A Microseismic Monitoring Method for Underground Invasion Behavior Based on DBSCAN Clustering

Abstract

The present invention discloses a microseismic monitoring method for underground intrusion behavior based on DBSCAN clustering. First, all microseismic events in the microseismic data are screened out, and the occurrence time and occurrence location of each microseismic event are determined. Then, the forward and reverse DBSCAN clustering methods are used to analyze and process the microseismic data. According to the occurrence time sorting of each microseismic event, the forward clustering method and the reverse clustering method are used to obtain two sets of cluster data and noise data respectively, and the noise data are distinguished. Then, the two sets of cluster data are subjected to intersection processing to further remove the noise data, so as to finally obtain the set of microseismic events generated by the excavation behavior during underground intrusion. According to the occurrence location and occurrence time of each microseismic event in this set, continuous monitoring of the underground intrusion excavation location in different time periods is realized; this method can not only ensure a large monitoring range, exclude interfering microseismic events, but also continuously and accurately locate the excavation location of underground intrusion behavior.

Description

Technical Field

[0001] The present invention relates to a microseismic monitoring method for underground intrusion excavation behavior, specifically a microseismic monitoring method for underground intrusion behavior based on DBSCAN clustering. Background Art

[0002] Lawbreakers need to illegally cross the national border for illegal purposes. Since there are defense personnel regularly patrolling around the national border, making it impossible for them to pass smoothly, some lawbreakers will choose suitable geological conditions to invade through underground excavation from outside the country until they reach the territory inside, so as to avoid defense personnel and cross the border illegally.

[0003] In order to detect underground intrusion behavior in time, detection devices need to be set up. Currently, microseismic monitoring technology can be used to monitor the excavation response during underground intrusion. However, microseismic monitoring technology may produce false alarms for other underground activities or vibrations. For example, walking, driving, farming, etc. may all produce underground vibrations, thus triggering the alarm of the microseismic monitoring system. These factors may lead to frequent occurrence of interfering microseismic events, increasing the difficulty of accurately monitoring and analyzing underground intrusion behavior. Therefore, how to provide a new method that can ensure a large monitoring range while monitoring underground intrusion behavior, and can exclude interfering microseismic events, so as to accurately locate the underground intrusion excavation location, is the research direction of this industry. Summary of the Invention

[0004] Aiming at the problems existing in the above-mentioned prior art, the present invention provides a microseismic monitoring method for underground intrusion based on DBSCAN clustering, which can ensure a large monitoring range while monitoring underground intrusion behavior, can exclude interfering microseismic events, and can continuously and accurately locate the excavation location of underground intrusion.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is: a microseismic monitoring method for underground intrusion based on DBSCAN clustering, and the specific steps are as follows:

[0006] Step 1: Layout the microseismic monitoring system: First, determine the position of the boundary line. One side of the boundary line is the monitoring area. Arrange two rows of seismic geophone groups on the boundary line. Each seismic geophone in each seismic geophone group is connected to the microseismic monitor, and the layout of the microseismic monitoring system is completed;

[0007] Step 2: Microseismic monitoring: Start the microseismic monitoring system to continuously monitor the monitoring area, and store the obtained microseismic monitoring data;

[0008] Step 3. Microseismic data processing: The microseismic monitoring host first screens the microseismic monitoring data collected in Step 2 to obtain the occurrence time T, and the occurrence position coordinates X and Y of all microseismic events. Then, the forward and reverse DBSCAN clustering analysis method is used to analyze and process all microseismic events, so as to distinguish the microseismic events generated during the excavation of underground intrusion behavior. Finally, the position set of the distinguished microseismic events is obtained, which is the microseismic event set of underground intrusion;

[0009] Step 4. Determine the underground intrusion excavation positions at different times: The microseismic event set of underground intrusion is divided into multiple time periods in chronological order, and the average values of the occurrence position coordinates X and Y of all microseismic events in each time period are calculated respectively, so as to monitor the excavation positions of underground intrusion in each time period.

