高分辨率快速反卷积声源成像技术研究

发布时间：2018-04-10 05:07

本文选题：反卷积声源成像　切入点：点扩展函数　出处：《合肥工业大学》2017年硕士论文

【摘要】：反卷积声源成像是一种建立在互谱成像波束形成技术基础上的高分辨率噪声源识别与定位技术。其空间分辨率远高于常规波束形成技术,因此反卷积声源成像算法在噪声治理和故障诊断工程中有着广泛的应用前景,值得进一步研究。本文首先回顾了反卷积声源成像技术的发展历程,并对研究现状和发展趋势进行了分析。分析发现已有的算法在性能优势和适用范围上有所不同。为在实际工程应用中合理选择反卷积声源成像算法,充分发挥各算法的性能优势,通过数值仿真和实验对现有算法在不同频率、不同测量距离、不同测量点位置和数目以及不同聚焦点数目情况下的空间分辨率和计算效率进行了研究,得出了如下结论:稀疏重构反卷积声源成像算法的空间分辨率最高,但由于需要计算完整的点扩展函数矩阵,因此算法的计算效率相对较低;而基于傅里叶变换的反卷积声源成像算法计算效率较高,但由于其获得的是最小二乘解,未利用解的稀疏先验,加之傅里叶变换引入的卷绕误差影响,其空间分辨率较之稀疏重构反卷积声源成像算法要低。论文根据上述结论给出了实际工程应用中合理选择反卷积声源成像算法的指导意见。通过对以上结论进行分析可以看出现有的六种反卷积声源成像算法在空间分辨率或计算效率上的均存在一定的局限。为了能够兼顾算法的空间分辨率和计算效率,本文在稀疏重构反卷积声源成像算法的基础上利用点扩展函数的近似平移不变性提出一种高分辨率快速反卷积声源成像算法,通过仿真对比本文方法和现有反卷积声源成像算法的声源识别性能,进一步利用半消声室中两个白噪声声源的识别实验验证了仿真结论的正确性,最后将本文所提方法用于某主机箱噪声源的识别定位,进一步验证了该方法的有效性和实用性,仿真和实验均表明本文提出的高分辨率快速反卷积声源成像算法既具有类似于稀疏重构反卷积声源成像算法的高空间分辨率,又具有远高于现有反卷积声源成像算法的计算效率,非常适合于实际应用。
[Abstract]:Deconvolution acoustic source imaging is a high resolution noise source recognition and location technique based on cross-spectral beamforming.Its spatial resolution is much higher than that of conventional beamforming technology. Therefore, deconvolution acoustic source imaging algorithm has a broad application prospect in noise control and fault diagnosis engineering, and deserves further study.In this paper, the development of deconvolution acoustic source imaging technology is reviewed, and the research status and development trend are analyzed.It is found that the existing algorithms have different performance advantages and applicable range.In order to reasonably select deconvolution sound source imaging algorithm in practical engineering application and give full play to the performance advantages of each algorithm, the existing algorithms are measured at different frequencies and different measuring distances through numerical simulation and experiments.The spatial resolution and computational efficiency of different positions and numbers of measurement points and different focal points are studied. The following conclusions are drawn: the spatial resolution of sparse reconstructed deconvolution source imaging algorithm is the highest.However, due to the need to compute the complete point spread function matrix, the computational efficiency of the algorithm is relatively low, while the computation efficiency of the deconvolution acoustic source imaging algorithm based on Fourier transform is higher, but the least square solution is obtained.The spatial resolution is lower than the sparse reconstruction deconvolution source imaging algorithm due to the influence of the sparse prior solution and the winding error introduced by Fourier transform.According to the above conclusions, the paper gives the guidance of how to select the deconvolution sound source imaging algorithm in practical engineering.By analyzing the above conclusions, we can see that there are some limitations in the spatial resolution or computational efficiency of the existing six deconvolution acoustic source imaging algorithms.In order to give consideration to the spatial resolution and computational efficiency of the algorithm, this paper proposes a high resolution fast deconvolution source imaging algorithm based on sparse reconstruction deconvolution source imaging algorithm using the approximate translation invariance of point spread function.Through the simulation and comparison of the methods in this paper and the existing deconvolution sound source imaging algorithms, the correctness of the simulation results is verified by using the recognition experiments of two white noise sources in the semi-anechoic chamber.Finally, the method proposed in this paper is used to identify and locate the noise source of a main case, which further verifies the effectiveness and practicability of the method.Simulation and experiments show that the high resolution fast deconvolution source imaging algorithm proposed in this paper has a high spatial resolution similar to that of sparse reconstructed deconvolution source imaging algorithm.Moreover, it is much more efficient than the existing deconvolution source imaging algorithm, so it is very suitable for practical application.
【学位授予单位】：合肥工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TB53

【相似文献】