Market fire-fighting linkage early warning method based on multi-source twinning

Abstract

The present application belongs to the technical field of fire-fighting early warning, and particularly relates to a market fire-fighting linkage early warning method based on multi-source twinning. The method comprises: obtaining market object information and multi-source monitoring data of a target market, and constructing a market object twin graph comprising market object nodes and object connection relationships based on the market object information; attributing the multi-source monitoring data to the corresponding market object nodes according to the market object twin graph to obtain object evidence information; generating a candidate fire risk event chain according to the object evidence information, and performing stage matching processing and risk source positioning processing to obtain a target fire risk event chain and a corresponding risk source object; determining a risk propagation path according to the target fire risk event chain and the risk source object, determining a target linkage action based on the risk propagation path, and outputting fire-fighting linkage early warning information. The present application can improve risk source positioning accuracy, reduce false positive and false negative probabilities, and improve the synergy of early warning and linkage disposal.

Description

Technical Field

[0005]

[0001] The present invention relates to the technical field of fire warning, and particularly relates to a market fire linkage warning method based on multi-source twins. Background Technique

[0002] Market operation spaces usually have characteristics such as dense stalls, many electrical equipment, many temporary piled-up objects, frequent personnel flow, and complex ventilation paths. Fire safety management has high requirements for early hidden danger identification, risk source positioning, and linkage disposal. Existing fire warning methods mostly rely on single or scattered data sources such as smoke sensors, temperature sensors, electrical sensors, video monitoring, or manual inspections. The data correlation degree between each system is insufficient, and it is easy to form the problem of "having data but being difficult to judge".

[0003] In actual operation, the alarm position of a certain sensor is not necessarily the risk source position. Smoke may migrate along the ventilation path, electrical abnormalities may conduct along the power supply circuit, and the occlusion of fire-fighting facilities does not necessarily mean that a fire is occurring. If only judged according to regional thresholds, single-point alarms, or simple comprehensive scoring, it is easy to出现 false alarms, missed alarms, deviation in risk source positioning, and lag in disposal actions. Especially in large markets, underground markets, or multi-format mixed markets, fire risks are often formed by the combined action of electrical, space, ventilation, combustibles, and personnel factors. Existing technologies are difficult to restore multi-source data to specific market objects and their associated paths, and it is also difficult to synchronously determine reasonable fire linkage actions during the warning stage. Summary of the Invention

[0004] The present invention provides a market fire linkage warning method based on multi-source twins, which is used to at least solve the problem of accurately positioning market fire risk sources and generating linkage warnings under the condition of fragmented multi-source monitoring data.

[0005] The present invention provides a market fire linkage warning method based on multi-source twins, and the method includes: Obtain the market object information and multi-source monitoring data of the target market, and construct a market object twin graph based on the market object information. The market object twin graph includes market object nodes and object connection relationships. The object connection relationships include spatial adjacency relationships, electrical connection relationships, and ventilation propagation relationships; According to the market object twin graph, attribute the multi-source monitoring data to the corresponding market object nodes to obtain object evidence information; Generate a candidate fire risk event chain according to the object evidence information, and perform stage matching processing and risk source positioning processing on the candidate fire risk event chain to obtain a target fire risk event chain and its corresponding risk source object; Determine the risk propagation path according to the target fire risk event chain and the risk source object, determine the target linkage action according to the risk propagation path, and output fire linkage warning information.

[0006] In one possible implementation, market object nodes include at least two of the following: stall objects, electrical equipment objects, gas equipment objects, power distribution equipment objects, ventilation equipment objects, fire protection facilities objects, and evacuation route objects. Spatial adjacency relationships are used to indicate the positional association between different market object nodes, electrical connection relationships are used to indicate the power supply association between different market object nodes, and ventilation propagation relationships are used to indicate the airflow propagation association between different market object nodes.

[0007] In one possible implementation, based on the market object twin graph, multi-source monitoring data is assigned to corresponding market object nodes to obtain object evidence information. This includes: for each monitoring data item in the multi-source monitoring data, determining the data source location corresponding to the monitoring data; determining the attribution matching degree between the monitoring data and each candidate market object node based on the data source location and object connection relationship, whereby the attribution matching degree indicates the degree of matching between the monitoring data and the candidate market object node in terms of location association, power supply association, or airflow propagation association; identifying the candidate market object node with the highest attribution matching degree that is higher than a preset attribution threshold as the attribution object corresponding to the monitoring data; and generating object evidence information based on the attribution object and the monitoring data.

[0008] In one possible implementation, the object evidence information includes the attribution object, evidence type, evidence relationship, and evidence time information. The evidence type is used to indicate the anomaly category corresponding to the multi-source monitoring data. The evidence relationship is used to indicate the support or exclusion relationship between the object evidence information and the candidate fire risk event chain in the stage matching process. The evidence time information is used to indicate the generation time and validity time of the object evidence information.

[0009] In one possible implementation, generating a candidate fire risk event chain based on object evidence information includes: determining the object type corresponding to the attributing object; determining a candidate fire risk event chain matching the object evidence information from a preset fire risk event chain set based on the object type and evidence type; wherein the preset fire risk event chain set includes at least two of the following: electrical thermal runaway event chain, gas accumulation event chain, combustible material spread event chain, smoke migration event chain, and fire protection facility availability reduction event chain.

[0010] In one possible implementation, a stage matching process is performed on the candidate fire risk event chain, including: identifying multiple event stages in the candidate fire risk event chain; matching object evidence information to the corresponding event stages based on evidence type, evidence relationship, and evidence time information; and determining the stage matching result of the candidate fire risk event chain based on the object evidence information corresponding to each event stage. The stage matching result includes a stage matching value, which is used to indicate the degree of matching between the object evidence information and the event stage.

[0011] In one possible implementation, the risk source location process includes: determining the candidate risk source object corresponding to the candidate fire risk event chain based on the stage matching results, evidence time information, and object connection relationship; and determining the risk source object from the candidate risk source objects based on the consistency between the candidate risk source object and the object evidence information, where consistency is used to indicate the degree of matching between the candidate risk source object and the object evidence information in terms of time sequence and object connection relationship.

[0012] In one possible implementation, the method further includes: determining an active verification action when the difference between the highest stage matching value and the second highest stage matching value in at least two candidate fire risk event chains is less than a preset difference threshold; obtaining supplementary monitoring data based on the active verification action and updating the object evidence information based on the supplementary monitoring data; and performing stage matching processing and risk source location processing again based on the updated object evidence information; wherein the active verification action includes at least one of camera supplementary data acquisition, monitoring data acquisition frequency adjustment, linkage equipment status reading, and on-site verification tasks.

[0013] In one possible implementation, determining the risk propagation path based on the target fire risk event chain and the risk source object includes: determining the target propagation relationship from the object connection relationship based on the event chain type of the target fire risk event chain; determining the affected market object nodes based on the risk source object and the target propagation relationship; and generating the risk propagation path based on the affected market object nodes.

