A knowledge graph-driven integrated control and management decision-making method for multiple disasters and three-dimensional disaster prevention.
By constructing a multi-hazard coupling diagram constrained by ventilation status and setting a semi-stable window, the problem of misjudging the direction of disaster propagation when the mine ventilation status changes is solved, the credibility of the risk chain and the accuracy of control decisions are improved, and the observation results are ensured to meet the propagation qualification constraints.
Patent Information
- Application Number
- CN202611351148.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-02
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2046-09-02
AI Technical Summary
In mines, when ventilation conditions change rapidly, existing technologies struggle to accurately determine the risk direction of gas migration, smoke diffusion, or dust propagation, and the observation results after the response may affect the reliability of subsequent control decisions.
By converting the multi-source status data of ventilation, air conditioning, and fire prevention into object status records indexed by common decision-making time, ventilation status version, and smallest alleyway spatial unit, a multi-hazard coupling diagram of ventilation status constraints is constructed. A semi-stable window for ventilation reconstruction is set, and propagation qualification access processing is performed to generate inferable propagation relationships and evidence of permitted release. Control and management permission gating is implemented, and the multi-hazard coupling control and management decision results are output.
It solves the problem of misidentification of disaster propagation direction after changes in ventilation status, improves the technical reliability of multi-hazard coupled risk chain, ensures that the observation results after the response action meet the propagation qualification constraints, and avoids miswriting of risk relief and action effect sample updates.
Smart Images

Figure CN122864923B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of mine safety data processing technology, and in particular to a knowledge graph-driven method for coupled control and management of ventilation, fire prevention, and multiple disasters. Background Technology
[0002] Coal mine ventilation, dust, and gas control typically rely on multi-source data such as gas, wind speed, wind direction, temperature, smoke, dust, damper status, fan operation status, and dispatch records to comprehensively assess ventilation anomalies, gas accumulation, fire precursors, and dust diffusion. With the advancement of intelligent mine construction, related systems are now able to digitally link roadways, measuring points, ventilation facilities, disaster events, and response measures. They can also use knowledge graphs, spatiotemporal event chains, or risk rule bases to assist in generating early warning information and response suggestions. In actual production, the advancement of mining faces, roadway connection, temporary windbreak facilities, local fan switching, main fan frequency adjustment, and disaster-related wind control operations can all change the original airflow direction, upstream and downstream relationships, and bypass influence range. Especially during periods when ventilation reconstruction is not yet stable, different measuring points may simultaneously reflect the airflow status before and after the adjustment, and there may be upload delays between equipment logs and sensor data, making it difficult to maintain consistency in the digital representation of disaster propagation paths within the same decision-making window.
[0003] Existing technologies can express mine entity relationships based on knowledge graphs and conduct risk warnings or emergency response based on multi-source monitoring data. However, in scenarios where ventilation conditions change rapidly, there is still room for further refinement in the determination mechanism of whether the graph propagation relationship is still suitable for current inference. After the ventilation condition is reconstructed, the original roadway adjacency relationship, airflow influence relationship, and disaster propagation relationship are still retained in the graph structure, but their corresponding airflow direction, influence range, and propagation significance may have changed. If directly used for inference of gas migration, smoke diffusion, or dust propagation, it may affect the accuracy of the judgment of risk source and propagation direction. Furthermore, actions such as ventilation increase, spraying, power outage, closure, or local isolation will change the observation conditions. Normal readings, no alarm results, or a decrease in risk score may be due to dilution, shielding, sampling interruption, or temporary unobservability of the area. If such results are directly used as risk removal evidence and written back to the knowledge graph, it may affect the reliability of action effect samples and control recommendations in subsequent similar events. Therefore, it is necessary to provide a data processing method that can determine the qualification of graph propagation edge inference within the semi-stable window of ventilation reconstruction and determine whether normal observation results have the qualification of risk removal evidence after the action. Summary of the Invention
[0004] This application proposes a knowledge graph-driven method for integrated control and management of multiple disasters and disaster prevention, in order to address the problems mentioned in the background technology.
[0005] To achieve the above objectives, this application adopts the following technical solution: a knowledge graph-driven method for integrated control and management of multiple disasters, comprising the following steps: S1. Obtain multi-source status data of ventilation, fire prevention, and gas control. Use versioned attribute processing of ventilation status to convert the multi-source status data of ventilation, fire prevention, and gas control into status records of ventilation, fire prevention, and gas control objects indexed by common decision time, ventilation status version and smallest roadway spatial unit. Based on the status records of ventilation, fire prevention, and gas control objects, construct a multi-hazard coupling diagram of ventilation status constraints. In the multi-hazard coupling diagram of ventilation status constraints, form candidate relationships for disaster propagation, relationship for measurement point response and relationship for control action. S2, when the ventilation status record of the three-proof object represents the ventilation status from a stable state to a reconstructed state, within the semi-stable window of ventilation reconstruction, the candidate relationship for disaster propagation is processed for propagation qualification based on the ventilation status version and the measurement point response succession relationship, and a reasonable propagation relationship is generated. S3, along the reasonable propagation relationship, takes over the candidate nodes of disaster events to generate a multi-hazard coupling risk chain. After the control action is executed, the normal observation results after the control action are assigned to the multi-hazard coupling risk chain according to the relationship of the control action. The evidence of permission to release is generated through the elimination of control disturbances and the verification of chain fallback. S4, based on the multi-hazard coupling risk chain, inferable propagation relationship and permitted release evidence, implements control and permission gates, outputs multi-hazard coupling control and decision results, and prohibits normal observation results after handling that have not formed permitted release evidence or have not formed a multi-hazard coupling control and decision result that allows risk release from entering the risk release write-back and action effect sample update.
[0006] Furthermore, the specific operation of the ventilation status versioning and attribution process is as follows: Based on the upload time, sampling period, and communication link verification record of the data record, the maximum allowable delay time is formed. Data records exceeding the maximum allowable delay time are excluded from the visible range at the time of public decision-making. For data records within the visible range, dimensionless conversion is performed according to the stable center value, stable fluctuation scale determined by the historical stable record under the same working condition, and the zero-prevention lower limit determined by the minimum resolution of the measuring point sensor to form a standardized observation description. The stable range under the same working condition is formed by the stable center value and stable fluctuation scale. The standardized observation description is assigned to the smallest roadway spatial unit according to the roadway topology and measuring point coverage boundary to form the status record of the ventilation and three-proof object.
[0007] Furthermore, the specific operation of the ventilation state constraint multi-hazard coupling diagram is as follows: using the smallest roadway spatial unit as the graph index, the ventilation state version is written into the graph index, the propagation start point and propagation end point of the disaster propagation candidate relationship are determined according to the roadway connection direction, the direction driving basis of the disaster propagation candidate relationship is formed according to the ventilation direction benchmark and the pressure difference symbol benchmark, and the measurement point response acceptance relationship formed by the first effective deviation time of the measurement point is bound to the disaster propagation candidate relationship. The action type, effective time, spatial action boundary and expected impact direction of the control action are bound to the smallest roadway spatial unit to form the control action action relationship.
[0008] Furthermore, the specific operation of the ventilation reconstruction semi-stable window is as follows: when the status record of the ventilation and three-proof object shows that at least one of the ventilation direction reference, differential pressure symbol reference, and ventilation facility status has changed relative to the previous ventilation status version, a ventilation reconstruction semi-stable window is generated. During the effective period of the ventilation reconstruction semi-stable window, the disaster propagation candidate relationship is prohibited from being used directly as a propagation relationship that can be inferred. If the ventilation direction reference and differential pressure symbol reference do not change again within three to six consecutive effective confirmation cycles, the effective confirmation cycle is counted in units of the public decision time, and the ventilation direction reference, differential pressure symbol reference, and ventilation facility status all have data records within the visible range at the corresponding public decision time, and the standardized observation description corresponding to the air volume change is within the stable range of the same working condition, the ventilation reconstruction semi-stable window ends. The three to six consecutive effective confirmation cycles are calibrated by the historical ventilation adjustment record and are subject to the mine operation procedures.
[0009] Furthermore, the specific operation of the propagation qualification access processing is as follows: within the ventilation reconstruction semi-stable window, read the directional driving state and the first effective deviation time of the measuring point at both ends of the disaster propagation candidate relationship. Take the measuring point on the propagation starting point side as the upstream measuring point and the measuring point on the propagation ending point side as the downstream measuring point. Verify whether the disaster propagation candidate relationship conforms to the airflow influence direction under the current ventilation state version, and verify whether the first effective deviation of the upstream measuring point is earlier than the first effective deviation of the downstream measuring point and the deviation interval does not exceed the maximum allowable response lag. The maximum allowable response lag is calibrated by the historical similar ventilation state record and is limited by the mine safety regulations. The disaster propagation candidate relationship that passes the verification is written into the inferable propagation relationship. When the disaster propagation candidate relationship conflicts with the current airflow influence direction and the conflict basis comes from the data record that can be read at the time of public decision, write the suspended propagation relationship. When the response acceptance requirement is not met, write the candidate review relationship.