[0010] Further, in Step 3, the forward and reverse DBSCAN clustering analysis method is used to analyze and process all microseismic events. The specific process is as follows:

[0011] ① Forward time clustering: During the DBSCAN clustering process, start clustering from the point with the earliest occurrence time among all microseismic events, and only select points that occur later or slightly earlier than the core point for neighborhood diffusion during clustering;

[0012] ② Reverse time clustering: During the DBSCAN clustering process, start clustering from the point with the latest occurrence time among all microseismic events, and only select points that occur earlier or slightly later than the core point for neighborhood diffusion during clustering;

[0013] ③ Intersection of clustering results: Intersect the largest cluster after removing noise in Step ① forward time clustering with the largest cluster after removing noise in Step ② reverse time clustering. The microseismic events in the intersection of the two are the microseismic events generated during underground intrusion excavation.

[0014] Further, the forward time clustering in Step ① is specifically as follows:

[0015] 1) Parameter setting: Set three parameters: spatial neighborhood radius ε s and time neighborhood ε t , and the minimum neighborhood point number MinPts;

[0016] Among them, the spatial neighborhood radius ε s : Defines the neighborhood range of a point on the plane (X, Y), that is, a circular area with this point as the center and a radius of ε s ;

[0017] The time neighborhood ε t : Defines a T neighborhood range in time. During forward clustering, the occurrence time of other events is not earlier than the occurrence time of the microseismic event at this point minus the time neighborhood ε t; When performing reverse clustering, the occurrence time of other events is not later than the occurrence time of the microseismic event at this point plus the time neighborhood ε t ;

[0018] Minimum neighborhood points MinPts: Defines the minimum number of points required to simultaneously satisfy the time neighborhood condition within the spatial neighborhood radius ε s ;

[0019] 2) Classify data points: According to the set parameters, classify data points into core points, border points, and noise points;

[0020] Core point: If within the spatial neighborhood radius ε of a point E s contains at least MinPts microseismic event points that satisfy the following time neighborhood condition, then this point is a core point:

[0021] T α >T E -ε t

[0022] where T α is the occurrence time of a microseismic event point within the spatial domain, and T E is the occurrence time of the microseismic event point E;

[0023] Border point: If the number of points within the spatial neighborhood radius ε of a point that satisfy the above time neighborhood condition is less than MinPts, but this point is within the spatial neighborhood radius ε of other core points s and satisfies the time neighborhood condition of other core points, then this point is a border point; s ;

[0024] Noise point: A point that is neither a core point nor a border point;

[0025] 3) Initial selection:

[0026] Select the point with the earliest occurrence time among all microseismic events;

[0027] 4) Expand the cluster: If the initial point is a core point, expand a new cluster starting from this point;

[0028] Mark the initial point as visited;

[0029] Add all core points and border points within the spatial neighborhood radius ε of the initial point that satisfy the above time neighborhood condition to the cluster; For each core point added to the cluster, repeat step 4) to expand the cluster until the cluster no longer grows; s ;

[0030] 5) Mark border points: If the initial point is a border point, mark it as visited but do not expand the cluster;

[0031] 6) Process noise points: If the initial point is a noise point, mark it as a noise point.

[0032] 7) Repeat process: Repeat steps 3) to 6) until all points have been visited.

[0033] 8) Output results: Output all formed clusters and noise points.

[0034] Furthermore, the reverse-time clustering in step ② is specifically as follows:

[0035] 1) Classify data points: According to the set parameters, classify data points into core points, border points, and noise points.

[0036] Core point: If within the spatial neighborhood radius ε of a point E, there are at least MinPts microseismic event points that satisfy the following time neighborhood condition, then this point is a core point: s T

[0037] α <T E t +ε t

[0038] where T α is the occurrence time of a certain microseismic event point within the spatial domain, and T E is the occurrence time of the microseismic event point E.

[0039] Border point: If the number of points within the spatial neighborhood radius ε of a point that satisfy the above time neighborhood condition is less than MinPts, but this point is within the spatial neighborhood radius ε of other core points and satisfies the time neighborhood condition of other core points, then this point is a border point. s s s Noise point: A point that is neither a core point nor a border point.

[0040] 2) Initial selection:

[0041] Select the point with the latest occurrence time among all microseismic events.

[0042]

[0043] 3) Expand the cluster: If the initial point is a core point, expand a new cluster starting from this point.

[0044] Mark the initial point as visited.