[0014] In one possible implementation, determining the target linkage action based on the risk propagation path includes: generating candidate linkage actions based on the risk propagation path; determining the target linkage action from the candidate linkage actions based on the impact of the candidate linkage actions on the risk propagation range or risk propagation time corresponding to the risk propagation path, and the linkage constraints corresponding to the candidate linkage actions; wherein the linkage constraints include at least one of fire equipment power supply constraints, smoke exhaust safety constraints, and evacuation route constraints.

[0015] Compared with the prior art, the advantages and beneficial effects of the present invention are as follows: By constructing a twin map of market objects based on market object information, a unified representation of objects such as stalls, electrical equipment, gas equipment, power distribution equipment, ventilation equipment, fire protection facilities, and evacuation routes is achieved, so that fire risk assessment is no longer limited to a rough regional level.

[0016] By setting spatial adjacency, electrical connection, and ventilation propagation relationships in the twin graph of market objects, the association model between multi-source monitoring data and physical objects, power supply paths, and airflow paths is realized, reducing the misjudgment of risk sources caused by downstream smoke alarms, electrical cross-zone anomalies, and video false detections.

[0017] By attributing multi-source monitoring data to the corresponding market object nodes and generating object evidence information, the monitoring data is transformed from single-point alarms to object-level evidence, giving subsequent risk assessments a clear data source and object orientation.

[0018] By generating a candidate fire risk event chain based on the evidence information of the object, and performing stage matching and risk source location processing, the staged identification of the fire risk development process is realized, avoiding the direct interpretation of disordered anomalies as fire conclusions.

[0019] By determining the risk propagation path based on the target fire risk event chain and the risk source object, the ability to predict the affected objects and the direction of propagation in advance is achieved.

[0020] By determining target-linked actions based on risk propagation paths, the system achieves the synchronous generation of early warning information and fire-fighting response, reducing response delays caused by manual analysis after an alarm is triggered. Attached Figure Description

[0021] Figure 1 This is a schematic flowchart of the method of the present invention; Figure 2 This is a schematic diagram of partial relationships in the twin graph of market objects in an embodiment of the present invention; Figure 3 This is a time-series variation diagram of key monitoring data in an embodiment of the present invention; Figure 4 This is a comparison chart of matching values ​​for candidate fire risk event chain stages in an embodiment of the present invention; Figure 5 This is a comparison chart of risk propagation time before and after linkage in an embodiment of the present invention. Detailed Implementation

[0022] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the invention. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of the invention for ease of explanation. However, it will be apparent that one or more embodiments may also be practiced without these specific details. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.

[0023] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the invention. The terms “comprising,” “including,” etc., as used herein indicate the presence of relevant features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0024] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0025] Multi-source twinning refers to mapping data from different sensing systems, management systems, and field equipment onto a digital representation structure corresponding to the actual location, according to a unified object relationship. This allows dispersed data to no longer exist as individual values, single-point alarms, or independent image results, but rather to form a dynamic representation around specific objects, object states, and the relationships between objects. In fire early warning scenarios, multi-source twinning is not merely a visual reconstruction of the site space, nor simply a centralized display of sensor data. Instead, it represents objects such as stalls, electrical equipment, gas equipment, power distribution equipment, ventilation equipment, fire protection facilities, and evacuation routes within the market as nodes, using spatial adjacency, power supply connections, and airflow propagation as the basis for connecting these objects. Based on this representation method, multi-source monitoring data can be attributed to specific market objects and their associated paths, thus providing a unified data foundation for fire risk event chain identification, risk source location, risk propagation judgment, and linkage action determination. Based on this, this invention proposes a market fire linkage early warning method based on multi-source twinning.

[0026] like Figure 1 As shown, a market fire alarm linkage early warning method based on multi-source twins is proposed, which includes: Acquire market object information and multi-source monitoring data for the target market, and construct a market object twin graph based on the market object information. The market object twin graph includes market object nodes and object connection relationships. The object connection relationships include spatial adjacency relationships, electrical connection relationships, and ventilation propagation relationships. In constructing a market object twin map, market object information can include market layout, stall distribution, location of electrical equipment, gas equipment, power distribution equipment, ventilation equipment, fire protection facilities, and evacuation routes. Multi-source monitoring data can include electrical monitoring data, environmental monitoring data, video recognition data, fire protection facility status data, and personnel access data. Based on this market object information, a market object twin map is built. This map converts physical objects in the target market into market object nodes and transforms the spatial locations, power lines, and airflow paths between physical objects into object connection relationships. Object connection relationships include spatial adjacency, electrical connections, and ventilation propagation relationships. The completed market object twin map establishes a mapping foundation with the multi-source monitoring data, enabling subsequent assignment of monitoring data according to market object nodes and further forming object evidence information.

[0027] Market object nodes include at least two of the following: stall objects, electrical equipment objects, gas equipment objects, power distribution equipment objects, ventilation equipment objects, fire protection facilities objects, and evacuation route objects. Spatial adjacency relationships are used to indicate the positional association between different market object nodes, electrical connection relationships are used to indicate the power supply association between different market object nodes, and ventilation propagation relationships are used to indicate the airflow propagation association between different market object nodes.

[0028] In one embodiment, the system generates an object base table based on the market floor plan, power distribution system diagram, ventilation system diagram, fire protection facility register, and on-site inspection records to define the category range of market object nodes and the generation method of different object connection relationships. The target market can be a farmers' market, a comprehensive market, a wholesale market, or an underground commercial market. The system can generate the object base table based on the market floor plan, power distribution system diagram, ventilation system diagram, fire protection facility register, and on-site inspection records. The object base table records the object number, object name, object type, spatial coordinates, region, associated equipment, and status update time. Stall objects can be generated according to stall number and business type; electrical equipment objects can be generated according to equipment number, rated power and access location; gas equipment objects can be generated according to gas stove, gas valve and gas pipeline location; power distribution equipment objects can be generated according to distribution box, branch circuit and socket port; ventilation equipment objects can be generated according to fan, exhaust vent and ventilation opening; fire protection facility objects can be generated according to fire extinguisher, fire hydrant, sprinkler terminal, fire door and fireproof roller shutter; evacuation route objects can be generated according to route entrance, route width and connection direction.

[0029] Spatial adjacency relationships are used to indicate the positional associations between different market object nodes. The system can determine spatial adjacency relationships based on the spatial coordinates, object boundaries, and access connections of market object nodes. For example, a spatial adjacency relationship can be generated if two stall objects are within a preset distance threshold in planar distance, or if an electrical equipment object is located within the object boundary of a stall object. The preset distance threshold can be set according to the target market's aisle width, stall spacing, and fire safety management requirements, or it can be configured by administrators during system initialization. Spatial adjacency relationships can record adjacent object numbers, adjacent distances, whether there is physical obstruction, and whether they are on the same access path.

