Rockburst incubation process multi-source information analysis method and system based on improved delay-dsgn

Abstract

The application relates to the technical field of data processing, and discloses a rock burst incubation process multi-source information analysis method and system based on an improved Delay-DSGN, which aims to solve the problem of poor reliability of analysis results of an existing method, and mainly comprises the following steps: obtaining rock burst incubation monitoring data and forming monitoring unit time sequence data, generating a dynamic pulse diagram with monitoring units as nodes and rock mass energy response states as pulse states; determining an energy migration reachable delay time window based on monitoring point positions, microseismic event migration and stress displacement response directions, transforming a Gaussian delay kernel according to the delay time window, respectively determining the effectiveness of microseismic and non-microseismic pulses and determining aggregation contribution, and then updating node state propagation rock burst incubation states; and determining an incubation stage and a risk level according to a propagation result. The application improves the accuracy of rock burst incubation state propagation analysis and is suitable for deep mines, underground caverns, tunnel excavation and the like.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, specifically to a method and system for multi-source information analysis of rockburst incubation process based on an improved Delay-DSGN. Background Technology

[0002] Rockbursts are common dynamic hazards in deep mines, underground caverns, and tunnel excavation, typically occurring under conditions of high ground stress, intense excavation disturbance, and continuous local energy accumulation. During rockburst incubation, the rock mass undergoes microcrack propagation, energy release, stress redistribution, and deformation response. These processes can be observed using monitoring methods such as microseismic activity, acoustic emission, stress, and displacement. Because rockburst incubation is not a single-point instantaneous event but exhibits spatial migration and temporal lag between different monitoring areas, a single monitoring indicator cannot fully reflect the process of rock mass energy propagation and enhancement from local accumulation. Therefore, identifying the rock mass energy response state from multi-source monitoring data and analyzing the rockburst incubation process by combining the spatial relationships and response sequence between monitoring units has become a crucial technical aspect of underground engineering safety monitoring.

[0003] Application publication number CN120656302A discloses a method, equipment, and medium for early rockburst identification based on physical information injection. It mainly discloses the collection of acoustic emission waveform data, microseismic waveform data, triaxial stress data, and tunnel deformation and displacement data within underground engineering areas. This data is then combined with static geological attribute data for time synchronization and spatial coordinate alignment to generate high-dimensional feature cube data with a unified spatiotemporal reference. Subsequently, the high-dimensional feature cube data is input into a physical information injection transformer for modeling, obtaining intermediate feature vector data such as energy accumulation rate, stress concentration factor, and fracture evolution potential, as well as three-dimensional hazard confidence distribution data. Based on this, risk heat maps, potential fracture surface spatial location data, and risk assessment results are generated for early rockburst identification and warning.

[0004] However, in the analysis scenario where rockburst propagation occurs along different monitoring units, existing methods mainly organize multi-source monitoring data into unified spatiotemporal features and perform overall modeling. They haven't yet addressed whether the historical responses of the source monitoring units and the current responses of the target monitoring units conform to the rock mass energy migration direction and hysteresis time. Microseismic events, stress responses, and displacement responses may have similar intensities at different locations, but they may not belong to the same energy migration path. If historical responses that conform to reachable hysteresis relationships are not distinguished from those that do not, the model is prone to including physically unreachable historical pulses in the propagation state aggregation, causing the rockburst propagation state analysis to deviate from the actual energy migration process, thus resulting in poor reliability of the rockburst propagation state analysis results. Summary of the Invention

[0005] This invention aims to address the problem of poor reliability of analysis results in existing rockburst incubation state analysis methods, and proposes a multi-source information analysis method and system for rockburst incubation process based on an improved Delay-DSGN.

[0006] The technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows:

[0007] In a first aspect, the present invention provides a multi-source information analysis method for rockburst incubation process based on an improved Delay-DSGN, the method comprising:

[0008] Step 1: Obtain rockburst incubation monitoring data, and form monitoring unit time series data based on the rockburst incubation monitoring data. The rockburst incubation monitoring data includes microseismic events, acoustic emission response, stress response, displacement response, and monitoring point location.

[0009] Step 2: Generate a dynamic pulse diagram based on the time series data of the monitoring unit. The nodes of the dynamic pulse diagram are rockburst incubation monitoring units, and the pulse state of the nodes of the dynamic pulse diagram represents the rock mass energy response state of the rockburst incubation monitoring unit.

[0010] Step 3: Based on the monitoring point locations, microseismic event migration relationships, stress response directions, and displacement response directions of the source rockburst incubation monitoring unit and the target rockburst incubation monitoring unit in the dynamic pulse diagram, determine the energy migration reachable delay time window from the source rockburst incubation monitoring unit to the target rockburst incubation monitoring unit. The energy migration reachable delay time window is used to limit the hysteresis time that conforms to the rock mass energy migration.

[0011] Step 4: Modify the Gaussian delay kernel neighborhood aggregation link of Delay-DSGN according to the energy migration reachable delay time window to form a reachable time window limited delay kernel; the reachable time window limited delay kernel determines the aggregation contribution of the historical pulse of the source rockburst incubation monitoring unit to the current state of the target rockburst incubation monitoring unit according to the relationship between the historical pulse time interval and the energy migration reachable delay time window.

[0012] Step 5: Based on the dynamic pulse diagram, the reachable time window-limited delay kernel, and the aggregated contribution, perform Delay-DSGN node state updates to obtain the propagation results of rockburst incubation state;

[0013] Step 6: Determine the rockburst incubation stage and rockburst risk status based on the propagation results of the rockburst incubation state, and output the rockburst incubation analysis results.

[0014] Further, in step 1, the formation of monitoring unit time-series data based on the rockburst incubation monitoring data specifically includes:

[0015] The microseismic events, acoustic emission responses, stress responses, displacement responses, and monitoring point locations in the rockburst incubation monitoring data are matched according to the acquisition time to form a multi-source response sequence;

[0016] Based on the location of the monitoring points, the microseismic events, acoustic emission responses, stress responses, and displacement responses in the multi-source response sequence are assigned to the corresponding rockburst incubation monitoring units to form a monitoring unit response sequence.

[0017] The monitoring unit response sequence is divided into time slices according to a preset time slice, and the order of responses is retained within the same time slice to form a monitoring unit time slice response sequence.

[0018] The microseismic events, acoustic emission responses, stress responses, displacement responses, and state fields corresponding to the monitoring point locations are extracted from the time-slice response sequence of the monitoring units and arranged according to the rockburst incubation monitoring units and time slices to form the monitoring unit time-series data.

[0019] Further, in step 2, generating a dynamic pulse diagram based on the time-series data of the monitoring unit specifically includes:

[0020] Based on the monitoring point locations in the time-series data of the monitoring unit, the monitoring point locations are matched with the spatial boundaries of the preset rockburst incubation monitoring unit to determine the correspondence between the rockburst incubation monitoring unit and the nodes of the dynamic pulse diagram.

[0021] Based on the correspondence, the micro-seismic events, acoustic emission responses, stress responses, and displacement responses of each node in the dynamic pulse map within the time slice are extracted from the time series data of the monitoring unit to form a node multi-source response sequence;

[0022] Based on the arrival time of microseismic events in the nodal multi-source response sequence, as well as the acquisition time of acoustic emission response, stress response and displacement response, the response sequence of the nodal multi-source response sequence is calibrated to form a nodal hysteresis response sequence.

[0023] Based on the microseismic event energy release, acoustic emission response energy, stress response direction, displacement response direction, and monitoring point location in the node hysteresis response sequence, the rock mass energy response state of the rockburst incubation monitoring unit within the time slice is determined.

[0024] The energy response state of the rock mass is bound to the node of the corresponding dynamic pulse diagram, and the node pulse state is calibrated with the response occurrence time to form a node pulse state with the response occurrence time.

[0025] Based on the monitoring point location corresponding to the node in the dynamic pulse diagram and the node pulse state with the response occurrence time, an adjacent propagation relationship is established between the nodes in the dynamic pulse diagram to generate a dynamic pulse diagram in which the node is a rockburst incubation monitoring unit and the node pulse state represents the energy response state of the rock mass.

[0026] Furthermore, the method for establishing the adjacent propagation relationship is as follows:

[0027] If node With nodes The corresponding preset rockburst incubation monitoring unit spatial boundary satisfies the following spatial adjacency conditions: there is a common boundary or the minimum Euclidean distance between two spatial boundaries does not exceed the preset spatial adjacency threshold; and the nodes With nodes If there are node pulse states with response times within the same time slice, then at node... With nodes Establish between nodes Pointing to node The adjacent propagation relationship.

[0028] Further, in step 3, determining the energy migration reachable delay window from the source rockburst incubation monitoring unit to the target rockburst incubation monitoring unit specifically includes:

[0029] Define source rockburst incubation monitoring unit In the source historical time slice To the target rockburst incubation monitoring unit In the target time slice spatial distance for: Define directional consistency measure for:

[0030] ;

[0031] The energy migration can reach the delayed time window. lower boundary and upper boundary They are respectively:

[0032] ;

[0033] ;

[0034] When the temporal migration condition, spatial migration condition, directional consistency condition, and boundary validity condition are simultaneously satisfied... ;otherwise :

[0035] The timing transition conditions are as follows: ;

[0036] The spatial migration conditions are as follows: ,and ;

[0037] The condition for directional consistency is: ;

[0038] The boundary validity condition is: ;

[0039] in, Indicates the source rockburst incubation monitoring unit In the source historical time slice To the target rockburst incubation monitoring unit In the target time slice Energy migration can reach a time delay window; Index of source rockburst incubation monitoring unit; Indicates the index of the target rockburst incubation monitoring unit; Indicates the source historical time slice index; Indicates the target time slice index; This indicates that the source rockburst incubation monitoring unit is located in the source history time slice. Location of monitoring points; This indicates that the target rockburst incubation monitoring unit is in the target time segment. Location of monitoring points; Denotes the Euclidean norm; Indicates the arrival time of microseismic events in the source rock burst incubation monitoring unit; Indicates the arrival time of microseismic events for the target rockburst incubation monitoring unit; Indicates the direction of location change of microseismic events; Indicates the stress response direction of the rockburst incubation monitoring unit; Indicates the stress response direction of the target rockburst incubation monitoring unit; Indicates the direction of displacement response of the source rockburst incubation monitoring unit; Indicates the direction of displacement response of the target rockburst incubation monitoring unit; This represents the preset threshold corresponding to the directional consistency value; This represents the minimum delay time allowed to enter the Gaussian delayed kernel neighborhood aggregation stage; This represents the maximum delay time allowed to enter the Gaussian delayed kernel neighborhood aggregation stage; This represents the minimum hysteresis coefficient per unit distance. This indicates the maximum hysteresis coefficient per unit distance. This indicates the preset upper boundary contraction coefficient; This indicates an indicator function; it takes the value 1 if the condition within the parentheses is true, and takes the value 0 if the condition is false. Represents the maximum value function; Describes the minimum value function; This indicates that no effective energy migration can be achieved within a certain time window.