[0045] Add all core points and border points within the spatial neighborhood radius ε of the initial point that satisfy the above time neighborhood condition to the cluster; for each core point added to the cluster, repeat step 4) to expand the cluster until the cluster no longer grows. s

[0046]

[0046] 4) Mark boundary points: If the initial point is a boundary point, mark it as visited but do not expand the cluster;

[0047] 5) Process noise points: If the initial point is a noise point, mark it as a noise point;

[0048] 6) Repeat the process: Repeat steps 2) to 5) until all points have been visited;

[0049] 7) Output the results: Output all formed clusters and noise points.

[0050] The principle of determining the excavation location of underground intrusion behavior is as follows: In terms of microseismicity, the inventor has studied and found that during the underground intrusion excavation process, a large number of microseismic events will continuously occur near the excavation location. Therefore, microseismic events will present a dense "intrusion path" in the monitoring area. Disturbance events such as walking, driving, and farming will also generate microseismic events, which are used as noise data. By using the traditional DBSCAN method to perform clustering analysis on the two parameters X and Y of the excavated microseismic events, the results can remove noise data to a certain extent and extract real microseismic events. The obtained image is as Figure 1 shown. From a planar perspective, the effect is good. However, if the time axis of the microseismic occurrence time is added, it will be found that this method will still retain some disturbance events. These disturbance events are very close to the excavation path in terms of planar position (X, Y). The DBSCAN method that only clusters X and Y cannot eliminate such Figure 2 as shown. If these disturbance events are retained for the average of microseismic events within a time period, the position of the cavity will be distorted. And the actual underground intrusion excavation situation may not be continuous in time (that is, in order to prevent the discovery of underground intrusion, excavation is generally not carried out continuously, but intermittently). If the traditional DBSCAN (X, Y, T clustering) with time T is used, the same excavation intrusion path may be clustered into two tunnels as Figure 3 shown. Based on this, the inventor further studied and obtained a positive and negative DBSCAN clustering method that can achieve time and space clustering of microseismic events that are disconnected in time and continuous in space, and finally more accurately determine the location of real-time underground intrusion.

[0051] Compared with the prior art, based on the research of the inventors, the present invention first screens out all microseismic events in the microseismic data, determines the occurrence time and location of each microseismic event, then uses the forward and reverse DBSCAN clustering methods to analyze and process the microseismic data. According to the occurrence time sorting of each microseismic event, the forward clustering method and the reverse clustering method are used to obtain two sets of cluster data and noise data respectively, distinguish the noise data, and then perform an intersection process on the two sets of cluster data to further remove the noise data. Thus, finally, the set of microseismic events generated by excavation during underground intrusion is obtained. According to the occurrence location and occurrence time of each microseismic event in this set, continuous monitoring of the excavation location of underground intrusion behavior in different time periods is realized. This method can not only ensure a large monitoring range, exclude interfering microseismic events, but also continuously and accurately locate the excavation location of underground intrusion. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 is a plan view of the XY position of microseismic data at a certain moment before and after clustering by the traditional DBSCAN method;

[0053] Among them, Figure 1 a is the plan view of the original microseismic data before clustering; Figure 1 b is the plan view after clustering;

[0054] Figure 2 is a data graph of the XY position of microseismic data with a time axis before and after clustering by the traditional DBSCAN method;

[0055] Among them, Figure 2 a is the original microseismic data graph before clustering; Figure 2 b is the data graph after clustering;

[0056] Figure 3 is a data graph of the XY position and time T of microseismic data before and after clustering by the traditional DBSCAN method;

[0057] Among them, Figure 3 a is the original microseismic data graph before clustering; Figure 3 b is the data graph after clustering;

[0058] Figure 4 is the data graph after clustering the XY position and time T of microseismic data by the forward and reverse DBSCAN methods of the present invention;

[0059] Among them, Figure 4 a is the data graph after forward clustering; Figure 4 b is the data graph after reverse clustering; Figure 4 c is the data graph after the intersection of forward and reverse clustering. DETAILED DESCRIPTION OF THE INVENTION

[0060] The present invention will be further described below.