[0030] Electrical connection relationships are used to indicate the power supply associations between different market object nodes. The system can establish a power supply hierarchy for power distribution equipment, electrical equipment, and stall objects based on the power distribution system diagram and field circuit identification. For example, a distribution box object connects to multiple branch circuit objects, a branch circuit object connects to multiple socket objects or electrical equipment objects, and an electrical equipment object belongs to a corresponding stall object. The electrical connection relationships can record the upstream power supply object, the downstream electrical equipment object, the circuit number, the switch status, and the rated load capacity. This relationship is used to determine which power supply path the abnormal data should be preferentially attributed to when electrical monitoring data is abnormal.

[0031] Ventilation propagation relationships are used to indicate the airflow propagation connections between different market object nodes. The system can establish ventilation propagation relationships based on the location of ventilation equipment, fan operating direction, exhaust vent orientation, passageway connection direction, and spatial enclosure degree. For example, a ventilation propagation relationship can be established between a cooked food stall and the exhaust vent above it, and a ventilation propagation relationship can also be established between a passageway object and an adjacent exhaust vent object. The ventilation propagation relationship can record airflow direction, propagation sequence, propagation distance, and ventilation equipment operating status. This relationship is used to determine whether an anomaly in smoke concentration, combustible gas concentration, or temperature is likely to propagate along the airflow direction, and provides a relational basis for subsequent risk source object location. The market object twin map obtained through the above method can continuously receive multi-source monitoring data and is updated when the location of market objects, power supply lines, or ventilation status change.

[0032] Based on the market object twin graph, multi-source monitoring data are assigned to the corresponding market object nodes to obtain object evidence information; During the generation of object evidence information, the system matches each monitoring data point with candidate market object nodes. Multi-source monitoring data can be organized according to collection time, data source location, data type, and collection device number. The system reads the market object nodes and object connection relationships in the market object twin graph and matches each monitoring data point with candidate market object nodes. During the matching process, electrical monitoring data is prioritized based on electrical connection relationships; smoke concentration, combustible gas concentration, and temperature monitoring data are prioritized based on spatial adjacency and ventilation propagation relationships; and video recognition data is prioritized based on camera field of view and spatial adjacency. The system designates market object nodes that meet the matching criteria as the assigned objects and integrates the assigned object, monitoring data, anomaly category, and time information into object evidence information. This object evidence information proceeds to the next stage to generate candidate fire risk event chains.

[0033] Based on the market object twin graph, multi-source monitoring data are assigned to corresponding market object nodes to obtain object evidence information, including: for each monitoring data item in the multi-source monitoring data, determining the data source location corresponding to the monitoring data; determining the attribution matching degree between the monitoring data and each candidate market object node based on the data source location and object connection relationship, the attribution matching degree is used to indicate the degree of matching between the monitoring data and the candidate market object node in terms of location association, power supply association, or airflow propagation association; determining the candidate market object node with the highest attribution matching degree and higher than the preset attribution threshold as the attribution object corresponding to the monitoring data; and generating object evidence information based on the attribution object and the monitoring data.

[0034] In one embodiment, after receiving multi-source monitoring data, the system can classify it according to data type into electrical monitoring data, environmental monitoring data, video recognition data, and fire protection facility status data. Each monitoring data item generates a record to be assigned, which includes at least the acquisition device number, acquisition time, data type, original value, anomaly marker, and data source location. The data source location can be determined by the installation coordinates of the acquisition device in the market object twin map, or by the location of the corresponding power distribution circuit, camera field of view, or ventilation path. For video recognition data, the data source location can be the coordinate area of ​​the identified object mapped onto the market floor plan in the camera view; for electrical monitoring data, the data source location can be the location of the corresponding power distribution box, branch circuit, or socket.

[0035] The system selects candidate market object nodes from the market object twin graph. Candidate market object nodes can be those adjacent to the data source location, or those within the same power supply path, airflow path, or camera field of view as the data acquisition device. To avoid an overly broad candidate pool, candidate filtering rules can be set according to data type. For electrical monitoring data, candidate market object nodes can be limited to those within the same branch circuit and its downstream electrical equipment, stalls, and distribution equipment; for smoke concentration and combustible gas concentration data, candidate market object nodes can be limited to upstream objects and adjacent spatial objects within the same ventilation path; for video recognition data, candidate market object nodes can be limited to stalls, fire protection facilities, and evacuation routes within the camera's field of view.

[0036] The attribution matching degree can be determined according to the following expression:

[0037] in, For the first The monitoring data and the first The degree of attribution matching between candidate market object nodes; For the first The monitoring data and the first The matching value of each candidate market object node in terms of spatial adjacency; For the first The monitoring data and the first The matching value of each candidate market object node in terms of electrical connection relationship; For the first The monitoring data and the first The matching value of each candidate market object node in terms of ventilation propagation relationship; , , These are weighting coefficients for spatial adjacency, electrical connection, and ventilation propagation relationships, respectively. Each weighting coefficient can be set according to the type of monitored data. For electrical monitoring data, the weighting coefficient for electrical connection relationships is higher than that for spatial adjacency and ventilation propagation relationships. For smoke concentration, temperature, and combustible gas concentration data, the weighting coefficients for ventilation propagation relationships and spatial adjacency relationships are higher than those for electrical connection relationships. Each matching value can range from 0 to 1, with higher values ​​indicating a higher degree of matching. The preset attribution threshold can be determined based on the density of market object nodes, sensor deployment density, and historical misattribution records, or it can be configured by administrators during system initialization. The system identifies candidate market object nodes with the highest attribution matching degree that exceeds the preset attribution threshold as the assigned objects. If no candidate market object nodes with a degree higher than the preset attribution threshold exist, the system can mark the corresponding monitoring data as unassigned data and retain the collection time, data source location, and data type of the unassigned data for subsequent re-matching with supplementary monitoring data.

[0038] The object evidence information includes the attribution object, evidence type, evidence relationship, and evidence time information. The evidence type is used to indicate the anomaly category corresponding to the multi-source monitoring data. The evidence relationship is used to indicate the support or exclusion relationship between the object evidence information and the candidate fire risk event chain in the stage matching process. The evidence time information is used to indicate the generation time and validity time of the object evidence information.

[0039] In one embodiment, the system configures object evidence information as data records that can be directly read by subsequent stages, used to establish the judgment relationship between the attributed object and the fire risk event chain. The attributed object indicates the market object node to which the monitoring data is assigned. The evidence type indicates the anomaly category corresponding to the monitoring data, which may include electrical anomalies, temperature anomalies, smoke anomalies, combustible gas anomalies, video anomalies, fire protection facility status anomalies, and personnel access anomalies. The evidence type can be determined based on the data type of the monitoring data, anomaly markers, and threshold judgment results. For example, a sudden increase in branch circuit current accompanied by leakage current anomalies can constitute an electrical anomaly; a smoke sensor concentration value exceeding the smoke concentration warning threshold can constitute a smoke anomaly; and video recognition results showing a fire extinguisher being obstructed by goods can constitute a fire protection facility status anomaly. Different evidence types are not directly equivalent to fire conclusions but serve as inputs for subsequent event chain judgments.

