CN115327613A - Mine micro-seismic waveform automatic classification and identification method in multilayer multistage mode - Google Patents
Mine micro-seismic waveform automatic classification and identification method in multilayer multistage mode Download PDFInfo
- Publication number
- CN115327613A CN115327613A CN202210694880.1A CN202210694880A CN115327613A CN 115327613 A CN115327613 A CN 115327613A CN 202210694880 A CN202210694880 A CN 202210694880A CN 115327613 A CN115327613 A CN 115327613A
- Authority
- CN
- China
- Prior art keywords
- waveform
- identification
- waveforms
- microseismic
- mine
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 78
- 239000011435 rock Substances 0.000 claims abstract description 19
- 238000012544 monitoring process Methods 0.000 claims abstract description 17
- 238000012545 processing Methods 0.000 claims abstract description 17
- 238000012512 characterization method Methods 0.000 claims abstract description 7
- 230000007246 mechanism Effects 0.000 claims abstract description 6
- 238000007781 pre-processing Methods 0.000 claims abstract description 5
- 239000011159 matrix material Substances 0.000 claims description 32
- 239000013598 vector Substances 0.000 claims description 30
- 238000012549 training Methods 0.000 claims description 24
- 238000004364 calculation method Methods 0.000 claims description 16
- 238000010276 construction Methods 0.000 claims description 16
- 238000000605 extraction Methods 0.000 claims description 15
- 238000005457 optimization Methods 0.000 claims description 13
- 238000003860 storage Methods 0.000 claims description 12
- 238000012360 testing method Methods 0.000 claims description 12
- 238000009826 distribution Methods 0.000 claims description 10
- 238000012847 principal component analysis method Methods 0.000 claims description 7
- 238000005422 blasting Methods 0.000 claims description 6
- 238000004422 calculation algorithm Methods 0.000 claims description 6
- 238000013145 classification model Methods 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 6
- 238000007433 macroscopic evaluation Methods 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 6
- 238000000513 principal component analysis Methods 0.000 claims description 6
- 230000009467 reduction Effects 0.000 claims description 6
- 230000003595 spectral effect Effects 0.000 claims description 6
- 238000007619 statistical method Methods 0.000 claims description 6
- 230000006870 function Effects 0.000 claims description 5
- 238000000354 decomposition reaction Methods 0.000 claims description 4
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 claims description 3
- 238000007689 inspection Methods 0.000 claims description 3
- SBYHFKPVCBCYGV-UHFFFAOYSA-N quinuclidine Chemical compound C1CC2CCN1CC2 SBYHFKPVCBCYGV-UHFFFAOYSA-N 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000000926 separation method Methods 0.000 claims description 3
- 238000012706 support-vector machine Methods 0.000 claims description 3
- 230000009466 transformation Effects 0.000 claims description 3
- 230000001960 triggered effect Effects 0.000 claims description 3
- 230000019771 cognition Effects 0.000 claims description 2
- 230000001186 cumulative effect Effects 0.000 claims 1
- 238000005065 mining Methods 0.000 abstract description 10
- 230000008859 change Effects 0.000 abstract description 3
- 230000001052 transient effect Effects 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 13
- 238000005516 engineering process Methods 0.000 description 8
- 238000011160 research Methods 0.000 description 7
- 239000003245 coal Substances 0.000 description 6
- 238000005070 sampling Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 5
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000009412 basement excavation Methods 0.000 description 2
- 238000005336 cracking Methods 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 238000005553 drilling Methods 0.000 description 2
- 238000010079 rubber tapping Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
- G01V1/28—Processing seismic data, e.g. for interpretation or for event detection
- G01V1/288—Event detection in seismic signals, e.g. microseismics
Landscapes
- Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Environmental & Geological Engineering (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- Emergency Management (AREA)
- Business, Economics & Management (AREA)
- Acoustics & Sound (AREA)
- Geology (AREA)
- General Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Geophysics (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
The invention relates to a microseismic waveform automatic classification and identification method in a multilayer multistage mode. The method starts with the information mining of the microseismic signals, establishes a characteristic model of a typical microseismic event, establishes a method system of single-wave multi-layer recognition and multi-wave joint recognition, and finally establishes a mine microseismic waveform recognition system based on a multi-layer multi-level mode. On one hand, the problems that a mine site seismic source mechanism is complex, interference factors are many, micro-seismic waveforms are various, randomly change and suddenly transient, the identification difficulty is high, the manual processing difficulty is high, and monitoring and early warning are influenced are solved; on the other hand, through signal preprocessing and multiple feature mining, a foundation is laid for classifying and identifying effective rock fracture signals and accurately positioning, and a theoretical basis and an application basis are provided for realizing problems such as automatic identification and rapid positioning of microseismic waveforms. Compared with the prior art, the method has innovativeness in aspects such as feature mining and quantitative characterization, and classification and identification modes and methods.
Description
Technical Field
The invention relates to the field of mine safety, in particular to a mine micro-seismic waveform automatic classification and identification method in a multi-layer and multi-stage mode.
Background
Coal is one of main energy sources in China, and has great demand in various fields of human life. With the rapid development of mining technology and the increasing demand of human coal, coal resources in shallow ground are gradually mined and exhausted, so that people begin to mine deeper coal resources, even reaching the depth of more than kilometers, and are still gradually deepened. However, the deeper mine environment is more complex and changeable, and especially, the instability of surrounding rocks such as a goaf, a roadway and the like is easily induced due to the micro-shock caused by the autonomous fracture or fluid disturbance of the rocks during the mining process, so that mine dynamic disasters and rock instability damage dynamic disasters seriously threatening the safety of mine mining and miners are brought.
In order to effectively prevent the threat and loss caused by mine dynamic disasters caused by micro-earthquakes, numerous monitoring technologies are successively proposed in the industry, wherein the micro-earthquake monitoring technology with the highest use frequency plays a key role in underground engineering monitoring and early warning. The microseism monitoring technology is a geophysical technology for monitoring the mine state by analyzing data of microseism events. The microseism monitoring technology can realize inversion positioning of the fracture position through vibration signals generated in the internal fracture process of the rock, further study the activity rule of the coal rock and the internal fracture mechanism, and lay a foundation for subsequent monitoring and early warning of mine disasters (coal-rock dynamic disasters such as impact ground pressure, coal and gas outburst and the like).