[0061] The specific steps of the present invention are as follows:

[0062] Step 1: Deploy a microseismic monitoring system: First, determine the position of the boundary line. One side of the boundary line is the monitoring area. Two rows of seismic geophone groups are deployed on the boundary line. Each seismic geophone in each seismic geophone group is connected to a microseismic monitor, and the deployment of the microseismic monitoring system is completed.

[0063] Step 2: Microseismic monitoring: Start the microseismic monitoring system to continuously monitor the monitoring area and store the obtained microseismic monitoring data.

[0064] Step 3: Microseismic data processing: The microseismic monitoring host first screens the microseismic monitoring data collected in Step 2 to obtain the occurrence time T and the occurrence position coordinates X, Y of all microseismic events. Then, the forward and reverse DBSCAN clustering analysis method is used to analyze and process all microseismic events, so as to distinguish the microseismic events generated during the excavation of underground intrusion behavior. Finally, the position set of the distinguished microseismic events is obtained, which is the microseismic event set of underground intrusion. The specific process is as follows:

[0065] ① Forward time clustering. During the DBSCAN clustering process, start clustering from the point with the earliest occurrence time among all microseismic events. The neighborhood diffusion only selects the points that occur later or slightly earlier than the core point for clustering. Specifically:

[0066] 1) Parameter setting: Set three parameters: the spatial neighborhood radius ε s and the time neighborhood ε t , and the minimum neighborhood point number MinPts;

[0067] Among them, the spatial neighborhood radius ε s : Defines the neighborhood range of a point in the plane (X, Y), that is, a circular area with the point as the center and a radius of ε s ;

[0068] The time neighborhood ε t : Defines a T neighborhood range in time. During forward clustering, the occurrence time of other events is not earlier than the occurrence time of the microseismic event at this point minus the time neighborhood ε t ; During reverse clustering, the occurrence time of other events is not later than the occurrence time of the microseismic event at this point plus the time neighborhood ε t ;

[0069] The minimum neighborhood point number MinPts: Defines the minimum number of points required for the core point to simultaneously meet the time neighborhood condition within the spatial neighborhood radius ε s ;

[0070] 2) Classify the data points: According to the set parameters, classify the data points into core points, boundary points, and noise points;

[0071] Core point: If within the spatial neighborhood radius ε of a point E, there are at least MinPts microseismic event points that satisfy the following temporal neighborhood condition, then this point is a core point: s Within it, there are at least MinPts microseismic event points that satisfy the following temporal neighborhood condition, then this point is a core point:

[0072] T α >T E -ε t

[0073] Where T α is the occurrence time of a microseismic event point within the spatial domain, and T E is the occurrence time of the microseismic event point E;

[0074] Boundary point: If the number of points within the spatial neighborhood radius ε of a point that satisfy the above temporal neighborhood condition is less than MinPts, but this point is within the spatial neighborhood radius ε of other core points and satisfies the temporal neighborhood condition of other core points, then this point is a boundary point; s Within it, the number of points that satisfy the above temporal neighborhood condition is less than MinPts, but this point is within the spatial neighborhood radius ε of other core points and satisfies the temporal neighborhood condition of other core points, then this point is a boundary point; s Noise point: A point that is neither a core point nor a boundary point;

[0075] Noise point: A point that is neither a core point nor a boundary point;

[0076] 3) Initial selection:

[0077] Select the point with the earliest occurrence time among all microseismic events;

[0078] 4) Expand the cluster: If the initial point is a core point, expand a new cluster starting from this point;

[0079] Mark the initial point as visited;

[0080] Add all core points and boundary points within the spatial neighborhood radius ε of the initial point that satisfy the above temporal neighborhood condition to the cluster; For each core point added to the cluster, repeat step 4) to expand the cluster until the cluster no longer grows; s Add all core points and boundary points within the spatial neighborhood radius ε of the initial point that satisfy the above temporal neighborhood condition to the cluster; For each core point added to the cluster, repeat step 4) to expand the cluster until the cluster no longer grows;

[0081] 5) Mark the boundary point: If the initial point is a boundary point, mark it as visited but do not expand the cluster;

[0082] 6) Process the noise point: If the initial point is a noise point, mark it as a noise point;

[0083] 7) Repeat the process: Repeat steps 3) to 6) until all points have been visited;

[0084] 8) Output the result: Output all the formed clusters and noise points as Figure 4 shown in a.

[0085] ② Reverse time clustering. During the DBSCAN clustering process, start clustering from the point with the latest occurrence time among all microseismic events. The neighborhood diffusion only selects points that occurred earlier or slightly later than the core point for clustering. Specifically:

[0086] 1) Classify data points: According to the set parameters, classify data points into core points, border points, and noise points;

[0087] Core point: If within the spatial neighborhood radius ε of a point E, there are at least MinPts microseismic event points that satisfy the following time neighborhood condition, then this point is a core point: s T

[0088] T α <T E +ε t

[0089] where T α is the occurrence time of a certain microseismic event point within the spatial domain, and T E is the occurrence time of the microseismic event point E;

[0090] Border point: If the number of points within the spatial neighborhood radius ε of a point that satisfy the above time neighborhood condition is less than MinPts, but this point is within the spatial neighborhood radius ε of other core points and satisfies the time neighborhood condition of other core points, then this point is a border point; s s s Noise point: A point that is neither a core point nor a border point;

[0091] 2) Initial selection:

[0092] Select the point with the latest occurrence time among all microseismic events;

[0093]

[0094]

[0094] 3) Expand the cluster: If the initial point is a core point, expand a new cluster starting from this point;

[0095] Mark the initial point as visited;

[0096] Add all core points and border points within the spatial neighborhood radius ε of the initial point that satisfy the above time neighborhood condition to the cluster; For each core point added to the cluster, repeat step 4) to expand the cluster until the cluster no longer grows; s

[0097] 4) Mark border points: If the initial point is a border point, mark it as visited but do not expand the cluster;

[0098] 5) Process noise points: If the initial point is a noise point, mark it as a noise point;

[0099] 6) Repeat the process: Repeat steps 2) to 5) until all points have been visited;

[0100] 7) Output the result: Output all the formed clusters and noise points as Figure 4 shown in b.

[0101] ③ Intersection of clustering results. Intersect the largest cluster after removing noise from the forward-time clustering in step ① with the largest cluster after removing noise from the backward-time clustering in step ②. The microseismic events in the intersection of the two are the microseismic events generated by excavation during underground intrusion, as Figure 4 shown in c.

[0102] Step Four: Determine the underground intrusion excavation positions at different times: Divide the set of microseismic events during underground intrusion into multiple time periods in chronological order, and calculate the average of the position coordinates X and Y of all microseismic events in each time period, so as to monitor the excavation positions of underground intrusion in each time period.

[0103] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A method for microseismic monitoring of underground intrusion behavior based on DBSCAN clustering, characterized in that: The specific steps are: Step 1: Deploy the microseismic monitoring system: first determine the boundary line position, one side of the boundary line is the monitoring area, and two rows of seismic detector groups are deployed on the boundary line. Each seismic detector in each seismic detector group is connected to the microseismic monitor to complete the deployment of the microseismic monitoring system; Step 2: Microseismic monitoring: Start the microseismic monitoring system to continuously monitor the monitoring area and store the obtained microseismic monitoring data; Step 3: Microseismic data processing: The microseismic monitoring host first screens the microseismic monitoring data collected in step 2 to obtain the occurrence time T and the occurrence position coordinates X, Y of all microseismic events, and then uses the forward and reverse DBSCAN clustering analysis method to analyze and process all microseismic events. The specific process is as follows: ① Forward time clustering: In the DBSCAN clustering process, clustering starts from the earliest point in all microseismic events, and neighborhood diffusion only selects points that occur later than the core point for clustering; ② Reverse time clustering: In the DBSCAN clustering process, clustering starts from the point with the latest occurrence time among all microseismic events, and neighborhood diffusion only selects points that occur earlier than the core point for clustering; ③ Intersection of clustering results: the largest cluster after removing noise by forward time clustering in step ① is intersected with the largest cluster after removing noise by reverse time clustering in step ②. The microseismic events at the intersection of the two are the microseismic events generated by excavation during underground intrusion. Finally, the location set of the distinguished microseismic events is obtained, which is the microseismic event set of underground intrusion. Step 4: Determine the underground intrusion excavation location at different times: Divide the underground intrusion microseismic event set into multiple time periods in chronological order, and average the X and Y coordinates of all microseismic events in each time period, so as to monitor the underground intrusion excavation location in each time period.