[0040] Evidence relationships are used in stage matching to indicate the support or exclusion relationships between object evidence information and candidate fire risk event chains. A support relationship indicates that the object evidence information corresponds to a specific event stage in the candidate fire risk event chain. For example, abnormal current, contact temperature rise, and leakage current changes in electrical equipment can support the early electrical anomaly stage in an electrical thermal runaway event chain. An exclusion relationship indicates that the object evidence information weakens a candidate fire risk event chain. For example, a red-orange bright area identified in video, but located at a fixed advertising screen location, and with no abnormal smoke, temperature, or personnel avoidance behavior in the corresponding area, can serve as object evidence information to exclude the actual flame stage. Evidence relationships can be determined through a preset rule table, which records the correspondence between evidence type, attributing object type, candidate fire risk event chain, and event stage. The system can query the preset rule table based on the attributing object's object type and evidence type to obtain support or exclusion relationships.

[0041] Evidence time information indicates the generation and validity period of object evidence information. The generation time can be the time of monitoring data acquisition or the time of video recognition result formation. The validity period can be set according to the evidence type. The validity period for electrical transient anomalies can be set to a shorter duration, but can be extended if they recur on the same attributed object; temperature anomalies have thermal inertia, and their validity period can be longer than that of electrical transient anomalies; fire protection facility obstruction is usually persistent, and its validity period can continue until the next video recognition result or on-site verification result changes the state. After the validity period expires, the object evidence information no longer participates in the stage matching process unless the same evidence type reappears on the same attributed object. The system can configure a status flag for each piece of object evidence information, including valid, expired, and pending verification. Valid data records enter the candidate fire risk event chain generation process, expired data records are only used for historical tracing, and pending verification data records await confirmation from supplementary monitoring data or on-site verification results. Through the above data structure, object evidence information can maintain consistent semantics and transmission paths in attribute processing, stage matching processing, and risk source location processing.

[0042] Candidate fire risk event chains are generated based on object evidence information, and stage matching and risk source location processing are performed on the candidate fire risk event chains to obtain the target fire risk event chain and its corresponding risk source object. In the fire risk event chain identification process, the system reads the object evidence information formed in the previous stage, identifies the attributing object, evidence type, evidence relationship, and evidence time information, and matches it with a preset set of fire risk event chains. Candidate fire risk event chains represent the possible risk evolution process in the target market, and event stages represent different development nodes in this risk evolution process. The system matches object evidence information to the corresponding event stages, obtaining stage matching results. The system combines evidence time information and object connection relationships to determine the degree of correlation between candidate risk source objects and object evidence information, and identifies the risk source object from the candidate risk source objects. The obtained target fire risk event chain and risk source object then enter the risk propagation path determination process.

[0043] Generate candidate fire risk event chains based on object evidence information, including: determining the object type corresponding to the object; determining candidate fire risk event chains that match the object evidence information from a preset fire risk event chain set based on the object type and evidence type; wherein the preset fire risk event chain set includes at least two of the following: electrical thermal runaway event chain, gas accumulation event chain, combustible material spread event chain, smoke migration event chain, and fire protection facility availability reduction event chain.

[0044] In one embodiment, the system can establish a set of preset fire risk event chains during the initialization phase. Each preset fire risk event chain can record the applicable object type, acceptable evidence type, event stage sequence, and exclusion conditions. The system can establish a set of preset fire risk event chains during the initialization phase, which includes at least two of the following: electrical thermal runaway event chains, gas accumulation event chains, combustible material spread event chains, smoke migration event chains, and fire protection facility availability reduction event chains. Each preset fire risk event chain can record the applicable object type, acceptable evidence type, event stage sequence, and exclusion conditions. For example, the electrical thermal runaway event chain can be applied to electrical equipment, power distribution equipment, and stalls, and can accept evidence types such as electrical anomalies, temperature anomalies, and smoke anomalies. The gas accumulation event chain can be applied to food stalls, gas equipment, and ventilation equipment, and can accept evidence types such as combustible gas anomalies, ventilation status anomalies, and video anomalies. The combustible material spread event chain can be applied to stalls, goods storage areas, and evacuation routes, and can accept evidence types such as video anomalies, temperature anomalies, and personnel passage anomalies. The event chain for reduced availability of fire protection facilities can be applied to both fire protection facilities and evacuation routes, and can accept evidence types such as obstruction anomalies, equipment status anomalies, and passage obstruction anomalies.

[0045] When generating candidate fire risk event chains, the system can first read the attributing object from the object evidence information and query the corresponding object type in the market object twin graph. The object type can be written when the market object node is created or determined by the business attributes in the object base table. The system uses the object type and evidence type as search conditions to find matching risk event chains in the preset fire risk event chain set. If the attributing object is a power distribution equipment object and the evidence type is electrical anomaly, the system can generate an electrical thermal runaway event chain as a candidate fire risk event chain. If the attributing object is a ventilation equipment object and the evidence type is smoke anomaly or combustible gas anomaly, the system can generate a smoke migration event chain or a gas accumulation event chain as a candidate fire risk event chain. If the same object evidence information can match multiple preset fire risk event chains simultaneously, the system retains multiple candidate fire risk event chains and distinguishes them in subsequent matching processes by evidence relationship and evidence time information. If neither the object type nor the evidence type can match any preset fire risk event chain, the system can mark the object evidence information as a record to be verified and not directly proceed to risk source location processing. The preset fire risk event chain set can be configured by managers based on market conditions, fire management regulations, and historical risk records, and can also be updated during system operation based on on-site verification results.

[0046] The process of performing stage matching on the candidate fire risk event chain includes: identifying multiple event stages in the candidate fire risk event chain; matching the object evidence information to the corresponding event stage based on the evidence type, evidence relationship, and evidence time information; and determining the stage matching result of the candidate fire risk event chain based on the object evidence information corresponding to each event stage. The stage matching result includes a stage matching value, which is used to indicate the degree of matching between the object evidence information and the event stage.

[0047] In one embodiment, each candidate fire risk event chain may include multiple event stages, arranged in order of risk evolution. For example, an electrical thermal runaway event chain may include an electrical anomaly stage, a localized heat accumulation stage, an insulation degradation stage, and a smoke appearance stage. Similarly, a gas accumulation event chain may include a combustible gas anomaly stage, an insufficient ventilation dilution stage, a localized gas accumulation stage, and an ignition source proximity stage. The system reads the evidence type, evidence relationship, and evidence time information from the object evidence information and determines the event stage corresponding to the object evidence information based on a preset stage mapping table. The stage mapping table records the correspondence between evidence type, object type, candidate fire risk event chains, and event stages. For example, an electrical anomaly can match the electrical anomaly stage in an electrical thermal runaway event chain, a temperature anomaly can match the localized heat accumulation stage, and a smoke anomaly can match the smoke appearance stage. When the evidence relationship is supportive, the corresponding object evidence information serves as a positive matching basis for the event stage; when the evidence relationship is exclusionary, the corresponding object evidence information is used to reduce the matching degree of the event stage.