Compared with the traditional microseismic positioning and early warning method, the microseismic monitoring technology can acquire various signal waves generated by different seismic sources such as seismic waves, mechanical vibration, explosion vibration and the like. However, since there is no effective automatic identification means and method for microseismic events in the mine site, the microseismic monitoring system cannot automatically identify and record the effective events, and can only identify and extract the effective signals by means of manual processing, which results in that the microseismic data cannot be processed in time and the identification efficiency is low while a lot of time and energy are spent, and thus the popularization and the popularization of the microseismic monitoring technology are severely restricted.
The research of mine microseismic waveform identification is still in a starting stage, and the limitations of the research are mainly embodied in the following aspects:
(1) The waveform attribute is less, and the method is single. The existing identification method mostly adopts single attribute identification, and the attributes are mostly conventional attributes such as amplitude, duration, frequency and other conventional characteristics.
(2) The identification accuracy rate is low and the efficiency is low. The current identification method mostly adopts a pure manual mode or a semi-automatic mode, and the identification efficiency is very low. Because the difficulty of waveform identification is high, and the randomness of a manual identification mode is high, the identification result of each person may be different.
(3) The degree of automation is not high. Because the characteristics of the waveform cannot be obtained and the quantitative expression of the waveform cannot be obtained, the automatic identification of the mine microseismic waveform cannot be realized, and the automation degree of the mine waveform identification is low.
(4) The recognition system is imperfect. At present, the identification of mine microseismic waveforms is still in the exploration stage, and the research on the aspects is in the research stage, so that a complete and accurate model still needs to be explored.
In a mine field, due to the fact that effective automatic microseismic event identification means and methods are not available, a microseismic monitoring system cannot automatically identify and record effective events, technicians depend on a manual processing mode, face a large amount of microseismic data to be analyzed and processed urgently every day, the efficiency is low, and the situations of error processing, processing omission, untimely processing and the like often occur.
Disclosure of Invention
The invention aims to solve the technical problem of designing a mine micro-seismic waveform automatic classification and identification method in a 'multilayer multistage' mode so as to solve the problems of difficult identification, low labor efficiency and low identification rate of the micro-seismic waveform of the middlings introduced in the background introduction and further improve the automation degree of micro-seismic monitoring, positioning and early warning.
In order to solve the technical problem, the automatic classification and identification method of the mine microseismic waveform in the multilayer multistage mode comprises the following steps:
step 1: leading the microseismic events into an identification system, and constructing a microseismic waveform database;
and 2, step: preprocessing the acquired micro-seismic waveform, and performing band-pass filtering processing by adopting a wavelet packet filtering method to obtain a de-noised micro-seismic waveform; the microseismic waveform is mixed with a large amount of interference components, such as background noise, pulse interference, background noise and the like, and the components are suppressed and removed, so that the extraction and feature excavation of the microseismic waveform feature are facilitated;
and step 3: carrying out feature extraction on different mine microseismic waveforms by adopting different methods, establishing corresponding quantitative characterization models, and providing features of typical microseismic waveforms; considering the particularity of mine microseismic signals, mine microseismic waveform characteristics are divided into five types according to different representation modes or different research objects, namely time domain characteristics, frequency domain derivative characteristics, time frequency characteristics, mathematical statistics characteristics and macroscopic evaluation characteristics, and corresponding effective criteria are established for judgment so as to extract the characteristics of the mine microseismic waveforms;
and 4, step 4: building mine microseismic waveform feature vectors according to the features of the typical microseismic waveforms provided in the step 3;
and 5: performing dimensionality reduction on the mine microseismic waveform feature vector by using a principal component analysis method, and establishing an optimized feature low-dimensional vector;
step 6: the method comprises the following steps of establishing a hierarchical identification system of mine microseismic waveforms, wherein the hierarchical identification system comprises a hierarchical identification mode and a hierarchical identification mode: the layered identification mode refers to three stages of data acquisition, storage and optimized positioning, different judgment and identification methods are implemented according to different stages, and as shown in fig. 7, the layered judgment and positioning calculation of the microseismic waveform mainly comprises four stages of triggering judgment, classified storage, optimized judgment and positioning calculation; the step-by-step identification mode is step-by-step identification, namely, interference signals are firstly divided into three types of random interference signals A, regular interference signals B and difficult-to-identify signals C, then the interference signals are stripped by using a mode identification method according to the characteristic difference of each type of signals, and final effective signals are reserved;
according to field data analysis, mine field interference waveforms are divided into three classes according to the difficulty of waveform identification, as shown in the following:
(1) Common waveforms of class a: bottom noise, electromagnetic interference, high-power equipment interference and the like;
(2) B type: mechanical vibration (drilling), man-made tapping, and rail transport;
(3) Class C: blasting vibration, natural earthquake, etc.;
the rock fracture effective event is set to O class.
And 7: the mine microseismic waveform classification and identification mode is constructed, and the identification methods are different due to different characteristics of O, A, B and C four types of waveforms, and mainly comprise three methods of single-channel multi-stage classification identification, multi-channel combined classification identification and single-channel and multi-channel mixed identification mechanisms;
and step 8: the method comprises the steps of training known training samples, classifying unknown samples by utilizing the cognition and memory of a network to various event characteristics in a training mode, and constructing a micro-seismic waveform automatic identification method based on a support vector machine, namely micro-seismic event automatic identification based on an SVM network, wherein the construction of a model comprises two parts of characteristic vector construction and network classifier construction;
and step 9: the construction and implementation of the automatic identification system based on the mine microseismic waveform comprise the following steps:
(1) Reading all files in the time period, calculating the number of waveforms, and analyzing by taking a single waveform as a unit;
(2) Carrying out single-channel classification and identification according to a pre-flow, carrying out denoising pretreatment on waveform data, then carrying out feature extraction and vector construction on microseismic waveforms, and finally extracting rock fracture O-type waveforms by adopting a layered and graded single-channel waveform classification and identification method, namely, removing A-type waveforms at the first stage, removing B-type waveforms at the second stage and removing C-type waveforms at the third stage;
(3) Judging the number of effective waveforms in the dictionary, recombining the effective waveforms according to time, and writing the waveforms triggered at the same time into the dictionary named by the time; calculating the total number of events in the dictionary, analyzing by taking the events as a unit, entering a next multi-channel waveform joint identification module when the number of effective waveforms is not less than 4, and otherwise, processing as an ineffective event;
(4) And after multi-channel joint identification, reserving the optimized and judged channel, and bringing the channel into a final positioning model for positioning calculation.