[0010] Furthermore, candidate relationships for disaster propagation include gas migration relationships, dust propagation relationships, and smoke diffusion relationships; The specific operation for verifying candidate relationships of disaster propagation is as follows: For gas migration and dust propagation relationships, verify the consistency between the propagation direction and the current airflow influence direction; for smoke diffusion relationships, based on the verification of the consistency of the propagation direction, form a temperature gradient according to the temperature components in the temperature and humidity observation records, and form the response sequence according to the smoke response data. When both the temperature gradient direction and the smoke response sequence are consistent with the current airflow influence direction, write the smoke diffusion relationship into the inferable propagation relationship; when either the temperature gradient direction or the smoke response sequence is inconsistent with the current airflow influence direction, write the smoke diffusion relationship into the candidate verification relationship.
[0011] Furthermore, the specific operation for generating a multi-hazard coupled risk chain is as follows: read the smallest roadway spatial unit connected by the inferable propagation relationship, take the candidate nodes of disaster events whose standardized observation descriptions exceed the stable range of the same working condition under the same ventilation state version as the chain receiving objects, and form the abnormal deviation direction of the candidate nodes of disaster events with the deviation direction of the standardized observation descriptions relative to the stable center value. When the abnormal deviation direction of the candidate nodes of disaster events is consistent with the direction of the inferable propagation relationship, the candidate nodes of disaster events are connected to the same multi-hazard coupled risk chain. The candidate verification relationship does not enter the main propagation path of the current multi-hazard coupled risk chain, and the paused propagation relationship does not participate in the generation of the current multi-hazard coupled risk chain.
[0012] Furthermore, the specific operation of attributing the post-treatment normal observation results to the multi-hazard coupling risk chain based on the relationship of control actions is as follows: After the control action reaches the effective time, the post-treatment observation window is opened. The shortest observation confirmation time is determined based on the action type, the sampling cycle of the measuring points, and the historical treatment review records. The post-treatment observation records within the coverage of the control action relationship are read. At the end of the post-treatment observation window, when the post-treatment observation records within two consecutive public decision times are both within the visible range of the public decision time, both within the stable range of the same working conditions, and have not triggered the abnormal reporting rules determined by the mine operation procedures, a post-treatment normal observation result is formed. When the sampling cycle of the measuring points is longer than the public decision cycle, the two consecutive visible measuring point records at the end of the post-treatment observation window are used as the judgment object. When a new ventilation status version reconstruction occurs within the post-treatment observation window, the post-treatment normal observation result is written into the candidate review negative evidence. The candidate review negative evidence does not enter the risk relief write-back and action effect sample update.
[0013] Furthermore, the specific operations for handling disturbance elimination and chain fallback verification are as follows: Disturbances are identified based on the relationship between control actions. Situations where the standardized observation description of wind speed after a booster action exceeds the stable range under the same operating conditions, or where measuring points within the control action's coverage area are within the stable range under the same operating conditions but do not meet the chain fallback verification conditions, are identified as dilution disturbances. Situations where the standardized observation description of humidity after a spraying action exceeds the stable range under the same operating conditions, or where measuring points with visible records among smoke and dust measuring points are within the stable range under the same operating conditions but do not meet the chain fallback verification conditions, are identified as shielding disturbances. Sampling interruption in the target area after a closure / isolation action is identified as an isolation unobservable disturbance. Situations where the generation after a power outage action is identified as... Production source status cutoff is identified as a production source change disturbance. Power outages, communication interruptions, and sampling link interruptions at measurement points after a power outage are identified as acquisition link interruption disturbances. If normal observation results after the intervention are identified as intervention disturbances, they are not included in the evidence of permitted release. If normal observation results after the intervention are not identified as intervention disturbances and both upstream and downstream measurement points are within the same stable range under the same working conditions within the same post-intervention observation window, and the consistency of similar measurement points meets the minimum support requirements, the normal observation results after the intervention are included in the evidence of permitted release. The minimum support requirements are calibrated by the measurement point deployment record and are subject to the mine operation procedures. If the consistency of similar measurement points does not meet the minimum support requirements, the normal observation results after the intervention are included in the candidate negative evidence for review.
[0014] Furthermore, the specific operation of the control and permission gate is as follows: the multi-hazard coupling risk chain, inferable propagation relationship, evidence of permitted release, and control action relationship are uniformly converted to the smallest roadway space unit, while maintaining the original ventilation status version, public decision time, and post-disposal observation window source for each of the multi-hazard coupling risk chain, inferable propagation relationship, evidence of permitted release, and control action relationship. Records under different public decision times, different ventilation status versions, or different spatial action boundaries are not merged into the same release judgment object; the key roadway space unit is formed by the unit where the disaster event candidate node is located, the endpoint unit of the inferable propagation relationship, and the control action coverage unit, and the main propagation path is formed by the inferable propagation relationship that connects the disaster event candidate nodes; Specifically, the units requiring evidence to be removed in the key roadway spatial unit are formed by the unit where the candidate node of the disaster event is located and the endpoint unit of the inferable propagation relationship. The action coverage unit in the key roadway spatial unit is formed by the control action coverage unit. When the inferable propagation relationship covers the main propagation path, the permitted removal evidence covers the unit requiring evidence to be removed in the key roadway spatial unit, and the control action relationship covers the action coverage unit in the key roadway spatial unit, a multi-hazard coupled control decision result allowing risk removal is generated. When the inferable propagation relationship does not cover the main propagation path, a continued control result is generated. When the inferable propagation relationship covers the main propagation path and the permitted removal evidence does not cover the unit requiring evidence to be removed in the key roadway spatial unit, a review decision result is generated. When the inferable propagation relationship covers the main propagation path, the permitted removal evidence covers the unit requiring evidence to be removed in the key roadway spatial unit, and the control action relationship does not cover the action coverage unit in the key roadway spatial unit, a continued control result is generated. When generating the continued control result and the review decision result, normal observation results after disposal are prohibited from entering the risk removal write-back and action effect sample update.
[0015] The beneficial effects of this invention are as follows: 1. This invention converts multi-source status data of ventilation, air conditioning, and fire prevention into a status record of the ventilation, air conditioning, and fire prevention object, indexed by a common decision-making time, ventilation status version, and minimum roadway spatial unit. Simultaneously, in the ventilation status-constrained multi-hazard coupling diagram, it constrains the candidate relationships of disaster propagation, the relationship of measurement point response, and the relationship of control actions. This prevents changes in airflow direction, the first deviation of measurement points, and the boundary of control actions from being judged separately as independent monitoring results. Instead, they are confined to the same ventilation status version and the same minimum roadway spatial unit for subsequent reasoning. This solves the problem of mixing observation records, propagation direction, and response actions in terms of time version before and after ventilation system adjustments. It also ensures that subsequent disaster propagation judgments no longer rely on static roadway connectivity relationships but have an object-oriented reasoning basis with the current ventilation status as the boundary.
[0016] 2. This invention sets a semi-stable window for ventilation reconstruction when the ventilation state transitions from a stable state to a reconstruction state. Within this window, based on the ventilation state version and the measurement point response sequence, it performs propagation qualification access processing on disaster propagation candidate relationships. This further distinguishes spatial candidate relationships that might otherwise be directly used for risk chain generation into inferable propagation relationships, candidate verification relationships, and suspended propagation relationships. Before a consistent sequence is formed in the airflow direction, pressure difference sign, ventilation facility status, and measurement point response order, this processing prevents disaster propagation candidate relationships from entering the main propagation path of the multi-hazard coupling risk chain. This solves the problem that the disaster propagation direction is easily misidentified by the static diagram structure during the ventilation reconstruction stage, ensuring that the multi-hazard coupling risk chain is formed only along relationships that are qualified for propagation under the current ventilation state version, thereby improving the technical credibility of the chain source.
[0017] 3. This invention addresses the issue of misinterpreting normal post-treatment observations by sequentially limiting them to the relationships between control actions, multi-hazard coupling risk chains, post-treatment observation windows, and key roadway spatial units. This involves eliminating disturbances caused by the control actions, verifying the chain's return to normal operation, and implementing control permission gating. This prevents post-treatment phenomena such as increased ventilation and dilution, spray masking, unobservable isolation, changes in production sources, and interruptions in the data acquisition chain from entering risk clearance simply because the measurement point readings are normal. This solves the problem in existing control processes where localized normal observations, cross-version clearance evidence, or control actions not covering key roadway spatial units are easily written into risk clearance write-back and action effect sample updates. It achieves gating closure between permitted clearance evidence, risk clearance records, and action effect samples, ensuring that subsequent updates of control action effect samples have clear propagation eligibility constraints, clearance evidence constraints, and action coverage constraints. Attached Figure Description
[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort: Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a flowchart of the S2 process of the present invention. Detailed Implementation
[0019] 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.
[0020] like Figure 1 and Figure 2 As shown, this invention discloses a knowledge graph-driven method for integrated control and management of multiple disasters, including the following specific steps: In S1, the multi-source status data of ventilation, gas, fire prevention and control required for mine ventilation, gas, fire prevention and control are acquired, and the ventilation status versioning and attribution processing is performed on the multi-source status data of ventilation, gas, fire prevention and control. The multi-source status data of ventilation, gas, fire prevention and control with different sampling periods, different spatial locations and different data sources are converted into the status records of ventilation, gas, fire prevention and control objects indexed by the common decision time, ventilation status version and the smallest roadway spatial unit. Through this processing, it is possible to avoid the situation where the ventilation status has entered the reconstruction state but the historical measurement point records, ventilation facility records and control action records are still mixed in the same risk reasoning process, and to provide a unified data foundation for the subsequent disaster propagation candidate relationship propagation qualification access processing.
