Study On Identification And Automatic Identification Method Of Rock Fracture Signal

Rock fracture signals contain abundant information of rock failure,and have been widely applied in rock burst prediction and rock burst monitoring.However,the environment is complex and changeable,and the interference factors are numerous.It is very important for the engineering management to identify the rock burst signal from other microseismic signals for disaster warning and prediction.At present,most of the rock burst signals rely on manual processing,and the workload is large and it takes more time.It is easy to delay the prediction of disaster warning information.Therefore,it is an urgent problem how to quickly and automatically identify the signal signal of rock burst signal and improve the timeliness and accuracy of the early warning of rock burst and so on.Based on the microseismic signal of the underground powerhouse of Baihe beach columnar jointed basalt,this paper sums up the difference and basic characteristics of four kinds of microseismic signals,such as rock burst signal,blasting signal and electrical signal,and determines the feature extraction methods and characteristic parameters of four kinds of microseismic signals,and analyses the duration,rising time and P-of the four kinds of signals.S wave to the time difference,maximum amplitude,maximum spectrum value,the main frequency of the six characteristic parameters of the distribution of the interval and law,simple analysis of the six parameters as the following signal classification recognition basis is applicable.A sample set of rock burst signals is composed of blasting signals,rock burst signals,anchor rig signals and electrical signals,and the BP neural network recognition model with 6 input nodes,4 output layer nodes and 2 hidden layers is identified,and the number of nodes in the hidden layer is determined by the optimization method,based on the data of white crane beach.The identification model of rock burst signal of Baihe beach columnar joint basalt is established,and the recognition result of four kinds of signals for training data of network structure is completely correct.Through random selection of 37 blasting waveforms,27 rock burst waveforms,20 electrical signal waveforms,and 24 bolt rig waveforms,108 signals are predicted,and 37 blasting waves,27 rock burst waveforms,20 electrical signal waveforms and 24 rock bolt rig waveforms show that the model model of third chapters can be formed.It meets the requirements of actual engineering.By analyzing the misjudgement results of the four kinds of signals,it shows that the duration has a great influence on the classification and recognition of the four kinds of microseismic signals,and can be used as the main basis for the classification of the four kinds of signal.It provides a method and way for intelligent identification of signal types for microseismic monitoring system.