[0048] The stage matching value can be determined according to the following expression:

[0049] in, For the first The first in the candidate fire risk event chain Stage matching value for each event stage; To match the first The first event phase and the one with supporting relationships The evidentiary strength of the evidence information of each item; To match the first The first event phase and the one with exclusionary relationships The evidentiary strength of the evidence information of each item; For the first The first in the candidate fire risk event chain The normalization parameters correspond to each event stage. Evidence strength can be determined based on the magnitude of the anomaly in the monitoring data, the effective time, and the reliability of the data source. Normalization parameters can be determined based on the number of necessary evidence items corresponding to each event stage and the stage weights configured by management personnel. This formula is used to uniformly convert support and exclusion relationships into stage matching values ​​between 0 and 1. The closer the stage matching value is to 1, the higher the degree of matching between the object evidence information and the corresponding event stage. The system can calculate the stage matching value for each event stage in the candidate fire risk event chain and record the stage matching value, the event stage order, and the object evidence information involved in the matching as the stage matching result. The stage confirmation threshold can be determined based on the number of necessary evidence items corresponding to the event stage, historical false alarm records, and the stage weights configured by management personnel. If the stage matching value of a preceding event stage is lower than the stage confirmation threshold, while the stage matching value of a subsequent event stage is higher, the system can retain the candidate fire risk event chain but mark it as a stage sequence pending verification to avoid a single subsequent anomaly directly triggering risk source location.

[0050] The risk source location process includes: determining the candidate risk source objects corresponding to the candidate fire risk event chain based on the stage matching results, evidence time information, and object connection relationships; and determining the risk source objects from the candidate risk source objects based on the consistency between the candidate risk source objects and the object evidence information. The consistency is used to indicate the degree of matching between the candidate risk source objects and the object evidence information in terms of time sequence and object connection relationships.

[0051] In one embodiment, the system can select candidate fire risk event chains whose stage matching values ​​reach the stage confirmation threshold from the stage matching results, and use the attributing objects corresponding to the evidence information of the objects participating in the matching as the initial object set. For electrical thermal runaway event chains, the system prioritizes searching for power supply associations between upstream power distribution equipment objects, branch circuit objects, and downstream power consumption equipment objects in the initial object set along electrical connection relationships. For smoke migration event chains, the system prioritizes searching for propagation associations between upstream airflow objects, downstream airflow objects, and ventilation equipment objects in the initial object set along ventilation propagation relationships. For combustible material spread event chains, the system prioritizes searching for location associations between combustible material concentration objects, heat source objects, and adjacent stall objects along spatial adjacency relationships. Through the above methods, the system obtains candidate risk source objects corresponding to the candidate fire risk event chains.

[0052] Consistency is used to indicate the degree of matching between candidate risk source objects and object evidence information in terms of time sequence and object connectivity, and can be determined by the following expression:

[0053] in, For the first Consistency among candidate risk source objects; For the first The time-series matching value between candidate risk source objects and object evidence information; For the first The matching value between each candidate risk source object and the object evidence information in the object connection relationship; These are the weighting coefficients corresponding to the time-order matching values; The weight coefficients corresponding to the values ​​matched for object connection relationships. and The event chain type can be set according to the candidate fire risk event chain, and it can be configured as a non-negative number. The time sequence matching value is used to determine whether the generation time of the object evidence information conforms to the stage sequence of the corresponding risk event chain. The object connection relationship matching value is used to determine whether the candidate risk source object is located on the reasonable propagation path or power supply path of the object evidence information corresponding to the object's attribution. For example, if the smoke anomaly first appears downstream of the ventilation system and then upstream, the upstream ventilation object can obtain higher consistency; if the electrical anomaly first appears in a branch circuit and then appears near the downstream electrical equipment, the corresponding branch circuit object can obtain higher consistency. The risk source confirmation threshold can be determined based on the sensor deployment density of the target market, the completeness of the object connection relationship, and historical positioning errors. The system identifies the candidate risk source object with the highest consistency value that reaches the risk source confirmation threshold as the risk source object. If the consistency of multiple candidate risk source objects is close, the system can mark the candidate fire risk event chain as pending verification and enter the active verification process.

[0054] The method further includes: determining an active verification action when the difference between the highest stage matching value and the second highest stage matching value in at least two candidate fire risk event chains is less than a preset difference threshold; obtaining supplementary monitoring data based on the active verification action and updating the object evidence information based on the supplementary monitoring data; and performing stage matching processing and risk source location processing again based on the updated object evidence information; wherein the active verification action includes at least one of camera supplementary acquisition, monitoring data acquisition frequency adjustment, linkage equipment status reading, and on-site verification tasks.

[0055] In one embodiment, the system can compare the stage matching values ​​of at least two candidate fire risk event chains. When the difference between the highest stage matching value and the second highest stage matching value is less than a preset difference threshold, it indicates that the current object's evidence information is insufficient to stably distinguish different risk evolution processes. The preset difference threshold can be set based on historical false alarm records, equipment sensitivity, and market management requirements. For markets with high sensor density and few historical false alarms, the preset difference threshold can be set lower; for markets with more visual interference and frequent occurrences of fumes or steam, the preset difference threshold can be set higher to increase the triggering opportunities for proactive verification actions. Proactive verification actions can include at least one of the following: supplementary camera data acquisition, adjustment of monitoring data acquisition frequency, reading of linked equipment status, and on-site verification tasks.

[0056] When determining an active verification action, the system can select based on the type of the current candidate fire risk event chain and the type of missing evidence. To distinguish between real flames and advertising screen reflections, the system can select supplementary camera data collection, calling adjacent cameras or adjusting camera presets to re-collect suspected areas from different perspectives. To distinguish between normal equipment start-up / shutdown and electrical thermal runaway, the system can adjust the monitoring data collection frequency, increasing the collection frequency of current, voltage, leakage current, and temperature for the corresponding circuit within a limited time. To determine whether smoke is propagating along the ventilation path, the system can read the status of linked equipment such as fans, exhaust vents, fire doors, and fireproof roller shutters. If the equipment data still cannot distinguish the candidate fire risk event chain, the system can issue an on-site verification task to the on-site management terminal. The on-site verification task can include the location of the verification object, verification content, shooting requirements, and safety distance requirements. After receiving the supplementary monitoring data, the system updates the object evidence information according to the object attribution processing rules and performs stage matching and risk source location processing again based on the updated object evidence information. If the supplementary monitoring data forms an exclusion relationship, the system lowers the stage matching value of the corresponding candidate fire risk event chain; if the supplementary monitoring data forms a support relationship, the system increases the stage matching value of the corresponding event stage and redetermines the target fire risk event chain and its corresponding risk source object.