Further, in the step 3, the step of extracting the characteristics of the mine microseismic waveform comprises the following steps:
(1) Establishing a waveform characteristic mark for time domain type discrimination by using conventional characteristics such as the duration, amplitude, time-frequency characteristics and the like of mine micro seismic waves;
(2) By means of time domain and frequency domain analysis and other methods derived from the signal field, solving frequency domain characteristic indexes of the microseismic signals, including characteristics such as corner frequency, power spectral density and spectral ratio, and establishing a corresponding judgment model;
(3) Extracting time-frequency domain characteristics hidden in the microseismic signals, including energy distribution characteristics and fractal characteristics of different frequency bands of the signals, through decomposition and reconstruction of wavelet packets, wherein the characteristics are mainly applied to classification of blasting vibration signals and rock fracture signals;
(4) Carrying out statistical analysis on various typical events in the mine range based on a mathematical statistical method, seeking statistical characteristics of the typical events, such as zero crossing rate, threshold value statistics, amplitude distribution statistics and the like, and establishing corresponding type discrimination statistical characteristic marks;
(5) By establishing underground monitoring and manual inspection, combining underground abnormal phenomena (falling, inclined slope and the like) and macroscopic phenomena (strong earthquake feeling, roadway deformation and the like), a macroscopic evaluation sign of a large-energy event based on comprehensive evaluation is constructed.
Based on the method, four types of typical characteristics (time domain, frequency domain, time-frequency domain and statistical characteristics) of the mine microseismic waveform are established, a corresponding quantitative characterization model is established, and the characteristics of the typical microseismic waveform are provided.
Further, in step 5, after performing feature optimization extraction on the hierarchical waveform, performing PCA principal component analysis on each level of features, and setting the raw data X = [ X ] to obtain the hierarchical waveform 1 ,x 2 ,…,x n ]Wherein x is n For the nth class of feature vectors, the basic steps of the principal component analysis method are as follows:
(1) Forming a sample matrix X, centralizing the sample, and solving a covariance matrix S of X, wherein the matrix S is expressed as:
(2) Solving an eigenvalue lambda and an eigenvalue U of the covariance matrix S, wherein I is an identity matrix, (lambda I-S) U =0;
(3) Arranging the eigenvalues in a descending order, selecting eigenvectors corresponding to the largest m eigenvalues (the sum of the eigenvalues reaches 95%) to form a projection matrix, and measuring the information storage degree of the m components on the original data by using an accumulated contribution rate P:
(4) Taking out the m characteristics, constructing a new vector Y, and constructing a transformation matrix A = U T Calculating a principal component matrix, Y = U T X;
(5) And projecting the sample matrix to obtain a new sample matrix Z after dimension reduction.
Further, the triggering judgment in step 6 includes corresponding triggering parameters, such as amplitude ratio, signal-to-noise ratio and STA/LTA method; the classified storage is to carry out primary screening on the acquired data from the aspect of improving the recognition operation rate and eliminate a large amount of interference and invalid waveforms in the acquired data; the optimization judgment is to judge the transferred data by taking the positioning as the purpose, and carry out four-four combined positioning calculation on the finally reserved waveform event.
Further, in step 6, ABCO is a typical microseismic waveform of four classes, ABC is an interference waveform, O is an effective waveform, and the flow of waveform identification is changed to extract an O-class effective waveform from the four classes of waveforms, assuming that the identification difficulty of the A, B, C-class three-class waveforms is gradually increased, that is, the separability is H AO >H BO >H CO And H is a class interval, the class interval between the H and the O is shown to be smaller and smaller in the clustering result, the identification is carried out in a hierarchy mode, firstly, the A is removed, then, the B is used, and finally, two types of identifications C, O are formed, wherein the structure tree of the identification is shown in figure 8.
Furthermore, in order to reduce the accumulated error of the classification mode, two concepts of separability of class intervals and geometric intervals among classes are introduced, so that the error meets the field requirement. The method for calculating the class interval measures the separation difficulty of the waveform by adopting a clustering Euclidean distance algorithm, wherein the clustering Euclidean distance algorithm is as follows:
further, in step 7, single-channel multi-stage classification and identification are divided into three stages, wherein A-type waveforms are removed in the first stage, B-type waveforms are removed in the second stage, and C-type waveforms are removed in the third stage; the multi-channel joint classification recognition mainly comprises three parts of contents, namely, elimination of invalid waveforms (primary), extraction judgment of effective waveforms (secondary) and optimization of positioning precision (tertiary), wherein the effective waveforms are extracted from interference signals, and the main purpose of waveform recognition is achieved; the single-channel and multi-channel mixed recognition mode is characterized in that the two modes are utilized for carrying out combined recognition, wherein single-channel multi-stage recognition is carried out, overall classification judgment is carried out, the effectiveness of an event is judged, and recording is carried out; and (4) multi-channel joint identification, local optimization discrimination is carried out, effective channels in effective events are discriminated, effective channel waveforms are optimized, and therefore the final participation in positioning calculation of channel numbers is determined.
Further, step 8 includes:
(1) Carrying out normalization pretreatment on the training set and test set samples, and processing data by adopting a [0 1] interval normalization method;
(2) Selecting classifier parameters, wherein the classifier of the SVM network selects RBF (Radial Basis Function) Radial Basis Function:wherein, γ>0;
(3) And training the training set by using an SVM network to obtain a classification model, and then performing class label prediction on the test set by using the classification model.
