A method for mine roller fault acoustic source localization based on function deconvolution optimization was proposed to address the problems of frequency aliasing and large localization error in mine belt conveyor systems. The cross-spectral matrix function and the point spread function in the traditional fault acoustic source imaging function were reconstructed and analyzed. By optimizing the acoustic source localization function, the resolution of fault signals was improved, thereby enhancing localization accuracy and ensuring accurate identification of mine roller fault positions. First, the peak cross-spectral matrix function caused by sidelobes in the initial localization acoustic source map was calculated. The corrected cross-spectral matrix function, obtained by removing the peak cross-spectral matrix, was used as the target cross-spectral matrix function for reconstruction and optimization. Then, the reconstructed cross-spectral matrix function was decomposed into eigenvalues, and its eigenvectors and eigenvalues were used to form a unitary matrix and a diagonal matrix. Finally, an element-wise power transformation was performed on the acoustic source localization function, and an element-wise root transformation was applied to the cross-spectral matrix function to obtain the reconstructed and analyzed point spread function. A function deconvolution-optimized equation set for mine roller fault acoustic source localization was then established. To improve the computational efficiency of the function deconvolution optimization, an accelerated greedy iterative update strategy was adopted to achieve fast convergence to the localization solution. To address the problem of frequency aliasing in fault acoustic sources, a variational mode decomposition optimized by an improved sparrow search algorithm and an improved multi-scale dual-threshold wavelet cooperative hierarchical adaptive denoising method were applied to enhance the extraction of fault feature frequencies. Experimental results show that, compared with common mine roller fault acoustic source localization methods, the proposed method effectively suppresses sidelobe interference in acoustic imaging maps and improves the localization accuracy of mine roller faults. The optimized algorithm achieves a localization accuracy of 94.42% and increases computational efficiency by approximately 3.8 times. The signal-to-noise ratio and peak signal-to-noise ratio of the cooperative hierarchical adaptive denoising are improved by 88.25% and 56.06%, respectively, indicating significant gains in both accuracy and efficiency.
HU Kun, LI Jia, ZHANG Le.
Mine Roller Fault Acoustic Source Localization Based on Function Deconvolution Optimization[J].
Science Technology and Engineering, 2026, 26(13): 5501-5512 DOI:10.12404/j.issn.1671-1815.2504103
目前常见的故障声源定位算法有波束形成、高分辨率谱估计和时延估计等。首先,波束形成算法分为清除法(CLEAN)[5]、反卷积法(deconvolution approach for the mapping of acoustic source,DAMAS)[6]、互谱矩阵分解重构[7]以及改进算法,该方法常使用声学传感器阵列且适用于空间较大的环境。冀科伟等[8]引入反卷积声学成像算法来确定噪声源位置;褚志刚等[9]与刘扬等[10]通过设计各种波束形成算法来优化定位算法;针对高频滚动轴承故障定位,江文涛等[11]对故障声源采用共振解调与近场波束形成结合对故障轴承进行定位。其次,高分辨率谱估计方法包括多重信号分类(multiple signal classification,MUSIC)算法[12]和旋转不变技术信号参数估计(estimation of signal parameters via rotational invariance techniques,ESPRIT)算法[13]及其改进算法,该算法适合多个声源同时定位,能区分特征频率接近的声源。刘缘等[14]引入了蚁群算法与MUSIC算法相结合形成两级相控的声源定位算法。
最后,常用的时延(delay and sum,DAS)估计方法包括最小均方误差时延估计算法[15]和互相关[16]类算法,此模型计算较简单且易操作,需要时钟高精度同步,适用于短距离或室内定位。Wang等[17]提出了改进广义互相关和多传感器融合的协同定位方法,消除了声源的多径效应;Liu等[18]采用改进的二次相关延迟估计算法用于跟踪移动声源目标;曾庆宁等[19]采用将互相关序列与神经网络结合进行声源定位。同时也有其他声源故障定位方法,沈田等[20]利用门控循环单元网络和赤池信息准则对声源到达方向进行计算; Ojha等[21]对驱动端的轴承采用近场声压法检测噪声源的具体位置;吴文臻等[22]使用变分模态分解和神经网络结合,根据声传感器编号确定故障托辊位置。
由上述可知,目前故障声源定位算法集中于轴承故障定位,用于矿用托辊故障声源定位算法的较少。另外,矿用托辊故障声源定位在空间分辨率和动态范围等方面要求较高,现有方法在这两方面仍有优化空间。对于带式输送机复杂的运行环境,需要采取针对性的去噪方法。为此,基于课题组研发的无轨式巡检机器人平台,同时为模拟人工巡检作业和实现高精度的智能故障自主定位,现引入贪婪快速迭代收缩阈值算法,并提出一种基于函数反卷积优化(function deconvolution approach for the mapping of acoustic source-greedy fast iterative shrinkage threshold algorithm,FDAMAS-GFISTA)的矿用带式输送机托辊故障声源定位方法,旨在提高托辊故障声源的定位精度与计算效益。
HuKun, JiangHao, ChenZhuo, et al. Cross coupling collaborative control of magnetic electric hybrid suspension belt conveyor support system based on self coupling PID[J]. Journal of Coal Industry, 2025, 50(2): 1409-1424.
TianLiyong, TangRui, YuNing, et al. Research and application of robot for continuous replacement of roller in belt conveyor[J]. China Mechanical Engineering, 2024, 35(5): 938-949.
ChenXiangyuan, QinWei, LiuYanchi, et al. Audio recognition method for belt conveyor roller faults by integrating convolutional neural network and linear regression[J]. Coal Science and Technology, 2025, 53(S1): 389-398.
[9]
YeJ, LiY. Application of ROSI+CLEAN-SC algorithm on the measurement of rotating blade noise and its control[J]. Applied Acoustics, 2025, 240: 110866.
[10]
GillesC, JoséP, FrancoisO. Theoretical analysis of the DAMAS algorithm and efficient implementation of the covariance matrix fitting method for large-scale problems[J]. Journal of Sound and Vibration, 2021, 508: 116208.
[11]
WangR, YuM, ShanT, et al. Fast cyclostationary beamforming algorithm for localization of cyclostationary sound sources[J]. Mechanical Systems and Signal Processing, 2025, 230: 112604.
JiKewei, DuYingjun, WeiGuang, et al. Diagnostic and localization methods of acoustic imaging and deep learning[J]. Noise and Vibration Control, 2023, 43(1): 105-109, 134.
ChuZhigang, DuanYunyang, ShenLinbang, et al. Performance analysis and application of function beamforming sound source recognition[J]. Journal of Mechanical Engineering, 2017, 53(4): 67-76.
LiuYang, ZhaoJingyu, ZhangChuanying, et al. Design of controllable beamformer based on rectangular array[J]. Science Technology and Engineering, 2024, 24(32): 13862-13871.
JiangWentao, GuoYongjie, WuXing, et al. Research on fault localization method of rolling bearings based on resonance demodulation beamforming[J]. Mechanical Design, 2024, 41(4): 42-48.
[20]
LiH, LiuA, YangQ, et al. Improved source localization method of the small-aperture array based on the parasitic fly’s coupled ears and MUSIC-Like algorithm: regular section[J]. IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences, 2024, 107(8): 1355-1359.
[21]
DuS, ShiL, QiuS, et al. Broadband VHF lightning radiation sources localization by ESPRIT algorithm[J]. Atmospheric Research, 2025, 314: 107812.
XingYuhua, GuLiming. Research on chaotic spread spectrum time delay estimation algorithm in pulse noise environment[J]. Journal of Electronic Measurement and Instrumentation, 2023, 37(3): 230-237.
[26]
CuiX, LiX, YuK. A systematic solution to 3 danchorless direction estimation using TDOA measurements[J]. Signal Processing, 2024, 218: 109363.
[27]
WangX, LiJ. Cable external breakage source localization method based on improved generalized cross-correlation phase transform with multi-sensor fusion[J]. Energies, 2025, 18(10): 2628.
[28]
LiuM, ZengQ, JianZ, et al. A sound source localization method based on improved second correlation time delay estimation[J]. Measurement Science and Technology, 2023, 34(4): 045102.
ZengQingning, SuPan, WangHongli, et al. Sound source localization algorithm based on cross-correlation sequence and BP network[J]. Science Technology and Engineering, 2023, 23(16): 6945-6951.
ShenTian, LiuZongyang, LiHao, et al. Characterization mechanism and localization of bearing fault acoustic emission information based on gate controlled cycle unit[J]. Journal of Vibration Engineering, 2024, 37(8): 1442-1450.
[33]
OjhaS, JangidN, ShelkeA, et al. Probabilistic impact localization in composites using wavelet scattering transform and multi-output Gaussian process regression[J]. Measurement, 2024, 236: 115078.