[0021] The multi-source status data for ventilation, dust control, and gas prevention includes mine roadway topology data, measuring point layout data, gas observation data, wind speed observation data, wind direction observation data, differential pressure observation data, temperature and humidity observation data, smoke response data, dust observation data, damper status data, fan operation data, local ventilation facility status data, control action execution records, and scheduling records. Mine roadway topology data is used to determine the connectivity between the smallest roadway spatial units and, when cross-sectional records are included, for the conversion of airflow change status. Measuring point layout data is used to determine the smallest roadway spatial unit covered by the measuring points. Damper status data, fan operation data, and local ventilation facility status data are used to determine the ventilation status version. Control action execution records are used to determine the action type, effective time, spatial boundary, and expected direction of influence of control actions. Scheduling records are used to verify the consistency of control action execution records, maintenance records, and work switching records, and are used for historical calibration. The humidity component in the temperature and humidity observation data is used to identify the observation impact of spray, water mist, and high humidity environments on smoke and dust measuring points, and is not directly used as a basis for judging the magnitude of disaster risk.
[0022] The public decision-making time is determined according to a unified public decision-making cycle. The public decision-making cycle is determined based on the mine's monitoring sampling capacity, ventilation and dust control scheduling requirements, and stable communication link records, with a range of 10 to 60 seconds. For areas sensitive to changes in gas, smoke, dust, and airflow, the cycle is 10 to 30 seconds. For areas with stable ventilation and used only for routine monitoring, the cycle is 30 to 60 seconds. The same public decision-making cycle is used for the same decision-making chain. When the same multi-hazard coupling risk chain may cross different monitoring sensitive areas, the public decision-making cycle is determined by the area with the highest response demand in the areas involved in the chain. Different data sources are judged whether the data record is within the visible range of the public decision-making time according to the maximum allowable delay time.
[0023] For each data record, the upload time, sampling period, and communication link verification record are read to determine the maximum allowable delay time. This maximum allowable delay time is based on the equipment sampling period, corrected by the communication link verification record, and is subject to mine operation regulations. For gas observation data, wind speed observation data, wind direction observation data, and differential pressure observation data, the maximum allowable delay time is 2 to 3 times the corresponding sampling period, but not exceeding 60 seconds. For temperature and humidity observation data, smoke response data, and dust observation data, the maximum allowable delay time is 2 to 4 times the corresponding sampling period, but not exceeding 180 seconds. For damper status data, fan operation data, and local ventilation facility status data, the maximum allowable delay time is 30 seconds. For control action execution records, the maximum allowable delay time is 30 to 120 seconds. This maximum allowable delay time is only used to determine whether the control action execution record is within the visible range of the public decision-making time. The effective time of the control action is based on the action confirmation time in the control action execution record. For dispatch records, the maximum allowable delay time is 1 to 5 minutes. Dispatch records do not replace the control action execution record in determining whether the action is effective at the current public decision-making time. If a data record exceeds the maximum allowable delay time, the data record will be excluded from the visible range of the public decision-making time. Excluded data records will not participate in the ventilation status versioning, dissemination qualification access processing, and evidence release processing under the current public decision-making time.
[0024] For data records within the visible range at the time of public decision-making, dimensionless conversion is performed. The historical stable data under the same working conditions for the smallest roadway spatial unit containing the data record is retrieved. The stable center value and stable fluctuation scale are determined from the historical stable data under the same working conditions. "Same working conditions" refers to working conditions where the ventilation facility status, operational status, and measuring point layout remain consistent. The historical stable data under the same working conditions are data from the most recent 7 to 30 days that have been confirmed by the mine operation records as having no abnormal reports, no control actions implemented, and stable ventilation status. For gas observation data, wind speed observation data, and differential pressure observation data with high sampling frequencies, priority is given to… First, take data from 7 to 14 days. For temperature and humidity observation data, smoke response data, and dust observation data with low sampling frequency, take data from 14 to 30 days. When there are no less than 30 sets of historical stable records under the same working conditions, the stability center value is formed by the median value of the historical stable records under the same working conditions, and the stability fluctuation scale is formed by converting the interquartile range of the historical stable records under the same working conditions. When there are fewer than 30 sets of historical stable records under the same working conditions, the initial stability center value and the initial stability fluctuation scale are determined by the mine operation procedures, sensor calibration records, and the most recent ventilation commissioning record, and are corrected after the subsequent stable operation records reach 30 sets.
[0025] The zero-limit is determined by the minimum resolution of the sensor at the corresponding measuring point. The zero-limit for gas concentration measuring points is 0.01%CH4; for wind speed measuring points it is 0.01m / s to 0.05m / s; for differential pressure measuring points it is 1Pa to 5Pa; for temperature measuring points it is 0.1℃; and for humidity measuring points it is 1%RH. The zero-limit for smoke response data and dust observation data is determined according to the minimum resolution in the sensor protocol. The deviation between the current data record and the stable center value is converted according to the scale after superimposing the zero-limit on the stable fluctuation scale to form a standardized observation description. The standardized observation description retains the deviation direction and deviation magnitude. The deviation direction is used to judge the abnormal deviation direction of the candidate nodes of disaster events, and the deviation magnitude is used to judge whether the data record is within the stable range of the same operating conditions.
[0026] The stable range under the same working conditions is formed by using the stable center value and the stable fluctuation scale. The stable range under the same working conditions is then converted into a dimensionless judgment boundary for the standardized observation description. Subsequent comparisons are based on whether the deviation of the standardized observation description falls within this dimensionless judgment boundary. The stable range under the same working conditions is determined with the stable center value as the center and according to 1.5 to 3 times the stable fluctuation scale. Gas observation data, smoke response data, and dust observation data are taken at 2 to 3 times the stable fluctuation scale. Wind speed observation data and pressure difference observation data are taken at 1.5 to 2.5 times the stable fluctuation scale. The values are based on historical stable records under the same working conditions, mine operation procedures, and sensor calibration records. Wind direction observation data are projected according to the direction of roadway connection of the smallest roadway spatial unit where the measuring point is located to form a ventilation direction benchmark with forward, reverse, or no effective direction. This benchmark is not used as a regular continuous value and is not averaged with other observation data.
[0027] After dimensionless conversion, standardized observation descriptions are assigned to the smallest roadway spatial unit based on the roadway topology and the measurement point coverage boundary, forming a status record of the ventilation and fire prevention object. The smallest roadway spatial unit is jointly determined by the roadway topology, the functional boundary of the ventilation facility, and the measurement point coverage range. The roadway segment between adjacent ventilation nodes, air doors, air duct outlets, roadway intersections, and measurement point coverage boundaries is used as the basis for division. When a measurement point covers multiple smallest roadway spatial units, the standardized observation description of the measurement point is associated with the multiple covered smallest roadway spatial units according to the coverage range and roadway connection direction in the measurement point layout record, and the measurement point coverage boundary is retained. When the same measurement point record is associated with multiple smallest roadway spatial units, it is only used as evidence of the same measurement point coverage and is not split into multiple independent measurement point evidences. When multiple measurement points cover the same smallest roadway spatial unit, the standardized observation description and the first effective deviation time of each measurement point are retained separately, and the results of multiple measurement points are not directly averaged.
[0028] The ventilation status version is determined based on the status records of the ventilation, fire prevention, and gas control objects. The ventilation status version is primarily classified according to the ventilation direction reference, differential pressure sign reference, and ventilation facility status. The standardized observation description corresponding to airflow changes is used as the basis for confirming the stable state of the version. The ventilation direction reference is derived from wind direction observation data and roadway topology; the differential pressure sign reference is derived from differential pressure observation data between adjacent smallest roadway spatial units; the ventilation facility status is derived from damper status data, fan operation data, and local ventilation facility status data. When the mine roadway topology data includes cross-sectional records, the standardized observation description corresponding to airflow changes is formed by converting wind speed observation data with cross-sectional records. When the mine roadway topology data does not include cross-sectional records, the airflow change status is formed by the standardized observation description corresponding to wind speed observation data, based on airflow direction combinations, differential pressure sign combinations, damper opening / closing status, fan operating speed, and local ventilation... The ventilation status version is generated based on the facility start / stop status. When the directional relationship given by wind direction observation data and differential pressure observation data for the same smallest roadway spatial unit is inconsistent, or when the ventilation facility status data is not within the visible range at the time of public decision-making, the ventilation status version is not directly confirmed based on the conflict record. Instead, the affected disaster propagation candidate relationship is retained as an object that needs to be processed for propagation eligibility in S2. If the historical ventilation adjustment record contains no less than 10 similar ventilation adjustment events and covers the current relevant ventilation facility combination, the version division boundary is calibrated using the historical ventilation adjustment record. If the historical ventilation adjustment record contains fewer than 10 similar ventilation adjustment events, or does not cover the current relevant ventilation facility combination, the mine operation procedures and the most recent ventilation commissioning record are used as the basis for version division. The ventilation status version is only used for data attribution and subsequent propagation eligibility processing, and is not directly used as the result of judging the magnitude of disaster risk.