The invention provides a systematic and feasible method for researching the characteristics of various types of microseismic signals and extracting effective microseismic waveforms in a complex mine environment by the aid of the automatic classification and identification method of the microseismic waveforms in the multi-layer and multi-level mode. The invention can solve the following problems: on one hand, the problems that a mine site seismic source mechanism is complex, interference factors are many, micro-seismic waveforms are various, randomly change and suddenly transient, the identification difficulty is high, the manual processing difficulty is high, and monitoring and early warning are influenced are solved; on the other hand, through signal preprocessing and various feature mining, a foundation is laid for classifying and identifying effective rock fracture signals and accurately positioning, and a theoretical basis and an application basis are provided for realizing automatic identification, quick positioning and other problems of microseismic waveforms. The mine micro-seismic signal feature mining and automatic classification and identification method is a method for systematically and completely researching mine micro-seismic signal classification and identification, and compared with the prior art, the mine micro-seismic signal feature mining and automatic classification and identification method has innovation in the aspects of feature mining and quantitative characterization, and classification and identification modes and models.
Drawings
The following further explains embodiments of the present invention with reference to the drawings.
FIG. 1 is an attribute diagram of mine microseismic signals in the present invention.
FIG. 2 is a multi-dimensional feature presentation of four typical types of waveforms in the present invention.
FIG. 3 is a diagram of the construction and application of the characteristic vectors of microseismic waveforms in the present invention.
Fig. 4 is a (first-order) diagram of the three-dimensional spatial sample distribution formed by the principal components z1, z2, z3 in the present invention.
Fig. 5 is a (second-level) diagram of the three-dimensional spatial sample distribution formed by the principal components z1, z2, z3 in the present invention.
Fig. 6 is a diagram of the distribution (third level) of three-dimensional samples formed by the principal components z1, z2, z3 in the present invention.
Fig. 7 is a diagram of hierarchical judgment and positioning calculation of waveforms in the present invention.
FIG. 8 is a structural tree diagram of a hierarchical recognition scheme in accordance with the present invention.
Fig. 9 is a single-channel waveform one-level classification recognition diagram in the invention.
Fig. 10 is a two-stage classification recognition diagram of a single-channel waveform in the invention.
Fig. 11 is a diagram of three-level classification recognition of a single-channel waveform in the present invention.
FIG. 12 is a flow chart of multi-channel joint identification in the present invention.
FIG. 13 is a multi-channel and single-channel mixed classification decision machine diagram of the present invention.
Fig. 14 is a flow chart of SVM classification prediction in the present invention.
FIG. 15 is a flowchart of a binary SVM tree test in the present invention.
Fig. 16 is a data flow chart of the mine waveform automatic identification system according to the present invention.
FIG. 17 is a flow chart of an automatic mine microseismic waveform identification and positioning system in the present invention.
Detailed Description
The invention discloses a mine microseismic waveform automatic classification and identification method in a multilayer multistage mode, which comprises the following steps:
step 1: importing the microseismic event into an identification system, and constructing a microseismic waveform database;
the data source collected by the embodiment is the field collection of the BMS microseismic collection system. In order to research the characteristics of mine microseismic signals, a corresponding program module is compiled by means of an MATLAB platform, and the signals are analyzed and processed. The microseismic signal is read based on a pre-programmed 'FileRead.m' module, information such as a sampling channel, sampling frequency, sampling length, waveform data and the like in an original signal is taken out, and one complete microseismic event data comprises two parts of a header file and waveform data. The data acquisition parameters are set as: the sampling frequency is 1000Hz, the number of lagging points is 1024, the number of sampling points is changed with different monitored objects, and the sampling length is 2-7 s within the range of 2000-7000.
And 2, step: preprocessing the acquired micro-seismic waveform, and performing band-pass filtering processing by adopting a wavelet packet filtering method to obtain a de-noised micro-seismic waveform; the microseism waveform is mixed with a large amount of interference components, such as background noise, pulse interference, background noise and the like, and the components are suppressed and removed, so that the extraction and feature excavation of the microseism waveform are facilitated;
and step 3: carrying out feature extraction on different mine microseismic waveforms by adopting different methods, establishing corresponding quantitative characterization models, and providing features of typical microseismic waveforms; considering the particularity of the mine microseismic signals, mine microseismic waveform characteristics are divided into five types according to different representation modes or different research objects, namely time domain characteristics, frequency domain derivative characteristics, time frequency characteristics, mathematical statistics characteristics and macroscopic evaluation characteristics, corresponding effective criteria are established for judgment, and further the mine microseismic waveforms are subjected to characteristic extraction, wherein the attribute diagram of the mine microseismic signals is shown in figure 1;
and 4, step 4: building a mine microseismic waveform feature vector according to the features of the typical microseismic waveform provided in the step 3, as shown in fig. 3;
and 5: performing dimensionality reduction on the mine microseismic waveform feature vector by using a principal component analysis method, and establishing an optimized feature low-dimensional vector;
step 6: and (3) constructing a layered and graded recognition system of mine microseismic waveforms, which comprises a layered recognition mode and a graded recognition mode: the layered recognition mode refers to three stages of data acquisition, storage and optimized positioning, different judgment and recognition methods are carried out according to different stages, and as shown in fig. 7, the layered judgment and positioning calculation of the microseismic waveform mainly comprises four stages of triggering judgment, classified storage, optimized judgment and positioning calculation; the step-by-step identification mode is step-by-step identification, namely, interference signals are firstly divided into three types of random interference signals A, regular interference signals B and difficult-to-identify signals C, then the interference signals are stripped by using a mode identification method according to the characteristic difference of each type of signals, and final effective signals are reserved;
according to field data analysis, mine field interference waveforms are divided into three categories according to the difficulty of waveform identification, as shown in the following:
(1) Common waveforms of class a: bottom noise, electromagnetic interference, high-power equipment interference and the like;
(2) B type: mechanical vibration (drilling), man-made tapping, and rail transport;
(3) Class C: blasting vibration, natural earthquake, etc.;
the rock fracture effective event is set to O class.
