三大技术难题；研究了多次分离聚类优选的盲源分离增强机制，综合考虑多参数对信号分离的影响特性，构造了稳定可靠的高精度增强盲源分离模型；研究了壳体结构振动信源混合模型，剖析了振源混合机理并基于盲分离框架，提出了潜艇振动噪声定量溯源理论与方法；搭建了精密壳体结构振动噪声综合分析试验台，并进一步完善为多几何体振动噪声综合分析试验台，开展了多方面试验研究验证了提出方法的有效性和精确性。项目研究可为现役潜艇振动噪声监测与控制，以及新型潜艇的减振降噪设计提供科学依据，具有重要的学术意义和工程应用价值。