[0029] After obtaining the status records of the three-prevention objects, a ventilation status-constrained multi-hazard coupling diagram is constructed. The ventilation status-constrained multi-hazard coupling diagram uses the smallest roadway spatial unit as the graph index. The ventilation status version is written into the graph index, so that the measurement point observation, ventilation facility status and control actions at the same public decision time are limited to the same ventilation status version. The propagation start point and propagation end point of the disaster propagation candidate relationship are determined according to the roadway connection direction. The direction driving basis of the disaster propagation candidate relationship is formed according to the ventilation direction benchmark and the pressure difference symbol benchmark. The disaster propagation candidate relationship in S1 only represents the spatial relationship with the possibility of candidate propagation, and does not indicate that it has the qualification to participate in risk reasoning. The reasonable propagation relationship cannot be directly written into S1.
[0030] The ventilation state constraint multi-hazard coupling diagram serves as the knowledge graph representation of this invention. It uses the smallest alleyway spatial unit, disaster event candidate nodes, measurement point records, and control action records as objects in the diagram. It uses disaster propagation candidate relationships, measurement point response acceptance relationships, control action effect relationships, inferable propagation relationships, candidate verification relationships, and paused propagation relationships as relationships in the diagram. The visibility range of objects and relationships in the diagram is limited by the public decision time and ventilation state version.
[0031] The measurement point records covering the propagation start and end points are used to form a measurement point response succession relationship according to the first effective deviation time. The measurement point response succession relationship is then bound to the disaster propagation candidate relationship. The first effective deviation time refers to the measurement point record within the visible range of the public decision time. Its standardized observation description is the public decision time when it first exceeds the stable range of the same working condition. When multiple consecutive public decision times of the same measurement point exceed the stable range of the same working condition, the public decision time when it first exceeds the stable range of the same working condition is taken. The measurement point response succession relationship is used in S2 to determine whether there is a response sequence between the measurement point on the propagation start side and the measurement point on the propagation end side. It is not used to directly determine whether the disaster propagation candidate relationship is established in S1.
[0032] The action type, effective time, spatial boundary, and expected impact direction of the control actions are bound to the smallest roadway spatial unit to form the control action relationship. The action types include ventilation, spraying, closure and isolation, and power outage. The effective time is determined by the control action execution record and is verified for consistency by the scheduling record. The spatial boundary is based on the action object in the control action execution record and is modified by the ventilation facility's range of action and the roadway topology. When the action object is inconsistent with the ventilation facility's range of action, the spatial boundary is not expanded, but the control action relationship is retained as input for the subsequent verification of the normal observation results after the treatment in S3. The expected impact direction is determined by the technical purpose of the control action. The control action relationship is used in subsequent S3 to assign the normal observation results after the treatment to the multi-hazard coupling risk chain and to identify treatment disturbances. S1 does not form normal observation results after the treatment, nor does it form evidence of permitted release, to avoid prematurely writing post-treatment observations that have not yet occurred into the ventilation state constraint multi-hazard coupling diagram.
[0033] Through the aforementioned processing, S1 outputs the status record of the ventilation and three-proof objects and the multi-hazard coupling diagram of ventilation status constraints. The multi-hazard coupling diagram of ventilation status constraints forms the candidate relationship of disaster propagation, the relationship of measurement point response and the relationship of control action. This provides the input basis for S2 to perform the propagation qualification access processing within the ventilation reconstruction semi-stable window, and also provides the reference basis for S3 to attribute the normal observation results after the control action is executed and generate the evidence of permission to be released.
[0034] In S2, the status records of the ventilation and three-proof objects and the multi-hazard coupling diagram of ventilation status constraints output by S1 are read. It is determined whether the ventilation status at the current public decision moment has changed from a stable state to a reconstructed state. When the status records of the ventilation and three-proof objects indicate that the ventilation status has changed from a stable state to a reconstructed state, a ventilation reconstructed semi-stable window is generated. Within the ventilation reconstructed semi-stable window, based on the ventilation status version and the measurement point response succession relationship, the propagation qualification admission processing is performed on the disaster propagation candidate relationship to generate a reasonable propagation relationship. This processing is used to prevent the disaster propagation candidate relationship from continuing to enter the risk reasoning in the propagation direction before the reconstruction after the ventilation status changes.
[0035] Read the status records of the ventilation, fire prevention, and gas control objects at the current public decision-making time, and read the ventilation direction reference, differential pressure sign reference, and ventilation facility status of the corresponding smallest roadway space unit under the previous ventilation status version. If there is no previous ventilation status version, the initial ventilation status version is formed based on the most recent ventilation commissioning record, mine operation procedures, and continuous stable observation records. When at least one of the ventilation direction reference, differential pressure sign reference, and ventilation facility status at the current public decision-making time changes relative to the previous ventilation status version, and the change is based on data records within the visible range at the public decision-making time, the ventilation status is determined to enter the reconstruction state from the stable state, and a ventilation reconstruction semi-stable window is generated. When the change is based on data delay, data missing, or directional relationship conflict, the reconstruction state is not directly confirmed, but the affected disaster propagation candidate relationship is retained as an object that needs to be processed for propagation qualification access.
[0036] The starting point of the ventilation reconfiguration semi-stable window is the public decision moment when the ventilation state is first determined to have transitioned from a stable state to a reconfigured state. During the effective period of the ventilation reconfiguration semi-stable window, disaster propagation candidate relationships cannot be directly used as inferable propagation relationships. The window ends when: within 3 to 6 consecutive effective confirmation periods, the ventilation direction reference and differential pressure sign reference have not changed again, and the standardized observation description corresponding to the air volume change is within the stable range of the same operating condition. The effective confirmation period is counted in units of the public decision moment, referring to the confirmation period in which the ventilation direction reference, differential pressure sign reference, and ventilation facility status all have data records within the visible range of the public decision moment. The 3 to 6 consecutive effective confirmation periods are calibrated by historical ventilation adjustment records and are limited by the mine operation procedures. In areas with frequent ventilation facility adjustments, local ventilation facility switching, or complex roadway connectivity, the nearest 6 effective confirmation periods are taken. In areas with stable ventilation facility status, simple roadway connectivity, and sufficient historical ventilation adjustment records, the nearest 3 effective confirmation periods are taken. When historical ventilation adjustment records are insufficient, the number of effective confirmation periods is determined by the mine operation procedures and the most recent ventilation commissioning record.
[0037] Within the semi-stable window of ventilation reconstruction, candidate disaster propagation relationships in the multi-hazard coupling diagram constrained by the ventilation state are read. For each candidate disaster propagation relationship, the propagation start point, propagation end point, direction driving basis, and measurement point response inheritance relationship are read. The direction driving basis is determined by the ventilation direction reference and pressure difference symbol reference formed by S1. The measurement point response inheritance relationship is formed by the measurement point records covering the propagation start point and propagation end point according to the first effective deviation time. The current airflow influence direction is first determined based on the current ventilation state version. When the propagation direction of the candidate disaster propagation relationship is consistent with the current airflow influence direction, the measurement point on the propagation start point side is taken as the upstream measurement point, and the measurement point on the propagation end point side is taken as the downstream measurement point. When the propagation direction of the candidate disaster propagation relationship is inconsistent with the current airflow influence direction, the upstream and downstream measurement point correspondence is not used to directly verify the measurement point response inheritance relationship. This processing is used to avoid continuing to use the upstream and downstream measurement point correspondence before reconstruction after the ventilation state is reconstructed.
[0038] Propagation qualification access processing is performed on disaster propagation candidate relationships. This process includes propagation direction verification and measurement point response acceptance relationship verification. When performing propagation direction verification, the directional driving states at both ends of the disaster propagation candidate relationship are read. The directional driving states are jointly formed by the airflow direction, pressure difference sign, and ventilation facility status under the current ventilation status version. When the propagation direction of the disaster propagation candidate relationship is consistent with the current airflow influence direction, the pressure difference sign benchmark between the propagation start point and the propagation end point is consistent with the pressure difference driving direction, and the relevant ventilation facility status does not block the propagation direction, the propagation direction verification passes. When the disaster propagation candidate relationship conflicts with the current airflow influence direction, and the conflict is based on data records within the visible range at the time of public decision-making, the disaster propagation candidate relationship is written into the suspended propagation relationship. When the airflow direction, pressure difference sign, or ventilation facility status used for propagation direction verification has data delays, missing data, or directional relationship conflicts that are insufficient to confirm the suspended propagation relationship, the disaster propagation candidate relationship is written into the candidate review relationship.
[0039] When verifying the connection between measuring point responses, the first effective deviation times of the upstream and downstream measuring points are read. It is determined whether the first effective deviation of the upstream measuring point is earlier than that of the downstream measuring point, and whether the deviation interval between the upstream and downstream measuring points does not exceed the maximum permissible response lag. The maximum permissible response lag is calibrated based on historical records of similar ventilation conditions and is subject to mine safety regulations. First, a transmission reference time is formed based on the roadway distance from the propagation start point to the propagation end point and the minimum effective wind speed under the same working conditions. Then, the transmission reference time is converted into the number of response lag cycles according to the common decision-making cycle, and corrected in conjunction with the measuring point sampling cycle. When the minimum effective wind speed under the same working conditions is lower than the minimum effective wind speed limit determined by the mine operation regulations, this transmission reference time is not used. Long-term confirmed propagation relationships are used to write corresponding disaster propagation candidate relationships into the candidate verification relationship. The maximum permissible response lag is usually 2 to 10 public decision cycles. When the roadway distance is long, the minimum effective wind speed under the same working conditions is close to the lower limit of the minimum effective wind speed determined by the mine operation regulations, and the historical similar ventilation status records cover the same roadway type, the maximum permissible response lag is 10 to 15 public decision cycles, with the mine safety regulations as the upper limit. If this condition is not met, disaster propagation candidate relationships with a response lag of more than 10 public decision cycles are written into the candidate verification relationship. When the historical similar ventilation status records are insufficient, the conservative upper limit determined by the mine safety regulations is adopted, and disaster propagation candidate relationships with insufficient response are written into the candidate verification relationship.