And 7: the mine microseismic waveform classification and identification mode is constructed, and the identification methods are different due to different characteristics of O, A, B and C four types of waveforms, and mainly comprise three methods of single-channel multi-stage classification identification, multi-channel combined classification identification and single-channel and multi-channel mixed identification mechanisms;
and 8: training known training samples, utilizing the network under the training mode to recognize and memorize the characteristics of various events, classifying unknown samples, and constructing a micro-seismic waveform automatic identification method based on a support vector machine, namely, the micro-seismic event automatic identification method based on the SVM network, wherein as shown in FIG. 14, the micro-seismic event automatic identification method is an SVM classification prediction flow chart, and the construction of a model comprises two parts, namely, the construction of a feature vector and the construction of a network classifier;
and step 9: the construction and implementation of the automatic identification system based on mine microseismic waveforms mainly comprise the following steps:
9.1 the data flow process of automatic identification of mine microseismic waveform is shown in fig. 16, and the data is processed totally;
9.2 as shown in fig. 17, the process of the automatic identification and positioning system for mine microseismic waveforms is implemented by the following steps:
(1) Reading all files in the time period, calculating the number of waveforms, and analyzing by taking a single waveform as a unit;
(2) Carrying out single-channel classification and identification according to a pre-flow, carrying out denoising pretreatment on waveform data, then carrying out feature extraction and vector construction on microseismic waveforms, and finally extracting rock fracture O-type waveforms by adopting a layered and graded single-channel waveform classification and identification method, namely, removing A-type waveforms at the first stage, removing B-type waveforms at the second stage and removing C-type waveforms at the third stage;
(3) Judging the number of effective waveforms in the dictionary, recombining the effective waveforms according to time, and writing the waveforms triggered at the same time into the dictionary named by the time; calculating the total number of events in the dictionary, analyzing by taking the events as a unit, entering a next multi-channel waveform joint identification module when the number of effective waveforms is not less than 4, and otherwise, processing as an ineffective event;
(4) And after multi-channel joint identification, reserving the channels after optimization judgment, and bringing the channels into a final positioning model for positioning calculation.
In this embodiment, preferably, the step of extracting the features of the mine microseismic waveform in step 3 is as follows:
(1) Establishing a waveform characteristic mark for time domain type discrimination by using conventional characteristics such as the duration, amplitude, time-frequency characteristics and the like of mine micro seismic waves;
(2) By means of time domain and frequency domain analysis and other methods derived from the signal field, solving frequency domain characteristic indexes of the microseismic signal, including characteristics such as corner frequency, power spectral density and spectral ratio, and establishing a corresponding judgment model;
(3) Extracting time-frequency domain characteristics hidden in the microseismic signals, including energy distribution characteristics and fractal characteristics of different frequency bands of the signals, through decomposition and reconstruction of wavelet packets, wherein the characteristics are mainly applied to classification of blasting vibration signals and rock fracture signals;
(4) Carrying out statistical analysis on various typical events in the mine range based on a mathematical statistical method, seeking statistical characteristics of the typical events, such as zero crossing rate, threshold value statistics, amplitude distribution statistics and the like, and establishing corresponding type discrimination statistical characteristic marks;
(5) By establishing underground monitoring and manual inspection, combining underground abnormal phenomena (caving, inclined slope and the like) and macroscopic phenomena (strong seismic sense, roadway deformation and the like), a macroscopic evaluation mark of a large-energy event based on comprehensive evaluation is constructed.
Based on the method, four types of typical characteristics (time domain, frequency domain, time-frequency domain and statistical characteristics) of the mine microseismic waveform are established, a corresponding quantitative characterization model is established, the characteristics of the typical microseismic waveform are provided, and the display results of the four types of waveform characteristics are shown in fig. 2.
Preferably, in this embodiment, in step 5, after performing feature optimization extraction on the hierarchical waveform, performing PCA principal component analysis on features of each level, and setting raw data X = [ X ]) 1 ,x 2 ,…,x n ]Wherein x is n If the n-th class of feature vectors is used, the principal component analysis method comprises the following basic steps:
(1) Forming a sample matrix X, centralizing the sample, and solving a covariance matrix S of the X, wherein the matrix S is expressed as:
(2) Solving an eigenvalue lambda and an eigenvalue U of the covariance matrix S, wherein I is an identity matrix, (lambda I-S) U =0;
(3) Arranging the eigenvalues in a descending order, selecting eigenvectors corresponding to the largest m eigenvalues (the sum of the eigenvalues reaches 95%) to form a projection matrix, and measuring the information storage degree of the m components on the original data by using an accumulated contribution rate P:
(4) Taking out the m characteristics, constructing a new vector Y, and constructing a transformation matrix A = U T Obtaining a principal component matrix, Y = U T X;
(5) And projecting the sample matrix to obtain a new sample matrix Z after dimension reduction.
After the hierarchical waveform is subjected to feature optimization extraction, PCA principal component analysis is carried out on the features of each level, and an optimized feature low-dimensional vector is established. And if yes, respectively forming a sample three-dimensional space distribution diagram by the principal components z1, z2 and z3 after the principal component analysis is carried out on the first-stage feature vector, the second-stage feature vector and the third-stage feature vector.
As can be seen from the figure, interference such as yellow bottom noise, electromagnetic interference waveform, green mechanical vibration and the like are obviously separated from the rock cracking waveform, and the waveform belongs to a waveform which is easy to distinguish. The blasting vibration waveform and the rock cracking waveform are crossed and mixed, the existing feature vectors are difficult to distinguish, and a classifier is required to be adopted for classification and identification.
The basic principle of the principal component analysis method is as follows:
suppose a sample space R N Presence sample set X = { X = ×) 1 ,X 2 ,…,X M X, N, X dimensions of each sample X k ={x i1 , x i2 ,…,x iN },X k ∈R M×N If, ifThen its covariance matrix is:
and solving the eigenvalue of the above formula to obtain the eigenvector corresponding to the eigenvalue. Introducing linear mapping function phi and high-dimensional feature space F to make sample point X of input space k Transformed to a sample point phi (X) in the feature space F k )。
for principal component analysis, the covariance matrix in the feature space F is:by performing eigenvalue decomposition on this equation, it can be seen that the solution of the principal components of PCA in the eigenspace F is essentially a solutionThe eigenvalue λ and the eigenvector ν. Therefore, there are:let v be expressed as:substituting the formula to obtain:
by solving the formula, the space vector v of the test sample is obtained k The projection on is expressed as:
preferably, in this embodiment, the triggering determination in step 6 includes corresponding triggering parameters, such as an amplitude ratio, a signal-to-noise ratio, and an STA/LTA method; the classified storage is to carry out primary screening on the acquired data from the aspect of improving the recognition operation rate and eliminate a large amount of interference and invalid waveforms in the acquired data; and the optimization judgment is to judge the transferred data by taking positioning as a purpose and carry out four-four combined positioning calculation on the finally reserved waveform event.