[0057] The risk propagation path is determined based on the target fire risk event chain and the risk source object. The target linkage action is determined based on the risk propagation path, and fire linkage early warning information is output.

[0058] During the coordinated early warning output process, the system reads the event chain type of the target fire risk event chain and, combined with the object connection relationships in the market object twin graph, determines the target propagation relationships involved in the propagation judgment. The system prioritizes electrical connection relationships for electrical thermal runaway event chains, ventilation propagation relationships for smoke migration event chains, and spatial adjacency relationships for combustible material spread event chains. Starting from the risk source object, the system searches for potentially affected market object nodes along the target propagation relationships, forming a risk propagation path that includes the propagation start point, propagation direction, affected market object nodes, and propagation order. Based on the risk propagation path, the system generates candidate coordinated actions and, under constraints such as fire equipment power supply, smoke exhaust safety, and evacuation routes, determines the target coordinated action, outputting fire coordinated early warning information.

[0059] Determining the risk propagation path based on the target fire risk event chain and risk source objects includes: determining the target propagation relationship from the object connection relationship based on the event chain type of the target fire risk event chain; determining the affected market object nodes based on the risk source objects and the target propagation relationship; and generating the risk propagation path based on the affected market object nodes.

[0060] In one embodiment, the event chain types of the target fire risk event chain may include electrical thermal runaway, gas accumulation, combustible material spread, smoke migration, and reduced availability of fire protection facilities. The system can establish a propagation relationship mapping table between event chain types and object connection relationships. The propagation relationship mapping table records the primary and secondary propagation relationships corresponding to different event chain types. For example, the primary propagation relationship for electrical thermal runaway is electrical connection, and the secondary propagation relationship is spatial adjacency; the primary propagation relationship for smoke migration is ventilation propagation, and the secondary propagation relationship is spatial adjacency; the primary propagation relationship for combustible material spread is spatial adjacency, and the secondary propagation relationship may include ventilation propagation; the primary propagation relationship for gas accumulation is ventilation propagation, and the secondary propagation relationship is spatial adjacency; the primary propagation relationship for reduced availability of fire protection facilities is the positional association between fire protection facility objects and evacuation route objects or stall objects.

[0061] The system starts with the risk source object and reads the market object nodes that are directly or indirectly connected to the risk source object in the target propagation relationship. For electrical connection relationships, the system can find affected market object nodes along the power supply path between power distribution equipment objects, branch circuit objects, socket objects, and electrical equipment objects. For ventilation propagation relationships, the system can find affected market object nodes along the airflow direction between exhaust vents, fans, passageways, and adjacent stalls. For spatial adjacency relationships, the system can find affected market object nodes based on the distance, obstruction status, and combustible material distribution between stall objects, goods stacking areas, heat source equipment, and evacuation routes. Affected market object nodes can include directly affected nodes and indirectly affected nodes. Directly affected nodes are market object nodes that have a single-segment target propagation relationship with the risk source object, while indirectly affected nodes are market object nodes connected to the risk source object via one or more intermediate market object nodes.

[0062] The risk transmission time can be determined according to the following expression:

[0063] in, For the risk to reach the first Risk propagation time for each affected market node; The first in the risk transmission path The propagation duration corresponding to the segment object connection relationship; For the risk source object to the first The number of object connection segments between affected market object nodes. The propagation duration can be set according to the object connection type. The propagation duration of electrical connection relationships can be determined based on circuit level, switch status, and protection device status; the propagation duration of ventilation propagation relationships can be determined based on ventilation distance, fan operating status, and airflow direction; the propagation duration of spatial adjacency relationships can be determined based on object spacing, combustible material type, and barrier status. The system writes the risk source object, target propagation relationship, affected market object node, and risk propagation time into the risk propagation path. If multiple propagation paths exist for a certain affected market object node, the system can retain the propagation path with the shorter risk propagation time, or retain multiple propagation paths for use in linkage action judgment. The risk propagation path enters the linkage action determination process to determine the market object node that should be dealt with first and the time margin for handling.

[0064] Determining the target linkage action based on the risk propagation path includes: generating candidate linkage actions based on the risk propagation path; determining the target linkage action from the candidate linkage actions based on the impact of the candidate linkage actions on the risk propagation scope or risk propagation time corresponding to the risk propagation path, and the linkage constraints corresponding to the candidate linkage actions; wherein the linkage constraints include at least one of fire equipment power supply constraints, smoke exhaust safety constraints, and evacuation route constraints.

[0065] In one embodiment, the system can automatically generate candidate linkage actions based on the risk propagation path. For electrical thermal runaway risk, candidate linkage actions may include cutting off non-critical branch circuits, maintaining power supply to fire-fighting equipment, sending verification tasks to the on-site management terminal, and prompting adjacent stalls to evacuate flammable materials. For smoke migration risk, candidate linkage actions may include adjusting the status of smoke exhaust fans, opening smoke exhaust valves, closing fire doors, activating local voice prompts, and guiding personnel to avoid affected passageways. For combustible material spread risk, candidate linkage actions may include prompting the removal of adjacent combustible materials, assigning on-site personnel to verify heat sources, activating status checks of nearby fire-fighting facilities, and issuing local evacuation notices. Candidate linkage actions can be provided by an action library, which records the action name, target, execution conditions, linked equipment, expected scope of effect, feedback method, and constraints.

[0066] Interlocking constraints are used to exclude candidate interlocking actions that may introduce secondary risks. Power supply constraints for fire equipment restrict power outage actions from affecting fire control equipment, emergency lighting, smoke extraction equipment, evacuation guidance equipment, and fire communication equipment. Smoke extraction safety constraints restrict smoke extraction actions from directing smoke to densely populated areas or safety exits. Escape route constraints restrict fire doors, fire-resistant roller shutters, and public address system guidance actions from blocking main evacuation routes or causing conflicts in evacuation routes. When judging candidate interlocking actions, the system can first verify whether the candidate interlocking actions meet the interlocking constraints. Candidate interlocking actions that do not meet the interlocking constraints are marked as unexecutable and will not be included in the target interlocking action selection.

[0067] The target-linked action can be determined through action scoring, which can be determined according to the following expression:

[0068] in, For the first Action score for each candidate linked action; For the first The reduction in the scope of risk propagation by each candidate coordinated action; For the first The extension of risk propagation time for each candidate coordinated action; For the first The constraint penalty value corresponds to each candidate linkage action. The reduction value of the risk propagation range can be determined based on the change in the number of affected market object nodes before and after the execution of the candidate linkage action. The extension value of the risk propagation time can be determined based on the change in the time it takes for the risk to reach the key market object node before and after the execution of the candidate linkage action. The constraint penalty value is used to indicate the degree of conflict between the candidate linkage action and the constraints of fire equipment power supply, smoke exhaust safety, or evacuation route. The system determines the candidate linkage action that meets the linkage constraint conditions and has the highest action score as the target linkage action. If the action scores of multiple candidate linkage actions are close, the system can select them in the order of priority of personnel safety, priority of fire equipment availability, and priority of risk source isolation. Fire linkage early warning information can include risk source object, target fire risk event chain, risk propagation path, affected market object nodes, target linkage action, action execution object, and feedback requirements. After output, the system can receive feedback from linkage equipment and on-site verification feedback, and incorporate the feedback content as new monitoring data into the aforementioned object attribution process.

