| Microseismic monitoring can evaluate the damage degree and safety status of monitoring objects by monitoring the vibration caused by rock mass rupture and structural change,so as to provide the basis for timely and accurate disaster prediction and control.Analysis and identification of microseismic signals generated by strata,bridges,buildings and other man-made structures can avoid or reduce the occurrence of major catastrophic accidents.However,the massive data collected by a large number of microseismic sensors in the monitoring area will have unpleasant effects on the data transmission and processing,energy consumption and system stability of the whole system.The compressed sensing theory can effectively solve this problem.In this dissertation,some key technologies in microseismic signal compression sensing are studied,which mainly solve the problems of large amount of data and high redundancy caused by high-speed signal sampling in microseismic monitoring system.Sparse representation,observation matrix,whole scheme of compressed sensing and Sparse Blind Source Separation of microseismic signals are chiefly studied.Firstly,the mathematical basis of compressed sensing is studied.The vector space,the theoretical framework of compressed sensing,the basic definition of sparsity and observation matrix,the reconstruction theory and algorithm are discussed,and the thinning characteristics of microseismic signals are analyzed.The acquisition model of microseismic signal is put forward,and according to the network topology structure of microseismic monitoring system and its time synchronization requirement,combining the advantages of TPSN and RBS algorithm,the hierarchical time synchronization algorithm ODLP,is proposed,which uses maximum likelihood estimation to improve the TPSN algorithm,and combines the broadcasting advantages of RBS algorithm.The simulation results show that the synchronization accuracy of the proposed algorithm is higher than that of the TPSN and RBS algorithms in the whole network.Based on the fact that microseismic signals are not sparse in time domain and sparse in frequency domain,a discrete Fourier transform method with adaptive hard threshold and soft threshold is proposed in this dissertation.Then,the adaptive AR basis of microseismic signal is proposed according to the theory that time series can be decomposed into several component models,in which each component can be represented by autoregressive model.According to the characteristics of microseismic signal,the AR basis is used as an over-complete dictionary.The adaptive sparse representation of microseismic signal is realized.The simulation results show that both of the two methods can achieve better sparse representation effect.In view of the criterion that the compressed sensing observation matrix should satisfy,and based on the actual conditions,the hardware-generated Toplitz observation matrix is used as the compressed sensing observation matrix of microseismic signals.Then in the light of Gold sequence and piecewise Logistic sequence,two kinds of Torplitz observation matrices for microseismic signals are constructed,and compared with other observation matrices in many aspects,their performances are analyzed,and good comprehensive results are obtained.Aiming at the main applications of source location and parameter inversion of microseismic signal,the importance of extracting the first break time of microseismic signal is analyzed,and the structure and model of spatial-temporal correlation of microseismic signal are constructed according to its characteristics,and the spatial correlation and temporal correlation are analyzed in detail.Then,based on the correlation of microseismic signals,the first break time extraction and signal alignment algorithm are proposed,and good results are obtained.At the same time,the distributed compression sensing model of microseismic signal and the whole scheme of compression sensing of microseismic signal based on correlation are proposed to further reduce the data amount of compression sensing of microseismic signal.Finally,the blind source separation problem of microseismic signals is studied for the possible mixing of microseismic signals.Blind source separation model for microseismic signals based on sparse representation and the linear delay hybrid model of microseismic signal are deduced.The sparse microseismic signal under-determined blind source separation algorithm based on fireworks algorithm is proposed.The mapping rules and explosion radius of fireworks algorithm are optimized.At the same time,fireworks algorithm is used in blind source separation data clustering.Simulation results show that the fireworks algorithm can achieve good results of blind source separation. |
PDF Full Text Request