A mine safety production risk management system and method

By calculating the regular distance and path search of monitored changes between mining areas, a risk migration chain is formed, which solves the problem of identifying the migration of hazardous factors in mine safety production, ensures that risk management is synchronized with the actual migration of hazardous factors, and improves the accuracy and safety of mine safety production.

CN123117798APending Publication Date: 2026-11-17SHANDONG GOLDSOFT TECH LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611318440.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-28
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In mine safety production, when local risk indicators decline, it is difficult to distinguish whether the dangerous factors have been eliminated or whether they have migrated spatially. This leads to a discrepancy between the risk management results and the actual destination of the dangerous factors, which may result in the premature termination of the risk management status while the dangerous factors are still migrating within the mine.

Method used

By calculating the regular distance of monitoring changes between regions, combined with regional path search and time-lapse data collection, a risk migration chain is formed. Based on the position at the end of the migration chain, the closure of risk events and the status of regional production control are controlled to ensure that the control is lifted only after the dangerous factors have completely migrated out of the mine.

Benefits of technology

This effectively avoids prematurely lifting risk management due to the restoration of normal local monitoring, ensuring that control measures are lifted only after hazardous factors have completely migrated out of the mine. This reduces the omission of migration relationships caused by monitoring a single area, and improves the accuracy and safety of risk management.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123117798A_ABST
    Figure CN123117798A_ABST
Patent Text Reader

Abstract

This invention discloses a mine safety production risk management system and method, specifically relating to the field of mine safety production risk management. The system includes acquiring risk events, corresponding regional monitoring value sequences, and channel connection records, with the corresponding region designated as the source region. The channel connection records indicate the starting and ending regions according to the direction of hazard factor transmission, using the increase / decrease relationship between the monitoring value at the time of risk occurrence and the previous value as the direction of risk change. By tracking the reduction of risk in the source region and the enhancement of risk in subsequent regions as the same migration process, and continuously forming a risk migration chain under edge computing processing, the production control status is only deactivated in reverse order when the end of the migration chain reaches outside the mine. This avoids directly shutting down the risk event after the monitoring value in the source region recovers, thereby alleviating the problem of the management status being prematurely deactivated while the hazard factor is still migrating underground.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of mine safety production risk management technology, and more specifically, to a mine safety production risk management system and method. Background Technology

[0002] Mine safety production risk management usually relies on monitoring data such as gas, wind speed, temperature, smoke and dust, and water inrush to determine the risk status of the work area. After an anomaly occurs, measures are taken to deal with it by adjusting ventilation, changing drainage paths, switching equipment operation status, or restricting operations in the relevant areas. The risk is then determined based on the monitoring results after the measures are taken. Taking the handling of local gas anomalies in underground mines as an example, after adjusting the air doors, fans or air volume distribution to reduce the gas concentration in the abnormal area, the values ​​of the original monitoring points can be restored to the normal range relatively quickly. However, there are connections between mine roadways, and the airflow direction and air volume distribution will change with the ventilation status. The gas in the original area may enter the adjacent roadway or downstream working area with the new airflow path. Similarly, adjustments to drainage paths and flue gas diversion may cause hazardous factors to shift from their original areas to other areas. Due to limitations such as underground space zoning, discrete deployment of monitoring points, and edge computing nodes processing field data by region, the observation of the same hazardous factor in different areas may occur at different times. As a result, the monitoring value in the original area may have decreased, while the migrating area may not yet have monitoring results sufficient to reflect the risk change. If the risk record is closed based solely on local data after the original area has been dealt with, a state fault will be created where the original risk has been eliminated but the actual hazardous factor continues to migrate within the mine space, making the risk management results inconsistent with the actual destination of the hazardous factor. The technical problem this application aims to solve is: how to distinguish between the elimination of hazardous factors and the spatial migration of hazardous factors when local risk indicators decline due to mine safety management, and how to maintain the risk status of hazardous factors until the termination of hazardous factors is confirmed, so as to avoid prematurely lifting risk management due to the restoration of local monitoring to normal. Summary of the Invention

[0003] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a mine safety production risk management system and method. By calculating the regular distance of monitoring changes between regions along the direction of risk factor transmission, and combining regional path search and time-tracking back to repeatedly correct the migration path, a risk migration chain is formed. Based on the position at the end of the migration chain, the risk event is closed and the regional production control status is controlled to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for mine safety production risk management, comprising: S1. Obtain the risk event, the corresponding regional monitoring value sequence, and the channel connection record, and take the corresponding region as the source region; the channel connection record records the starting and ending regions according to the direction of risk factor transmission, and takes the increase or decrease relationship between the monitoring value when the risk occurs and the previous value as the direction of risk change; S2. Calculate the difference between adjacent monitoring values ​​at both ends of each channel; when the risk change direction increases, take the previous value minus the next value at the starting point and the next value minus the previous value at the ending point; when it decreases, take the next value minus the previous value at the starting point and the previous value minus the next value at the ending point; subtract the two sets of results from each other and take the absolute value to form a distance array, take the current value as the cumulative value in the upper left corner, take the current value plus the previous cumulative value in the first row and first column, take the current value plus the cumulative values ​​in the upper left, upper left, and upper left corner of the remaining positions in ascending order, and take the cumulative value in the lower right corner as the normalized distance; S3. Using regions as nodes, channel connection records as directed edges, and regular distances as edge weights, run Dijkstra's algorithm from the source region, take the first node of the cumulative edge weight in ascending order to update adjacent nodes, and take the first path of the cumulative edge weight in ascending order without successor nodes as the region path. S4. Tracing back along the regional path to the cumulative position of the normalized distance, obtaining the collection time of adjacent areas and extracting the monitoring value sequence, recalculating the normalized distance and regional path; if the two paths are consistent, the current path is used as the risk migration chain, and if they are inconsistent, this step is repeated. S5. When the end of the risk migration chain is outside the mine, the regional production control status is deactivated in reverse order and the risk event is closed; when it is inside the mine, the regional production control status in the chain is maintained and the risk management result is output.

[0005] In a preferred embodiment, S1 includes: S1-1. Read the monitoring value corresponding to the time when the risk event occurs in the corresponding area and the monitoring value corresponding to the previous collection time. Subtract the monitoring value corresponding to the previous collection time from the monitoring value corresponding to the time when the risk event occurs to obtain the event difference value. Take the area corresponding to the risk event as the source area. Determine the direction of risk change by the sign of the event difference value. When the event difference value is positive, the corresponding monitoring value increases. When the event difference value is negative, the corresponding monitoring value decreases. Among them, the monitoring values ​​represent the numerical values ​​obtained in each area of ​​the mine at the corresponding collection time, which are used to reflect the status of hazardous factors, including gas concentration, flue gas concentration, dust concentration, temperature, wind speed and water inflow. The same monitoring index is used for each area corresponding to the same risk event, and multiple monitoring values ​​are arranged in order of collection time to form a monitoring value sequence. S1-2. Obtain the monitoring value sequence of each adjacent area that has a channel connection record with the source area. The monitoring value sequence is formed by arranging the monitoring values ​​corresponding to each collection time in the order of collection. Starting from the time of the risk event, subtract the monitoring value corresponding to the previous collection time from the monitoring value corresponding to the next collection time to obtain the monitoring difference. When the risk change direction is that the monitoring value increases, the next collection time corresponding to the first monitoring difference greater than zero is taken as the transmission time. When the risk change direction is that the monitoring value decreases, the next collection time corresponding to the first monitoring difference less than zero is taken as the transmission time. Among them, the channel connection record represents the connectivity data between two adjacent areas of the mine, including the starting area, the ending area and the direction of hazard factor transmission, which is used to determine the regional connection relationship of hazard factor transmission from the starting area to the ending area. S1-3. Taking the source region as the first starting region, read the adjacent regions connected to the first starting region. Use the first starting region as the starting region and each adjacent region as the ending region to form a channel connection record. Then, take each ending region as the new starting region, read the adjacent regions connected to the new starting region that have not yet formed a corresponding channel connection record, and use the new starting region and the adjacent regions as the starting region and the ending region respectively to form a channel connection record. Process step by step according to the method of using the ending region as the next starting region until there are no adjacent regions that have not yet formed a channel connection record, and obtain the channel connection record that connects the source region to the subsequent regions step by step.

[0006] In a preferred embodiment, S2 includes: S2-1. Obtain the starting area monitoring value sequence and the ending area monitoring value sequence for each channel connection record; when the risk change direction is increasing, subtract the next monitoring value from the previous monitoring value in the starting area to obtain the starting change amount, and subtract the previous monitoring value from the next monitoring value in the ending area to obtain the ending change amount; when the risk change direction is decreasing, subtract the previous monitoring value from the next monitoring value in the starting area to obtain the starting change amount, and subtract the next monitoring value from the previous monitoring value in the ending area to obtain the ending change amount, and take the acquisition time of the next monitoring value corresponding to each change amount as the corresponding acquisition time, and arrange them in order of acquisition to form the starting change sequence and the ending change sequence respectively. S2-2. Using the starting change amount in the starting change sequence as rows and the ending change amount in the ending change sequence as columns, the absolute value of the difference between each starting change amount and each ending change amount is taken as the distance value and written into the corresponding row and column intersection position to form a distance array. S2-3. Use the distance value of the top left corner of the distance array as the cumulative value; for the remaining positions in the first row, add the current position distance value to the cumulative value on the left to obtain the cumulative value, and record the left position as the source position; for the remaining positions in the first column, add the current position distance value to the side cumulative value to obtain the cumulative value, and record the top position as the source position; for the remaining positions, sort the left, top, and top left cumulative values ​​in ascending order, add the first cumulative value to the current position distance value to obtain the cumulative value, and record the position of the first cumulative value as the source position. If the values ​​are the same, record the source position in the order of top left, left, and top; use the cumulative value of the bottom right corner as the normalized distance.