[0069] A test area of ​​approximately 420 square meters was selected within the cooked food and refrigerated food section of a farmers' market for verification. This test area included 12 stalls, one primary distribution box, three branch circuits, four exhaust systems, two evacuation routes, and several fire-fighting facilities. The stall in question was designated as stall S12, the corresponding refrigerated freezer as freezer E12, the corresponding branch circuit as branch circuit R1, the corresponding exhaust vent as exhaust vent V3, and the corresponding evacuation route as evacuation route C2. Monitoring equipment included current sensors, leakage current sensors, surface temperature sensors, smoke sensors, video monitoring terminals, and linked equipment status acquisition terminals. The sampling period was one minute.

[0070] like Figure 2 As shown, Figure 2 This diagram was compiled and drawn based on the area's floor plan, electrical topology diagram, and ventilation layout diagram. The diagram shows key market nodes including distribution box A1, branch circuit R1, freezer E12, stall S12, adjacent stall S13, exhaust vent V3, evacuation route C2, and fire hydrant H2. Solid lines represent electrical connections, dashed lines represent spatial adjacency, and arrows represent ventilation propagation. The diagram shows that freezer E12 is connected to the electrical system via branch circuit R1. Freezer E12 is located inside stall S12. Stall S12 is spatially adjacent to adjacent stall S13 and evacuation route C2. There is an airflow propagation path between exhaust vent V3 and evacuation route C2.

[0071] The test period was from 10:00 to 10:09. During this period, the current in the circuit corresponding to freezer E12 was 11.8 A, 12.4 A, 13.1 A, 14.2 A, 15.6 A, 16.8 A, 17.5 A, 18.1 A, 18.4 A, and 18.6 A, respectively; the surface temperatures were 36.2°C, 38.5°C, 41.7°C, 47.6°C, 55.2°C, 63.8°C, and 72.1°C, respectively. Temperatures of 78.9°C, 84.3°C, and 88.4°C were recorded; smoke concentrations of 0.02, 0.03, 0.05, 0.07, 0.11, 0.16, 0.24, 0.31, 0.38, and 0.44 were recorded; and leakage currents of 8 mA, 9 mA, 11 mA, 14 mA, 18 mA, 23 mA, 29 mA, 34 mA, 39 mA, and 42 mA were recorded.

[0072] like Figure 3 As shown, Figure 3 The graph was created by organizing and normalizing the above four types of raw monitoring data in chronological order. It fully illustrates the changing trends of current, surface temperature, smoke concentration, and leakage current from 10:00 to 10:09. Notably, all four indicators showed a continuous increase after 10:04, with a significantly steeper upward slope after 10:06, indicating that the abnormal condition evolved from a single electrical anomaly to heat accumulation and smoke appearance.

[0073] In the object attribution processing, taking the abnormal current data at 10:07 as an example, the attribution matching degree is calculated for three candidate objects: freezer E12, stall S12, and distribution box A1. The weights for spatial association, electrical association, and ventilation association are set to 0.2, 0.7, and 0.1, respectively. The three matching values ​​corresponding to freezer E12 are 0.80, 1.00, and 0.20, respectively, and the attribution matching degree is:

[0074] The matching degree of booth S12 is:

[0075] The matching degree of distribution box A1 is:

[0076] Therefore, the abnormal current data is attributed to freezer E12. Taking the abnormal smoke data at 10:07 as another example, with weights of 0.4, 0.1, and 0.5 for spatial correlation, electrical correlation, and ventilation correlation respectively, the matching degree for stall S12 is 0.795, and the matching degree for exhaust vent V3 is 0.755. Therefore, the abnormal smoke data is attributed to stall S12. This forms the object evidence information, including the electrical abnormality, temperature abnormality, and leakage abnormality corresponding to freezer E12, and the smoke abnormality corresponding to stall S12.

[0077] During the candidate fire risk event chain generation phase, three candidate event chains were obtained based on the type of attributed object and the type of evidence: electrical thermal runaway event chain, smoke migration event chain, and reduced availability of fire protection facilities event chain. Initial phase matching calculations showed that the overall phase matching result for the electrical thermal runaway event chain was 0.72, and for the smoke migration event chain, it was 0.68. The difference between the two was only 0.04, lower than the preset difference threshold of 0.08, thus triggering an active verification action. The active verification action included supplementary camera data collection, increasing the monitoring frequency to once every 10 seconds, and reading the operating status of the exhaust system. Key data obtained after verification included: a freezer compressor casing hotspot temperature of 92.6 degrees Celsius, three consecutive insulation alarms in branch circuit R1, normal operation of exhaust vent V3, and no independent smoke source detected in adjacent stall S13. After updating the supplementary data, the phase matching values ​​for each candidate event chain were recalculated.

[0078] like Figure 4 As shown, Figure 4 The values ​​were calculated based on the updated object evidence information and stage mapping rules. The figure shows the stage matching values ​​of three candidate fire risk event chains, categorized into stages one through four: electrical thermal runaway event chain (0.86, 0.82, 0.79, and 0.89), smoke migration event chain (0.51, 0.58, 0.63, and 0.60), and reduced fire protection facility availability event chain (0.22, 0.29, 0.31, and 0.27). As can be seen from the figure, the electrical thermal runaway event chain is significantly higher than the other two candidate event chains at each stage, especially reaching 0.89 in stage four. Therefore, the electrical thermal runaway event chain is identified as the target fire risk event chain.

[0079] In the risk source location processing, freezer E12, distribution box A1, and stall S12 were selected as candidate risk source objects. Consistency was calculated using time sequence matching values ​​and object connection relationship matching values, with a time sequence weight of 0.6 and an object connection relationship weight of 0.4. Freezer E12 had a time sequence matching value of 0.94 and an object connection relationship matching value of 0.91, resulting in the following consistency:

[0080] The consistency of distribution box A1 is 0.762, and the consistency of stall S12 is 0.748. Therefore, freezer E12 is identified as a source of risk.

[0081] During the coordinated decision-making phase, the system generates three sets of candidate coordinated actions. The first set involves cutting off power to branch circuit R1, activating exhaust vent V3, and providing voice guidance; the second set involves activating only smoke extraction and broadcasting; and the third set involves closing fire compartmentation facilities and broadcasting. Action scores are calculated based on the risk propagation scope reduction value, risk propagation time extension value, and constraint penalty value. The score for the first set of actions is:

[0082] The second group of actions scored 3.0, and the third group of actions scored 1.3. Therefore, the first group was selected as the target linkage action.

