k-Means Clustering Method to the Identification of Rock Damage Precursors
In order to study the acoustic emission signals closely related to the rock damage precursors when the rock suddenly fails under external force disturbance,based on the characteristics of k-means clustering method,the uniaxial compression acoustic emission tests of red sandstone were carried out.The five parameters of energy,amplitude,rise time,duration and ringing count of acoustic emission signals were used as feature vectors,and k-means clustering analysis was carried out.Three different types of acoustic emission signals were obtained,and time-series evolution analysis was carried out.The results show that the first and second types of acoustic emission signals continue to appear densely from the beginning of the elastic stage until the end of the test,and there is no obvious stage change of evolution anomaly.The third type of signal appears sporadically in the elastic deformation stage,and suddenly changes to continuous dense appearance in the deformation localization stage.There are obvious evolution anomalies,which are consistent with the failure mechanism of rock,and the abnormal time point is the same as the mutation time point of scattered cloud map.This kind of signal also has the characteristics of large energy proportion and small signal quantity proportion,which can reflect the precursor information of rock damage and is suitable for rock damage early warning.
万方数据
维普
