Feature Recognition And Application Of Vibration Signal In Complex Field Environment

The ground intrusion early warning system based on ground vibration is a new early warning system in recent years,which mainly senses the activities of ground targets through vibration sensors,through the acquisition of the target signal processing analysis to determine the target type and location,it has the advantages of long detection distance,small volume,small power consumption,strong anti-interference ability,good concealment and so on.At present,most of the ground vibration target recognition systems only consider single geology,and the equipment can not take into account the low-cost and low-power characteristics required for large-area use.Therefore,a feature recognition system for ground vibration signals in complex field environment is designed,which can greatly reduce the power consumption and cost while keeping the system efficiency normal,the influence of various geological conditions on the signal is considered by means of multi-geological expert database.The main research contents of the thesis are as follows:(1)Firstly,the system requirements and waveform characteristics are analyzed,and the main research waveform and methods are determined.According to the application environment of the system and the characteristics of the seismic wave,the Rayleigh wave is selected as the detection wave.Through the analysis and experimental comparison of common nonlinear signal feature analysis methods,choosing HHT method,this method is more suitable for feature extraction of ground vibration signal under the environment requirement of this system.(2)The design of hardware and software of the identification system of ground vibration characteristics in complex outdoor environment is completed.In the hardware part,the functions of vibration signal acquisition,filtering and AD conversion are completed around the MCU module,and the alarm is realized by using the alarm lamp and the voice broadcasting chip,each terminal module uses Zigbee module to complete networking,which can transmit signal data to control center and Receive Early Warning Signal.In the software part,the functions of signal sending / receiving,signal early-warning judgment,sending early-warning signal and so on are completed.(3)Signal filtering,feature extraction and establishment of expert database.Firstly,the detailed method of signal processing is determined.The main work is to choose the method of denoising through the analysis of wavelet threshold denoising and the comparison of experiments,the validity of the feature extraction method based on energy-zero-crossing joint HHT method is verified by practical experiments.Then a large number of non-interference experiments are carried out,the specific characteristics of each signal are calculated,and an expert database is established.An adaptive expert database change method is proposed,and the effectiveness of this method is verified by experiments.On this basis,the SVM is trained by using the most feature vector of the edge spectrum obtained by HHT,and the vibration signals are recognized by means of more generalization ability.(4)Experimental analysis of different geological conditions on the impact of ground vibration signals and system verification experiment.The validity of the system is verified by the field experiment.The average recognition success rate of the whole system is 95%,and the lowest recognition rate is 85%,the average recognition success rate of SVM multi-class classifier is 98.33%,and the lowest recognition rate is 90%.At the same time,the characteristics of vibration wave patterns of different geology and different behavior are analyzed in detail,and the influence of each geology on vibration wave patterns and the characteristics of different behavior vibration wave patterns are obtained.