[0083] like Figure 5 As shown, Figure 5 The graph was drawn based on the calculation results of the risk propagation path. The graph shows the arrival times of the risk at stall S12, adjacent stall S13, exhaust vent V3, and evacuation route C2 before and after the linkage. Before the linkage, the risk arrival times were 1.2 minutes, 3.6 minutes, 1.5 minutes, and 3.6 minutes respectively; after the linkage, the corresponding times became 3.5 minutes, 8.4 minutes, 5.2 minutes, and 9.1 minutes. This shows that the risk arrival time at evacuation route C2 increased from 3.6 minutes to 9.1 minutes, an increase of approximately 152.8%; the average risk arrival time for the four affected objects increased from 2.48 minutes to 6.55 minutes, an increase of approximately 164.1%; and the number of affected market object nodes decreased from 4 to 1 within 4 minutes. These results indicate that the present invention can not only identify risk sources earlier but also select more targeted linkage actions based on the risk propagation path, thereby effectively compressing the scope of risk propagation and extending the response window.

[0084] As can be seen from this embodiment, the present invention achieves object-level attribution of multi-source monitoring data through market object twin graphs, then achieves risk assessment through candidate fire risk event chains, stage matching processing, and risk source location processing, and finally completes the selection of linkage actions by combining risk propagation paths. Compared with methods that rely solely on single-point threshold alarms or single-system discrimination, the present invention in this embodiment achieves clearer risk object location, more stable event chain determination, and more targeted linkage response.

[0085] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects.

[0086] The above are merely embodiments of the present invention and are not intended to limit the invention. Those skilled in the art can make various modifications, equivalent substitutions, and improvements within the spirit and principles of the present invention.

Claims

1. A market fire alarm linkage early warning method based on multi-source twins, characterized in that, The method includes: Obtain market object information and multi-source monitoring data of the target market, and construct a market object twin graph based on the market object information. The market object twin graph includes market object nodes and object connection relationships. The object connection relationships include spatial adjacency relationships, electrical connection relationships, and ventilation propagation relationships. Based on the market object twin graph, the multi-source monitoring data is assigned to the corresponding market object node to obtain object evidence information; Based on the object evidence information, a candidate fire risk event chain is generated, and the candidate fire risk event chain is subjected to stage matching and risk source location processing to obtain the target fire risk event chain and its corresponding risk source object. The risk propagation path is determined based on the target fire risk event chain and the risk source object. The target linkage action is determined based on the risk propagation path, and fire linkage early warning information is output.

2. The method according to claim 1, characterized in that, The market object nodes include at least two of the following: stall objects, electrical equipment objects, gas equipment objects, power distribution equipment objects, ventilation equipment objects, fire protection facilities objects, and evacuation passage objects. The spatial adjacency relationship is used to indicate the positional association between different market object nodes, the electrical connection relationship is used to indicate the power supply association between different market object nodes, and the ventilation propagation relationship is used to indicate the airflow propagation association between different market object nodes.

3. The method according to claim 1, characterized in that, The step of attributing the multi-source monitoring data to the corresponding market object nodes based on the market object twin graph to obtain object evidence information includes: For each piece of monitoring data in the multi-source monitoring data, determine the location of the data source corresponding to the monitoring data; Based on the data source location and the object connection relationship, the affiliation matching degree between the monitoring data and each candidate market object node is determined. The affiliation matching degree is used to indicate the degree of matching between the monitoring data and the candidate market object node in terms of location association, power supply association, or airflow propagation association. The candidate market object node with the highest attribution matching degree and higher than the preset attribution threshold is determined as the attribution object corresponding to the monitoring data; The object evidence information is generated based on the attributed object and the monitoring data.

4. The method according to claim 3, characterized in that, The object evidence information includes the attribution object, evidence type, evidence relationship, and evidence time information. The evidence type is used to indicate the anomaly category corresponding to the multi-source monitoring data. The evidence relationship is used to indicate the support or exclusion relationship between the object evidence information and the candidate fire risk event chain in the stage matching process. The evidence time information is used to indicate the generation time and validity time of the object evidence information.

5. The method according to claim 4, characterized in that, The step of generating a candidate fire risk event chain based on the object evidence information includes: Determine the object type corresponding to the attributed object; Based on the object type and the evidence type, a candidate fire risk event chain that matches the object evidence information is determined from a preset fire risk event chain set; The preset fire risk event chain set includes at least two of the following: electrical thermal runaway event chain, gas accumulation event chain, combustible material spread event chain, smoke migration event chain, and fire protection facility availability reduction event chain.

6. The method according to claim 5, characterized in that, Performing stage matching processing on the candidate fire risk event chain includes: Determining multiple event stages in the candidate fire risk event chain; Matching the object evidence information to the corresponding event stage according to the evidence type, the evidence relationship, and the evidence time information; Determining the stage matching result of the candidate fire risk event chain according to the object evidence information corresponding to each event stage, where the stage matching result includes a stage matching value, and the stage matching value is used to indicate the matching degree between the object evidence information and the event stage.

7. The method according to claim 6, characterized in that, The risk source location processing includes: Determining a candidate risk source object corresponding to the candidate fire risk event chain according to the stage matching result, the evidence time information, and the object connection relationship; Determining the risk source object from the candidate risk source objects according to the consistency between the candidate risk source object and the object evidence information, where the consistency is used to indicate the matching degree between the candidate risk source object and the object evidence information in terms of time sequence and object connection relationship.

8. The method according to claim 7, characterized in that, The method further includes: Determining an active verification action when the difference between the highest stage matching value and the second highest stage matching value in at least two of the candidate fire risk event chains is less than a preset difference threshold; Obtaining supplementary monitoring data according to the active verification action, and updating the object evidence information based on the supplementary monitoring data; Performing the stage matching processing and the risk source location processing again based on the updated object evidence information; Wherein, the active verification action includes at least one of supplementary camera acquisition, adjustment of monitoring data acquisition frequency, reading of linkage device status, and on-site verification task.

9. The method according to claim 1, characterized in that, Determining the risk propagation path according to the target fire risk event chain and the risk source object includes: Determining a target propagation relationship from the object connection relationship according to the event chain type of the target fire risk event chain; Determining the affected market object nodes based on the risk source object and the target propagation relationship; Generating the risk propagation path according to the affected market object nodes.

10. The method according to claim 9, characterized in that, Determining the target linkage action according to the risk propagation path includes: Generating candidate linkage actions according to the risk propagation path; Determining the target linkage action from the candidate linkage actions according to the influence of the candidate linkage action on the risk propagation range or risk propagation time corresponding to the risk propagation path, and the linkage constraint conditions corresponding to the candidate linkage action; ​

Images