[0040] When the propagation direction verification passes and the measurement point response acceptance verification passes, the corresponding disaster propagation candidate relationship is written into the inferable propagation relationship. When the propagation direction verification does not form a clear conflict, but the measurement point response acceptance relationship is insufficient, the upstream or downstream measurement point lacks a measurement point record within the visible range of the public decision-making time, or the first effective deviation time cannot form a valid sequence, the disaster propagation candidate relationship is written into the candidate verification relationship. When the disaster propagation candidate relationship appears for the first time at the current public decision-making time, the propagation qualification admission process is performed on the disaster propagation candidate relationship, and it is not directly written into the inferable propagation relationship because the relationship has a candidate connection in the ventilation state constraint multi-hazard coupling diagram.
[0041] Candidate relationships for disaster propagation include gas migration, dust propagation, and smoke diffusion. For gas migration and dust propagation, the consistency between the propagation direction and the current airflow influence direction is verified. The verification content includes airflow direction, pressure difference sign, and ventilation facility status. Gas migration and dust propagation relationships will only proceed to the measurement point response succession relationship verification if the propagation direction is consistent with the current airflow influence direction, and the pressure difference sign reference and ventilation facility status both support the propagation direction.
[0042] For smoke diffusion relationships, based on the consistency verification of propagation direction, temperature components from temperature and humidity observation records within the visible range and belonging to the same ventilation state version at the same public decision time are read to form a temperature gradient. Smoke response data is then read to form the response sequence. The temperature gradient is determined by the difference in temperature components between the propagation start point and the propagation end point. If the temperature component difference does not exceed the lower limit of zero at the temperature measuring point, the temperature gradient is not used as a basis to support the current airflow influence direction. The smoke response sequence is determined by the first effective deviation time of the smoke measuring points at the propagation start point and the propagation end point. If the smoke measuring points at the propagation start point and the propagation end point form the first effective deviation at the same public decision time, it is not considered to have a response sequence. When the smoke diffusion relationship has passed the propagation direction verification, and both the temperature gradient direction and the smoke response sequence are consistent with the current airflow influence direction, the smoke diffusion relationship is written into the inferable propagation relationship. When either the temperature gradient direction or the smoke response sequence is inconsistent with the current airflow influence direction, or if either the smoke measuring point at the propagation start point or the smoke measuring point at the propagation end point cannot form the first effective deviation time, the smoke diffusion relationship is written into the candidate verification relationship.
[0043] At the end of S2, the propagation qualification access processing results are output. The propagation qualification access processing results include inferable propagation relationships, candidate verification relationships, and suspended propagation relationships. Inferable propagation relationships enter the subsequent S3 to receive candidate nodes of disaster events and generate multi-hazard coupling risk chains. Candidate verification relationships do not enter the main propagation path of the current multi-hazard coupling risk chain. Suspended propagation relationships do not participate in the generation of the current multi-hazard coupling risk chain. S2 does not generate normal observation results after disposal, does not generate evidence of permission to be lifted, and does not perform risk lifting write-back and action effect sample update.
[0044] In S3, the ventilation state constraint multi-hazard coupling diagram formed by S1 and the propagation qualification admission processing results output by S2 are read. The candidate nodes of disaster events are taken along the inferable propagation relationship to generate a multi-hazard coupling risk chain. After the control action is executed, the normal observation results after the control action are assigned to the multi-hazard coupling risk chain according to the relationship of the control action. The evidence of permission to release is generated through the elimination of disturbances and the verification of chain fallback. This process is used to prevent the normal observation results after the control action from being used as the basis for risk release without the elimination of disturbances and the verification of chain fallback.
[0045] Read the inferable propagation relationship, candidate verification relationship and pause propagation relationship output by S2, and read the disaster event candidate nodes in the ventilation state constraint multi-hazard coupling diagram. The disaster event candidate nodes are derived from the state records of the ventilation and three-proof objects. The standardized observation descriptions of their corresponding measuring points have been assigned to the common decision time, ventilation state version and minimum roadway space unit. Read the minimum roadway space unit connected by the inferable propagation relationship, and find the disaster event candidate nodes whose standardized observation descriptions exceed the stable range of the same working condition under the same ventilation state version. Use the disaster event candidate nodes as the chain receiving objects.
[0046] For each candidate node of a disaster event, the deviation direction relative to the stable center value is described by standardized observations to form an abnormal deviation direction. When the abnormal deviation direction of the candidate node of a disaster event is consistent with the direction of the inferable propagation relationship, and the smallest roadway spatial unit where the candidate node of the disaster event is located can form a spatial connection with the adjacent candidate node of the disaster event along the inferable propagation relationship, the candidate node of the disaster event is connected to the same multi-hazard coupling risk chain. When the closure of the chain requires the support of the candidate verification relationship, the candidate verification relationship does not enter the main propagation path of the current multi-hazard coupling risk chain; when the closure of the chain requires the support of the pause propagation relationship, the pause propagation relationship does not participate in the generation of the current multi-hazard coupling risk chain.
[0047] After generating a multi-hazard coupled risk chain, the relationship between control actions is read. This relationship includes the action type, effective time, spatial boundary, and expected direction of impact. The smallest roadway spatial unit covered by the multi-hazard coupled risk chain is matched with the smallest roadway spatial unit covered by the control action relationship. When the spatial boundary of the control action relationship covers at least one smallest roadway spatial unit in the multi-hazard coupled risk chain, and the action type and expected direction of impact meet the requirements for handling abnormal deviations of candidate nodes of disaster events in the mine operation procedures, the control action relationship is used as the basis for attributing the normal observation results after handling within its spatial boundary. Smallest roadway spatial units not covered by the control action relationship do not form evidence of permitted release due to the control action relationship. Control action relationships that do not meet the handling requirements are not included in the process of judging the release evidence of the multi-hazard coupled risk chain.
[0048] After the control action reaches its effective time, a post-control observation window is opened. The starting point of the post-control observation window is the effective time of the control action, and the ending point is the public decision time after meeting the minimum observation confirmation time. The minimum observation confirmation time is measured in terms of the public decision cycle and is determined by the action type, the sampling cycle of the measuring point, and the historical control review records. It is also subject to the mine operation procedures. The minimum observation confirmation time for ventilation increase actions is 2 to 6 public decision cycles; the minimum observation confirmation time for spraying actions is 3 to 8 public decision cycles; and the minimum observation confirmation time for closure and isolation actions and power outage actions is 5 to 12 public decision cycles. When the sampling cycle is short, the measuring point response is timely, and the historical control review records are sufficient, the value closer to the lower limit is used. When the roadway distance is long, the measuring point sampling cycle is long, and the transmission delay of the control effect is significant, the value closer to the upper limit is used. When the historical control review records are insufficient, the conservative confirmation time determined by the mine operation procedures is adopted, and the post-control observation records with insufficient confirmation are retained as negative evidence for candidate review.
[0049] Within the post-disposal observation window, read the post-disposal observation records within the coverage area of the control action. At the end of the post-disposal observation window, if the post-disposal observation records within two consecutive public decision-making moments are both within the visible range of the public decision-making moment and are both within the stable range of the same working conditions without triggering the abnormal reporting rules determined by the mine operation procedures, a normal post-disposal observation result is formed. The two consecutive public decision-making moments are determined by the sampling period of the measuring points and the historical disposal review records. When the sampling period of the measuring points is longer than the public decision-making period, the two consecutive visible measuring point records at the end of the post-disposal observation window are used as the judgment object. If two consecutive visible records or two consecutive visible measuring point records cannot be formed for two public decision-making moments, no permitted release evidence is generated, and the corresponding post-disposal observation record is written into the candidate review negative evidence. When a new ventilation status version is reconstructed within the post-disposal observation window, the normal post-disposal observation result is written into the candidate review negative evidence.
[0050] After the normal observation results are generated after the treatment, the disturbance elimination is carried out according to the relationship of the control action. For the wind increase action, the standardized observation description corresponding to the wind speed and the post-treatment observation records of the measuring points within the coverage of the control action are read. When the standardized observation description corresponding to the wind speed after the wind increase action exceeds the stable range of the same working condition, the measuring points within the coverage of the control action are within the stable range of the same working condition, and the upstream and downstream measuring points in the same multi-hazard coupling risk chain are not simultaneously within the stable range of the same working condition, the normal observation results after the treatment are identified as a dilution disturbance and are not included in the evidence of permitted release. If the upstream and downstream measuring points in the same multi-hazard coupling risk chain are both within the stable range of the same working condition, the normal observation results after the treatment continue to enter the chain fallback acceptance verification.