[0040] Furthermore, in step 4, the relationship between the historical pulse time interval and the energy migration reachability delay window is represented by the validity of the delay response calculation, which is determined according to the following formula:

[0041] ;

[0042] in, Indicates the source rockburst incubation monitoring unit In the source historical time slice Historical pulses act on the target rockburst incubation monitoring unit In the target time slice The validity of the delayed response calculation; Index of source rockburst incubation monitoring unit; Indicates the index of the target rockburst incubation monitoring unit; Indicates the source historical time slice index; Indicates the target time slice index; Indicates the polymerization inhibition coefficient. ; This indicates an indicator function; it takes the value 1 if the condition within the parentheses is true, and takes the value 0 if the condition is false. Numerical labels representing adjacent propagation relationships, when nodes To the node In the When there is an adjacent propagation relationship within a target time slice ,otherwise ; Indicates the source rockburst incubation monitoring unit The corresponding nodes in the dynamic pulse diagram; Indicates the target rockburst incubation monitoring unit The corresponding nodes in the dynamic pulse diagram; This indicates that the energy migration reachability delay window is valid. hour ,otherwise ; This indicates that energy migration can reach the lower boundary of the delay time window; This indicates that energy migration can reach the upper boundary of the delay time window; Indicates the historical pulse occurrence time of the source rock burst incubation monitoring unit; This indicates the current state of the target rockburst incubation monitoring unit at the corresponding time. Indicates the historical pulse time interval;

[0043] when When the historical pulse time interval satisfies the energy migration reachable delay time window, the historical pulse of the source rockburst incubation monitoring unit is considered valid; when When the historical pulse time interval does not meet the energy migration reachable delay time window, an aggregation suppression mark is formed for the corresponding historical pulse.

[0044] Further, in step 4, the method for determining the aggregation contribution includes:

[0045] The historical pulses of the identified valid source rockburst inoculation monitoring units are grouped into a set, which serves as the effective calculation range of the delay kernel. The effective calculation range of the delay kernel is determined according to the following formula:

[0046] ;

[0047] in, Indicates the target rockburst incubation monitoring unit In the target time slice The corresponding effective computational range of the delayed kernel; Indicates the index of the target rockburst incubation monitoring unit; Indicates the target time slice index; Index of source rockburst incubation monitoring unit; Indicates the source historical time slice index; Indicates the number of rockburst incubation monitoring units; Indicates the source rockburst incubation monitoring unit In the source historical time slice Candidate historical pulses; Numerical markers representing adjacent propagation relationships; This indicates the effective marker of the energy migration reach delay time window; Indicates the historical pulse time interval; Indicates the historical pulse occurrence time of the source rock burst incubation monitoring unit; This indicates the current state of the target rockburst incubation monitoring unit at the corresponding time. This indicates that energy migration can reach the lower boundary of the delay time window; This indicates that energy migration can reach the upper boundary of the delay time window;

[0048] When the candidate historical pulse corresponds to Within the effective computational range of the delayed kernel At that time, the historical pulses of the source rock burst incubation monitoring unit retain the qualification for aggregation; when the candidate historical pulse corresponds to Not within the effective computational range of delayed kernels At that time, the historical pulses of the source rock burst incubation monitoring unit do not retain aggregation eligibility; historical pulses that retain aggregation eligibility participate in neighborhood aggregation; historical pulses that participate in neighborhood aggregation are aggregated by Gaussian delay kernel neighborhood aggregation to obtain the aggregation contribution.

[0049] Furthermore, in step 5, the Delay-DSGN node state update includes a Gaussian delay kernel neighborhood aggregation step and a node state update step, and a pulse coding step is included before the Gaussian delay kernel neighborhood aggregation step;

[0050] The pulse coding stage receives the node pulse state of each node in the dynamic pulse diagram and converts the node pulse state into a node pulse state code.

[0051] The Gaussian delay kernel neighborhood aggregation step receives historical pulses from the source rockburst incubation monitoring unit, the current state of the target rockburst incubation monitoring unit, the historical pulse time interval, and the reachable time window-defined delay kernel. In this step, multiple aggregation neurons determine the delay kernel weights based on the historical pulse time interval and the reachable time window-defined delay kernel, and generate constrained delay responses based on the historical pulses from the source rockburst incubation monitoring unit, the current state of the target rockburst incubation monitoring unit, and the delay kernel weights. All constrained delay responses from the same target rockburst incubation monitoring unit and the same target time slice are then aggregated in a neighborhood to obtain the aggregation contribution.

[0052] The node status update step receives the aggregated contribution and the current status of the target rockburst incubation monitoring unit, and generates the updated node status of the target rockburst incubation monitoring unit based on the aggregated contribution and the current status of the target rockburst incubation monitoring unit.

[0053] The process is recursively applied along the time slices of the dynamic pulse diagram. The updated node state of the current time slice is used as the current state of the target rockburst incubation monitoring unit in the next time slice. The Gaussian delay kernel neighborhood aggregation and node state update steps are then executed until the recursion of all time slices is completed, and the propagation result of the rockburst incubation state is obtained.

[0054] Furthermore, step 6 specifically includes:

[0055] Based on the propagation results of the rockburst incubation state, the propagation state of each rockburst incubation monitoring unit within the time slice is aggregated to form a propagation characterization of the rockburst incubation monitoring unit.

[0056] Based on the propagation characterization of the rockburst incubation monitoring unit, a rockburst incubation stage determination result is formed when... When the energy enhancement gestation stage is determined, The low-energy response gestation stage determination results are formed at that time; among them, This indicates the index of the rockburst incubation monitoring unit. Indicates the time slice index. Indicates rockburst incubation monitoring unit In time slice The stage response value, that is, the difference between the output component of the energy-enhanced incubation stage and the output component of the low-energy response incubation stage, Indicates the preset stage threshold;

[0057] Based on the propagation characterization of the rockburst incubation monitoring unit and the results of the rockburst incubation stage determination, the risk response value is calculated:

[0058] ;

[0059] in, Indicates the risk response value; The output component represents the low-risk state. The output component represents the medium-risk state; The output component represents a high-risk state; This indicates the pre-set risk bias. This indicates an indicator function; it takes the value 1 if the condition inside the parentheses is true, and takes the value 0 if the condition is false. This indicates the result of the rockburst incubation stage determination. This indicates the result of the energy enhancement gestation stage assessment. This indicates the result of determining the incubation stage of a low-energy response;

[0060] The risk response value With preset risk threshold Comparisons are made to generate a rockburst risk status assessment result. When a low-risk status determination result is formed, When a medium-risk status determination result is formed, A high-risk status determination result is generated at that time; among them, This indicates the boundary between low-risk and medium-risk states. This indicates the boundary between medium-risk and high-risk states. ;

[0061] The rockburst incubation stage is determined based on the rockburst incubation stage determination result, and the rockburst risk state is determined based on the rockburst risk state determination result. The rockburst incubation stage and rockburst risk state are output as the rockburst incubation analysis result according to the correspondence between the rockburst incubation monitoring unit and the time slice.

[0062] Secondly, the present invention provides a multi-source information analysis system for rockburst incubation processes based on an improved Delay-DSGN, used to implement the multi-source information analysis method for rockburst incubation processes based on an improved Delay-DSGN as described in the first aspect, the system comprising:

[0063] The data acquisition module is used to acquire rockburst incubation monitoring data and form monitoring unit time series data based on the rockburst incubation monitoring data. The rockburst incubation monitoring data includes microseismic events, acoustic emission response, stress response, displacement response, and monitoring point location.

[0064] The dynamic pulse diagram generation module is used to generate a dynamic pulse diagram based on the time series data of the monitoring unit. The nodes of the dynamic pulse diagram are rockburst incubation monitoring units, and the pulse state of the nodes of the dynamic pulse diagram represents the rock mass energy response state of the rockburst incubation monitoring unit.

[0065] The delay time window determination module is used to determine the energy migration reachable delay time window from the source rockburst incubation monitoring unit to the target rockburst incubation monitoring unit based on the monitoring point positions, microseismic event migration relationships, stress response direction, and displacement response direction of the source rockburst incubation monitoring unit and the target rockburst incubation monitoring unit in the dynamic pulse diagram. The energy migration reachable delay time window is used to limit the hysteresis time that conforms to the rock mass energy migration.

[0066] The delay core modification module is used to modify the Gaussian delay core neighborhood aggregation link of Delay-DSGN according to the energy migration reachable delay time window to form a reachable time window limited delay core; the reachable time window limited delay core determines the aggregation contribution of the historical pulse of the source rockburst incubation monitoring unit to the current state of the target rockburst incubation monitoring unit according to the relationship between the historical pulse time interval and the energy migration reachable delay time window;

[0067] The node state update module is used to update the Delay-DSGN node state based on the dynamic pulse diagram, the reachable time window-limited delay kernel, and the aggregate contribution, so as to obtain the propagation result of the rockburst incubation state.

[0068] The analysis results output module is used to determine the rockburst incubation stage and rockburst risk status based on the propagation results of the rockburst incubation state, and output the rockburst incubation analysis results.

[0069] The beneficial effects of this invention are as follows: The multi-source information analysis method and system for rockburst incubation process based on the improved Delay-DSGN provided by this invention constructs an energy migration reachability delay time window based on the monitoring point location, microseismic event migration relationship, stress response direction, and displacement response direction. This time window is then used to modify the Gaussian delay kernel neighborhood aggregation link of Delay-DSGN to form a reachability time window-limited delay kernel. This ensures that node state updates only allow historical pulses that conform to the energy migration direction and hysteresis time to participate in neighborhood aggregation. This solves the problem in the prior art where existing methods, due to the failure to distinguish between physically reachable and inaccessible historical responses, include irrelevant pulses in the propagation state aggregation, leading to analysis results deviating from the actual energy migration process. Therefore, this improves the accuracy of rockburst incubation state propagation analysis, reduces stage misjudgments and risk false alarms caused by erroneous propagation, and provides a more reliable decision-making basis for engineering safety monitoring. Attached Figure Description

[0070] Figure 1 A flowchart illustrating the multi-source information analysis method for rockburst incubation process based on the improved Delay-DSGN provided in this embodiment;

[0071] Figure 2 A schematic diagram illustrating the formation process of the dynamic pulse diagram provided in this embodiment;

[0072] Figure 3 A schematic diagram of the core propagation calculation process of the Delay-DSGN model provided for the embodiment;

[0073] Figure 4 A schematic diagram illustrating the output process of stage determination and risk determination provided in the embodiment;

[0074] Figure 5 A schematic diagram showing the comparison between the improved Delay-DSGN and the original Delay-DSGN in terms of rockburst risk identification timing, provided for an example.

[0075] Figure 6 A schematic diagram of the structure of a multi-source information analysis system for rockburst incubation process based on an improved Delay-DSGN, provided as an example. Detailed Implementation

[0076] The core technical problem to be solved by this invention is how to limit the reachable hysteresis time of the historical pulse of the source rockburst initiation monitoring unit to the current state of the target rockburst initiation monitoring unit based on the location of the monitoring point, the migration relationship of microseismic events, the direction of stress response and the direction of displacement response in a multi-source rockburst initiation monitoring scenario, so that the rockburst initiation state propagation analysis conforms to the rock mass energy migration process, thereby improving the reliability of the rockburst initiation state analysis results.

[0077] The core concept of this invention lies in embedding the rock mass energy migration mechanism into the core aggregation stage of a dynamic graph pulse propagation model. First, multi-source monitoring data is organized into time-series data of monitoring units according to monitoring units and time slices, generating a dynamic pulse graph with rockburst incubation monitoring units as nodes and node pulse states representing the rock mass energy response state, thus establishing a mapping from the original monitoring data to the graph structure data. Then, before propagation calculations begin, an energy migration reachability delay time window is constructed based on the monitoring point locations between the source and target monitoring units, the migration relationship of microseismic events, the direction of stress response, and the direction of displacement response. This window serves as a constraint to determine whether historical pulses conform to the energy migration path. Next, this time window is used to modify the Gaussian delay kernel neighborhood aggregation stage of the Delay-DSGN model, ensuring that node state updates only allow historical pulses whose time intervals fall within this time window to participate in neighborhood aggregation, excluding historical pulses that do not conform to the actual energy migration path from the propagation calculation. Finally, based on the constrained propagation results, the rockburst incubation stage and rockburst risk state are determined, ensuring that the analysis results strictly adhere to the physical process of rock mass energy accumulation from local accumulation to propagation enhancement.