Preferably, in step 6, ABCO is a typical microseismic waveform of four classes, ABC is an interference waveform, O is an effective waveform, and the waveform identification process is changed to extract an O-class effective waveform from the four classes of waveforms, assuming that the identification difficulty of the A, B, C-class three-class waveforms is gradually increased, that is, the separability is H AO >H BO >H CO And H is a class interval, the class interval between the H and the O is shown to be smaller and smaller in the clustering result, the identification is carried out in a hierarchical manner, firstly, A is removed, then, B is carried out, and finally, two types of identifications C, O are formed, wherein the structure tree of the identification is shown in fig. 8.
Preferably, in the embodiment, in order to reduce the accumulated error of the classification mode, two concepts of separability of class intervals and geometric intervals between classes are introduced, so that the error meets the field requirement. The method for calculating the class distance adopts a clustered Euclidean distance algorithm to measure the separation difficulty of the waveform, wherein the clustered Euclidean distance algorithm is as follows:
in this embodiment, preferably, in step 7, the single-channel multi-level classification and identification is divided into three levels, wherein a-type waveforms (as shown in fig. 9) are removed in the first level, B-type waveforms (as shown in fig. 10) are removed in the second level, and C-type waveforms (as shown in fig. 11) are removed in the third level; as shown in fig. 12, the multi-channel joint classification recognition mainly includes three contents, namely, removal of an invalid waveform (first level), extraction and judgment of an effective waveform (second level), and optimization of positioning accuracy (third level), wherein the effective waveform is extracted from an interference signal, and is a main purpose of waveform recognition; the single-channel and multi-channel mixed recognition mode is a combined recognition by using two modes, and the basic flow is shown in fig. 13, wherein single-channel multi-stage recognition is performed, overall classification judgment is performed, the validity of an event is judged, and recording is performed; and (4) multi-channel joint identification, local optimization discrimination is carried out, effective channels in effective events are discriminated, effective channel waveforms are optimized, and therefore the final participation in positioning calculation of channel numbers is determined.
This embodiment preferably includes, in step 8:
(1) Carrying out normalization pretreatment on the training set and test set samples, and processing data by adopting a [0 1] interval normalization method;
(2) Selecting classifier parameters, selecting a Radial Basis Function (RBF) by a classifier of the SVM network:wherein, γ>0;
(3) Training the training set by using an SVM network to obtain a classification model, and then performing class label prediction on the test set by using the classification model.
The binary SVM tree recognition model is shown in fig. 15. Dividing a sample into a test set and a training set, wherein the training set rainDS is obtained by manually processing field data and the representation category (label) of a known waveform is known; the test set testDS is a sample of unknown representative categories. The quality of the training set samples has a large influence on the recognition result, so that the integrity, the reasonability and the identifiability of the waveform are ensured as much as possible when the waveform is collected and recognized for the first time. The sample set comprises two parts of a feature vector and an identification label, wherein trainDS comprises trainFV and trainWL, and the testDS comprises testFV and testWL. The sample set and label vary with layer-by-layer identification. In the first layer, the training set has inconvenient data, labels are divided into 1 and 2, A is deleted in the first layer, the mark A is set as 1, and the rest are set as 2; the second layer, the A class in the training set is removed, the B class mark is set as 1, the rest are set as 2, at the moment, the part with the mark of 2 is reserved in the test result, and the sample with the mark of 1 is deleted; and in the third layer, removing the B type in the training set in the second layer, setting the C type identifier as 1, setting the rest as 2, similarly reserving the part of the test sample set as 2, and deleting the 1 identifier data. At the same time, different feature vectors are selected for each layer of the identified object. And after the classification is finished, obtaining a sample marked as 2, namely the effective waveform O class.
In the previous description, numerous specific details were set forth in order to provide a thorough understanding of the present invention. The foregoing description is only illustrative of the preferred embodiments of the invention, which can be embodied in many different forms than those herein described, and the invention is not limited to the specific embodiments disclosed above. And that those skilled in the art may, using the methods and techniques disclosed above, make numerous possible variations and modifications to the disclosed embodiments, or modify equivalents thereof, without departing from the scope of the claimed embodiments. Any simple modification, equivalent change and modification of the above embodiments according to the technical essence of the present invention are within the scope of the technical solution of the present invention.