[0007] In a preferred embodiment, when the Dijkstra algorithm in S3 is executed, it starts from the node corresponding to the source region, progressively accumulates the normalized distance of each directed edge, retains the lowest accumulated edge weight reaching each node and its corresponding predecessor node, and backtracks along the predecessor node to obtain the region path. S3 includes: S3-1. Take the starting region in each channel connection record as the starting node, the ending region as the ending node, the starting node pointing to the ending node as the directed edge, the normalized distance of the corresponding channel connection record as the edge weight of the directed edge, the node corresponding to the source region as the starting node, and record the cumulative edge weight of the starting node as zero. S3-2. Arrange the nodes that have been written into the cumulative edge weights but have not yet been processed as the current node in ascending order of cumulative edge weights. Take the first node as the current node, add the cumulative edge weight of the current node to the edge weight of each directed edge pointed out from the current node, and use the resulting value as the candidate cumulative edge weight of the corresponding endpoint node.

[0008] In a preferred embodiment, S3 further includes: S3-3. When the endpoint node has not been written with cumulative edge weights, write the candidate cumulative edge weights into the endpoint node and take the current node as the predecessor node of the endpoint node. When the endpoint node has been written with cumulative edge weights, arrange the candidate cumulative edge weights and the written cumulative edge weights in ascending order of value, take the first value as the cumulative edge weight of the endpoint node, and take the current node that generated the first value as the predecessor node of the endpoint node. After all the directed edges corresponding to the current node have been processed, mark the current node as processed, and continue to take the first node in ascending order of cumulative edge weight from the nodes that have been written with cumulative edge weights and have not yet been processed as the next current node, until each node that has been written with cumulative edge weights has completed one directed edge processing. S3-4. Nodes with accumulated edge weights that do not have directed edges starting from themselves are considered as nodes without successors. Arrange these nodes in ascending order of accumulated edge weights. Use the first node without successors as the end node of the path. Backtrack from the end node to the starting node according to the predecessor nodes recorded by each node. Arrange the backtracked nodes in the order from the starting node to the end node of the path. Arrange the regions corresponding to each node in order to obtain the region path.

[0009] In a preferred embodiment, S4 includes: S4-1. Take the preceding region in the sequentially adjacent region path as the starting region and the following region as the ending region. Starting from the cumulative value position at the lower right corner of the distance array between the two regions, trace back along the source position of each cumulative value record to the upper left corner. Pair the collection time of the starting region corresponding to the passed position with the collection time of the ending region to form a time pair sequence. The source position refers to the position of the previous cumulative value used when calculating the cumulative value from the current position of the array. It includes one of the positions to the left, top, and top left of the current position. It is used to trace back from the bottom right cumulative value position to the top left and determine the starting and ending point collection times of each cumulative position. S4-2. Subtract the monitoring value of the previous collection time from the monitoring value of the collection time in each time pair to obtain the starting difference and the ending difference; when the risk change direction is increasing, retain the time pair with the starting difference less than zero and the ending difference greater than zero; when the risk change direction is decreasing, retain the time pair with the starting difference greater than zero and the ending difference less than zero. Form a sequence of retained time pairs according to the collection order.

[0010] In a preferred embodiment, S4 further includes: S4-3. The start time of the first and last positions of the sequence is used as the start and end times of the interception of the start region monitoring value sequence, and the end time of the first and last positions is used as the start and end times of the interception of the end region monitoring value sequence, thus obtaining the start interception sequence and the end interception sequence. S4-4. When the direction of risk change is increasing, the starting point cutoff sequence is obtained by subtracting the previous value from the next value to get the starting point change, and the ending point cutoff sequence is obtained by subtracting the previous value from the next value to get the ending point change; when the direction of risk change is decreasing, the starting point cutoff sequence is obtained by subtracting the previous value from the next value to get the starting point change, and the ending point cutoff sequence is obtained by subtracting the previous value from the next value to get the ending point change. The absolute values ​​of the changes at each starting point and each ending point are subtracted from each other to form a new distance array. The distance value at the top left corner is used as the cumulative value. The first row and the first column are obtained by adding the cumulative values ​​of the left and top sides to the current position distance value and recording the corresponding source position. For the remaining positions, the cumulative value is obtained by adding the lowest value among the cumulative values ​​of the left, top, and top left sides to the current position distance value and recording the corresponding source position. When the values ​​are the same, the source positions are recorded in the order of top left, left, and top. The cumulative value at the bottom right corner is used as the new normalized distance and as the new edge weight of the corresponding directed edge.

[0011] In a preferred embodiment, S4 further includes: S4-5. Take the node corresponding to the source region as the starting node, clear the cumulative edge weights, predecessor nodes and processed marks of all nodes except the starting node, record the cumulative edge weight of the starting node as zero, and record the remaining nodes as unprocessed nodes. Take the node with the lowest cumulative edge weight from the nodes that have been recorded but not yet processed as the current node. Add the cumulative edge weight of the current node to the new edge weight of each directed edge pointed out from the current node to obtain the candidate cumulative edge weight of the corresponding endpoint node. When the endpoint node has not been written with cumulative edge weights, the candidate cumulative edge weight is used as the endpoint node's cumulative edge weight, and the current node is used as the predecessor node. When the endpoint node has been written with cumulative edge weights, the lower of the candidate cumulative edge weight and the original cumulative edge weight is used as the endpoint node's cumulative edge weight, and the node that generated the value is used as the predecessor node. After the directed edge corresponding to the current node is processed, the current node is marked as processed. Then, the node with the lowest cumulative edge weight value among the nodes that have been written with cumulative edge weights but have not been processed is taken as the new current node, until there are no nodes that have been written with cumulative edge weights but have not been processed. Nodes that have no directed edges starting from themselves and whose cumulative edge weights have been written are considered as nodes without successors. The node with the lowest cumulative edge weight is taken as the end node of the path. The path is then traced back to the corresponding node in the source region along the predecessor node. The corresponding regions of each node are arranged in the order from the source region to the end node of the path to obtain the current regional path. S4-6. When the number of regions and their arrangement are the same as the previous regional path, the new distance array of each adjacent region in the current regional path is traced back to obtain the time pair sequence. When the acquisition time of the last endpoint of the previous adjacent region is not later than the acquisition time of the first and second starting points of the next adjacent region, the current regional path is taken as the risk migration chain, and each region in the risk migration chain is set to the production control status. When the time order is reversed, the first acquisition time after the last acquisition time of the previous end is taken as the new starting point for the acquisition of the monitoring value sequence of the next adjacent region, and its new normalized distance and current regional path are recalculated; if the two regional paths are different, the current regional path is taken as the new previous regional path, and the cumulative position backtracking, time pair screening, sequence truncation, normalized distance recalculation and regional path recalculation are performed on the adjacent regions in sequence until the regional arrangement and acquisition time order are consistent.

[0012] In a preferred embodiment, S5 includes: S5-1. Obtain boundary records for the inside and outside of the mine by region number, and compare the end region number of the risk migration chain with the boundary records item by item; when the end region is outside the mine, arrange the mine internal regions in the risk migration chain in the order from the end to the source region to form a release sequence; when the end region is inside the mine, write each region in the risk migration chain into the unwritten sequence. Among them, the boundary record represents the regional location data used to distinguish between the internal area and the external area of ​​the mine, including the area number and the corresponding location category, which is used to match the area number at the end of the risk migration chain with the area number and determine whether the end of the risk migration chain is located inside or outside the mine. S5-2. Read the first region of the release sequence as the current region, change the production control status of the current region from the hold status to the release status and write the release mark; after the release mark is written to the current region, take the next region of the release sequence as the new current region and perform risk status release and release mark writing, until the release mark is written to the source region; The release marker represents the status record generated after the current region completes the release of production control status. It includes the current region number and the release status. It is written by the system after the production control status of the current region changes from the hold status to the release status. It is used to confirm that the current region has completed the risk release and serves as the basis for the next region in the release sequence to perform risk status release. S5-3: When the source region writes the de-marking, the risk event is closed and the de-marked risk management result is output; when an unde-marked sequence is formed, the risk event is kept valid and the production control status of each region in the unde-marked sequence is maintained, and the unde-marked risk management result is output.

[0013] A mine safety production risk management system, the system including a risk acquisition module, a distance normalization module, a path search module, a migration confirmation module, and a risk cancellation module; The risk acquisition module acquires risk events, corresponding regional monitoring value sequences, and channel connection records, using the corresponding regions as the source regions. The channel connection records record the starting and ending regions according to the direction of risk factor transmission, and use the increase or decrease relationship between the monitoring value at the time of risk occurrence and the previous value as the direction of risk change. The distance normalization module calculates the difference between adjacent monitoring values ​​at both ends of each channel. When the risk change direction increases, the starting point is the previous value minus the next value, and the ending point is the next value minus the previous value. When it decreases, the starting point is the next value minus the previous value, and the ending point is the previous value minus the next value. The absolute values ​​of the two sets of results are subtracted from each other to form a distance array. The cumulative value in the upper left corner is the current value. The first row and first column are the current value plus the previous cumulative value. The remaining positions are the current value plus the cumulative values ​​of the left, top, and upper left corners in ascending order. The cumulative value in the lower right corner is used as the normalized distance. The path search module uses regions as nodes, channel connection records as directed edges, and regular distances as edge weights. It runs the Dijkstra algorithm from the source region, takes the first node in ascending order of cumulative edge weights to update adjacent nodes, and takes the first path in ascending order of cumulative edge weights to the region path with no successor nodes. The migration confirmation module traces back along the regional path to the cumulative position of the normalized distance, obtains the collection time of adjacent areas and extracts the monitoring value sequence, and recalculates the normalized distance and regional path; if the two paths are consistent, the current path is used as the risk migration chain, and if they are inconsistent, this step is repeated. The risk write-off module, when the end of the risk migration chain is outside the mine, reverses the regional production control status and closes the risk event; when it is inside the mine, it maintains the regional production control status in the chain and outputs the risk management results by adopting the above technical solution.