[0078] The technical solutions in this embodiment will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0079] Figure 1 A flowchart illustrating a multi-source information analysis method for rockburst incubation processes based on an improved Delay-DSGN is shown. Please refer to [link / reference]. Figure 1 The method includes the following steps:

[0080] Step 1: Obtain rockburst incubation monitoring data, and form monitoring unit time series data based on the rockburst incubation monitoring data. The rockburst incubation monitoring data includes microseismic events, acoustic emission response, stress response, displacement response, and monitoring point location.

[0081] In this embodiment, the time-series data of the monitoring unit is formed based on the rockburst incubation monitoring data, specifically including:

[0082] The microseismic events, acoustic emission responses, stress responses, displacement responses, and monitoring point locations in the rockburst incubation monitoring data are matched according to the acquisition time to form a multi-source response sequence;

[0083] Based on the location of the monitoring points, the microseismic events, acoustic emission responses, stress responses, and displacement responses in the multi-source response sequence are assigned to the corresponding rockburst incubation monitoring units to form a monitoring unit response sequence.

[0084] The monitoring unit response sequence is divided into time slices according to a preset time slice, and the order of responses is retained within the same time slice to form a monitoring unit time slice response sequence.

[0085] The microseismic events, acoustic emission responses, stress responses, displacement responses, and state fields corresponding to the monitoring point locations are extracted from the time-slice response sequence of the monitoring units and arranged according to the rockburst incubation monitoring units and time slices to form the monitoring unit time-series data.

[0086] In this embodiment, the acquired microseismic events are denoted as... ,in Indicates the first The timing of data acquisition for each microseismic event. Indicates the first Content of the micro-earthquake event, This indicates the number of microseismic events recorded. The acoustic emission response is denoted as... ,in Indicates the acquisition time of the acoustic emission response. This indicates the content of the acoustic emission response. This indicates the number of acoustic emission response records. The stress response is denoted as... ,in Indicates the time of data acquisition for the stress response. Indicates the stress response content. This indicates the number of stress response records. The displacement response is denoted as... ,in Indicates the time of data acquisition for the displacement response. This indicates the content of the displacement response. This indicates the number of displacement response records. The location of the monitoring point is recorded as... ,in Indicates the data collection time at the monitoring point location. This indicates the horizontal position and depth position of the monitoring point. This indicates the number of monitoring point location records.

[0087] In practical applications, the first step is to determine the set of monitoring point locations from the monitoring point locations. , Indicates the first Location of each monitoring point This indicates the number of monitoring point locations. For each monitoring point location... The microseismic events, acoustic emission responses, stress responses, and displacement responses that meet the preset position matching thresholds are selected. The records at the same monitoring point location are used as the records to be matched. The acquisition times of the records to be matched are merged and sorted in ascending order to obtain the baseline acquisition time. , Indicates the first The number of reference acquisition times corresponding to each monitoring point location. For any reference acquisition time... In the preset matching window Within the data, microseismic events with the smallest time difference, acoustic emission responses, stress responses, displacement responses, and monitoring point locations are selected to form multi-source response records. The multi-source response records at the same monitoring point are arranged according to the baseline acquisition time to form a multi-source response sequence. A multi-source response sequence refers to a data sequence of microseismic events, acoustic emission responses, stress responses, and displacement responses arranged according to the acquisition time within the same rockburst incubation monitoring unit.

[0088] Then, based on the location of the monitoring points, the multi-source response sequences are assigned to rockburst incubation monitoring units. Let the set of rockburst incubation monitoring units be denoted as . , Indicates the first One rockburst incubation monitoring unit This indicates the number of rockburst incubation monitoring units; each rockburst incubation monitoring unit corresponds to a preset rockburst incubation monitoring unit spatial boundary. The spatial boundary of the pre-set rockburst incubation monitoring unit is determined by the engineering excavation area, the horizontal position range of the monitoring points, and the depth range of the monitoring points. When the monitoring point location... Falling into the spatial boundary of the pre-set rockburst incubation monitoring unit At that time, the multi-source response sequence Microseismic events, acoustic emission responses, stress responses, and displacement responses are incorporated into the rockburst incubation monitoring unit. The data are then arranged in ascending order according to the time of collection to form a monitoring unit response sequence. .

[0089] Next, the response sequence of the monitoring unit is divided into time slices. A preset time slice length is used. Constructing a unified timeline ,in Indicates the start time. This indicates the total number of time slices, which are arranged in chronological order. A time slice ( Corresponding time interval The monitoring unit response sequence The collection time falls into the first The multi-source response records of each time slice are included , Indicates rockburst incubation monitoring unit In the The monitoring unit time-slice response sequence refers to the data sequence formed by aligning multi-source response sequences according to time slices, and preserving the chronological order of each response within the same time slice. Within the monitoring unit time-slice response sequence, microseismic events are sorted by arrival time, and acoustic emission responses, stress responses, and displacement responses are sorted by acquisition time. When multiple responses have the same acquisition time, they are arranged according to the field order of microseismic event, acoustic emission response, stress response, and displacement response, thus preserving the chronological order of responses within the same time slice. Preserving the chronological order of responses within the same time slice means arranging microseismic events, acoustic emission responses, stress responses, and displacement responses falling within the same time slice according to their arrival time or acquisition time.

[0090] Finally, state fields are extracted from the time-slice response sequences of the monitoring units. State fields refer to the fields extracted from the time-slice response sequences of the monitoring units that characterize microseismic events, acoustic emission responses, stress responses, displacement responses, and monitoring point locations. This is done for each rockburst incubation monitoring unit. and each time slice Constructing a twelve-dimensional state vector Twelve-dimensional state vector The first dimension is the number of events, determined by the number of microseismic event records in the time-slot response sequence of the monitoring unit; the second dimension is the energy release, accumulated from the energy release records of microseismic events in the time-slot response sequence of the monitoring unit; the third dimension is the event arrival time, using the arrival time of the earliest microseismic event within the time slot; the fourth dimension is the event location change, read from the microseismic event content; if the microseismic event content does not provide the event location change, it is determined based on the location of the last microseismic event within the time slot relative to the location of the first microseismic event; the fifth dimension is the acoustic emission count, accumulated from the acoustic emission response count within the time slot; the sixth dimension is the acoustic emission energy... The quantity is obtained by accumulating the acoustic emission response energy within the time slice; the seventh dimension is the stress change, determined based on the last stress response value relative to the first stress response value within the time slice; the eighth dimension is the stress response direction, using the direction encoding value corresponding to the stress change; the ninth dimension is the displacement change, determined based on the last displacement response value relative to the first displacement response value within the time slice; the tenth dimension is the displacement response direction, using the direction encoding value corresponding to the displacement change; the eleventh dimension is the horizontal position of the monitoring point, and the eleventh dimension is the depth position of the monitoring point. The horizontal and depth positions of the monitoring points are determined by the monitoring point positions included in the rockburst incubation monitoring unit. The time-series data of the monitoring unit is denoted as... each For the node input of the dynamic pulse map received by Delay-DSGN, the order of the twelve status fields remains consistent across all rockburst incubation monitoring units and all time slices.

[0091] Step 2: Generate a dynamic pulse diagram based on the time-series data of the monitoring unit. The nodes of the dynamic pulse diagram are rockburst incubation monitoring units, and the pulse state of the nodes in the dynamic pulse diagram represents the rock mass energy response state of the rockburst incubation monitoring unit.

[0092] In this embodiment, generating a dynamic pulse diagram based on the time-series data of the monitoring unit specifically includes:

[0093] Based on the monitoring point locations in the time-series data of the monitoring unit, the monitoring point locations are matched with the spatial boundaries of the preset rockburst incubation monitoring unit to determine the correspondence between the rockburst incubation monitoring unit and the nodes of the dynamic pulse diagram.

[0094] Based on the correspondence, the micro-seismic events, acoustic emission responses, stress responses, and displacement responses of each node in the dynamic pulse map within the time slice are extracted from the time-series data of the monitoring unit to form a node multi-source response sequence;

[0095] Based on the arrival time of microseismic events in the nodal multi-source response sequence, as well as the acquisition time of acoustic emission response, stress response and displacement response, the response sequence of the nodal multi-source response sequence is calibrated to form a nodal hysteresis response sequence.

[0096] Based on the microseismic event energy release, acoustic emission response energy, stress response direction, displacement response direction, and monitoring point location in the node hysteresis response sequence, the rock mass energy response state of the rockburst incubation monitoring unit within the time slice is determined.

[0097] The energy response state of the rock mass is bound to the node of the corresponding dynamic pulse diagram, and the node pulse state is calibrated with the response occurrence time to form a node pulse state with the response occurrence time.

[0098] Based on the monitoring point location corresponding to the node in the dynamic pulse diagram and the node pulse state with the response occurrence time, an adjacent propagation relationship is established between the nodes in the dynamic pulse diagram to generate a dynamic pulse diagram in which the node is a rockburst incubation monitoring unit and the node pulse state represents the energy response state of the rock mass.

[0099] In this embodiment, the timing data of the monitoring unit is denoted as... , Indicates the first One rockburst incubation monitoring unit This indicates the index of the rockburst incubation monitoring unit. Indicates the number of rockburst incubation monitoring units. Indicates the time slice index. Indicates the number of time slices. Indicates rockburst incubation monitoring unit In the The state vector within each time slice. This is a twelve-dimensional vector, with fields including, in order: number of events, energy release, event arrival time, event location change, acoustic emission count, acoustic emission energy, stress change, stress response direction, displacement change, displacement response direction, horizontal position of the monitoring point, and depth position of the monitoring point. The horizontal position and depth position of the monitoring point together constitute the monitoring point location. The status field in the monitoring unit's time-series data is bound to the response occurrence time marker, which is represented as... , Indicates the arrival time of a microseismic event. Indicates the acquisition time of the acoustic emission response. Indicates the time of data acquisition for the stress response. This indicates the acquisition time of the displacement response. The response occurrence time marker is used to determine the sequence of responses and does not increase the node input dimension of the dynamic pulse diagram.

[0100] When determining the nodes of the dynamic pulse diagram based on the location of monitoring points, a set of rockburst incubation monitoring units is set up. And configure a preset spatial boundary for each rockburst incubation monitoring unit. The spatial boundary of the pre-set rockburst incubation monitoring unit is determined by the engineering excavation area, the horizontal location range of the monitoring points, and the depth location range of the monitoring points. The location of the monitoring points... Spatial boundary with the pre-set rockburst incubation monitoring unit Perform matching; when the monitoring point location Falling into the spatial boundary of the pre-set rockburst incubation monitoring unit At that time, the rockburst incubation monitoring unit Nodes bound to a dynamic pulse graph , Indicates the first The nodes of a dynamic pulse graph. The set of nodes in a dynamic pulse graph is denoted as . Each node in the dynamic pulse diagram corresponds to a rockburst incubation monitoring unit.

[0101] Based on the correspondence between the rockburst incubation monitoring units and the nodes of the dynamic pulse diagram, the microseismic events, acoustic emission responses, stress responses, and displacement responses of each node in the dynamic pulse diagram within each time slice are extracted from the time-series data of the monitoring units. For the nodes of the dynamic pulse diagram... , will the Microseismic events extracted within a time slice are denoted as Acoustic emission response is denoted as The stress response is denoted as The displacement response is denoted as And form a multi-source response record for the nodes. The node multi-source response records are arranged according to their time-slice index to form a node multi-source response sequence. A node multi-source response sequence refers to a data sequence formed by arranging microseismic events, acoustic emission responses, stress responses, and displacement responses corresponding to nodes in a dynamic pulse diagram according to time slices. The node multi-source response sequence retains the source fields of the microseismic events, acoustic emission responses, stress responses, and displacement responses.