Claims (8)
1. A mine micro-seismic waveform automatic classification and identification method in a multi-layer and multi-stage mode is characterized in that: the method comprises the following steps:
step 1: importing the microseismic event into an identification system, and constructing a microseismic waveform database;
step 2: preprocessing the acquired micro-seismic waveform, and performing band-pass filtering processing by adopting a wavelet packet filtering method to obtain a de-noised micro-seismic waveform;
and 3, step 3: carrying out feature extraction on different mine microseismic waveforms by adopting different methods, establishing corresponding quantitative characterization models, and providing features of typical microseismic waveforms; classifying mine micro-seismic waveform characteristics into five categories, namely time domain characteristics, frequency domain derivative characteristics, time frequency characteristics, mathematical statistics characteristics and macroscopic evaluation characteristics, establishing corresponding effective criteria for judgment, and further extracting the characteristics of the mine micro-seismic waveform;
and 4, step 4: building a mine micro-seismic waveform feature vector according to the features of the typical micro-seismic waveform provided in the step 3;
and 5: performing dimensionality reduction on the mine microseismic waveform feature vector by using a principal component analysis method, and establishing an optimized feature low-dimensional vector;
step 6: and (3) constructing a layered and graded recognition system of mine microseismic waveforms, which comprises a layered recognition mode and a graded recognition mode: the layered identification mode refers to three stages of data acquisition, storage and optimized positioning, different judgment and identification methods are implemented according to different stages, and the layered judgment and positioning calculation of the microseismic waveform mainly comprises four stages of triggering judgment, classified storage, optimized judgment and positioning calculation; the step-by-step identification mode is step-by-step identification, namely, the interference signals are divided into three types of random interference signals A, regular interference signals B and difficult-to-identify signals C, then the interference signals are stripped by using a mode identification method according to the characteristic difference of each type of signals, and final effective signals are reserved;
and 7: the mine microseismic waveform classification and identification mode is constructed, and the characteristics of O, A, B and C waveforms are different, so that the identification methods are different, the rock fracture effective event is O type, and the method mainly comprises three methods of single-channel multi-stage classification identification, multi-channel combined classification identification and single-channel and multi-channel mixed identification mechanisms;
and 8: the method comprises the steps of training known training samples, classifying unknown samples by utilizing the cognition and memory of a network to various event characteristics in a training mode, and constructing a micro-seismic waveform automatic identification method based on a support vector machine, namely micro-seismic event automatic identification based on an SVM network, wherein the construction of a model comprises two parts of characteristic vector construction and network classifier construction;
and step 9: the construction and implementation of the automatic identification system based on the mine microseismic waveform comprise the following steps:
(1) Reading all files in the time period, calculating the number of waveforms, and analyzing by taking a single waveform as a unit;
(2) Carrying out single-channel classification and identification according to a pre-flow, carrying out denoising pretreatment on waveform data, then carrying out feature extraction and vector construction on the microseismic waveform, and finally extracting a rock breaking O-type waveform by adopting a layered and graded single-channel waveform classification and identification method, namely, rejecting the A-type waveform at the first stage, rejecting the B-type waveform at the second stage and rejecting the C-type waveform at the third stage;
(3) Judging the number of effective waveforms in the dictionary, recombining according to time, and writing the waveforms triggered at the same time into the dictionary named by the time; calculating the total number of events in the dictionary, analyzing by taking the events as a unit, entering a next multi-channel waveform joint identification module when the number of effective waveforms is not less than 4, and otherwise, processing as an ineffective event;
(4) And after multi-channel joint identification, reserving the optimized and judged channel, and bringing the channel into a final positioning model for positioning calculation.
2. The automatic classification and identification method for mine microseismic waveforms in a multi-layer and multi-stage mode according to claim 1, which is characterized in that: the step 3 of extracting the characteristics of the mine microseismic waveform comprises the following steps:
(1) Establishing a waveform characteristic mark for time domain type discrimination by using the time length, amplitude and time-frequency characteristics of mine micro-seismic waves;
(2) By means of a time domain and frequency domain analysis method derived from the signal field, solving frequency domain characteristic indexes of the microseismic signal, including characteristics such as corner frequency, power spectral density and spectral ratio, and establishing a corresponding judgment model;
(3) Extracting time-frequency domain characteristics hidden in the microseismic signals through decomposition and reconstruction of wavelet packets, wherein the time-frequency domain characteristics comprise energy distribution characteristics and fractal characteristics of different frequency bands of the signals, and the characteristics are mainly applied to classification of blasting vibration signals and rock fracture signals;
(4) On the basis of a mathematical statistical method, performing statistical analysis on various typical events in a mine range, seeking statistical characteristics of the typical events, such as zero crossing rate, threshold value statistics and amplitude distribution statistics, and establishing corresponding type discrimination statistical characteristic marks;
(5) By establishing underground monitoring and manual inspection and combining underground abnormal phenomena and macroscopic phenomena, a macroscopic evaluation mark of a large-energy event based on comprehensive evaluation is established.
3. The automatic classification and identification method for mine microseismic waveforms in a multi-layer and multi-stage mode according to claim 1, which is characterized in that: in step 5, after the hierarchical waveform is subjected to feature optimization extraction, PCA principal component analysis is performed on each level of features, and original data X = [ X ] is set 1 ,x 2 ,…,x n ]Wherein x is n If the n-th class of feature vectors is used, the principal component analysis method comprises the following basic steps:
(1) Forming a sample matrix X, centralizing the sample, and solving a covariance matrix S of the X, wherein the matrix S is expressed as:
(2) Solving an eigenvalue lambda and an eigenvalue U of a covariance matrix S, wherein I is a unit matrix, and (lambda I-S) U =0;
(3) Arranging the eigenvalues in a descending order, selecting eigenvectors corresponding to the largest m eigenvalues to form a projection matrix, and measuring the information storage degree of the m components on the original data by using the cumulative contribution ratio P:
(4) Taking out the m characteristics, constructing a new vector Y, and constructing a transformation matrix A = U T Obtaining a principal component matrix, Y = U T X;
(5) And projecting the sample matrix to obtain a new sample matrix Z after dimension reduction.
4. The automatic classification and identification method for mine microseismic waveforms in a multi-layer and multi-stage mode according to claim 1, which is characterized in that: the triggering judgment in the step 6 comprises corresponding triggering parameters, such as amplitude ratio, signal-to-noise ratio and STA/LTA method; the classified storage is to carry out primary screening on the acquired data from the aspect of improving the recognition operation rate and eliminate a large amount of interference and invalid waveforms in the acquired data; and the optimization judgment is to judge the transferred data by taking positioning as a purpose and carry out four-four combined positioning calculation on the finally reserved waveform event.
5. The automatic mine microseismic waveform classification and identification method based on the multilayer multistage mode as recited in claim 1, which is characterized in that: in step 6, ABCO is a four-class typical microseismic waveform, ABC is an interference waveform, O is an effective waveform, and the waveform is identifiedThe flow is changed to extract O-type effective waveforms from the four-type waveforms, and the recognition difficulty of the A, B, C-type three-type waveforms is assumed to be gradually increased, namely the separability is H AO >H BO >H CO And H is a class interval, the class interval between the H and the O is shown to be smaller and smaller in the clustering result, the identification is carried out in a hierarchical manner, firstly, A is removed, then, B is removed, and finally, two types of identification C, O are formed.