[0014] The technical effects and advantages of this invention are as follows: By tracking the reduction of risk in the source area and the enhancement of risk in the subsequent area as the same migration process, and continuously forming a risk migration chain under edge computing processing, the production control status is only lifted in reverse order when the end of the migration chain reaches the outside of the mine. This can avoid directly shutting down the risk event after the monitoring value of the source area recovers, thereby alleviating the problem that the management status has been lifted in advance while the dangerous factors are still migrating underground. By transforming adjacent monitoring values ​​at both ends of the channel according to the direction of risk change, and then comparing the two sets of changes pairwise and accumulating them to obtain the regular distance, a numerical correspondence can be established when the monitoring changes in different areas have a time sequence, reducing the omission of migration relationships caused by comparing only at the same collection time; By using channel connection records as directed edges and regularized distances as edge weights, and employing Dijkstra's algorithm to accumulate the regularized distances of each channel, regional paths extending backward from the source region can be determined in multi-level connected regions, transforming risk propagation judgment from single-region comparison to continuous channel path analysis. By tracing back the source location in the distance array through the regional path, the normalized distance and regional path are recalculated after the corresponding collection period is extracted. This allows the spatial path results to correct the time data range in reverse, and the regional path is corrected again with the recalculated results, reducing the possibility of forming an incorrect migration path in a single calculation. By continuously processing monitoring values ​​and production control status in each area through edge computing, the production control status of the areas inside the mine in the chain is maintained before the risk migration chain reaches the outside of the mine. After reaching the outside of the mine, the control is lifted step by step in reverse order of migration, so that the order of control lifting corresponds to the actual migration order of the risk factors. Attached Figure Description

[0015] Figure 1 This is a flowchart of the present invention.

[0016] Figure 2 This is a system diagram of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention 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. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Refer to the instruction manual appendix Figure 1-2 The present invention provides a method for mine safety production risk management, comprising: S1. Obtain the risk event, the corresponding regional monitoring value sequence, and the channel connection record, and take the corresponding region as the source region; the channel connection record records the starting and ending regions according to the direction of risk factor transmission, and takes the increase or decrease relationship between the monitoring value when the risk occurs and the previous value as the direction of risk change; In this embodiment, S1 is used to determine the source area of ​​risk events, the direction of risk change, and the regional connection relationship that the risk factors can continue to transmit from mine safety monitoring data and channel data; risk events are generated by the mine safety monitoring system when the monitoring indicators reach the corresponding safety management conditions, and the time of occurrence of risk events, corresponding area numbers, and monitoring indicators are recorded; an area represents a mine spatial unit divided according to roadways, working faces, chambers, and monitoring locations and assigned an area number; After a risk event is processed, the monitoring values ​​of the corresponding monitoring indicators in each area are read. The monitoring values ​​are bound to the collection time and area number. Multiple monitoring values ​​in the same area are arranged according to the collection time to form a monitoring value sequence. The same monitoring indicators and units of measurement are used for the same risk event. The channel connection record is read from the existing channel data in the mine. It records the actual transmission direction of the hazard factors in two directly connected areas and at the time of the risk event. Among them, the transmission direction of gas, flue gas and dust is determined according to the airflow direction of the corresponding channel, and the transmission direction of water inrush is determined according to the water flow direction. In S1-1, the starting area and direction for subsequent monitoring value changes are determined; the area number in the risk event record is read, and the area corresponding to the area number is taken as the source area; the monitoring value corresponding to the time of the risk event and the monitoring value corresponding to the previous collection time are read, and the event difference is obtained by subtracting the latter from the former; when the event difference is greater than zero, the increase of the monitoring value is recorded as the direction of risk change. When the event difference is less than zero, the decrease in the monitored value is recorded as the direction of risk change, and the source area and the direction of risk change are written into the risk event record for reading by the monitoring value processing at both ends of the channel. The time of occurrence of the risk event is the time of the first collection of the risk event to avoid repeatedly determining the source area during the continuous risk period. If the monitoring value of the previous collection time is missing, the event difference is not calculated. It is calculated after the monitoring values ​​of the two adjacent collection times are obtained. When the event difference is equal to zero, the direction of risk change is uncertain. The collection position where the risk event first shows a non-zero monitoring change is read for calculation. In S1-2, the time and position of the first occurrence of a risk change in an adjacent area that is in the same direction as the risk change in the source area are recorded; the monitoring value sequence of each adjacent area directly connected to the source area and the adjacent area from the time of the risk event is read; for two monitoring values ​​collected in the same area with adjacent collection order, the subsequent monitoring value is subtracted from the previous monitoring value in turn to obtain the monitoring difference; when the risk change direction is that the monitoring value increases, the monitoring difference is read in the order of collection, and the collection time of the subsequent monitoring value corresponding to the first monitoring difference that is greater than zero is recorded as the transmission time; When the risk change direction is a decrease in the monitored value, the time of collection of the next monitored value corresponding to the first monitoring difference less than zero is recorded as the transmission time; the adjacent area numbers and transmission times are written into the corresponding channel connection records for subsequent judgment of the temporal relationship of the risk factors between areas; if there are less than two monitored values, the monitoring difference is not calculated; if no monitoring difference that matches the risk change direction appears, the transmission time is not written, but the corresponding channel connection relationship is retained so that physical channels that have not yet shown monitoring changes can still participate in subsequent area path calculations; In S1-3, a multi-level directed connection relationship extending from the source area to the subsequent area is obtained from the existing channel connection records in the mine. First, the source area is taken as the first starting area, and the channel connection records from the source area to the adjacent area are read. The source area is recorded as the starting area and the adjacent area to which it is pointed is recorded as the ending area. Then, the unprocessed ending areas are taken as the new starting areas, and the channel connection records from the new starting areas to other adjacent areas are read. The adjacent areas to which they are pointed are recorded as the new ending areas. During processing, the region number that has been read as the starting region and the combination of the starting region number and the ending region number that has been written are recorded. The same combination is written only once. When the ending region has completed the outward channel reading, it is not expanded again, so as to avoid the loop processing when A points to B, B points to C, and C points to A again. Until all the obtained ending regions have completed one outward channel reading and there are no new ending regions, the channel connection record connected step by step from the source region is output, so that S2 can read the starting region monitoring value sequence and the ending region monitoring value sequence of each record respectively. Through the above processing, the risk event is converted into the source area, the risk change direction, the monitoring value sequence of each area and the multi-level channel connection record. The source area determines the starting point for subsequent path calculation, the risk change direction determines the subtraction order of the monitoring values ​​at both ends of the channel in S2, the channel connection record limits the connection direction of the area that the risk factor can pass through, and the transmission time retains the time basis of the monitoring changes in adjacent areas. For example, if a gas risk event occurs in area A at 10:00, and the gas concentrations at 9:59 and 10:00 are 0.80% and 1.10% respectively, with an event difference of 0.30%, then area A is taken as the source area, and the increase in the monitoring value is taken as the direction of risk change. If the airflow flows from A to B and C at the time of the risk event, then the flow from A to B and from A to C is recorded. The downstream channels of B and C are then read. If B points to D, C points to E, and D points to F, then the flow from B to D, C to E, and D to F is recorded. This yields a multi-level channel connection record for subsequent regularization distance and regional path calculation.

[0019] S2. Calculate the difference between adjacent monitoring values ​​at both ends of each channel; when the risk change direction increases, take the previous value minus the next value at the starting point and the next value minus the previous value at the ending point; when it decreases, take the next value minus the previous value at the starting point and the previous value minus the next value at the ending point; subtract the two sets of results from each other and take the absolute value to form a distance array, take the current value as the cumulative value in the upper left corner, take the current value plus the previous cumulative value in the first row and first column, take the current value plus the cumulative values ​​in the upper left, upper left, and upper left corner of the remaining positions in ascending order, and take the cumulative value in the lower right corner as the normalized distance; In this embodiment, S2 is used to convert the changes in monitoring values ​​at both ends of each channel into a regularized distance; the starting and ending regions use the same monitoring index and the same unit of measurement; the monitoring values ​​are arranged according to the acquisition time, and each monitoring value is bound to the corresponding acquisition time; during the initial calculation, the monitoring values ​​between the acquisition time before the occurrence of the risk event and the current last acquisition time are read for both regions to form the change sequence at both ends, the distance array, and the regularized distance; this implementation process includes the following steps: In S2-1, the risk reduction change in the starting area and the risk enhancement change in the ending area are converted into directly comparable changes; a channel connection record is read to record the corresponding starting area monitoring value sequence and ending area monitoring value sequence; for two monitoring values ​​with adjacent collection order in each sequence, the monitoring value collected earlier is taken as the previous monitoring value and the monitoring value collected later is taken as the next monitoring value. When the direction of risk change is an increase in the monitored value, the change in the starting point is obtained by subtracting the next monitored value from the previous monitored value in the starting area, and the change in the ending point is obtained by subtracting the previous monitored value from the next monitored value in the ending area. When the direction of risk change is a decrease in the monitored value, the change in the starting point is obtained by subtracting the previous monitored value from the next monitored value in the starting area, and the change in the ending point is obtained by subtracting the next monitored value from the previous monitored value in the ending area. The time of the last monitoring value collected for each change is taken as the corresponding collection time of the change. The changes at the starting point and the ending point are arranged in order of collection time to obtain the starting point change sequence and the ending point change sequence respectively. When the risk factor migrates from the starting point area to the ending point area, the reverse change of the starting point area along the risk change direction and the continued change of the ending point area along the risk change direction are represented by the same numerical direction. If there are fewer than two monitoring values ​​in any area, no corresponding change sequence is formed, and the channel connection record is not included in the normalized distance calculation of this round. In S2-2, the numerical difference between each change in the starting change sequence and the ending change sequence is calculated; the rows of the distance array are set in the order of the starting change, and the columns of the distance array are set in the order of the ending change. Subtract each starting change from each ending change and take the absolute value of the difference. Write the absolute value at the intersection of the row containing the starting change and the column containing the ending change. After all changes are calculated pairwise, a distance array is formed. Each row of the distance array corresponds to a starting change and its acquisition time, and each column corresponds to an ending change and its acquisition time, so that the acquisition time of the two regions can be obtained later based on the position of the distance array. In S2-3, the cumulative distance from the top left corner to the bottom right corner is calculated along the distance array, and the previous position used for each cumulative value is recorded; the distance value of the top left corner is directly used as the cumulative value of the top left corner; for the remaining positions in the first row, the cumulative value of the current position is obtained by adding the current position distance value to the cumulative value on the left, and the left position is recorded as the source position; For the remaining positions in the first column, the current position's cumulative value is obtained by adding the current position's distance value to the cumulative value on the side, and the top position is recorded as the source position; for the remaining positions, the cumulative values ​​on the left, top, and top-left sides are read, arranged in ascending order of value, and the first cumulative value is added to the current position's distance value to obtain the current position's cumulative value, and the position of the first cumulative value is recorded as the source position; When the same cumulative value appears, the source position is determined sequentially by the upper left position, the left side position, and the upper side position. After calculating all positions of the distance array, the cumulative value at the lower right corner is used as the normalized distance of the channel connection record, and the source position of each cumulative value is retained. The normalized distance is obtained by adding non-negative distance values ​​one by one, so it is kept non-negative and used as the edge weight of the corresponding directed edge in S3. The source position is used by S4 to trace back from the lower right corner cumulative value position to the upper left corner step by step, and the acquisition time of the starting area and the ending area is obtained according to the row and column correspondence of the distance array. Through the above processing, each channel connection record that can participate in the calculation obtains a normalized distance, while retaining the distance array, cumulative value, and source location; the normalized distance is used to compare the cumulative difference of the monitored changes at both ends of different channel connection records, and serves as the edge weight for subsequent regional path calculation; the source location is used to subsequently restore the correspondence between the acquisition times of the changes at both ends; For example, if the risk change direction is an increase in the monitored value, and area A to area B constitutes a channel connection record, and the adjacent monitored value in area A decreases from 1.20 to 1.00, then the starting change is 1.20 minus 1.00, resulting in 0.20; if the adjacent monitored value in area B increases from 0.70 to 0.88, then the ending change is 0.88 minus 0.70, resulting in 0.18; the absolute value of the two is 0.02, and this is written to the corresponding distance array position; the same method is used to calculate all distance values ​​for the remaining starting and ending changes, and then the cumulative values ​​on the left, top, and upper left sides are calculated up to the lower right corner according to the recursive rule, and the cumulative value at the lower right corner is used as the normalized distance of the corresponding channel connection record from area A to area B.

[0020] S3. Using regions as nodes, channel connection records as directed edges, and regular distances as edge weights, run Dijkstra's algorithm from the source region, take the first node of the cumulative edge weight in ascending order to update adjacent nodes, and take the first path of the cumulative edge weight in ascending order without successor nodes as the region path. In this embodiment, S3 is used to convert the multi-level channel connection record obtained by S1 into a directed region graph, and the regularized distance calculated by S2 represents the cumulative difference of the monitoring changes at both ends when the hazard factor is transmitted along the corresponding channel. Then, the Dijkstra algorithm is used to accumulate the regularized distance from the source region level by level, retain the low cumulative edge weights and corresponding predecessor nodes that reach each node, and obtain the region path extending backward from the source region. Since the normalized distance is obtained by summing the distance values ​​one by one and is not less than zero, it can be directly used as the weight of the directed edge in Dijkstra's algorithm; the cumulative edge weight represents the sum of the weights of all directed edges traversed from the source region along the continuous channel to the current node, and is used to compare different connection routes; this implementation process includes the following steps: In S3-1, the connection relationships and regular distances of the mining area are converted into nodes, directed edges and edge weights that can be directly processed by the Dijkstra algorithm. The connection records of each channel formed by S1 are read, the starting area in the record is taken as the starting node, the ending area is taken as the ending node, the connection from the starting node to the ending node is taken as the directed edge, and the regular distance obtained by S2 for the connection records of the same channel is written into the corresponding directed edge as the edge weight. When the same region appears repeatedly in different channel connection records, only one node is established, and each directed edge is still retained separately; the node corresponding to the source region is taken as the starting node, the cumulative edge weight of the starting node is written to zero, the cumulative edge weight of the other nodes is not written for the time being, and all nodes are initially recorded as unprocessed; thus, the cumulative edge weight of each subsequent node is calculated from the source region, avoiding the use of historical path data that is unrelated to this risk event; In S3-2, the next calculation position is determined from the nodes that can be reached from the source region, and the cumulative edge weights required to reach each endpoint node via the current node are calculated; the nodes that have been written with cumulative edge weights but have not been processed are read and arranged in ascending order of cumulative edge weights, with the first node as the current node; when the cumulative edge weights are the same, they are arranged in ascending order of the region number corresponding to the node, with the node with the first region number as the current node. Read each directed edge pointed to by the current node, add the cumulative edge weight of the current node to the corresponding directed edge weight, and use the resulting value as the candidate cumulative edge weight of the endpoint node of that directed edge; for example, the edge weight from the source region to region B is 2.4, and the edge weight from region B to region D is 1.3, then when the cumulative edge weight of region B is 2.4, the candidate cumulative edge weight obtained by reaching region D via region B is 3.7; the candidate cumulative edge weight is written to the next step for comparison to determine whether the current route has a lower total normalized distance than the routes already recorded; if the current node does not have a pointed-to-point directed edge, no candidate cumulative edge weight is generated, and the processing of that node ends directly; In S3-3, when multiple paths can reach the same destination node, only the path with the lower cumulative edge weight value and its previous node are retained; when the destination node has not yet written the cumulative edge weight, the candidate cumulative edge weight is directly written to the destination node, and the current node that generated the candidate cumulative edge weight is written to the destination node as the predecessor node. When the endpoint node has been written into the cumulative edge weight, the candidate cumulative edge weight is compared with the original cumulative edge weight. If the candidate cumulative edge weight is lower than the original cumulative edge weight, the candidate cumulative edge weight replaces the original cumulative edge weight, and the current node replaces the original predecessor node. If the candidate cumulative edge weight is higher than the original cumulative edge weight, the original cumulative edge weight and the original predecessor node remain unchanged. If the two values ​​are the same, the original cumulative edge weight and the original predecessor node are also retained. After all directed edges pointed to by the current node have been processed, the current node is marked as processed. Then, the next current node is determined from the nodes that have been written into the cumulative edge weight but have not been processed, according to the sorting rule in S3-2. The calculation and comparison of candidate cumulative edge weights continue. When there are no nodes that have been written into the cumulative edge weight but have not been processed, the cumulative calculation ends. Nodes that have not obtained a continuous directed edge connection from the source region are never written into the cumulative edge weight and do not participate in the region path determination. In S3-4, the path with the lowest cumulative regularization distance is determined from all the channel termination positions that can be reached from the source region, and the order of the regions traversed by the path is restored; the nodes whose cumulative edge weights have been written are read, and nodes that do not have a directed edge with themselves as the starting node are treated as nodes without successors. When multiple nodes without successors exist, they are arranged in ascending order of cumulative edge weight. The node with the lowest cumulative edge weight without successors is taken as the end node of the path. When the cumulative edge weights are the same, the first node is taken in ascending order of region number. The predecessor node recorded by the end node of the path is read, and then the predecessor node recorded by the predecessor node itself is read. The same method is used to backtrack level by level until the starting node corresponding to the source region is reached. The nodes obtained by backtracking are arranged in reverse order, so that the node corresponding to the source region is in the first position and the end node of the path is in the last position. The regions corresponding to each node are arranged in this order to obtain the region path, and written to S4 for reading. If all nodes with accumulated edge weights have the specified directed edges but no unsuccessful nodes, then no regional path will be generated in this round. The risk events and the channel connection records that have been obtained will be retained and recalculated after subsequent monitoring data or channel transmission status updates, in order to avoid arbitrarily specifying the end of the path from the closed loop connection. Through the above processing, the region, channel connection record and normalized distance are respectively implemented as nodes, directed edges and edge weights, and the region connection order passed from the source region to the back is determined by comparing the cumulative edge weights; the predecessor node is used to save the previous node of the path with low cumulative edge weight, so that the final region path can be recovered from the end of the path without saving all candidate routes. For example, source region A has two directed edges, A to B and A to C, with normalized distances of 1.8 and 2.5 respectively. The normalized distances from B to D and from C to D are 1.1 and 0.9 respectively. Therefore, the cumulative edge weights of regions B and C are 1.8 and 2.5 respectively. First, process region B to obtain the candidate cumulative edge weight of region D, which is 2.9. Record B as the predecessor node of D. Then process region C to obtain the candidate cumulative edge weight of region D, which is 3.4. Since 3.4 is higher than the recorded 2.9, retain the cumulative edge weight of region D, which is 2.9, and the predecessor node B. When there is no specified directed edge in region D, treat D as a node without a successor. Backtrack from D along the predecessor node B to A, and then arrange the paths in the order of A, B, and D to obtain the region path.

[0021] S4. Tracing back along the regional path to the cumulative position of the normalized distance, obtaining the collection time of adjacent areas and extracting the monitoring value sequence, recalculating the normalized distance and regional path; if the two paths are consistent, the current path is used as the risk migration chain, and if they are inconsistent, this step is repeated. In this embodiment, S4 is used to reverse the time range of the monitoring values ​​at both ends of the channel by using the regional path obtained by S3, and then recalculate the normalized distance and regional path with the intercepted monitoring values, so that the regional connection order and the monitoring changes of the risk factors between adjacent regions are simultaneously established. Upon first entering S4, the region path obtained in S3 is used as the previous region path; when calculating the region path again in subsequent rounds, the current region path obtained in the previous round is used as the previous region path; each row of the distance array corresponds to a starting point change and its acquisition time, and each column corresponds to an ending point change and its acquisition time, so the cumulative position in the array can be restored to the combination of acquisition times of the two regions; this implementation process includes the following steps: In S4-1, the time correspondence of the monitoring changes in adjacent areas is obtained in reverse from the cumulative process of regular distance; two adjacent areas in the area path are read in sequence, with the former area as the starting area and the latter area as the ending area, and the distance array formed by the two areas in S2 and the source location of each cumulative value record are read. Starting from the position of the cumulative value in the lower right corner, read the source position of the current position record and move to that position. Move to the upper left corner in the same way. Each row of the distance array corresponds to a starting change and its collection time, and each column corresponds to an ending change and its collection time. The two collection times corresponding to the same position are combined into a time pair, and then arranged from the earliest to the latest according to the collection time of the starting area. When the starting point regions have the same collection time, they are arranged from earliest to latest according to the collection time of the ending point regions to form a time pair sequence. When the source position is missing, the backtracking of the corresponding adjacent regions is stopped, and the time pairs after the missing position are not used for sequence truncation to avoid data whose collection time source cannot be determined from entering the subsequent calculation. In S4-2, the collection times that conform to the relationship of "risk decreases in the starting area and risk increases in the ending area" are retained from the time pair sequence; for each time pair, the monitoring value corresponding to the collection time of the starting area and the monitoring value of the previous collection time are read, and the difference between the former and the latter is obtained. Read the monitoring value corresponding to the collection time of the endpoint region and the monitoring value of the previous collection time, and obtain the endpoint difference by subtraction in the same way; when the risk change direction is that the monitoring value increases, only the time pairs where the starting difference is less than zero and the endpoint difference is greater than zero are retained; when the risk change direction is that the monitoring value decreases, only the time pairs where the starting difference is greater than zero and the endpoint difference is less than zero are retained; time pairs where the difference is equal to zero are not retained; arrange the retained time pairs according to the order of the collection time of the starting region to form a sequence of retained time pairs, which is used to determine the subsequent interception range of the two regions; If no time pair is retained, it means that the migration time correspondence between the two regions was not obtained in this round of data; if only one time pair is retained, it is impossible to form the change between at least two monitoring values; in both cases, no corresponding new normalized distance is generated, and the current directed edge continues to use the normalized distance obtained by S2 as the edge weight in this round, and is reprocessed after obtaining new monitoring values; In S4-3, the acquisition time that conforms to the migration direction is converted into the monitoring value range of the two regions that actually participate in the re-regulation; the first and last time pairs of the retained time pair sequence are read, the acquisition time of the starting region in the first time pair is taken as the starting point of the interception, and the acquisition time of the starting region in the last time pair is taken as the starting point of the interception. All monitoring values ​​of the starting region between the two times are intercepted and arranged according to the acquisition time to obtain the starting point interception sequence. Following the same method, the collection time of the endpoint region in the first and last time pairs is used as the endpoint truncation start and endpoint, respectively, to obtain the endpoint truncation sequence; the truncation range includes the monitoring values ​​corresponding to the truncation start and endpoint; when any truncation sequence has fewer than two monitoring values, the new normalized distance calculation for the corresponding directed edge is not performed, and the original normalized distance is continued to be used as the edge weight for this round. In S4-4, the normalized distance between adjacent regions is recalculated based on the truncated time range, so that the weight of the corresponding directed edge only reflects the monitoring changes that are consistent with the current risk migration direction. When the risk change direction is an increase in the monitoring value, the change at the starting point is obtained by subtracting the previous value from the next value in the adjacent monitoring values ​​in the starting point truncated sequence, and the change at the ending point is obtained by subtracting the previous value from the next value in the ending point truncated sequence. When the risk change direction is a decrease in the monitored value, the starting point intercept sequence is obtained by subtracting the previous value from the subsequent value to obtain the starting point change amount, and the ending point intercept sequence is obtained by subtracting the previous value from the subsequent value to obtain the ending point change amount. Each starting point change amount is subtracted from each ending point change amount, and the absolute value of the difference is taken. A new distance array is formed with the starting point change amount as the row and the ending point change amount as the column. The distance value of the top left corner is directly used as the cumulative value. For the remaining positions in the first row, the cumulative value is obtained by adding the current position distance value to the cumulative value on the left and the left position is recorded. For the remaining positions in the first column, the cumulative value is obtained by adding the current position distance value to the cumulative value on the right and the top position is recorded. For the remaining positions, the cumulative value is obtained by adding the lowest value among the cumulative values ​​on the left, top, and top left sides to the current position distance value and the corresponding source position is recorded. For the same value, the source position is determined in the order of top left, left, and top. The cumulative value in the lower right corner is used as the new normalized distance and written into the corresponding directed edge as the new edge weight; the directed edges that have not been truncated and recalculated in this round continue to use the normalized distance obtained by S2 as the edge weight in this round, so that subsequent path calculations can still select channels outside the previous regional path. In S4-5, the rewritten edge weights are used to recalculate the regional path from the source region to the next region and eliminate the influence of the previous Dijkstra calculation result on the current path. The nodes and directed edges established in S3 are kept unchanged, and the cumulative edge weights, predecessor nodes and processed marks of each node except the starting node are cleared. The node corresponding to the source region is taken as the starting node and the cumulative edge weight is written to zero. From the nodes that have been written with cumulative edge weights but have not yet been processed, select the node with the lowest cumulative edge weight value as the current node. If the values ​​are the same, select the first node in ascending order of region number. Add the cumulative edge weight of the current node to the current weight of each directed edge pointed to by the current node in this round to obtain the candidate cumulative edge weight of the corresponding endpoint node. If the endpoint node has not been written with cumulative edge weights, directly write the candidate cumulative edge weights and record the current node as the predecessor node. If the cumulative edge weights have been written, compare the candidate cumulative edge weights with the original cumulative edge weights, write the one with the lower value and the predecessor node that generated the value. If the values ​​are the same, retain the original cumulative edge weights and the original predecessor node. After all directed edges of the current node have been processed, it is recorded as processed. Continue to select the next unprocessed node until there are no more nodes that have been written with cumulative edge weights but have not yet been processed. Then, from the nodes that have been written with cumulative edge weights and do not have outward directed edges, take the node with the lowest cumulative edge weight as the end node of the path, backtrack along the predecessor node to the corresponding node of the source region, and reverse the arrangement of the regions corresponding to each node to obtain the current regional path; if there is no reachable node without successor, the current regional path is not generated in this round, and the previous regional path and risk event are retained, and recalculated after the monitoring value or channel connection record is updated. In S4-6, the area connection order and the hazard propagation time order are checked simultaneously, and the interception range is adjusted in reverse and recalculated when the two are inconsistent. First, the current area path is compared with the previous area path. If the two paths contain the same number of areas and the area number of each arrangement position is the same, the area arrangement is determined to be the same. When a directed edge not included in the previous regional path appears in the current regional path, the original distance array and source position of the directed edge are read first. Then, the cumulative position backtracking, time pair filtering, monitoring value truncation and new normalized distance calculation are performed in sequence to generate the corresponding new distance array before the time order judgment is performed. The time order judgment is performed sequentially from the source region to the end of the regional path. The time of the end region of the retained time pair of the previous adjacent region is collected and compared with the time of the start region of the retained time pair of the next adjacent region. If the former is not later than the latter, the current regional path is taken as the risk migration chain, and the production control status of each area located inside the mine in the risk migration chain is set to "maintained". If the last endpoint collection time of the previous adjacent area is later than the first starting point collection time of the next adjacent area, the first time pair in the retained time pair sequence of the next adjacent area where the starting point collection time is later than the previous last endpoint collection time is found. The starting point collection time in this time pair is taken as the new starting point of the starting point monitoring value sequence, and the endpoint collection time in the same time pair is taken as the new starting point of the endpoint monitoring value sequence. The original endpoint remains unchanged, and the corresponding new normalized distance and the current regional path are recalculated. If there is no time pair that satisfies the time sequence relationship, the risk migration chain is not confirmed, and it is reprocessed after the new monitoring value is added. If the current regional path differs from the previous regional path, the current regional path will be used as the new previous regional path, and backtracking, screening, interception, and normalization distance calculation will be re-executed for each adjacent region. Each round will simultaneously record the arrangement of regional numbers and the interception start and end points of each adjacent region. If the same combination of records reappears and the regional arrangement or the order of collection time still does not meet the requirements, the path confirmation for this round will be stopped to maintain the validity of risk events and the status of internal regions of the mine that are already under production control, and to prevent the same data from being calculated repeatedly between different paths. Through the above processing, the spatial region path obtained by S3 is used to reverse the time correspondence of the monitoring values ​​of the two regions. The time correspondence then limits the range of monitoring values ​​and changes the weight of the directed edge. The changed edge weight then participates in the regional path calculation again until the regional arrangement and the order of collection time along the path are simultaneously established. The resulting risk migration chain not only indicates that there is a physical connection between regions, but also that the risk reduction in the starting region and the risk enhancement in the subsequent region can be sequentially connected in time. For example, S3 first obtains the paths A, B, and D. The time pairs obtained by backtracking from A to B cover 10:01 to 10:05, and the time pairs from B to D cover 10:04 to 10:09. After truncation, the new regular distances from A to B and from B to D are recalculated. If the path calculation is re-executed and A, C, and D are obtained, then A, C, and C, D are used as new adjacent region pairs for backtracking and truncation. If the recalculation still yields A, C, and D, and the collection time of the last endpoint from A to C is 10:06, and the collection time of the first starting point from C to D is 10:07, then the region arrangement and time sequence are both valid, and A, C, and D are used as the risk migration chain. If the collection time of the first starting point from C to D is 10:05, then the first time pair in the retained time pair sequence from C to D with the starting region collection time later than 10:06 is searched, and the starting points of the two ends are re-determined based on this time pair before the new regularized distance and regional path are calculated.

[0022] S5. When the end of the risk migration chain is outside the mine, the regional production control status is deactivated in reverse order and the risk event is closed; when it is inside the mine, the regional production control status in the chain is maintained, and the risk management result is output. In this embodiment, S5 is used to determine whether the risk event meets the closure conditions based on the location of the end of the risk migration chain, and to gradually remove the production control status of the internal area of ​​the mine along the risk migration chain from the end to the source area; the boundary record is read from the mine area division data, each area number corresponds to a location category, the location category includes the inside of the mine and the outside of the mine, and the same area number corresponds to only one location category; The risk event log includes the event status. When a risk event enters S1 processing, the event status is recorded as valid. After S4 forms a risk migration chain, the production control status of the internal areas of the mine in the risk migration chain is recorded as "maintained." The external areas of the mine are only used to indicate that the hazardous factors have left the mine, and no production control status is set there. This implementation process includes the following steps: In S5-1, it is used to determine whether to proceed with step-by-step de-escalation or continue production control based on whether the end of the risk migration chain has crossed the mine boundary; it reads the area number of the end area of ​​the risk migration chain, searches for the same area number in the boundary record, and reads the corresponding location category; When the location category is outside the mine, read each region in reverse from the end of the risk migration chain to the source region, remove the regions outside the mine, take the region inside the mine closest to the end of the mine as the first, and then arrange them in order to the source region to form the release sequence; when the location category is inside the mine, write each region inside the mine into the unwritten sequence according to the original region order of the risk migration chain. The release sequence is used to specify the order in which production control status is released, so that downstream areas are released first, followed by upstream areas; the unverified sequence is used to record areas that still need to maintain production control when the hazardous factors have not yet moved out of the mine; when the area number at the end of the risk migration chain does not exist in the boundary record, it is not determined that the end is outside the mine, the production control status of the risk event and the area inside the mine in the risk migration chain is maintained, and the missing area number is recorded for supplementation of area division data; In S5-2, the production control status of the internal areas of the mine is gradually deactivated in reverse order of the risk migration chain, and the processing of the next area is controlled by the deactivation results that have been completed; the first area of ​​the deactivation sequence is read as the current area, and after confirming that the production control status of the current area is in the hold state, the production control status is rewritten to the deactivation state, and the current area number and deactivation status are written into the status record, and the status record is used as the deactivation mark. After obtaining the release flag for the current region, the next region in the release sequence is read as the new current region. The production control status is rewritten from the maintained state to the released state in the same way, and the corresponding release flag is generated. This process is then applied to the source region. The release flag is used to prove that the current region has completed the release of the production control status and serves as the basis for the next region in the release sequence to begin the release process, thereby preventing the source region from prematurely resuming before the downstream migration of the hazardous factors has been completed. If the current region is not in a held state, no release flag will be generated again; if any region in the release sequence lacks a production control status record, the release of subsequent regions will be stopped, the production control status of the regions that have not yet been released will be maintained, and the corresponding region number will be recorded. In S5-3, it is used to close a risk event after the source area has been cleared, or to keep the risk event valid while the hazard is still inside the mine; read the clearing mark of the source area. When the source area has been cleared, the event status in the risk event record is changed from valid to closed, and the risk event identifier, risk migration chain, and production control status of each mine internal area in the risk migration chain are written into the cleared risk management results. When an unverified sequence is formed in S5-1, the effective status of the risk event is not changed, and the production control status of each area in the unverified sequence is maintained. The risk event identifier, risk migration chain, unverified sequence, and production control status of each area are written into the unverified risk management results. The risk event is not closed when the source area has not obtained the release mark. The closure of the risk event must be based on the premise that all mine internal areas to be released in the risk migration chain have completed the release of production control status in sequence. Through the above processing, the end position of the risk migration chain is transformed into a risk event closure condition, and the production control status is gradually deactivated in the opposite direction of the risk factor migration through the deactivation sequence, avoiding the direct closure of risk events based solely on the decrease in monitoring values ​​in the source area; For example, if the risk migration chain consists of region A, region B, region D, and external region E, and the boundary records indicate that A, B, and D belong to the inside of the mine, while E belongs to the outside of the mine, then a release sequence is formed in the order of D, B, and A. First, the production control status of D is changed from "maintained" to "released," and a release marker for D is generated. Then, release markers for B and A are generated in sequence. After A generates a release marker, the corresponding risk event status is changed from "valid" to "closed." If the end of the risk migration chain is still the internal region D of the mine, then A, B, and D are written into the unverified sequence, maintaining the production control status and the valid status of the risk event in the three regions, without executing the risk event closure.

[0023] Figure 1 It should be noted that: This diagram illustrates the overall process of a risk event from input, migration path calculation to risk write-off; after a risk event is input, the source region and the direction of risk change are first determined, then the normalized distance between the monitored changes at both ends of the channel is calculated, and the normalized distance is used as the edge weight to search for the region path; subsequently, the collection time is traced back and the monitoring value sequence is extracted, and the normalized distance and region path are recalculated. When the region arrangement is consistent with the order of the collection time, the risk migration chain is determined; otherwise, the extraction range is adjusted and the calculation is recalculated; finally, it is determined whether the end of the risk migration chain is located outside the mine. If it is located outside the mine, the production control status of the internal area of ​​the mine is deactivated in reverse order and the risk event is closed; if it is still located inside the mine, the production control status and the risk event remain valid. In this diagram, solid arrows indicate the execution direction between each step, diamond boxes represent conditional judgments, and "yes" or "no" indicates the corresponding judgment result; Figure 2It should be noted that this diagram illustrates the data input, risk calculation and migration determination, and control output relationships in mine safety production risk management from a system architecture perspective. The data input layer provides risk events, regional monitoring value sequences, channel connection records, and boundary records, which are then organized by edge computing nodes according to risk events and sent to subsequent modules. The risk calculation and migration determination layer is the core processing layer. The risk acquisition module first determines the source region and the direction of risk change and forms hierarchical channel connection records. The distance normalization module forms a distance array based on the changes in monitoring values ​​at both ends of the channel and obtains the normalized distance. The path search module uses the region as a node, the channel as a directed edge, and the normalized distance as the edge weight to run Dijkstra's algorithm to obtain the regional path. The migration confirmation module then backtracks based on the data collection time and performs sequence truncation. The risk migration chain is confirmed by recalculating the distance and regional path. The risk cancellation module determines whether to lift or maintain production control status based on the comparison between the end of the migration chain and the boundary record. The control output layer is used to output the final risk migration chain, the production control status of each region, and the effective or closed status of risk events. The solid arrows in the diagram indicate the transfer of data or processing results between adjacent modules in the direction of the arrow, and the text next to the arrow indicates the data object being transferred. The dashed arrows indicate the feedback relationship of retrospectively checking, correcting, or recalculating existing calculation results during the migration confirmation process, and do not represent the addition of new processing modules. The outer frames between each layer represent different functional levels, and each module frame represents the corresponding functional processing unit, which is used to reflect the complete system relationship from data input, migration judgment to control result output.

[0024] The present invention also includes a mine safety production risk management system, the system comprising a risk acquisition module, a distance normalization module, a path search module, a migration confirmation module, and a risk verification module; The risk acquisition module acquires risk events, corresponding regional monitoring value sequences, and channel connection records, using the corresponding regions as the source regions. The channel connection records record the starting and ending regions according to the direction of risk factor transmission, and use the increase or decrease relationship between the monitoring value at the time of risk occurrence and the previous value as the direction of risk change. The distance normalization module calculates the difference between adjacent monitoring values ​​at both ends of each channel. When the risk change direction increases, the starting point is the previous value minus the next value, and the ending point is the next value minus the previous value. When it decreases, the starting point is the next value minus the previous value, and the ending point is the previous value minus the next value. The absolute values ​​of the two sets of results are subtracted from each other to form a distance array. The cumulative value in the upper left corner is the current value. The first row and first column are the current value plus the previous cumulative value. The remaining positions are the current value plus the cumulative values ​​of the left, top, and upper left corners in ascending order. The cumulative value in the lower right corner is used as the normalized distance. The path search module uses regions as nodes, channel connection records as directed edges, and regular distances as edge weights. It runs the Dijkstra algorithm from the source region, takes the first node in ascending order of cumulative edge weights to update adjacent nodes, and takes the first path in ascending order of cumulative edge weights to the region path with no successor nodes. The migration confirmation module traces back along the regional path to the cumulative position of the normalized distance, obtains the collection time of adjacent areas and extracts the monitoring value sequence, and recalculates the normalized distance and regional path; if the two paths are consistent, the current path is used as the risk migration chain, and if they are inconsistent, this step is repeated. The risk write-off module, when the end of the risk migration chain is outside the mine, reverses the regional production control status and closes the risk event; when it is inside the mine, it maintains the regional production control status in the chain and outputs the risk management results.

[0025] Working principle: This solution takes the entire process of risk factors migrating from the risk-occurring area to subsequent areas as the management object, and continuously processes the monitoring data of each area through edge computing; First, determine the source region and the direction of risk change based on the risk event. Then, establish a multi-level channel connection relationship extending backward from the source region according to the actual transmission direction of the hazard factor, so that the subsequent calculation has a clear spatial range. Subsequently, the continuous changes of the monitoring values ​​at both ends of each channel are compared. The normalized distance is obtained by distance array and cumulative calculation. It is used to measure the degree of correspondence between the risk reduction in the starting area and the risk enhancement in the ending area over time. The normalized distance is then used as the regional connectivity cost and input into the Dijkstra algorithm to obtain the regional path corresponding to the monitoring change relationship from the multi-level channels. After the regional path is formed, edge computing continues to obtain the collection times at both ends along the cumulative source position of the distance array, and filters out the time relationship that matches the risk migration from the previous area to the next area. Based on this, the range of monitoring values ​​involved in the calculation is narrowed and the regularized distance and regional path are recalculated, so that the spatial path in turn constrains the time data, and the time data re-corrects the spatial path. When the regional arrangement and the timing of data collection are both stable, the regional path is determined as the risk migration chain. Finally, production control is implemented based on the end position of the risk migration chain. After the hazardous factor moves out of the mine, the production control status of the internal area of ​​the mine is gradually lifted along the risk migration chain. When the hazardous factor is still inside the mine, the production control status of the relevant area is maintained, so that the risk is lifted on the basis of the actual termination of the migration of the hazardous factor. For example, if there is a risk of increased gas concentration in area A, the edge computing node reads the gas concentration data and real-time airflow direction of area A and adjacent areas, identifies area A as the source area, and forms channel connections from A to B, from A to C, and from B and C to continue downstream. If the gas concentration in region A begins to decrease while that in region B subsequently increases, the regularized distance from A to B is calculated by regularizing the changes in the two regions. This regularized distance is then used in conjunction with the regularized distances from other channels to calculate the Dijkstra path. First, the regional paths for A, B, and D are obtained. Then, the acquisition time is traced back from the distance arrays corresponding to A to B and B to D to check whether B increases after A decreases and whether D continues to increase after B decreases. Based on the monitoring values ​​that conform to this propagation sequence, the channel regularized distance and regional path are recalculated. If the recalculation still yields A, B, and D, and the changes in each area occur in the order of A, B, and D, then A, B, and D are identified as the risk migration chain. When the gas continues to migrate along the channel after D to the outside of the mine, the system gradually deactivates production control and shuts down the risk event, starting from the internal area near the mine exit and moving towards area A. If the gas remains within area D, the production control status of the internal areas of the mine in the risk migration chain is maintained to avoid prematurely ending risk management based solely on the decrease in concentration in area A.

[0026] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for managing safety production risks in mines, characterized in that, include: S1. Obtain risk events, corresponding regional monitoring value sequences, and channel connection records, and use the corresponding region as the source region; The channel connection record records the starting and ending areas according to the direction of hazard factor transmission, and uses the increase or decrease relationship between the monitoring value when the risk occurs and the previous value as the direction of risk change; S2. Calculate the difference between adjacent monitoring values ​​at both ends of each channel; when the risk change direction increases, take the previous value minus the next value at the starting point and the next value minus the previous value at the ending point; when it decreases, take the next value minus the previous value at the starting point and the previous value minus the next value at the ending point; subtract the two sets of results from each other and take the absolute value to form a distance array, take the current value as the cumulative value in the upper left corner, take the current value plus the previous cumulative value in the first row and first column, take the current value plus the cumulative values ​​in the upper left, upper left, and upper left corner of the remaining positions in ascending order, and take the cumulative value in the lower right corner as the normalized distance; S3. Using regions as nodes, channel connection records as directed edges, and regular distances as edge weights, run Dijkstra's algorithm from the source region, take the first node of the cumulative edge weight in ascending order to update adjacent nodes, and take the first path of the cumulative edge weight in ascending order without successor nodes as the region path. S4. Tracing back along the regional path to the cumulative position of the normalized distance, obtaining the collection time of adjacent areas and extracting the monitoring value sequence, recalculating the normalized distance and regional path; if the two paths are consistent, the current path is used as the risk migration chain, and if they are inconsistent, this step is repeated. S5. When the end of the risk migration chain is outside the mine, the regional production control status is deactivated in reverse order and the risk event is closed; when it is inside the mine, the regional production control status in the chain is maintained and the risk management result is output.

2. The mine safety production risk management method according to claim 1, characterized in that: S1 includes: S1-1. Read the monitoring value corresponding to the time when the risk event occurs in the corresponding area and the monitoring value corresponding to the previous collection time. Subtract the monitoring value corresponding to the previous collection time from the monitoring value corresponding to the time when the risk event occurs to obtain the event difference value. Take the area corresponding to the risk event as the source area. Determine the direction of risk change by the sign of the event difference value. When the event difference value is positive, the corresponding monitoring value increases. When the event difference value is negative, the corresponding monitoring value decreases. S1-2. Obtain the monitoring value sequence of each adjacent area that has a channel connection record with the source area. The monitoring value sequence is formed by arranging the monitoring values ​​corresponding to each collection time in the order of collection. Starting from the time of the risk event, subtract the monitoring value corresponding to the previous collection time from the monitoring value corresponding to the next collection time to obtain the monitoring difference. When the risk change direction is that the monitoring value increases, the next collection time corresponding to the first monitoring difference greater than zero is taken as the transmission time. When the risk change direction is that the monitoring value decreases, the next collection time corresponding to the first monitoring difference less than zero is taken as the transmission time. S1-3. Taking the source region as the first starting region, read the adjacent regions connected to the first starting region. Use the first starting region as the starting region and each adjacent region as the ending region to form a channel connection record. Then, take each ending region as the new starting region, read the adjacent regions connected to the new starting region that have not yet formed a corresponding channel connection record, and use the new starting region and the adjacent regions as the starting region and the ending region respectively to form a channel connection record. Process step by step according to the method of using the ending region as the next starting region until there are no adjacent regions that have not yet formed a channel connection record, and obtain the channel connection record that connects the source region to the subsequent regions step by step.

3. The mine safety production risk management method according to claim 2, characterized in that: S2 includes: S2-1. Obtain the starting area monitoring value sequence and the ending area monitoring value sequence for each channel connection record; when the risk change direction is increasing, subtract the next monitoring value from the previous monitoring value in the starting area to obtain the starting change amount, and subtract the previous monitoring value from the next monitoring value in the ending area to obtain the ending change amount; when the risk change direction is decreasing, subtract the previous monitoring value from the next monitoring value in the starting area to obtain the starting change amount, and subtract the next monitoring value from the previous monitoring value in the ending area to obtain the ending change amount, and take the acquisition time of the next monitoring value corresponding to each change amount as the corresponding acquisition time, and arrange them in order of acquisition to form the starting change sequence and the ending change sequence respectively. S2-2. Using the starting change amount in the starting change sequence as rows and the ending change amount in the ending change sequence as columns, the absolute value of the difference between each starting change amount and each ending change amount is taken as the distance value and written into the corresponding row and column intersection position to form a distance array. S2-3. Use the distance value of the top left corner of the distance array as the cumulative value; for the remaining positions in the first row, add the current position distance value to the cumulative value on the left to obtain the cumulative value, and record the left position as the source position; for the remaining positions in the first column, add the current position distance value to the side cumulative value to obtain the cumulative value, and record the top position as the source position; for the remaining positions, sort the left, top, and top left cumulative values ​​in ascending order, add the first cumulative value to the current position distance value to obtain the cumulative value, and record the position of the first cumulative value as the source position. If the values ​​are the same, record the source position in the order of top left, left, and top; use the cumulative value of the bottom right corner as the normalized distance.

4. The mine safety production risk management method according to claim 3, characterized in that: When the Dijkstra algorithm in S3 is executed, it starts from the node corresponding to the source region, accumulates the normalized distance of each directed edge level by level, retains the lowest accumulated edge weight and the corresponding predecessor node to each node, and backtracks along the predecessor node to obtain the region path. S3 includes: S3-1. Take the starting region in each channel connection record as the starting node, the ending region as the ending node, the starting node pointing to the ending node as the directed edge, the normalized distance of the corresponding channel connection record as the edge weight of the directed edge, the node corresponding to the source region as the starting node, and record the cumulative edge weight of the starting node as zero. S3-2. Arrange the nodes that have been written into the cumulative edge weights but have not yet been processed as the current node in ascending order of cumulative edge weights. Take the first node as the current node, add the cumulative edge weight of the current node to the edge weight of each directed edge pointed out from the current node, and use the resulting value as the candidate cumulative edge weight of the corresponding endpoint node.

5. A method for managing mine safety production risks according to claim 4, characterized in that: S3 further includes: S3-3. When the endpoint node has not been written with cumulative edge weights, write the candidate cumulative edge weights into the endpoint node and take the current node as the predecessor node of the endpoint node. When the endpoint node has been written with cumulative edge weights, arrange the candidate cumulative edge weights and the written cumulative edge weights in ascending order of value, take the first value as the cumulative edge weight of the endpoint node, and take the current node that generated the first value as the predecessor node of the endpoint node. After all the directed edges corresponding to the current node have been processed, mark the current node as processed, and continue to take the first node in ascending order of cumulative edge weight from the nodes that have been written with cumulative edge weights and have not yet been processed as the next current node, until each node that has been written with cumulative edge weights has completed one directed edge processing. S3-4. Nodes with accumulated edge weights that do not have directed edges starting from themselves are considered as nodes without successors. Arrange these nodes in ascending order of accumulated edge weights. Use the first node without successors as the end node of the path. Backtrack from the end node to the starting node according to the predecessor nodes recorded by each node. Arrange the backtracked nodes in the order from the starting node to the end node of the path. Arrange the regions corresponding to each node in order to obtain the region path.

6. The mine safety production risk management method according to claim 5, characterized in that: S4 includes: S4-1. Take the preceding region in the sequentially adjacent region path as the starting region and the following region as the ending region. Starting from the cumulative value position at the lower right corner of the distance array between the two regions, trace back along the source position of each cumulative value record to the upper left corner. Pair the collection time of the starting region corresponding to the passed position with the collection time of the ending region to form a time pair sequence. S4-2. Subtract the monitoring value of the previous collection time from the monitoring value of the collection time in each time pair to obtain the starting difference and the ending difference; when the risk change direction is increasing, retain the time pair with the starting difference less than zero and the ending difference greater than zero; when the risk change direction is decreasing, retain the time pair with the starting difference greater than zero and the ending difference less than zero. Form a sequence of retained time pairs according to the collection order.

7. A method for managing mine safety production risks according to claim 6, characterized in that: S4 further includes: S4-3. The start time of the first and last positions of the sequence is used as the start and end times of the interception of the start region monitoring value sequence, and the end time of the first and last positions is used as the start and end times of the interception of the end region monitoring value sequence, thus obtaining the start interception sequence and the end interception sequence. S4-4. When the direction of risk change is increasing, the starting point cutoff sequence is obtained by subtracting the previous value from the next value to get the starting point change, and the ending point cutoff sequence is obtained by subtracting the previous value from the next value to get the ending point change; when the direction of risk change is decreasing, the starting point cutoff sequence is obtained by subtracting the previous value from the next value to get the starting point change, and the ending point cutoff sequence is obtained by subtracting the previous value from the next value to get the ending point change. The absolute values ​​of the changes at each starting point and each ending point are subtracted from each other to form a new distance array. The distance value at the top left corner is used as the cumulative value. The first row and the first column are obtained by adding the cumulative values ​​of the left and top sides to the current position distance value and recording the corresponding source position. For the remaining positions, the cumulative value is obtained by adding the lowest value among the cumulative values ​​of the left, top, and top left sides to the current position distance value and recording the corresponding source position. When the values ​​are the same, the source positions are recorded in the order of top left, left, and top. The cumulative value at the bottom right corner is used as the new normalized distance and as the new edge weight of the corresponding directed edge.

8. A method for managing mine safety production risks according to claim 7, characterized in that: S4 further includes: S4-5. Take the node corresponding to the source region as the starting node, clear the cumulative edge weights, predecessor nodes and processed marks of all nodes except the starting node, record the cumulative edge weight of the starting node as zero, and record the remaining nodes as unprocessed nodes. Take the node with the lowest cumulative edge weight from the nodes that have been recorded but not yet processed as the current node. Add the cumulative edge weight of the current node to the new edge weight of each directed edge pointed out from the current node to obtain the candidate cumulative edge weight of the corresponding endpoint node. When the endpoint node has not been written with cumulative edge weights, the candidate cumulative edge weight is used as the endpoint node's cumulative edge weight, and the current node is used as the predecessor node. When the endpoint node has been written with cumulative edge weights, the lower of the candidate cumulative edge weight and the original cumulative edge weight is used as the endpoint node's cumulative edge weight, and the node that generated the value is used as the predecessor node. After the directed edge corresponding to the current node is processed, the current node is marked as processed. Then, the node with the lowest cumulative edge weight value among the nodes that have been written with cumulative edge weights but have not been processed is taken as the new current node, until there are no nodes that have been written with cumulative edge weights but have not been processed. Nodes that have no directed edges starting from themselves and whose cumulative edge weights have been written are considered as nodes without successors. The node with the lowest cumulative edge weight is taken as the end node of the path. The path is then traced back to the corresponding node in the source region along the predecessor node. The corresponding regions of each node are arranged in the order from the source region to the end node of the path to obtain the current regional path. S4-6. When the number of regions and their arrangement are the same as the previous regional path, the new distance array of each adjacent region in the current regional path is traced back to obtain the time pair sequence. When the acquisition time of the last endpoint of the previous adjacent region is not later than the acquisition time of the first and second starting points of the next adjacent region, the current regional path is taken as the risk migration chain, and each region in the risk migration chain is set to the production control status. When the time order is reversed, the first acquisition time after the last acquisition time of the previous end is taken as the new starting point for the acquisition of the monitoring value sequence of the next adjacent region, and its new normalized distance and current regional path are recalculated; if the two regional paths are different, the current regional path is taken as the new previous regional path, and the cumulative position backtracking, time pair screening, sequence truncation, normalized distance recalculation and regional path recalculation are performed on the adjacent regions in sequence until the regional arrangement and acquisition time order are consistent.

9. A method for managing mine safety production risks according to claim 8, characterized in that: S5 includes: S5-1. Obtain boundary records for the inside and outside of the mine by region number, and compare the end region number of the risk migration chain with the boundary records item by item; when the end region is outside the mine, arrange the mine internal regions in the risk migration chain in the order from the end to the source region to form a release sequence; when the end region is inside the mine, write each region in the risk migration chain into the unwritten sequence. S5-2. Read the first region of the release sequence as the current region, change the production control status of the current region from the hold status to the release status and write the release mark; after the release mark is written to the current region, take the next region of the release sequence as the new current region and perform risk status release and release mark writing, until the release mark is written to the source region; S5-3: When the source region writes the de-marking, the risk event is closed and the de-marked risk management result is output; when an unde-marked sequence is formed, the risk event is kept valid and the production control status of each region in the unde-marked sequence is maintained, and the unde-marked risk management result is output.

10. A mine safety production risk management system, used to implement the mine safety production risk management method according to any one of claims 1-9, the system comprising a risk acquisition module, a distance normalization module, a path search module, a migration confirmation module, and a risk verification module, characterized in that: The risk acquisition module acquires risk events, corresponding regional monitoring value sequences, and channel connection records, using the corresponding regions as the source regions. The channel connection records record the starting and ending regions according to the direction of risk factor transmission, and use the increase or decrease relationship between the monitoring value at the time of risk occurrence and the previous value as the direction of risk change. The distance normalization module calculates the difference between adjacent monitoring values ​​at both ends of each channel. When the risk change direction increases, the starting point is the previous value minus the next value, and the ending point is the next value minus the previous value. When it decreases, the starting point is the next value minus the previous value, and the ending point is the previous value minus the next value. The absolute values ​​of the two sets of results are subtracted from each other to form a distance array. The cumulative value in the upper left corner is the current value. The first row and first column are the current value plus the previous cumulative value. The remaining positions are the current value plus the cumulative values ​​of the left, top, and upper left corners in ascending order. The cumulative value in the lower right corner is used as the normalized distance. The path search module uses regions as nodes, channel connection records as directed edges, and regular distances as edge weights. It runs the Dijkstra algorithm from the source region, takes the first node in ascending order of cumulative edge weights to update adjacent nodes, and takes the first path in ascending order of cumulative edge weights to the region path with no successor nodes. The migration confirmation module traces back along the regional path to the cumulative position of the normalized distance, obtains the collection time of adjacent areas and extracts the monitoring value sequence, and recalculates the normalized distance and regional path; if the two paths are consistent, the current path is used as the risk migration chain, and if they are inconsistent, this step is repeated. The risk write-off module, when the end of the risk migration chain is outside the mine, reverses the regional production control status and closes the risk event; when it is inside the mine, it maintains the regional production control status in the chain and outputs the risk management results.