[0102] When calibrating the response sequence of a node's multi-source response sequence, the response is marked according to the time of occurrence. The microseismic events, acoustic emission responses, stress responses, and displacement responses of nodes in the same dynamic pulse diagram within the same time slice are sorted. The sorting rule is as follows: according to... , , and The responses are arranged from earliest to latest; when multiple responses occur at the same time, they are arranged in the order of microseismic event, acoustic emission response, stress response, and displacement response; if a response's occurrence time is marked as invalid by a preset flag, the corresponding response is not included in the response sequence calibration. The node multi-source response sequence after response sequence calibration forms the node hysteresis response sequence. , Nodes representing dynamic pulse diagrams In the Microseismic events, acoustic emission responses, stress responses, and displacement responses are arranged according to their occurrence times within a time slice. Response chronology calibration refers to determining the temporal order of responses in the nodal multi-source response sequence based on the arrival times of microseismic events and the acquisition times of the acoustic emission, stress, and displacement responses. The nodal hysteresis response sequence is a data sequence formed after response chronology calibration of the nodal multi-source response sequence, used to represent the hysteresis relationships between different responses within the same dynamic pulse diagram.

[0103] When determining the energy response state of rock mass based on the nodal hysteresis response sequence, the nodal hysteresis response sequence is used. and monitoring point locations Construct node pulse state Node pulse state A twelve-dimensional vector The twelve fields sequentially represent the number of events, energy release, event arrival time, event location change, acoustic emission count, acoustic emission energy, stress change, stress response direction, displacement change, displacement response direction, horizontal position of the monitoring point, and depth position of the monitoring point. The energy release and acoustic emission response energy of the microseismic event characterize the energy response intensity of the rockburst incubation monitoring unit; the stress response direction and displacement response direction characterize the rock mass response direction of the rockburst incubation monitoring unit; the monitoring point position maintains the spatial boundary of the rockburst incubation monitoring unit; and the node pulse state... The rock mass energy response state of the rockburst incubation monitoring unit within the time slice. The rock mass energy response state refers to the energy response state of the rockburst incubation monitoring unit within the time slice, characterized by microseismic event energy release, acoustic emission response energy, stress response direction, displacement response direction, and monitoring point location. The node pulse state refers to the rock mass energy response state of the nodes in the dynamic pulse diagram within the time slice, used for historical pulse transmission and Delay-DSGN node state updates.

[0104] When forming a node pulse state with the response occurrence time, the event arrival time of the microseismic event is recorded. Corresponding node pulse state As the object being updated, it will be later than the event arrival time. The acoustic emission response, stress response, and displacement response are sequentially written into the nodal pulse state according to the time of response occurrence. The direction of the change in location of the microseismic event is denoted as... The direction of stress response is denoted as The direction of the displacement response is denoted as , , and Both represent unit direction vectors obtained from a preset direction mapping table within the same engineering coordinate system. When the location change field of a microseismic event is a direction-coded value, it is obtained from the preset direction mapping table. When the location change field of a microseismic event is a location increment, it is matched against a preset direction mapping table based on the location increment. .Will and Direction angle, and The directional angles are respectively related to the preset threshold. Comparison, This indicates the allowable angle for directional consistency. The included directional angle must not exceed a preset threshold. At that time, retain the corresponding response pair node pulse state. Update; the directional angle exceeds the preset threshold. When this occurs, the corresponding direction field is not updated, and the direction field before the update or the preset default value is retained. The time of the response is marked. With the updated node pulse state Binding creates node pulse states with the timing of the response.

[0105] Adjacent propagation relationships are established based on the monitoring point locations corresponding to the nodes in the dynamic pulse diagram and the node pulse states with response times. Adjacent propagation relationships refer to the propagation relationships between nodes in the dynamic pulse diagram that define the transmission of historical pulses from the source rockburst incubation monitoring unit to the target rockburst incubation monitoring unit. These adjacent propagation relationships are denoted as... , Indicates the first Within a given time slice, an ordered set of node relationships allows the transmission of historical pulses from the source rockburst incubation monitoring unit to the target rockburst incubation monitoring unit. This includes the nodes of the dynamic pulse diagram. and nodes of dynamic pulse diagram , When node Corresponding preset rockburst incubation monitoring unit spatial boundaries and nodes The corresponding preset rockburst incubation monitoring unit spatial boundaries have a common boundary, or the minimum Euclidean distance between the spatial boundaries of two preset rockburst incubation monitoring units does not exceed the preset spatial adjacency threshold. And nodes and nodes In the When there are node pulse states with response occurrence times in each time slice, in Add ordered node pairs Preset spatial adjacency threshold Used to define the spatial range of adjacent rockburst incubation monitoring units, the adjacent propagation relationship does not calculate the energy migration reach delay time window.

[0106] The dynamic pulse diagram is represented as The dynamic pulse diagram of a single time slice is represented as follows: , Indicates the first The dynamic pulse diagram is a set of node pulse states for all nodes within a given time slice. Each node in the dynamic pulse diagram represents a rockburst incubation monitoring unit, and the node pulse states indicate the rock mass energy response state of that monitoring unit.

[0107] Figure 2This illustration demonstrates the process of generating a dynamic pulse map from rockburst incubation monitoring data in this embodiment. The left side shows the input preprocessing section, which uses microseismic events, acoustic emission responses, stress responses, displacement responses, and monitoring point locations as multi-source inputs, and performs time matching according to the acquisition time. Subsequently, it assigns units based on monitoring point locations and divides time according to preset time slices, forming time-series data of monitoring units arranged according to rockburst incubation monitoring units and time slices. The middle section displays the twelve-dimensional node pulse state formed by each monitoring unit within each time slice. The twelve state fields include the number of events, energy release, event arrival time, event location change, acoustic emission count, acoustic emission energy, stress change, stress response direction, displacement change, displacement response direction, horizontal position of the monitoring point, and depth position of the monitoring point. The right side shows the construction of the dynamic pulse map, where nodes represent rockburst incubation monitoring units, edges represent adjacent propagation relationships, and node pulse states represent the rock mass energy response state. This figure illustrates that this embodiment can unify multi-source heterogeneous monitoring information into node inputs that can be received by a dynamic graph model, while preserving the time slice order, response sequence, and spatial unit relationship, so that subsequent Delay-DSGN node status updates have a clear data foundation and spatial propagation foundation.

[0108] Step 3: Based on the monitoring point locations, microseismic event migration relationships, stress response directions, and displacement response directions of the source rockburst incubation monitoring unit and the target rockburst incubation monitoring unit in the dynamic pulse diagram, determine the energy migration reachable delay time window from the source rockburst incubation monitoring unit to the target rockburst incubation monitoring unit. The energy migration reachable delay time window is used to limit the hysteresis time that conforms to the energy migration of the rock mass.

[0109] In this embodiment, the dynamic pulse diagram is represented as follows: The dynamic pulse diagram of a single time slice is represented as follows: , This represents the set of nodes in a dynamic pulse graph. Indicates the first Adjacent propagation relationships within a time slice Indicates the first The set of node pulse states within a time slice Indicates the number of time slices. Node pulse state set. Include , Indicates the number of rockburst incubation monitoring units. Indicates the first The rockburst incubation monitoring unit was in the first Node pulse state within a time slice. This is a twelve-dimensional vector, with twelve fields representing the number of events, energy release, event arrival time, event location change, acoustic emission count, acoustic emission energy, stress change, stress response direction, displacement change, displacement response direction, horizontal position of the monitoring point, and depth position of the monitoring point. The horizontal position and depth position of the monitoring point together constitute the monitoring point location. .

[0110] Regarding the propagation relationship between adjacent neighbors ordered node pairs in , will node Corresponding rockburst incubation monitoring unit The node was identified as a source rock burst incubation monitoring unit. Corresponding rockburst incubation monitoring unit This unit has been identified as the target rockburst incubation monitoring unit. (Source: Historical time slice) In the set of integers The value is taken from the inner time slice, and the target time slice is... From node pulse state Read the monitoring point location of the source rock burst incubation monitoring unit Time of arrival of microseismic events Event location changes, stress response direction, and displacement response direction; from nodal pulse state Read the monitoring point location of the target rockburst incubation monitoring unit. Time of arrival of microseismic events The event location change, stress response direction, and displacement response direction are considered. When the arrival time or location change of a microseismic event is a preset invalid marker, no effective candidate hysteresis time range is formed from the source rockburst incubation monitoring unit to the target rockburst incubation monitoring unit.

[0111] Based on the location of the monitoring points and monitoring point locations Determine the spatial distance and spatial pointing relationship. The spatial distance is obtained by the Euclidean distance corresponding to the difference in the horizontal position and the difference in the depth position of the monitoring points. The spatial pointing relationship is determined by the location of the monitoring points. Pointing to the location of the monitoring point The unit direction vector representation. Spatial distance and spatial pointing relationships are written into the spatial migration datum. Spatial migration benchmark Used to match the migration relationships of microseismic events.

[0112] The migration relationship of microseismic events is determined based on the microseismic events of the source rockburst incubation monitoring unit and the target rockburst incubation monitoring unit. Earlier At the same time, the sequence of microseismic events corresponds to the migration sequence from the source rockburst incubation monitoring unit to the target rockburst incubation monitoring unit. Mapping the event location changes of the source rockburst incubation monitoring unit and the target rockburst incubation monitoring unit to the same engineering coordinate system forms the direction of the microseismic event location changes. When the event location change field is a direction-encoded value, a unit direction vector is obtained according to a preset direction mapping table; when the event location change field is a location increment, the location increment is converted into a unit direction vector in the same engineering coordinate system. The direction of the location change of the microseismic event is then determined. The direction consistency is compared with the spatial pointing relationship, and the direction consistency value meets the preset threshold. At that time, the direction of the change in location of the microseismic event The direction of energy release migration was determined and written into the space migration reference. To form a microseismic constraint migration benchmark .

[0113] Based on the stress response direction of the source rockburst incubation monitoring unit and the target rockburst incubation monitoring unit, a reference for the migration of stress response direction relative to microseismic constraint is formed. The directional consistency value. The stress response direction of the source rock burst incubation monitoring unit is denoted as... The stress response direction of the target rockburst incubation monitoring unit is denoted as... When the stress response direction field is a direction-encoded value, a unit direction vector is obtained according to a preset direction mapping table; when the stress response direction field is a direction increment, the direction increment is converted into a unit direction vector in the same engineering coordinate system. Compare the dot product of the unit direction vectors with the energy release migration direction, and... The stress response direction is compared with the energy release migration direction using a unit direction vector inner product, and the smaller of the two inner product results is taken as the directional consistency value. The directional consistency value corresponding to the stress response direction is not less than a preset threshold. At that time, the directional consistency value corresponding to the stress response direction is written into the microseismic constraint migration datum. To form a stress-constrained migration benchmark .

[0114] Based on the displacement response direction of the source rockburst incubation monitoring unit and the displacement response direction of the target rockburst incubation monitoring unit, a displacement response direction relative to the stress constraint migration reference is formed. The directional consistency value. The displacement response direction of the source rock burst incubation monitoring unit is denoted as... The displacement response direction of the target rockburst incubation monitoring unit is denoted as When the displacement response direction field is a direction-encoded value, a unit direction vector is obtained according to a preset direction mapping table; when the displacement response direction field is a direction increment, the direction increment is converted into a unit direction vector in the same engineering coordinate system. Compare the dot product of the unit direction vectors with the energy release migration direction, and... The displacement response direction is compared with the energy release migration direction using a unit direction vector inner product, and the smaller of the two inner product results is taken as the directional consistency value. The directional consistency value corresponding to the displacement response direction is not less than a preset threshold. At that time, the directional consistency value corresponding to the displacement response direction is written into the stress constraint migration datum. To form a displacement-constrained migration datum .

[0115] Based on displacement constraint migration datum The spatial distance, energy release and migration direction, and the order of occurrence of microseismic events are used to determine the candidate hysteresis time range. The preset minimum hysteresis coefficient per unit distance is denoted as... The preset maximum hysteresis coefficient per unit distance is denoted as , and The units are time per distance, and the values ​​are determined by historical monitoring calibration values ​​for the same engineering area. The preset time window boundary is denoted as... , This represents the minimum delay time allowed to enter the Gaussian delayed kernel neighborhood aggregation stage. This represents the maximum hysteresis time allowed to enter the Gaussian delayed kernel neighborhood aggregation stage. The preset upper boundary contraction coefficient is denoted as... , It is a dimensionless value, and .