6. The automatic mine microseismic waveform classification and identification method based on the multilayer multistage mode as recited in claim 5, which is characterized in that: the method for calculating the class interval measures the separation difficulty of the waveform by adopting a clustering Euclidean distance algorithm, wherein the clustering Euclidean distance algorithm is as follows:
7. the automatic mine microseismic waveform classification and identification method based on the multilayer multistage mode as recited in claim 1, which is characterized in that: in step 7, single-channel multi-stage classification and identification are divided into three stages, wherein A-type waveforms are removed in the first stage, B-type waveforms are removed in the second stage, and C-type waveforms are removed in the third stage; the multi-channel joint classification recognition mainly comprises three parts of contents, wherein invalid waveforms are removed, effective waveforms are extracted and judged, positioning accuracy is optimized, effective waveforms are extracted from interference signals, and the main purpose of waveform recognition is achieved; the single-channel and multi-channel mixed recognition mode is characterized in that two modes are utilized for carrying out combined recognition, wherein single-channel multi-stage recognition is carried out, overall classification judgment is carried out, the effectiveness of an event is judged, and recording is carried out; and (4) multi-channel joint identification, local optimization discrimination is carried out, effective channels in effective events are discriminated, and effective channel waveforms are optimized, so that the final participation in positioning calculation of channel numbers is determined.
8. The automatic classification and identification method for mine microseismic waveforms in a multi-layer and multi-stage mode according to claim 1, which is characterized in that: the step 8 comprises the following steps:
(1) Carrying out normalization pretreatment on the training set and test set samples, and processing data by adopting a [0 1] interval normalization method;
(2) Selecting classifier parameters, wherein the classifier of the SVM network selects an RBF radial basis kernel function:wherein, γ>0;
(3) And training the training set by using an SVM network to obtain a classification model, and then performing class label prediction on the test set by using the classification model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210694880.1A CN115327613A (en) | 2022-06-20 | 2022-06-20 | Mine micro-seismic waveform automatic classification and identification method in multilayer multistage mode |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210694880.1A CN115327613A (en) | 2022-06-20 | 2022-06-20 | Mine micro-seismic waveform automatic classification and identification method in multilayer multistage mode |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115327613A true CN115327613A (en) | 2022-11-11 |
Family
ID=83916318
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210694880.1A Pending CN115327613A (en) | 2022-06-20 | 2022-06-20 | Mine micro-seismic waveform automatic classification and identification method in multilayer multistage mode |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115327613A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116482752A (en) * | 2023-04-26 | 2023-07-25 | 四川大学 | Rapid three-dimensional positioning method for microseismic event in complex rock mass engineering structure |
CN116580309A (en) * | 2023-07-13 | 2023-08-11 | 湖北省地质环境总站 | Surface mine stope extraction method combining deep learning and object-oriented analysis |
-
2022
- 2022-06-20 CN CN202210694880.1A patent/CN115327613A/en active Pending
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116482752A (en) * | 2023-04-26 | 2023-07-25 | 四川大学 | Rapid three-dimensional positioning method for microseismic event in complex rock mass engineering structure |
CN116482752B (en) * | 2023-04-26 | 2024-01-23 | 四川大学 | Rapid three-dimensional positioning method for microseismic event in complex rock mass engineering structure |
CN116580309A (en) * | 2023-07-13 | 2023-08-11 | 湖北省地质环境总站 | Surface mine stope extraction method combining deep learning and object-oriented analysis |
CN116580309B (en) * | 2023-07-13 | 2023-09-15 | 湖北省地质环境总站 | Surface mine stope extraction method combining deep learning and object-oriented analysis |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110674841B (en) | Logging curve identification method based on clustering algorithm | |
CN115327613A (en) | Mine micro-seismic waveform automatic classification and identification method in multilayer multistage mode | |
Dong et al. | Discrimination of mine seismic events and blasts using the fisher classifier, naive bayesian classifier and logistic regression | |
Peng et al. | Automatic classification of microseismic signals based on MFCC and GMM-HMM in underground mines | |
Dong et al. | Discrimination of mining microseismic events and blasts using convolutional neural networks and original waveform | |
Scarpetta et al. | Automatic classification of seismic signals at Mt. Vesuvius volcano, Italy, using neural networks | |
CN107576772B (en) | Method for quantitatively evaluating coal body structure type by using logging data | |
CN105697002A (en) | Method for recognizing coal measure strata lithology | |
CN109613610B (en) | Automatic picking method for microseism signal arrival time difference | |
CN107389571B (en) | A kind of fine discrimination method of mineral constituent high-spectrum remote-sensing | |
CN105986811A (en) | Five-dimension lithologic character judgment and recognition method based on conventional logging data | |
CN114037020A (en) | Microseism event identification and classification method, device, equipment and storage medium | |
Qin et al. | Evaluation of goaf stability based on transfer learning theory of artificial intelligence | |
CN106990449B (en) | multi-dimensional data-based continental facies sequence stratigraphic framework construction characterization method | |
US9121256B2 (en) | Method of constructing a flow model associated with capillary pressure curves and/or relative permeability curves | |
Li et al. | Classification of seismic events and quarry blasts using singular value decomposition and support vector machine | |
CN112668872B (en) | Drilling block falling depth reduction method based on comprehensive similarity evaluation | |
CN114154542A (en) | Microseism event classification method, device, equipment and readable storage medium | |
Liang et al. | Shearer cutting pattern recognition based on multi-scale fuzzy entropy and support vector machine | |
CN109581489B (en) | Nested seismic facies extraction method and system | |
Parimontonsakul | An Analytical Approach to Automate Stratigraphic Correlation using Well Logging Information | |
Cheng et al. | Fuzzy weights of evidence method implemented in GeoDAS GIS for information extraction and integration for prediction of point events | |
CN112324426B (en) | Method for rapidly judging size of condensate gas reservoir oil ring based on gas measurement data | |
Narvekar | Distinguishing Earthquakes and Noise Using Random Forest Algorithm | |
MANDITA et al. | A PROTOTYPE FOR THE DETECTION AND CLASSIFICATION OF SEISMIC EVENTS USING STA/LTA AND MACHINE LEARNING |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |