小区域浅层地下爆破震动信号盲源分离算法研究

发布时间：2018-06-29 18:29

本文选题：地下震动 + 特征提取　；参考：《中北大学》2015年硕士论文

【摘要】：小区域浅层地下爆破震动信号特征分析是震源定位技术的关键，提取混合震动信号中的有用分量是进行信号特征分析的重要基础。地下固体介质的复杂性导致了爆破震动信号成分多样，各分量混合模式复杂，分离难度较高等问题，，针对这些问题，本文提出将盲源分离技术引入震动信号处理过程，实现小区域浅层地下爆破震动信号各分量分离。首先，利用基于多变量贝叶斯、基于径向基函数神经网络及基于插值神经元网络层结构的三种非线性盲源分离算法对多路模拟混合震动信号进行信噪分离，根据分离结果中相似系数及信噪比SNR的值验证基于插值神经元网络层结构算法在震动信号信噪分离方面的可行性及优越性；利用EEMD算法与基于插值神经元网络层结构算法相结合的方案对单路模拟混合震动信号进行信噪分离，验证该方案有效性。其次，利用基于扩展联合对角化、基于峭度及基于时间延迟的三种不同盲源分离算法对多路模拟混合震动信号进行横、纵波的分离，验证基于时间延迟算法在横、纵波分离方面的可行性及优越性；利用Radon变换与基于时间延迟算法相结合的方案对单路模拟混合震动信号进行横、纵波的分离，验证该方案有效性。利用以上方案对实测浅层小区域地下爆破震动信号进行分析处理，对得到的P波信号进行特征分析。结果表明，最终得到的P波信号速度特征及频率特性都与理论值相符，该结果证明本文方案可以实现单路小区域浅层地下爆破震动信号的横、纵波分离。利用该P波信号进行震源定位，定位精度可达0.5148m。
[Abstract]:The feature analysis of shallow blasting vibration signal in small area is the key of the focal location technology, and the extraction of useful components from the mixed vibration signal is an important basis for signal feature analysis. The complexity of underground solid media leads to various components of blasting vibration signal, complex mixing mode of each component and high difficulty of separation. In view of these problems, the blind source separation technology is introduced into the vibration signal processing process in this paper. Each component of vibration signal of shallow underground blasting in small area can be separated. Firstly, three nonlinear blind source separation algorithms based on multivariable Bayes, radial basis function neural network and interpolated neural network layer are used to separate the signals from multichannel analog mixed vibration signals. The feasibility and superiority of the interpolated neural network layer structure algorithm in vibration signal noise separation are verified according to the values of similar coefficients and SNR in the separation results. The EEMD algorithm and the algorithm based on interpolation neural network layer structure are used to separate the signal and noise of the single channel analog mixed vibration signal, and the validity of the scheme is verified. Secondly, based on extended joint diagonalization, three different blind source separation algorithms based on kurtosis and time delay are used to separate the transverse and longitudinal waves of multi-channel analog mixed vibration signals. The feasibility and superiority of P-wave separation, the combination of Radon transform and time-delay algorithm, is used to separate the transverse and P-wave signals of single channel analog mixed vibration signal, and the validity of the scheme is verified. The vibration signal of underground blasting in shallow and small area is analyzed and processed by the above scheme, and the characteristic of P wave signal is analyzed. The results show that the velocity and frequency characteristics of the P-wave signal obtained are in agreement with the theoretical values. The results show that the proposed scheme can achieve the separation of transverse and longitudinal waves of the vibration signals of shallow underground blasting in a single channel and small area. Using the P-wave signal to locate the source, the positioning accuracy can reach 0.5148m.
【学位授予单位】：中北大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P631.4;TN911.7

【参考文献】