[0051] For spraying actions, the standardized observation description corresponding to humidity, the post-treatment observation records of smoke measuring points, and the post-treatment observation records of dust measuring points are read. When the standardized observation description corresponding to humidity after spraying exceeds the stable range under the same operating conditions, and all measuring points with visible records in the smoke and dust measuring points are within the stable range under the same operating conditions, at least one measuring point has a visible record, and the upstream and downstream measuring points in the same multi-hazard coupling risk chain are not simultaneously within the stable range under the same operating conditions, the normal observation results after treatment are identified as masking disturbances and are not included in the evidence of permitted removal. When neither the smoke nor the dust measuring points can form visible records, the post-treatment observation records are included in the candidate negative evidence for review. If both the upstream and downstream measuring points in the same multi-hazard coupling risk chain are within the stable range under the same operating conditions, the normal observation results after treatment continue to enter the chain fallback acceptance verification.
[0052] For the closure and isolation action, the sampling continuity of the target area is read. The sampling continuity of the target area is determined by whether the measurement points in the target area within the observation window after the action form continuous measurement point records within the visible range at the time of public decision. When the measurement points in the target area do not form continuous measurement point records within the visible range at the time of public decision after the closure and isolation action, the corresponding post-action observation records are identified as unobservable disturbances of isolation, are not written into the evidence of permission to lift the restrictions, and are written into the candidate negative evidence for review.
[0053] For power outage actions, the production source status and measurement point acquisition status are read. When the production source status is cut off after the power outage action, the normal observation results after the action are identified as production source change disturbances and are not included in the evidence of permission cancellation. When any of the following conditions are met after the power outage action: measurement point power supply interruption, measurement point communication interruption, and sampling link interruption, the normal observation results after the action are identified as acquisition link interruption disturbances and are not included in the evidence of permission cancellation.
[0054] If the normal observation results after the treatment are not identified as a disturbance, a chain fallback acceptance verification is performed. Based on the upstream and downstream measuring points corresponding to the multi-hazard coupling risk chain and the inferable propagation relationship in S2, the fallback status of the upstream and downstream measuring points within the same post-treatment observation window is read. If both the upstream and downstream measuring points are within the same stable operating condition range, and the fallback status occurs under the same ventilation state version, the chain fallback acceptance verification is considered to have passed. If only the measuring points within the coverage of the control action relationship are within the same stable operating condition range, while the associated measuring points on the same multi-hazard coupling risk chain are not within the same stable operating condition range, then the normal observation results after the treatment will not be written into the permitted release evidence.
[0055] The consistency of similar measuring points is verified. Similar measuring points refer to measuring points of the same type, belonging to the same ventilation state version, and spatially associated with the same multi-hazard coupling risk chain. When the consistency of similar measuring points meets the minimum support requirements, the normal observation results after the treatment are allowed to be written into the evidence of permitted release. The minimum support requirements are calibrated by the measuring point deployment record and are subject to the mine operation procedures. There should be no less than 2 similar measuring points within the visible range at the time of public decision-making, and the proportion of the visible similar measuring points within the stable range of the same working condition should be no less than 70%. The corresponding minimum pass number is determined by multiplying the total number of visible similar measuring points by 70% and then rounding up. If there are less than 2 similar measuring points within the visible range at the time of public decision-making, the normal observation results after the treatment are written into the candidate negative evidence for review.
[0056] When the normal observation results after the treatment are not identified as a treatment disturbance, the chain fallback acceptance verification passes, and the consistency of similar measuring points meets the minimum support requirements, the normal observation results after the treatment will be written into the evidence of permitted release. The evidence of permitted release only indicates that the normal observation results after the treatment are qualified to enter the subsequent S4 execution control permission gate, and is not directly equivalent to the risk release permission result. If the normal observation results after the treatment meet the treatment disturbance identification conditions, the chain fallback acceptance verification is not met, the consistency of similar measuring points does not meet the minimum support requirements, or a new ventilation status version reconstruction occurs within the post-treatment observation window, the normal observation results after the treatment will be written into the candidate review negative evidence.
[0057] At the end of S3, the multi-hazard coupling risk chain, the evidence of permitted removal, and the negative evidence for candidate review are output. The multi-hazard coupling risk chain enters S4 for management permission gating; the evidence of permitted removal enters S4 to determine whether there is evidence to support removal; the negative evidence for candidate review is not used as evidence of permitted removal, and is not included in the risk removal write-back and action effect sample update. S3 does not generate the multi-hazard coupling management decision result that allows risk removal. This result is generated by S4 after completing the management permission gating.
[0058] In S4, the inferable propagation relationship output by S2, the multi-hazard coupling risk chain output by S3, and the permitted release evidence are read. Combined with the control action action relationship formed by S1, control permission gating is performed, and the multi-hazard coupling control decision result is output. This process unifies the propagation path, release evidence, and control action coverage to the smallest roadway space unit before gating judgment. This is used to prevent control actions that are normal after local treatment, cross-version observation results, or do not cover key roadway space units from entering the risk release write-back and action effect sample update.
[0059] The multi-hazard coupling risk chain, inferable propagation relationship, permitted release evidence, and control action effect relationship are all converted to the smallest roadway spatial unit. The multi-hazard coupling risk chain is used to determine the unit where the candidate node of the disaster event is located; the inferable propagation relationship is used to determine the endpoint unit in the propagation path; the permitted release evidence is used to determine the unit covered by the permitted release evidence; and the control action effect relationship is used to determine the unit covered by the control action. This conversion uses the smallest roadway spatial unit formed by S1 and the ventilation status version as the index, without changing the original ventilation status version of each object, and without merging records under different public decision times, different ventilation status versions, or different spatial action boundaries into the same release judgment object.
[0060] The critical roadway spatial unit is formed by the unit where the candidate node of the disaster event is located, the endpoint unit of the inferable propagation relationship, and the control action coverage unit. Among them, the unit in the critical roadway spatial unit that needs to be deactivated is formed by the unit where the candidate node of the disaster event is located and the endpoint unit of the inferable propagation relationship. The action coverage unit in the critical roadway spatial unit is formed by the control action coverage unit corresponding to the control action relationship to be used for control permission gating. When there are multiple control action relationships, a corresponding action coverage unit is formed for each control action relationship. The coverage of the current control action relationship is not supplemented by the coverage of other control action relationships. Through this division, the spatial range that needs to be proven to fall back in the multi-hazard coupling risk chain is judged separately from the spatial range of the actual effect of the control action. This avoids extending the normal observation results after the treatment of a single measuring point within the action coverage area to the basis for deactivation of the entire multi-hazard coupling risk chain.
[0061] The main propagation path is formed by the inferable propagation relationship of candidate nodes of a series of disaster events. The multi-hazard coupling risk chain includes at least two candidate nodes of disaster events that can be connected through inferable propagation relationships. Candidate anomalies that cannot form a node connection relationship are not considered as the main propagation path in the current control and permission gate. The main propagation path is based on the inferable propagation relationship under the same ventilation state version. The starting point and ending point are determined by the smallest alleyway spatial unit where adjacent candidate nodes of disaster events are located in the multi-hazard coupling risk chain. If the connection of the multi-hazard coupling risk chain requires the support of candidate verification relationship, or if there is a pause propagation relationship in the connection, then the connection is not considered as part of the main propagation path. The candidate verification relationship does not enter the main propagation path of the current control and permission gate, and the pause propagation relationship does not participate in the current control and permission gate.
[0062] After determining the key roadway spatial units and the main propagation path, the propagation coverage judgment, the evidence coverage removal judgment, and the control action coverage judgment are performed. The propagation coverage judgment is used to confirm whether the inferable propagation relationship covers the main propagation path. When each adjacent disaster event candidate node in the main propagation path is connected by an inferable propagation relationship, it is determined that the inferable propagation relationship covers the main propagation path. When there is no inferable propagation relationship between any adjacent disaster event candidate nodes, or only a candidate verification relationship or a suspended propagation relationship exists, it is determined that the inferable propagation relationship does not cover the main propagation path.
[0063] The determination of evidence coverage for removal is used to confirm whether the permitted removal evidence covers the units in the key roadway space units that require evidence support for removal. It involves reading the smallest roadway space unit, ventilation status version, post-treatment observation window, and whether it is within the visible range at the time of public decision-making corresponding to the permitted removal evidence. This information is then matched one-to-one with the unit where the candidate node of the disaster event is located and the endpoint unit of the inferable propagation relationship. Only when the permitted removal evidence and the unit requiring evidence support belong to the same ventilation status version, are within the visible range at the corresponding time of public decision-making, originate from the post-treatment observation window corresponding to the control action relationship intended for use in this control permitting, and cover the corresponding unit, is the unit considered to have evidence support for removal. If the permitted removal evidence only covers a local unit within the coverage area of the control action relationship, but does not cover the unit where the candidate node of the disaster event is located or the endpoint unit of the inferable propagation relationship in the multi-hazard coupling risk chain, then the permitted removal evidence is not considered to cover the units in the key roadway space units that require evidence support for removal.
[0064] The control action coverage judgment is used to confirm whether the control action relationship covers the action coverage unit in the critical roadway spatial unit. It reads the spatial action boundary and action coverage unit in the control action relationship and matches them with the action coverage unit in the critical roadway spatial unit. Only when the control action relationship covers the action coverage unit that it should act on, the action coverage unit does not exceed the spatial action boundary of the control action relationship, and the control action relationship and the corresponding multi-hazard coupling risk chain are in the same ventilation state version, is the control action relationship considered to cover the action coverage unit in the critical roadway spatial unit. If the control action relationship does not cover the unit that it should act on, the action coverage unit exceeds the spatial action boundary of the control action relationship, or a new ventilation state version is reconstructed after the action is executed, then the control action relationship is not used as the basis for allowing risk to be lifted.
[0065] Preferably, a control permission coefficient is used to formally determine the control permission threshold. This calculation does not use weighted summation, averaging, or penalty terms, but instead uses binary product gating, making dissemination eligibility, evidence of dissolution, and object closure relationships all necessary conditions for allowing risk dissolution. The control permission coefficient is calculated as follows: ; in, This indicates the number of the multi-hazard coupling risk chain, corresponding to the multi-hazard coupling risk chain generated by S3; This indicates the control action number, which corresponds to the control action relationship formed in S1 and used in S3 to assign the normal observation results after the treatment; This indicates the moment of public decision-making, corresponding to the moment of public decision-making determined by S1; Indicates the number is A multi-hazard coupling risk chain; Indicates a multi-hazard coupled risk chain Control actions At the moment of public decision-making The control and permit coefficient below, This indicates the threshold for dissemination eligibility, indicated by its superscript. This is a gating identifier for propagation eligibility, not an exponential operation; the input source of this gating quantity is the inferable propagation relationship output by S2 and the multi-hazard coupling risk chain output by S3. When the multi-hazard coupling risk chain... The main communication path in all of them is the moment of public decision-making. The value is 1 if the inferable propagation relation is covered, and 0 otherwise. Indicates the amount of evidence removed from the threshold, its superscript To remove the evidence gating flag, it does not represent an exponential operation; the input source for this gating quantity is the permitted removal evidence output by S3 and the key roadway space unit formed in this step, when the permitted removal evidence covers the multi-hazard coupling risk chain. In control actions The corresponding key lane space unit requires the removal of supporting evidence, and the permitted removal evidence is available at the time of public decision-making. The value is 1 if the value is within the visible range, and 0 otherwise. The superscript indicates the gating quantity of the object. This is a closed-loop gating identifier for an object, not representing exponential operations; the input sources for this gating quantity are multi-hazard coupling risk chains, the relationship between control actions, and evidence of permitted deregulation. Control actions The corresponding control actions and the evidence of permitted release are relevant in the smallest alleyway space unit, ventilation status version, and public decision-making time. The following forms a closed correspondence of the same object, and the evidence of permission revocation originates from the control action. The value is 1 for the corresponding post-treatment observation window, and 0 otherwise.
[0066] The aforementioned three gating quantities are all dimensionless binary quantities, with values ranging from 0 to 1, and do not involve denominators, normalization boundaries, or summation ranges. When multiple multi-hazard coupled risk chains or multiple control action relationships exist at the same public decision-making moment, the control permission coefficient is calculated separately for each multi-hazard coupled risk chain and each related control action relationship. Cross-chain or cross-action averaging is not performed. When any gating quantity is 0, the control permission coefficient is 0, and a multi-hazard coupled control decision result allowing risk removal is not generated. Instead, a continued control result or a review decision result is generated based on the unmet gating conditions. The object closure gating quantity is only used to represent the spatial, version, temporal, and evidence source closure relationship between multi-hazard coupled risk chains, control action relationships, and permitted removal evidence. It does not repeatedly judge whether the removal evidence coverage is sufficient, nor is it used as an independent business output result.
[0067] When the inferable propagation relationship covers the main propagation path, the permitted release evidence covers the units in the key lane space unit that require release evidence support, and the control action relationship covers the action coverage units in the key lane space unit, a multi-hazard coupling control decision result that allows risk release is generated. The multi-hazard coupling control decision result that allows risk release indicates that the multi-hazard coupling risk chain has the conditions for release permission at the current public decision moment. Subsequent risk release write-back and action effect sample update are still subject to the write-back restriction conditions of this step.
[0068] When the inferable propagation relationship does not cover the main propagation path, a continued control result is generated. The continued control result is used to maintain the current control status of the multi-hazard coupling risk chain. Normal observation results after disposal are not allowed to be included in the risk removal write-back and action effect sample update. When the inferable propagation relationship covers the main propagation path, but the permitted removal evidence does not cover the units in the key roadway space unit that require removal evidence support, a review decision result is generated. The review decision result is used to indicate that the key roadway space unit not covered by the permitted removal evidence needs to continue to be reviewed and observed. Normal observation results after disposal without the formation of permitted removal evidence are not allowed to be used as the basis for removal. When the inferable propagation relationship covers the main propagation path and the permitted removal evidence covers the units in the key roadway space unit that require removal evidence support, but the control action effect relationship does not cover the action coverage units in the key roadway space unit, a continued control result is generated to avoid writing the area that is not actually affected by the control action into the action effect sample update.
[0069] Regarding the restrictions on writing back risk clearance, risk clearance records are only allowed to be written into the ventilation status constraint multi-hazard coupling diagram when a multi-hazard coupling control decision result that allows risk clearance is generated. The risk clearance record includes the risk chain identifier, ventilation status version identifier, inferable propagation relationship, evidence of permitted clearance, control action effect relationship, corresponding key roadway spatial unit, public decision time, and clearance permission time. If the normal observation result after the treatment is candidate negative evidence for review, or no evidence of permitted clearance is formed, or no multi-hazard coupling control decision result that allows risk clearance is formed, then it will not be written into the risk clearance record, nor will it enter the action effect sample update. The risk clearance record shall not change the already generated inferable propagation relationship, multi-hazard coupling risk chain, and evidence of permitted clearance, and shall only serve as the record basis for subsequent queries and reviews.
[0070] Restrictions are applied to updating action effect samples. Only when the same multi-hazard coupled risk chain achieves a multi-hazard coupled control decision result allowing risk removal, and the coverage judgment of the control action relationship in the key roadway spatial unit passes, is it permitted to write the control action relationship and its corresponding permitted removal evidence into the action effect sample update. The sample content of the action effect sample update includes action type, spatial action boundary, expected impact direction, ventilation status version, minimum observation confirmation time, permitted removal evidence coverage unit, and multi-hazard coupled control decision result. If a continued control result or review decision result is generated, or if the normal observation result after the treatment does not form permitted removal evidence, the corresponding normal observation result after the treatment is prohibited from entering the action effect sample update. The action effect sample update only serves as a record of subsequent action effect samples and does not reverse the currently generated inferable propagation relationship, multi-hazard coupled risk chain, and permitted removal evidence.
[0071] At the end of S4, the multi-hazard coupling control decision results are output. These results include multi-hazard coupling control decisions that allow risk removal, continued control results, and review decisions. Decisions that allow risk removal are used for risk removal write-back and action effect sample updates. Continued control results are used to maintain the current control status of the multi-hazard coupling risk chain. Review decisions are used to identify key roadway spatial units that are not covered by permitted removal evidence and require further review. Normal observation results after treatment without permitted removal evidence are prohibited from entering risk removal write-back and action effect sample updates, ensuring that risk removal results and action effect samples only originate from control permission gating results constrained by propagation eligibility, removal evidence, and action coverage.
[0072] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A knowledge graph-driven method for coupled control and management of multiple disasters and three-dimensional disaster prevention, characterized in that, Includes the following steps: The system acquires multi-source status data for ventilation, fire prevention, and gas control. It adopts version-based attribution processing for ventilation status and converts the multi-source status data for ventilation, fire prevention, and gas control into status records of ventilation, fire prevention, and gas control objects indexed by common decision time, ventilation status version, and smallest roadway spatial unit. Based on the status records of ventilation, fire prevention, and gas control objects, a ventilation status constraint multi-hazard coupling diagram is constructed. In the ventilation status constraint multi-hazard coupling diagram, disaster propagation candidate relationships, measurement point response acceptance relationships, and control action effect relationships are formed. The specific operation of the ventilation status constraint multi-hazard coupling diagram is as follows: using the smallest roadway spatial unit as the graph index, the ventilation status version is written into the graph index, the propagation start point and propagation end point of the disaster propagation candidate relationship are determined according to the roadway connection direction, the direction driving basis of the disaster propagation candidate relationship is formed according to the ventilation direction benchmark and the pressure difference symbol benchmark, and the measurement point response acceptance relationship formed by the first effective deviation time of the measurement point is bound to the disaster propagation candidate relationship, and the action type, effective time, spatial action boundary and expected impact direction of the control action are bound to the smallest roadway spatial unit to form the control action action relationship; When the status record of the ventilation object represents the ventilation status from a stable state to a reconstructed state, within the semi-stable window of ventilation reconstruction, the candidate relationship for disaster propagation is processed for propagation qualification based on the ventilation status version and the measurement point response succession relationship, and a reasonable propagation relationship is generated. Among them, candidate relationships for disaster propagation include gas migration relationships, dust propagation relationships, and smoke diffusion relationships; The specific operation for verifying candidate relationships of disaster propagation is as follows: For gas migration and dust propagation relationships, verify the consistency between the propagation direction and the current airflow influence direction; for smoke diffusion relationships, based on the verification of the consistency of the propagation direction, form a temperature gradient according to the temperature components in the temperature and humidity observation records, and form the response sequence according to the smoke response data. When both the temperature gradient direction and the smoke response sequence are consistent with the current airflow influence direction, write the smoke diffusion relationship into the inferable propagation relationship; when either the temperature gradient direction or the smoke response sequence is inconsistent with the current airflow influence direction, write the smoke diffusion relationship into the candidate verification relationship. The multi-hazard coupled risk chain is generated by taking over the candidate nodes of disaster events along the reasonable propagation relationship. After the control actions are implemented, the normal observation results after the control actions are assigned to the multi-hazard coupled risk chain according to the relationship of the control actions. The evidence of permission to release is generated by eliminating the disturbances of the control and verifying the chain fallback. Based on the multi-hazard coupling risk chain, inferable propagation relationship, and evidence of permitted risk release, the system implements control and management gates, outputs multi-hazard coupling control and management decision results, and prohibits normal observation results after handling that have not formed evidence of permitted risk release or have not formed multi-hazard coupling control and management decision results that allow risk release from entering the risk release write-back and action effect sample update.
2. The knowledge graph-driven integrated control and management decision-making method for multiple disasters and three-dimensional prevention as described in claim 1, characterized in that, The specific operation of the ventilation status versioning and attribution process is as follows: Based on the upload time, sampling period, and communication link verification record of the data record, the maximum allowable delay time is formed. Data records exceeding the maximum allowable delay time are excluded from the visible range at the time of public decision-making. For data records within the visible range, dimensionless conversion is performed according to the stable center value, stable fluctuation scale determined by the historical stable record under the same working condition, and the zero-prevention lower limit determined by the minimum resolution of the measuring point sensor to form a standardized observation description. The stable range under the same working condition is formed by the stable center value and stable fluctuation scale. The standardized observation description is assigned to the smallest roadway spatial unit according to the roadway topology and measuring point coverage boundary to form the status record of the ventilation and three-proof object.
3. The knowledge graph-driven integrated control and management decision-making method for multiple disasters and three-dimensional disaster prevention according to claim 2, characterized in that, The specific operation of the ventilation reconstruction semi-stable window is as follows: When the status record of the ventilation and three-proof object shows that at least one of the ventilation direction reference, differential pressure symbol reference, and ventilation facility status has changed relative to the previous ventilation status version, a ventilation reconstruction semi-stable window is generated. During the effective period of the ventilation reconstruction semi-stable window, the disaster propagation candidate relationship is prohibited from being used directly as a propagation relationship that can be inferred. If the ventilation direction reference and differential pressure symbol reference do not change again within three to six consecutive effective confirmation cycles, and the effective confirmation cycle is counted at the time of public decision, and the ventilation direction reference, differential pressure symbol reference, and ventilation facility status all have data records within the visible range at the corresponding time of public decision, and the standardized observation description corresponding to the air volume change is within the stable range of the same working condition, the ventilation reconstruction semi-stable window ends. The three to six consecutive effective confirmation cycles are calibrated by the historical ventilation adjustment record and are subject to the mine operation procedures.
4. The knowledge graph-driven integrated control and management decision-making method for multiple disasters and three-dimensional disaster prevention according to claim 3, characterized in that, The specific operation of the propagation qualification access processing is as follows: within the ventilation reconstruction semi-stable window, read the directional driving state and the first effective deviation time of the measuring point at both ends of the disaster propagation candidate relationship. Take the measuring point on the propagation starting point side as the upstream measuring point and the measuring point on the propagation ending point side as the downstream measuring point. Verify whether the disaster propagation candidate relationship conforms to the airflow influence direction under the current ventilation state version, and verify whether the first effective deviation of the upstream measuring point is earlier than the first effective deviation of the downstream measuring point and the deviation interval does not exceed the maximum allowable response lag. The maximum allowable response lag is calibrated by the historical similar ventilation state record and is limited by the mine safety regulations. The disaster propagation candidate relationship that passes the verification is written into the inferable propagation relationship. When the disaster propagation candidate relationship conflicts with the current airflow influence direction and the conflict basis comes from the data record that can be read at the time of public decision, write the suspended propagation relationship. When the response acceptance requirement is not met, write the candidate review relationship.
5. The knowledge graph-driven integrated control and management decision-making method for multiple disasters and three-dimensional prevention as described in claim 4, characterized in that, The specific operation for generating a multi-hazard coupled risk chain is as follows: read the smallest roadway spatial unit connected by the inferable propagation relationship, take the candidate nodes of disaster events whose standardized observation descriptions exceed the stable range of the same working condition under the same ventilation state version as the chain receiving objects, and form the abnormal deviation direction of the candidate nodes of disaster events with the deviation direction of the standardized observation descriptions relative to the stable center value. When the abnormal deviation direction of the candidate nodes of disaster events is consistent with the direction of the inferable propagation relationship, the candidate nodes of disaster events are connected to the same multi-hazard coupled risk chain. The candidate verification relationship does not enter the main propagation path of the current multi-hazard coupled risk chain, and the paused propagation relationship does not participate in the generation of the current multi-hazard coupled risk chain.
6. The knowledge graph-driven integrated control and management decision-making method for multiple disasters and three-dimensional disaster prevention according to claim 5, characterized in that, The specific operation of attributing the post-disaster normal observation results to the multi-hazard coupling risk chain based on the relationship of control actions is as follows: After the control action reaches the effective time, the post-disaster observation window is opened. The shortest observation confirmation time is determined based on the action type, the sampling cycle of the measuring points, and the historical handling review records. The post-disaster observation records within the coverage of the control action relationship are read. At the end of the post-disaster observation window, when the post-disaster observation records within two consecutive public decision times are both within the visible range of the public decision time, both within the stable range of the same working conditions, and have not triggered the abnormal reporting rules determined by the mine operation procedures, a post-disaster normal observation result is formed. When the sampling cycle of the measuring points is longer than the public decision cycle, the two consecutive visible measuring point records at the end of the post-disaster observation window are used as the judgment object. When a new ventilation status version reconstruction occurs within the post-disaster observation window, the post-disaster normal observation result is written into the candidate review negative evidence. The candidate review negative evidence does not enter the risk relief write-back and action effect sample update.
7. The knowledge graph-driven integrated control and management decision-making method for multiple disasters and three-dimensional disaster prevention according to claim 6, characterized in that, The specific operations for handling disturbance elimination and chain fallback verification are as follows: Disturbances are identified based on the relationship between control actions. Situations where the standardized observation description of wind speed after a booster action exceeds the stable range under the same operating conditions, or where measuring points within the control action's coverage area are within the stable range under the same operating conditions but do not meet the chain fallback verification conditions, are identified as dilution disturbances. Situations where the standardized observation description of humidity after a spraying action exceeds the stable range under the same operating conditions, or where measuring points with visible records among smoke and dust measuring points are within the stable range under the same operating conditions but do not meet the chain fallback verification conditions, are identified as shielding disturbances. Sampling interruptions in the target area after a closure / isolation action are identified as unobservable isolation disturbances. Production sources after a power outage action are identified as... Status cutoff is identified as a disturbance caused by changes in the production source. Power outages, communication interruptions, and sampling link interruptions at measurement points are identified as disturbances caused by interruptions in the acquisition link. If normal observation results after the action are identified as disturbances after the action, they are not included in the evidence of permission to remove the action. If normal observation results after the action are not identified as disturbances after the action, and both upstream and downstream measurement points are within the same stable range under the same working conditions within the same observation window after the action, and the consistency of similar measurement points meets the minimum support requirements, the normal observation results after the action are included in the evidence of permission to remove the action. The minimum support requirements are calibrated by the measurement point deployment record and are subject to the mine operation procedures. If the consistency of similar measurement points does not meet the minimum support requirements, the normal observation results after the action are included in the candidate negative evidence for review.
8. The knowledge graph-driven integrated control and management decision-making method for multiple disasters and three-dimensional disaster prevention according to claim 7, characterized in that, The specific operation of the control and management permit gate is as follows: the multi-hazard coupling risk chain, inferable propagation relationship, evidence of permitted release, and control action relationship are uniformly converted to the smallest roadway space unit, while maintaining the original ventilation status version, public decision time, and post-disposal observation window source for each of the multi-hazard coupling risk chain, inferable propagation relationship, evidence of permitted release, and control action relationship. Records under different public decision times, different ventilation status versions, or different spatial action boundaries are not merged into the same release judgment object; the key roadway space unit is formed by the unit where the disaster event candidate node is located, the endpoint unit of the inferable propagation relationship, and the control action coverage unit, and the main propagation path is formed by the inferable propagation relationship that connects the disaster event candidate nodes; Specifically, the units requiring evidence to be removed in the key roadway spatial unit are formed by the unit where the candidate node of the disaster event is located and the endpoint unit of the inferable propagation relationship. The action coverage unit in the key roadway spatial unit is formed by the control action coverage unit. When the inferable propagation relationship covers the main propagation path, the permitted removal evidence covers the unit requiring evidence to be removed in the key roadway spatial unit, and the control action relationship covers the action coverage unit in the key roadway spatial unit, a multi-hazard coupled control decision result allowing risk removal is generated. When the inferable propagation relationship does not cover the main propagation path, a continued control result is generated. When the inferable propagation relationship covers the main propagation path and the permitted removal evidence does not cover the unit requiring evidence to be removed in the key roadway spatial unit, a review decision result is generated. When the inferable propagation relationship covers the main propagation path, the permitted removal evidence covers the unit requiring evidence to be removed in the key roadway spatial unit, and the control action relationship does not cover the action coverage unit in the key roadway spatial unit, a continued control result is generated. When generating the continued control result and the review decision result, normal observation results after disposal are prohibited from entering the risk removal write-back and action effect sample update.
Citation Information
Patent Citations
Coal mine multi-disaster risk area intelligent division method based on graph neural network
CN121235456A
Mine disaster tracing method based on knowledge graph
WO2024221780A1