[0116] Define source rockburst incubation monitoring unit In the source historical time slice To the target rockburst incubation monitoring unit In the target time slice spatial distance for: Define directional consistency measure for:

[0117] ;

[0118] The energy migration can reach the delayed time window. lower boundary and upper boundary They are respectively:

[0119] ;

[0120] ;

[0121] The left endpoint of the formula is determined by the spatial distance, the preset minimum hysteresis coefficient per unit distance, and the preset time window boundary. These factors are collectively defined to form the lower boundary of the candidate hysteresis time range and to perform lower boundary comparisons. The right endpoint in the formula is determined by the spatial distance, the preset maximum hysteresis coefficient per unit distance, the preset upper boundary contraction coefficient, the directional consistency value of the stress response direction, the directional consistency value of the displacement response direction, and the preset time window boundary. These parameters are jointly defined to form the upper boundary of the candidate hysteresis time range and to perform upper boundary comparisons. The directional consistency values ​​corresponding to both the stress response direction and the displacement response direction are not less than a preset threshold. When the upper boundary of the candidate hysteresis time range is retained, the directional consistency value corresponding to the stress response direction or the directional consistency value corresponding to the displacement response direction is less than a preset threshold. At that time, a preset upper boundary shrinkage coefficient is used. Shrink the upper boundary of the candidate hysteresis time range.

[0122] When the temporal migration condition, spatial migration condition, directional consistency condition, and boundary validity condition are simultaneously satisfied... ;otherwise :

[0123] The timing transition conditions are as follows: ;

[0124] The spatial migration conditions are as follows: ,and ;

[0125] The condition for directional consistency is: ;

[0126] The boundary validity condition is: ;

[0127] in, Indicates the source rockburst incubation monitoring unit In the source historical time slice To the target rockburst incubation monitoring unit In the target time slice Energy migration can reach a time delay window; Index of source rockburst incubation monitoring unit; Indicates the index of the target rockburst incubation monitoring unit; Indicates the source historical time slice index; Indicates the target time slice index; This indicates that the source rockburst incubation monitoring unit is located in the source history time slice. Location of monitoring points; This indicates that the target rockburst incubation monitoring unit is in the target time segment. Location of monitoring points; Denotes the Euclidean norm; Indicates the arrival time of microseismic events in the source rock burst incubation monitoring unit; Indicates the arrival time of microseismic events for the target rockburst incubation monitoring unit; Indicates the direction of location change of microseismic events; Indicates the stress response direction of the rockburst incubation monitoring unit; Indicates the stress response direction of the target rockburst incubation monitoring unit; Indicates the direction of displacement response of the source rockburst incubation monitoring unit; Indicates the direction of displacement response of the target rockburst incubation monitoring unit; This represents the preset threshold corresponding to the directional consistency value. The value is dimensionless. This represents the minimum delay time allowed to enter the Gaussian delayed kernel neighborhood aggregation stage; This represents the maximum delay time allowed to enter the Gaussian delayed kernel neighborhood aggregation stage; This represents the minimum hysteresis coefficient per unit distance. This indicates the maximum hysteresis coefficient per unit distance. This indicates the preset upper boundary contraction coefficient; This indicates an indicator function; it takes the value 1 if the condition inside the parentheses is true, and takes the value 0 if the condition is false. Represents the maximum value function; Describes the minimum value function; This indicates that no effective energy migration can be achieved within a certain time window.

[0128] Energy transfer can be delayed within a time window. The ordered node relationship between the source rockburst incubation monitoring unit and the target rockburst incubation monitoring unit is written into the data. The energy migration reachability delay time window is a one-dimensional time range quantity used to limit the historical pulse time interval that simultaneously satisfies the constraints of monitoring point location, microseismic event migration relationship, stress response direction, and displacement response direction, and serves as the delay constraint input for the reachability time window limiting delay kernel in subsequent steps.

[0129] Step 4: Modify the Gaussian delay kernel neighborhood aggregation link of Delay-DSGN according to the energy migration reachable delay time window to form a reachable time window limited delay kernel; the reachable time window limited delay kernel determines the aggregation contribution of the historical pulse of the source rockburst incubation monitoring unit to the current state of the target rockburst incubation monitoring unit according to the relationship between the historical pulse time interval and the energy migration reachable delay time window.

[0130] In this embodiment, the Gaussian delay kernel neighborhood aggregation stage utilizes adjacent propagation relationships, node pulse state encoding formed by the pulse coding layer, historical pulse time intervals, and energy migration reachability delay time windows. The node pulse states in the dynamic pulse graph are also considered. Converted into node pulse state codes via pulse coding layer , This indicates the index of the rockburst incubation monitoring unit. Indicates the time slice index. Indicates rockburst incubation monitoring unit In the The node pulse state encoding participates in the Gaussian delay kernel neighborhood aggregation process within each time slice. Each dimension of the node pulse state encoding corresponds to the output of a neuron in the pulse coding layer. The pulse coding layer consists of thirty-two neurons, each fully connected to a twelve-dimensional node pulse state. Each neuron contains weights, biases, and correction linear units.

[0131] Target rockburst incubation monitoring unit In the target time slice The current state will point to the target rockburst incubation monitoring unit in the adjacent propagation relationship. The source rock burst incubation monitoring unit is denoted as Source historical timeline In the set of integers Internal value retrieval. Historical pulses of the source rockburst incubation monitoring unit are used. It indicates that the target rockburst incubation monitoring unit is currently in use. This indicates that when a node pulse state is bound to multiple response occurrence times, the last response occurrence time that completes a valid write to the node pulse state is used as the corresponding node pulse state time; when the node pulse state only contains microseismic events, the arrival time of the microseismic event is used as the corresponding node pulse state time. The historical pulse occurrence times of the source rock burst incubation monitoring unit are... This indicates that the current state of the target rockburst incubation monitoring unit corresponds to the time specified in the text. The historical pulse time interval is denoted as... ,Depend on and The time difference is determined, that is .

[0132] The energy migration reach delay time window is denoted as When the energy migration reachability delay window is valid, the energy migration reachability delay window is written as... , This indicates that energy migration can reach the lower boundary of the delay time window. This represents the upper boundary of the energy migration reachability delay window. When the energy migration reachability delay window is an empty set, the valid flag for the energy migration reachability delay window is set to invalid, and... and Recorded as satisfied Invalid boundaries.

[0133] The relationship between the historical pulse time interval and the energy migration reachability delay window is represented by the validity of the delay response calculation, which is determined according to the following formula:

[0134] ;

[0135] in, Indicates the source rockburst incubation monitoring unit In the source historical time slice Historical pulses act on the target rockburst incubation monitoring unit In the target time slice The effectiveness of delay response calculation The value is dimensionless. The polymerization inhibition coefficient is a dimensionless value that satisfies the following conditions: ; This indicates an indicator function; it takes the value 1 if the condition inside the parentheses is true, and takes the value 0 if the condition is false. Numerical labels representing adjacent propagation relationships, when nodes To the node Belongs to the The adjacent propagation relationship within a target time slice ,otherwise ; Indicates the source rockburst incubation monitoring unit The corresponding nodes in the dynamic pulse diagram; Indicates the target rockburst incubation monitoring unit The corresponding nodes in the dynamic pulse diagram; This indicates that the energy migration reachability delay window is valid. hour ,otherwise .

[0136] when When the historical pulse time interval meets the energy migration reachable delay time window, the historical pulse of the source rock burst incubation monitoring unit is considered valid; when When the time interval between historical pulses does not meet the energy migration reachable delay time window, a corresponding aggregation suppression mark for the historical pulse is formed.

[0137] The historical pulses of the identified valid source rockburst inoculation monitoring units are grouped into a set, which serves as the effective calculation range of the delay kernel. The effective calculation range of the delay kernel is determined according to the following formula:

[0138] ;

[0139] in, Indicates the target rockburst incubation monitoring unit In the target time slice The corresponding effective computational range of the delayed kernel; Indicates the number of rockburst incubation monitoring units; Indicates the source rockburst incubation monitoring unit In the source historical time slice The candidate historical pulse. When the candidate historical pulse corresponds to Within the effective computational range of the delayed kernel At that time, the historical pulses of the source rock burst incubation monitoring unit retain the qualification for aggregation; when the candidate historical pulse corresponds to Not within the effective computational range of delayed kernels At that time, the historical pulses of the source rock burst incubation monitoring unit do not retain the qualification for aggregation.

[0140] The Gaussian delayed nucleus neighborhood aggregation stage consists of sixty-four aggregation neurons, with the aggregation neuron index denoted as . , Each aggregated neuron receives historical pulses from thirty-two-dimensional source rockburst gestation monitoring units. Current status of the 32-dimensional target rockburst incubation monitoring unit and one-dimensional historical pulse time interval Each aggregation neuron contains adjacent aggregation weights, delayed kernel weights, bias, and hyperbolic tangent nonlinear units. Each aggregation neuron is configured with a delayed nucleus center. and delayed kernel width , For time quantity, For time quantity and The original Gaussian delay response of the delay kernel weights is formed according to the following rules: calculate the historical pulse time intervals. With delayed nuclear center The time difference is squared and then the delay kernel width is applied. Normalize the square of , and take the negative exponent of the normalization result to obtain the th . The original Gaussian delayed response of a merging neuron.

[0141] Based on the effective computational range of the delay kernel, the energy migration reachable delay time window is embedded in the delay kernel neighborhood aggregation step of the Gaussian delay kernel weight calculation process. The candidate historical pulses are selected, preserving the complete delay kernel weights of the historical pulses within the Gaussian delay kernel. For The candidate historical pulses are multiplied by the aggregation suppression coefficient of the original Gaussian delay response according to the aggregation suppression label. The delay kernel weights corresponding to historical pulses are reduced. After energy transfer and reaching the delay time window limit, the delay kernel weights form the reachable time window limited delay kernel, and the action boundary of the Gaussian delay kernel neighborhood aggregation step is updated.

[0142] Based on the reachable time window, a delay kernel is defined, and the delay response is calculated for the historical pulse time intervals in the delay kernel input. Each aggregated neuron will transmit the historical pulses of the source rockburst gestation monitoring unit. Current status of the target rockburst incubation monitoring unit Historical pulse time interval Validity of delayed response calculation Adjacency aggregation weights, delay kernel weights, and bias input hyperbolic tangent nonlinear units form the first... Sixty-four aggregate neurons form sixty-four-dimensional restricted delay response vectors. The first of the sixty-four-dimensional restricted delay response vectors is... Dimension corresponding to the first The constrained delayed response generated by the aggregated neurons.

[0143] For the same target rockburst incubation monitoring unit and the same target time slice The constrained delayed responses from different source rockburst incubation monitoring units and different source historical time slices are aggregated in a neighborhood according to their adjacent propagation relationships to form a 64-dimensional aggregated contribution. Aggregated contributions The Veyddi The aggregated response is obtained by aggregating the constrained delay responses of each aggregated neuron. The aggregated contribution represents the neighborhood aggregated influence of the historical pulses of the source rockburst incubation monitoring unit on the current state of the target rockburst incubation monitoring unit after being calculated by the reachable time window-constrained delay kernel.

[0144] Step 5: Update the Delay-DSGN node state based on the dynamic pulse diagram, the reachable time window-limited delay kernel, and the aggregate contribution to obtain the propagation result of the rockburst incubation state.

[0145] In this embodiment, the Delay-DSGN node state update includes a Gaussian delay kernel neighborhood aggregation stage and a node state update stage, and a pulse coding stage is included before the Gaussian delay kernel neighborhood aggregation stage. The dynamic pulse diagram is represented as follows: The dynamic pulse diagram of a single time slice is represented as follows: , This represents the set of nodes in a dynamic pulse graph. Indicates the first Adjacent propagation relationships within a time slice Indicates the first The set of node pulse states within a time slice. Node pulse states It is a twelve-dimensional vector, with twelve fields in the following order: number of events, energy release, event arrival time, event location change, acoustic emission count, acoustic emission energy, stress change, stress response direction, displacement change, displacement response direction, horizontal position of monitoring point, and depth position of monitoring point.

[0146] The pulse coding layer receives the node pulse states of each node in the dynamic pulse graph. Convert node pulse states into node pulse state codes. , This indicates the index of the rockburst incubation monitoring unit. Indicates the time slice index. Indicates rockburst incubation monitoring unit In the Node pulse state encoding for participating in Delay-DSGN node state updates within a time slice. The vector is a 32-dimensional vector, with each dimension corresponding to the output of a neuron in the pulse coding layer. The pulse coding layer consists of 32 neurons, each fully connected to a 12-dimensional node pulse state. Each neuron contains weights, biases, and corrected linear units. The pulse coding layer converts the 12-dimensional node pulse states into 32-dimensional node pulse state codes, which are used as inputs for the subsequent Gaussian delay kernel neighborhood aggregation step.

[0147] The Gaussian delay kernel neighborhood aggregation stage receives historical pulses from the source rockburst incubation monitoring unit, the current state of the target rockburst incubation monitoring unit, the historical pulse time interval, and the reachable time window to define the delay kernel. This is specifically designed for the target rockburst incubation monitoring unit. In the target time slice The current state will point to the target rockburst incubation monitoring unit in the adjacent propagation relationship. The source rock burst incubation monitoring unit is denoted as Source historical timeline In the set of integers Internal value retrieval. Historical pulses of the source rockburst incubation monitoring unit are used. It indicates that the target rockburst incubation monitoring unit is currently in use. This indicates that when a node pulse state is bound to multiple response occurrence times, the last response occurrence time after which the node pulse state is validly written is used as the corresponding node pulse state time; when the node pulse state only contains microseismic events, the arrival time of the microseismic events is used as the corresponding node pulse state time. The historical pulse occurrence times of the source rock burst incubation monitoring unit are... This indicates that the current state of the target rockburst incubation monitoring unit corresponds to the time specified in the text. The historical pulse time interval is denoted as... ,Depend on and The time difference is determined, that is .

[0148] In the Gaussian delay kernel neighborhood aggregation step, multiple aggregation neurons determine the delay kernel weights based on historical pulse time intervals and reachable time windows, and generate constrained delay responses based on the historical pulses of the source rockburst incubation monitoring unit, the current state of the target rockburst incubation monitoring unit, and the delay kernel weights. The Gaussian delay kernel neighborhood aggregation step comprises sixty-four aggregation neurons, with the aggregation neuron index denoted as... , Each aggregated neuron receives historical pulses from thirty-two-dimensional source rockburst gestation monitoring units. Current status of the 32-dimensional target rockburst incubation monitoring unit and one-dimensional historical pulse time interval Each aggregation neuron contains adjacent aggregation weights, delayed kernel weights, bias, and hyperbolic tangent nonlinear units. Each aggregation neuron is configured in the center of the delay nucleus. and delayed kernel width , For time quantity, For time quantity and .

[0149] The original Gaussian delay response of the delayed kernel weights is formed according to the following rules: calculate the historical pulse time intervals. With delayed nuclear center The time difference is squared and then the delay kernel width is applied. Normalize the square of , and take the negative exponent of the normalization result to obtain the th . The original Gaussian delay response of a clustered neuron. The reachable time window limits the energy transfer of the delay kernel. The reachable delay time window is applied to the delay kernel weight calculation process, ensuring that the energy falls within the effective computational range of the delay kernel. The historical impulses are preserved to maintain the validity of the delay response calculation, ensuring that those not within the effective calculation range of the delay kernel are included. The historical impulses are reduced in efficiency for delayed response calculation by using aggregation inhibition markers. Sixty-four aggregation neurons form sixty-four-dimensional restricted delayed response vectors, and the first of these vectors... Dimension corresponding to the first The constrained delayed response generated by the aggregated neurons.

[0150] The aggregated contribution is obtained by performing neighborhood aggregation on all constrained delay responses of the same target rockburst incubation monitoring unit and the same target time slice. For the same target rockburst incubation monitoring unit... and the same target time slice The constrained delayed responses from different source rockburst incubation monitoring units and different source historical time slices are aggregated in a neighborhood according to their adjacent propagation relationships to form a 64-dimensional aggregated contribution. Aggregated contributions The Dimension indicates the first The neighborhood aggregation effect corresponding to each aggregation neuron When the target rockburst incubation monitoring unit does not have a source history time slice that satisfies the energy migration reachability delay window within the target time slice, the contribution will be aggregated. Set it to a 64-dimensional zero vector.

[0151] The node status update process receives the aggregated contribution and the current status of the target rockburst incubation monitoring unit, and generates the updated node status of the target rockburst incubation monitoring unit based on the aggregated contribution and the current status of the target rockburst incubation monitoring unit. The current status of the target rockburst incubation monitoring unit is denoted as... This represents the state of a 64-dimensional node. When the first time slice has no source historical time slice, the second time slice will be... Node pulse state encoding for each time slice The first dynamic graph propagation layer is input, and the first dynamic graph propagation layer outputs the 64-dimensional node states, which are then used as... .right The time slice will be the first The updated node status formed by each time slice Write to the The current state cache of the target rockburst incubation monitoring unit for each time slice is used as the 64-dimensional node state in the current state cache of the target rockburst incubation monitoring unit. Current pulse code of the target rockburst incubation monitoring unit. Participation in the reachable time window-limited delayed nuclear aggregation, current status of the target rockburst incubation monitoring unit. Participate in the state update of the node state update layer.

[0152] The node state update layer receives 64-dimensional aggregated contributions. Current status of the 64-dimensional target rockburst incubation monitoring unit The node state update layer consists of sixty-four update neurons, each of which is connected to... The sixty-four components and The system consists of 64 fully connected components, with each update neuron containing state update weights, biases, and corrected linear units. The outputs of these 64 update neurons represent the update node states of the target rockburst gestation monitoring unit. , For a 64-dimensional vector, the 64-dimensional vector's 64th dimension... Dimension corresponding to the first The state propagation component of each updated neuron output .

[0153] Following the time-slice sequence of the dynamic pulse diagram, the updated node state of the current time slice is used as the current state of the target rockburst incubation monitoring unit in the next time slice. The Gaussian delay kernel neighborhood aggregation and node state update processes continue until all time slices are recursively processed, yielding the rockburst incubation state propagation result. Specifically, the... The aggregate contribution of each time slice is set as a 64-dimensional zero vector and generated through the node state update layer. .right Time slices, As the first The state input for reaching time window-limited delay kernel aggregation is performed for each time slice. Each time slice generates an aggregation contribution based on the node pulse state in the dynamic pulse graph, adjacent propagation relationships, historical pulse time intervals, and the reaching time window-limited delay kernel, and generates the updated node state for the corresponding time slice. Time slice recursion is performed on all rockburst incubation monitoring units and all time slices to obtain the 64-dimensional state propagation vector of each rockburst incubation monitoring unit in each time slice. This is a result of the propagation of the rockburst gestation state.

[0154] Figure 3This diagram illustrates the core propagation calculation process of the Delay-DSGN model in this embodiment. The left side represents the pulse coding stage, where the twelve-dimensional node pulse states are converted into thirty-two-dimensional node pulse state codes via a pulse coding layer, serving as input for the subsequent Gaussian delay kernel neighborhood aggregation stage. The middle section represents the Gaussian delay kernel neighborhood aggregation stage, where historical pulses from the source rockburst gestation monitoring unit, the current state of the target rockburst gestation monitoring unit, the historical pulse time interval, and the energy migration reachability delay time window are all entered into the time window determination. Historical pulse time intervals that fall within the energy migration reachability delay time window retain aggregation eligibility; otherwise, an aggregation suppression marker is formed, thus creating a reachability time window-limited delay kernel. Subsequently, sixty-four aggregation neurons aggregate the retained or suppressed historical pulses to obtain a sixty-four-dimensional aggregation contribution. The right side represents the node state update stage, where the sixty-four-dimensional aggregation contribution and the sixty-four-dimensional target rockburst gestation monitoring unit's current state are input into the node state update layer to form a sixty-four-dimensional updated state, which is then recursively used along the time slices to obtain the rockburst gestation state propagation result. The figure clearly illustrates that in this embodiment, the energy migration reachability delay time window is embedded into a Gaussian delay kernel, so that the propagation calculation conforms to the hysteresis characteristics of rock mass energy migration.

[0155] Step 6: Determine the rockburst incubation stage and rockburst risk status based on the propagation results of the rockburst incubation state, and output the rockburst incubation analysis results.

[0156] In this embodiment, the propagation result of the rockburst incubation state is represented by a 64-dimensional state propagation vector formed by each rockburst incubation monitoring unit at each time slice. The index of the rockburst incubation monitoring unit is denoted as... , , Indicates the number of rockburst incubation monitoring units; the time slice index is denoted as , , Indicates the number of time slices. Rockburst incubation monitoring unit. In the The state propagation vector formed after each time slice undergoes state updates by the Delay-DSGN node is denoted as . , , Indicates the first The state propagation component of each updated neuron output The state propagation component already includes the aggregate contribution of the current state of the target rockburst incubation monitoring unit and the formation of the reachable time window-limited delay core.

[0157] In practical applications, firstly, based on the propagation results of rockburst incubation state, the propagation state of each rockburst incubation monitoring unit within a time slice is aggregated to form a propagation characterization of the rockburst incubation monitoring unit. For each rockburst incubation monitoring unit... and each time slice ,Will The sixty-four state propagation components are aggregated at the component level according to the updated neuron index to form the propagation characterization of the rockburst gestation monitoring unit. Propagation characterization of rockburst incubation monitoring unit It is a 64-dimensional vector. ,in correspond When multiple source rockburst incubation monitoring units have historical pulses affecting the same rockburst incubation monitoring unit within the same time slice, the influence of multiple historical pulses has already contributed through aggregation. Write Propagation characterization of rockburst incubation monitoring unit use The sixty-four state propagation components are formed, and the historical pulses of the source rock burst incubation monitoring unit are not repeatedly aggregated again.

[0158] Propagation characterization of the output layer receiving rockburst incubation monitoring unit The output layer is a fully connected layer with a 64-dimensional input and a 5-dimensional output. The output layer consists of five output neurons, each connected to... The system is fully connected to 64 components, with each output neuron containing output layer weights and biases. The output layer weights and biases are fixed model parameters of Delay-DSGN. Each output neuron performs a weighted sum of the 64 components and adds the output layer bias to obtain one output component. The system consists of five output components:

[0159] ;

[0160] in, This indicates the output component during the low-energy response incubation phase. This indicates the amount of energy output during the enhanced gestation stage. The output component represents the low-risk state. This represents the output component indicating a medium-risk state. This represents the output component indicating a high-risk state.

[0161] Then, the rockburst incubation stage determination result is formed based on the propagation characterization of the rockburst incubation monitoring unit. The difference between the output component of the energy-enhanced incubation stage and the output component of the low-energy response incubation stage is taken as the stage response value. ,Right now:

[0162] ;

[0163] Stage response value This is a dimensionless numerical value. The preset stage threshold is denoted as... This is a dimensionless threshold value, determined by calibration samples from the rockburst incubation stage or by engineering early warning level configuration. The stage response value will be... With preset stage threshold When comparing, When the energy enhancement gestation stage is determined, The low-energy response incubation stage is determined by the time of formation. The rockburst incubation stage determination result is denoted as... , This indicates the result of the energy enhancement gestation stage assessment. This indicates the result of determining the incubation stage of a low-energy response.

[0164] Next, based on the propagation characterization of the rockburst incubation monitoring unit and the results of the rockburst incubation stage determination, the risk response value is calculated. The low-risk state output component is then... Risk weights are set as follows Output components of medium-risk status Risk weights are set as follows Output components of high-risk states Risk weights are set as follows The three rockburst risk state output components are weighted and summarized according to their corresponding risk weights. When the rockburst incubation stage determination result... (i.e., the energy enhancement and incubation phase) will have a pre-set risk bias. Add to weighted summary results; when determining the rockburst incubation stage (i.e., the low-energy response incubation phase) does not incorporate the preset phase risk bias. The risk response value is calculated using the following formula:

[0165] ;

[0166] in, Indicates the risk response value; This represents the preset stage risk bias, which is a dimensionless value and is determined by the engineering early warning level configuration. This indicates an indicator function; the value is 1 if the condition within the parentheses is true, and 0 if the condition is false.

[0167] The preset risk threshold is denoted as , This indicates the boundary between low-risk and medium-risk states. This indicates the boundary between medium-risk and high-risk states. , and All are dimensionless thresholds. The risk response value... With preset risk threshold Compare and generate rockburst risk status assessment results: when When a low-risk status determination result is formed, When a medium-risk status determination result is formed, A high-risk condition determination result is generated at that time. The rockburst risk condition determination result is denoted as... .

[0168] Finally, the rockburst incubation stage is determined based on the rockburst incubation stage assessment results, and the rockburst risk status is determined based on the rockburst risk status assessment results. The rockburst incubation stage and rockburst risk status are then output as rockburst incubation analysis results according to the correspondence between rockburst incubation monitoring units and time slices. (The analysis is then performed according to the rockburst incubation monitoring unit index.) and time slice index The results of the rockburst incubation stage determination Rockburst risk status assessment results Binding, generating rockburst incubation analysis results .right and Generate rockburst incubation analysis results one by one The analysis results for all monitoring units and all time slices are output.

[0169] Figure 4 This embodiment illustrates the output process for stage and risk determination based on the propagation results of rockburst incubation state. The left input is a 64-dimensional state propagation vector formed by each rockburst incubation monitoring unit at each time slice. This vector serves as the propagation representation of the rockburst incubation monitoring unit and enters the output layer. The middle layer is a fully connected output layer with a 64-dimensional input and a 5-dimensional output. The five output components correspond to the low-energy response incubation stage, the energy-enhancing incubation stage, the low-risk state, the medium-risk state, and the high-risk state, respectively. The five-dimensional output components are then split into stage output components and risk output components. The stage response part uses the difference between the energy-enhancing incubation stage output component and the low-energy response incubation stage output component to form a stage response value, which is compared with a preset stage threshold to obtain the rockburst incubation stage determination result. The risk response part uses the low-risk, medium-risk, and high-risk state output components, combined with the rockburst incubation stage determination result, to form a risk response value, which is then compared with a preset risk threshold to obtain the rockburst risk state determination result. This figure illustrates that this embodiment outputs the incubation stage determination and risk status determination in a hierarchical manner. It can reflect the stage changes in the rockburst incubation process and also provide low-risk, medium-risk, and high-risk status results. The output results correspond to the monitoring unit and time slice, which facilitates the formation of rockburst incubation analysis results.

[0170] Figure 5This diagram compares the performance of the improved Delay-DSGN (using this embodiment) with the original Delay-DSGN in terms of rockburst risk identification timing. The horizontal axis represents time slices in hours, and the vertical axis represents the rockburst risk response value, ranging from 0 to 1. The diagram sets low-risk, medium-risk, and high-risk zones, with a low-risk threshold of 0.00, a medium-risk threshold of 0.35, and a high-risk threshold of 0.65. The actual rockburst or intense energy release time is marked as 30 hours. The solid line represents the risk response result of the improved Delay-DSGN, i.e., after introducing the energy migration reachability delay constraint; the dashed line represents the result of the original Delay-DSGN, i.e., without the energy migration reachability delay constraint. It can be seen that the solid line gradually rises before the actual rockburst or intense energy release time, entering the medium-to-high-risk zone 1-6 hours before the event and maintaining a relatively stable high-risk response; the dashed line only shows a significant rise after the event and exhibits large fluctuations. The figure shows that this embodiment can limit the effective aggregation range of historical pulses by limiting the energy migration reach delay time window, which can reduce the hysteresis effect that does not conform to the energy migration law of rock mass, make the risk response more continuous and stable, and improve the early identification capability in the rockburst incubation process.

[0171] Based on the above technical solutions, this embodiment also proposes a multi-source information analysis system for rockburst incubation processes based on an improved Delay-DSGN, used to implement the multi-source information analysis method for rockburst incubation processes based on an improved Delay-DSGN as described in the embodiment. Please refer to [link to relevant documentation]. Figure 6 The system includes:

[0172] The data acquisition module is used to acquire rockburst incubation monitoring data and form monitoring unit time series data based on the rockburst incubation monitoring data. The rockburst incubation monitoring data includes microseismic events, acoustic emission response, stress response, displacement response, and monitoring point location.

[0173] The dynamic pulse diagram generation module is used to generate a dynamic pulse diagram based on the time series data of the monitoring unit. The nodes of the dynamic pulse diagram are rockburst incubation monitoring units, and the pulse state of the nodes of the dynamic pulse diagram represents the rock mass energy response state of the rockburst incubation monitoring unit.

[0174] The delay time window determination module is used to determine the energy migration reachable delay time window from the source rockburst incubation monitoring unit to the target rockburst incubation monitoring unit based on the monitoring point positions, microseismic event migration relationships, stress response direction, and displacement response direction of the source rockburst incubation monitoring unit and the target rockburst incubation monitoring unit in the dynamic pulse diagram. The energy migration reachable delay time window is used to limit the hysteresis time that conforms to the rock mass energy migration.

[0175] The delay core modification module is used to modify the Gaussian delay core neighborhood aggregation link of Delay-DSGN according to the energy migration reachable delay time window to form a reachable time window limited delay core; the reachable time window limited delay core determines the aggregation contribution of the historical pulse of the source rockburst incubation monitoring unit to the current state of the target rockburst incubation monitoring unit according to the relationship between the historical pulse time interval and the energy migration reachable delay time window;

[0176] The node state update module is used to update the Delay-DSGN node state based on the dynamic pulse diagram, the reachable time window-limited delay kernel, and the aggregate contribution, so as to obtain the propagation result of the rockburst incubation state.

[0177] The analysis results output module is used to determine the rockburst incubation stage and rockburst risk status based on the propagation results of the rockburst incubation state, and output the rockburst incubation analysis results.

[0178] It is understood that the multi-source information analysis system for rockburst incubation process based on the improved Delay-DSGN described in this embodiment is a system for implementing the multi-source information analysis method for rockburst incubation process based on the improved Delay-DSGN described in the embodiment. As the system disclosed in the embodiment corresponds to the method disclosed in the embodiment, the description is relatively simple. For relevant parts, please refer to the description of the method. It will not be repeated here.

Claims

1. A multi-source information analysis method for rockburst incubation process based on an improved Delay-DSGN, characterized in that, The method includes: Step 1: Obtain rockburst incubation monitoring data, and form monitoring unit time series data based on the rockburst incubation monitoring data. The rockburst incubation monitoring data includes microseismic events, acoustic emission response, stress response, displacement response, and monitoring point location. Step 2: Generate a dynamic pulse diagram based on the time series data of the monitoring unit. The nodes of the dynamic pulse diagram are rockburst incubation monitoring units, and the pulse state of the nodes of the dynamic pulse diagram represents the rock mass energy response state of the rockburst incubation monitoring unit. Step 3: Based on the monitoring point locations, microseismic event migration relationships, stress response directions, and displacement response directions of the source rockburst incubation monitoring unit and the target rockburst incubation monitoring unit in the dynamic pulse diagram, determine the energy migration reachable delay time window from the source rockburst incubation monitoring unit to the target rockburst incubation monitoring unit. The energy migration reachable delay time window is used to limit the hysteresis time that conforms to the rock mass energy migration. Step 4: Modify the Gaussian delay kernel neighborhood aggregation link of Delay-DSGN according to the energy migration reachable delay time window to form a reachable time window limited delay kernel; the reachable time window limited delay kernel determines the aggregation contribution of the historical pulse of the source rockburst incubation monitoring unit to the current state of the target rockburst incubation monitoring unit according to the relationship between the historical pulse time interval and the energy migration reachable delay time window. Step 5: Based on the dynamic pulse diagram, the reachable time window-limited delay kernel, and the aggregated contribution, perform Delay-DSGN node state updates to obtain the propagation results of rockburst incubation state; Step 6: Determine the rockburst incubation stage and rockburst risk status based on the propagation results of the rockburst incubation state, and output the rockburst incubation analysis results.

2. The multi-source information analysis method for rockburst incubation process based on improved Delay-DSGN as described in claim 1, characterized in that, In step 1, time-series data for the monitoring unit is generated based on the rockburst incubation monitoring data, specifically including: The microseismic events, acoustic emission responses, stress responses, displacement responses, and monitoring point locations in the rockburst incubation monitoring data are matched according to the acquisition time to form a multi-source response sequence; Based on the location of the monitoring points, the microseismic events, acoustic emission responses, stress responses, and displacement responses in the multi-source response sequence are assigned to the corresponding rockburst incubation monitoring units to form a monitoring unit response sequence. The monitoring unit response sequence is divided into time slices according to a preset time slice, and the order of responses is retained within the same time slice to form a monitoring unit time slice response sequence. The microseismic events, acoustic emission responses, stress responses, displacement responses, and state fields corresponding to the monitoring point locations are extracted from the time-slice response sequence of the monitoring units and arranged according to the rockburst incubation monitoring units and time slices to form the monitoring unit time-series data.

3. The multi-source information analysis method for rockburst incubation process based on improved Delay-DSGN as described in claim 1, characterized in that, Step 2 involves generating a dynamic pulse diagram based on the time-series data from the monitoring unit, specifically including: Based on the monitoring point locations in the time-series data of the monitoring unit, the monitoring point locations are matched with the spatial boundaries of the preset rockburst incubation monitoring unit to determine the correspondence between the rockburst incubation monitoring unit and the nodes of the dynamic pulse diagram. Based on the correspondence, the micro-seismic events, acoustic emission responses, stress responses, and displacement responses of each node in the dynamic pulse map within the time slice are extracted from the time series data of the monitoring unit to form a node multi-source response sequence; Based on the arrival time of microseismic events in the nodal multi-source response sequence, as well as the acquisition time of acoustic emission response, stress response and displacement response, the response sequence of the nodal multi-source response sequence is calibrated to form a nodal hysteresis response sequence. Based on the microseismic event energy release, acoustic emission response energy, stress response direction, displacement response direction, and monitoring point location in the node hysteresis response sequence, the rock mass energy response state of the rockburst incubation monitoring unit within the time slice is determined. The energy response state of the rock mass is bound to the node of the corresponding dynamic pulse diagram, and the node pulse state is calibrated with the response occurrence time to form a node pulse state with the response occurrence time. Based on the monitoring point location corresponding to the node in the dynamic pulse diagram and the node pulse state with the response occurrence time, an adjacent propagation relationship is established between the nodes in the dynamic pulse diagram to generate a dynamic pulse diagram in which the node is a rockburst incubation monitoring unit and the node pulse state represents the energy response state of the rock mass.

4. The multi-source information analysis method for rockburst incubation process based on improved Delay-DSGN according to claim 3, characterized in that, The method for establishing the adjacent propagation relationship is as follows: If node With nodes The corresponding preset rockburst incubation monitoring unit spatial boundary satisfies the following spatial adjacency conditions: there is a common boundary or the minimum Euclidean distance between two spatial boundaries does not exceed the preset spatial adjacency threshold; and the nodes With nodes If there are node pulse states with response times within the same time slice, then at node... With nodes Establish between nodes Pointing to node The adjacent propagation relationship.

5. The multi-source information analysis method for rockburst incubation process based on improved Delay-DSGN according to claim 1, characterized in that, In step 3, determining the energy migration reachable delay window from the source rockburst incubation monitoring unit to the target rockburst incubation monitoring unit specifically includes: Define source rockburst incubation monitoring unit In the source historical time slice To the target rockburst incubation monitoring unit In the target time slice spatial distance for: Define directional consistency measure for: ； The energy migration can reach the delayed time window. lower boundary and upper boundary They are respectively: ； ； When the temporal migration condition, spatial migration condition, directional consistency condition, and boundary validity condition are simultaneously satisfied... ;otherwise : The timing transition conditions are as follows: ; The spatial migration conditions are as follows: ,and ; The condition for directional consistency is: ; The boundary validity condition is: ; in, Indicates the source rockburst incubation monitoring unit In the source historical time slice To the target rockburst incubation monitoring unit In the target time slice Energy migration can reach a time delay window; Index of source rockburst incubation monitoring unit; Indicates the index of the target rockburst incubation monitoring unit; Indicates the source historical time slice index; Indicates the target time slice index; This indicates that the source rockburst incubation monitoring unit is located in the source history time slice. Location of monitoring points; This indicates that the target rockburst incubation monitoring unit is in the target time segment. Location of monitoring points; Denotes the Euclidean norm; Indicates the arrival time of microseismic events in the source rock burst incubation monitoring unit; Indicates the arrival time of microseismic events for the target rockburst incubation monitoring unit; Indicates the direction of location change of microseismic events; Indicates the stress response direction of the rockburst incubation monitoring unit; Indicates the stress response direction of the target rockburst incubation monitoring unit; Indicates the direction of displacement response of the source rockburst incubation monitoring unit; Indicates the direction of displacement response of the target rockburst incubation monitoring unit; This represents the preset threshold corresponding to the directional consistency value; This represents the minimum delay time allowed to enter the Gaussian delayed kernel neighborhood aggregation stage; This represents the maximum delay time allowed to enter the Gaussian delayed kernel neighborhood aggregation stage; This represents the minimum hysteresis coefficient per unit distance. This indicates the maximum hysteresis coefficient per unit distance. This indicates the preset upper boundary contraction coefficient; This indicates an indicator function; it takes the value 1 if the condition within the parentheses is true, and takes the value 0 if the condition is false. Represents the maximum value function; Describes the minimum value function; This indicates that no effective energy migration can be achieved within a certain time window.

6. The multi-source information analysis method for rockburst incubation process based on improved Delay-DSGN as described in claim 1, characterized in that, In step 4, the relationship between the historical pulse time interval and the energy migration reachability delay window is represented by the validity of the delay response calculation, which is determined according to the following formula: ； in, Indicates the source rockburst incubation monitoring unit In the source historical time slice Historical pulses act on the target rockburst incubation monitoring unit In the target time slice The validity of the delayed response calculation; Index of source rockburst incubation monitoring unit; Indicates the index of the target rockburst incubation monitoring unit; Indicates the source historical time slice index; Indicates the target time slice index; Indicates the polymerization inhibition coefficient. ; This indicates an indicator function; it takes the value 1 if the condition within the parentheses is true, and takes the value 0 if the condition is false. Numerical labels representing adjacent propagation relationships, when nodes To the node In the When there is an adjacent propagation relationship within a target time slice ,otherwise ; Indicates the source rockburst incubation monitoring unit The corresponding nodes in the dynamic pulse diagram; Indicates the target rockburst incubation monitoring unit The corresponding nodes in the dynamic pulse diagram; This indicates that the energy migration reachability delay window is valid. hour ,otherwise ; This indicates that energy migration can reach the lower boundary of the delay time window; This indicates that energy migration can reach the upper boundary of the delay time window; Indicates the historical pulse occurrence time of the source rock burst incubation monitoring unit; This indicates the current state of the target rockburst incubation monitoring unit at the corresponding time. Indicates the historical pulse time interval; when When the historical pulse time interval satisfies the energy migration reachable delay time window, the historical pulse of the source rockburst incubation monitoring unit is considered valid; when When the historical pulse time interval does not meet the energy migration reachable delay time window, an aggregation suppression mark is formed for the corresponding historical pulse.

7. The multi-source information analysis method for rockburst incubation process based on improved Delay-DSGN according to claim 6, characterized in that, In step 4, the method for determining the aggregation contribution includes: The historical pulses of the identified valid source rockburst inoculation monitoring units are grouped into a set, which serves as the effective calculation range of the delay kernel. The effective calculation range of the delay kernel is determined according to the following formula: ； in, Indicates the target rockburst incubation monitoring unit In the target time slice The corresponding effective computational range of the delayed kernel; Indicates the index of the target rockburst incubation monitoring unit; Indicates the target time slice index; Index of source rockburst incubation monitoring unit; Indicates the source historical time slice index; Indicates the number of rockburst incubation monitoring units; Indicates the source rockburst incubation monitoring unit In the source historical time slice Candidate historical pulses; Numerical markers representing adjacent propagation relationships; This indicates the effective marker of the energy migration reach delay time window; Indicates the historical pulse time interval; Indicates the historical pulse occurrence time of the source rock burst incubation monitoring unit; This indicates the current state of the target rockburst incubation monitoring unit at the corresponding time. This indicates that energy migration can reach the lower boundary of the delay time window; This indicates that energy migration can reach the upper boundary of the delay time window; When the candidate historical pulse corresponds to Within the effective computational range of the delayed kernel At that time, the historical pulses of the source rock burst incubation monitoring unit retain the qualification for aggregation; when the candidate historical pulse corresponds to Not within the effective computational range of delayed kernels At that time, the historical pulses of the source rock burst incubation monitoring unit do not retain aggregation eligibility; historical pulses that retain aggregation eligibility participate in neighborhood aggregation; historical pulses that participate in neighborhood aggregation are aggregated by Gaussian delay kernel neighborhood aggregation to obtain the aggregation contribution.

8. The multi-source information analysis method for rockburst incubation process based on improved Delay-DSGN according to claim 1, characterized in that, In step 5, the Delay-DSGN node state update includes a Gaussian delay kernel neighborhood aggregation step and a node state update step, and a pulse coding step is included before the Gaussian delay kernel neighborhood aggregation step; The pulse coding stage receives the node pulse state of each node in the dynamic pulse diagram and converts the node pulse state into a node pulse state code. The Gaussian delay kernel neighborhood aggregation step receives historical pulses from the source rockburst incubation monitoring unit, the current state of the target rockburst incubation monitoring unit, the historical pulse time interval, and the reachable time window-defined delay kernel. In this step, multiple aggregation neurons determine the delay kernel weights based on the historical pulse time interval and the reachable time window-defined delay kernel, and generate constrained delay responses based on the historical pulses from the source rockburst incubation monitoring unit, the current state of the target rockburst incubation monitoring unit, and the delay kernel weights. All constrained delay responses from the same target rockburst incubation monitoring unit and the same target time slice are then aggregated in a neighborhood to obtain the aggregation contribution. The node status update step receives the aggregated contribution and the current status of the target rockburst incubation monitoring unit, and generates the updated node status of the target rockburst incubation monitoring unit based on the aggregated contribution and the current status of the target rockburst incubation monitoring unit. The process is recursively applied along the time slices of the dynamic pulse diagram. The updated node state of the current time slice is used as the current state of the target rockburst incubation monitoring unit in the next time slice. The Gaussian delay kernel neighborhood aggregation and node state update steps are then executed until the recursion of all time slices is completed, and the propagation result of the rockburst incubation state is obtained.

9. The multi-source information analysis method for rockburst incubation process based on improved Delay-DSGN according to claim 1, characterized in that, Step 6 specifically includes: Based on the propagation results of the rockburst incubation state, the propagation state of each rockburst incubation monitoring unit within the time slice is aggregated to form a propagation characterization of the rockburst incubation monitoring unit. Based on the propagation characterization of the rockburst incubation monitoring unit, a rockburst incubation stage determination result is formed when... When the energy enhancement gestation stage is determined, The low-energy response gestation stage determination results are formed at that time; among them, This indicates the index of the rockburst incubation monitoring unit. Indicates the time slice index. Indicates rockburst incubation monitoring unit In time slice The stage response value, that is, the difference between the output component of the energy-enhanced incubation stage and the output component of the low-energy response incubation stage, Indicates the preset stage threshold; Based on the propagation characterization of the rockburst incubation monitoring unit and the results of the rockburst incubation stage determination, the risk response value is calculated: ； in, Indicates the risk response value; The output component represents the low-risk state. The output component represents the medium-risk state; The output component represents a high-risk state; This indicates the pre-set risk bias. This indicates an indicator function; it takes the value 1 if the condition within the parentheses is true, and takes the value 0 if the condition is false. This indicates the result of the rockburst incubation stage determination. This indicates the result of the energy enhancement gestation stage assessment. This indicates the result of determining the incubation stage of a low-energy response; The risk response value With preset risk threshold Comparisons are made to generate a rockburst risk status assessment result. When a low-risk status determination result is formed, When a medium-risk status determination result is formed, A high-risk status determination result is generated at that time; among them, This indicates the boundary between low-risk and medium-risk states. This indicates the boundary between medium-risk and high-risk states. ; The rockburst incubation stage is determined based on the rockburst incubation stage determination result, and the rockburst risk state is determined based on the rockburst risk state determination result. The rockburst incubation stage and rockburst risk state are output as the rockburst incubation analysis result according to the correspondence between the rockburst incubation monitoring unit and the time slice.

10. A multi-source information analysis system for rockburst incubation process based on an improved Delay-DSGN, characterized in that, The system is used to implement the multi-source information analysis method for rockburst incubation process based on the improved Delay-DSGN as described in any one of claims 1 to 9, the system comprising: The data acquisition module is used to acquire rockburst incubation monitoring data and form monitoring unit time series data based on the rockburst incubation monitoring data. The rockburst incubation monitoring data includes microseismic events, acoustic emission response, stress response, displacement response, and monitoring point location. The dynamic pulse diagram generation module is used to generate a dynamic pulse diagram based on the time series data of the monitoring unit. The nodes of the dynamic pulse diagram are rockburst incubation monitoring units, and the pulse state of the nodes of the dynamic pulse diagram represents the rock mass energy response state of the rockburst incubation monitoring unit. The delay time window determination module is used to determine the energy migration reachable delay time window from the source rockburst incubation monitoring unit to the target rockburst incubation monitoring unit based on the monitoring point positions, microseismic event migration relationships, stress response direction, and displacement response direction of the source rockburst incubation monitoring unit and the target rockburst incubation monitoring unit in the dynamic pulse diagram. The energy migration reachable delay time window is used to limit the hysteresis time that conforms to the rock mass energy migration. The delay core modification module is used to modify the Gaussian delay core neighborhood aggregation link of Delay-DSGN according to the energy migration reachable delay time window to form a reachable time window limited delay core; the reachable time window limited delay core determines the aggregation contribution of the historical pulse of the source rockburst incubation monitoring unit to the current state of the target rockburst incubation monitoring unit according to the relationship between the historical pulse time interval and the energy migration reachable delay time window; The node state update module is used to update the Delay-DSGN node state based on the dynamic pulse diagram, the reachable time window-limited delay kernel, and the aggregate contribution, so as to obtain the propagation result of the rockburst incubation state. The analysis results output module is used to determine the rockburst incubation stage and rockburst risk status based on the propagation results of the rockburst incubation state, and output the rockburst incubation analysis results.

Images