2. The underground intrusion behavior microseismic monitoring method based on DBSCAN clustering according to claim 1 is characterized in that: The forward time clustering in step ① is specifically as follows: 1) Parameter setting: Set three parameters: spatial neighborhood radius ε s and the temporal neighborhood ε t , and the minimum number of neighborhood points MinPts; The spatial neighborhood radius ε s :Defines the neighborhood of a point on the plane (X, Y), that is, the point is the center and the radius is ε s The circular area of Temporal Neighborhood ε t :Defines a time neighborhood range of T. When performing forward clustering, the occurrence time of other events is no earlier than the occurrence time of the microseismic event at this point minus the time neighborhood ε t ; In reverse clustering, the occurrence time of other events is no later than the occurrence time of the microseismic event at this point plus the time neighborhood ε t ; Minimum number of neighborhood points MinPts: defines the core point in the spatial neighborhood radius ε s The minimum number of points required to simultaneously satisfy the temporal neighborhood conditions; 2) Classify data points: Classify data points into core points, boundary points and noise points according to the set parameters; Core point: If the spatial neighborhood radius ε of a point E s If there are at least MinPts microseismic event points that meet the following time neighborhood conditions, then this point is a core point: T α >T E -ε t Among them, T α is the occurrence time of a microseismic event point in the spatial domain, T E is the occurrence time of the microseismic event point E; Boundary point: If the spatial neighborhood radius of a point is ε s There are less than MinPts points that meet the above time neighborhood conditions, but the point is within the spatial neighborhood radius ε of other core points. s If the point is within the time range of other core points and meets the time neighborhood conditions of other core points, then the point is a boundary point; Noise point: a point that is neither a core point nor a boundary point; 3) Initial selection: Select the earliest point among all microseismic events; 4) Expanding the cluster: If the initial point is a core point, a new cluster is expanded with this point as the starting point; Mark the initial point as visited; Set the spatial neighborhood radius ε of the initial point s All core points and boundary points that meet the above time neighborhood conditions are added to the cluster; for each core point added to the cluster, step 4) is repeated to expand the cluster until the cluster no longer grows; 5) Marking boundary points: If the initial point is a boundary point, mark it as visited but do not expand the cluster; 6) Processing noise points: If the initial point is a noise point, mark it as a noise point; 7) Repeat the process: Repeat steps 3) to 6) until all points have been visited; 8) Output results: Output all formed clusters and noise points.

3. The underground intrusion behavior microseismic monitoring method based on DBSCAN clustering according to claim 1 is characterized in that: The reverse time clustering in step ② is specifically as follows: 1) Classify data points: Classify data points into core points, boundary points and noise points according to the set parameters; Core point: If the spatial neighborhood radius ε of a point E s If there are at least MinPts microseismic event points that meet the following time neighborhood conditions, then this point is a core point: T α <T E +e t Among them, T α is the occurrence time of a microseismic event point in the spatial domain, T E is the occurrence time of the microseismic event point E; Boundary point: If the spatial neighborhood radius of a point is ε s There are less than MinPts points that meet the above time neighborhood conditions, but the point is within the spatial neighborhood radius ε of other core points. s If the point is within the time range of other core points and meets the time neighborhood conditions of other core points, then the point is a boundary point; Noise point: a point that is neither a core point nor a boundary point; 2) Initial selection: Select the point with the latest occurrence time among all microseismic events; 3) Expanding the cluster: If the initial point is a core point, a new cluster is expanded with this point as the starting point; Mark the initial point as visited; Set the spatial neighborhood radius ε of the initial point s All core points and boundary points that meet the above time neighborhood conditions are added to the cluster; for each core point added to the cluster, step 4) is repeated to expand the cluster until the cluster no longer grows; 4) Marking boundary points: If the initial point is a boundary point, mark it as visited but do not expand the cluster; 5) Processing noise points: If the initial point is a noise point, mark it as a noise point; 6) Repeat the process: Repeat steps 2) to 5) until all points have been visited; 7) Output results: Output all formed clusters and noise points.

Images