Safety management and control method and system applied to emergency material storage park

Abstract

The application provides a safety management and control method and system applied to an emergency material storage park, relates to the technical field of emergency material storage management, and first constructs a safety state link based on multi-source monitoring information, containing material storage nodes, environment nodes and activity nodes and forming a net structure through associated edges; then traces the evolution process, records the state information of each node and associated edge changing with time to generate link evolution data; then adapts the control action set corresponding to the node state based on the link evolution data; then sends the control action set to the park management and control system driving execution device for execution, and collects feedback data after execution; finally, the evolution trend of the safety state link is calibrated according to the feedback data, so that the safety management and control level of the emergency material storage park can be comprehensively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of emergency material warehouse management, in particular to a safety management and control method and system applied to an emergency material warehouse park. BACKGROUND

[0002] In the management of an emergency material warehouse park, safety management and control is a crucial link. Traditional safety management and control methods for emergency material warehouse parks often rely on manual regular patrol and simple monitoring equipment. Manual patrol has the problem of low efficiency, and it is difficult to conduct comprehensive and real-time monitoring of the entire park, and is easily affected by human factors such as negligence and fatigue, so that safety hazards cannot be discovered in time. Simple monitoring equipment can usually only monitor a single factor, such as temperature or humidity, and lacks comprehensive perception and comprehensive analysis of the overall safety state of the park. Each monitoring equipment is independent of each other, and the data cannot be effectively integrated and associated, making it difficult to form a complete understanding of the safety state of the park. In addition, the traditional management and control method lacks a fast and accurate response strategy generation mechanism after discovering safety hazards, and cannot adjust the management and control actions in time according to the dynamic changes of the safety state, so that safety hazards cannot be handled in time and effectively, which seriously threatens the storage safety of emergency materials. SUMMARY

[0003] In view of the above-mentioned problems, in combination with the first aspect of the present application, the present application provides a safety management and control method applied to an emergency material warehouse park, the method comprising:

[0004] constructing a safety state link of the emergency material warehouse park, the safety state link being based on multi-source monitoring information and containing material storage nodes, environment nodes and activity nodes, each node being connected to form a mesh structure through associated edges;

[0005] tracking the evolution process of the safety state link, recording the state information of each node and associated edge changing with time, and generating link evolution data;

[0006] adapting a set of management and control actions based on the link evolution data, the set of management and control actions forming a corresponding relationship with the node state of the safety state link;

[0007] sending the set of management and control actions to a park management and control system, driving an execution device to execute the management and control actions, and collecting feedback data of the safety state link after execution by the park management and control system;

[0008] calibrating the evolution trend of the safety state link according to the feedback data.

[0009] In still another aspect, the present application also provides a safety management and control system applied to an emergency material storage park, comprising a processor, a machine readable storage medium, the machine readable storage medium being connected with the processor, the machine readable storage medium being used for storing programs, instructions or codes, and the processor being used for executing the programs, instructions or codes in the machine readable storage medium to realize the above method.

[0010] Based on the above aspects, the present application realizes comprehensive and systematic modeling of the safety state of the park by constructing a safety state link based on multi-source monitoring information, connecting the material storage nodes, the environment nodes and the activity nodes in a mesh structure, accurately reflecting the complex correlation between the nodes, tracking the evolution process of the safety state link and generating link evolution data, mastering the dynamic changes of the safety state of the park in real time, adapting the control actions based on the link evolution data, making the control actions accurately correspond to the node state of the safety state link, and improving the pertinence and effectiveness of the control actions. The control action set is sent to the park control system driving execution device for execution, and the feedback data after execution is collected, realizing closed-loop control of the control process. The evolution trend of the safety state link is calibrated according to the feedback data, the safety management and control strategy is continuously optimized, the dynamic changes of the safety state of the park are adapted, the safety management and control level of the emergency material storage park is effectively improved, the risk of safety accidents is reduced, and the safety storage of emergency materials is ensured. BRIEF DESCRIPTION OF DRAWINGS

[0011] Figure 1 is an execution flow schematic diagram of the safety management and control method applied to the emergency material storage park provided by an embodiment of the present application.

[0012] Figure 2 is a schematic diagram of exemplary hardware and software components of the safety management and control system applied to the emergency material storage park provided by an embodiment of the present application. DETAILED DESCRIPTION

[0013] The present application will be specifically described below in combination with the drawings of the specification, Figure 1 is a flow schematic diagram of the safety management and control method applied to the emergency material storage park provided by an embodiment of the present application, and the safety management and control method applied to the emergency material storage park will be described in detail below.

[0014] Step S110: constructing a safety state link of the emergency material storage park, the safety state link being based on multi-source monitoring information, containing material storage nodes, environment nodes and activity nodes, and each node being connected to form a mesh structure through an associated edge.

[0015] This embodiment takes a comprehensive emergency material storage park as the application scenario, which covers a cold-chain medical material warehouse, an explosion-proof disaster relief equipment warehouse, a constant-temperature food storage warehouse, and a comprehensive management area, and needs to construct a safety state link with dynamic correlation capability based on multi-dimensional monitoring data, and realize structured construction through steps S111 to S118.

[0016] Step S111: Collect multi-source monitoring information of the emergency material storage park, which covers material storage related information, park environment related information, and personnel activity related information.

[0017] A distributed monitoring network is deployed, and three types of professional monitoring subsystems are integrated: the material storage monitoring subsystem uses ultra-high frequency RFID tags and shelf pressure sensors to collect information such as SKU codes of materials, storage location coordinates, stacking layers, packaging tightness, and shelf load distribution; the environmental monitoring subsystem is configured with temperature and humidity transmitters, gas detection modules, wind speed and direction sensors, and illuminance meters to collect information such as temperature and humidity gradient distribution, volatile organic compound concentration, flammable gas concentration, air replacement rate, and external meteorological parameters; the personnel activity monitoring subsystem integrates video behavior analysis and UWB positioning technology to collect information such as personnel identity authentication information, real-time positioning coordinates, regional stay time, operation action sequence, and path deviation.

[0018] The monitoring data is transmitted to the edge computing gateway through industrial Ethernet, and after data cleaning and format conversion, it is uploaded to the central control platform, which establishes a distributed database according to the three categories of "storage-environment-activity".

[0019] Step S112: Extract core representation information of material storage nodes from multi-source monitoring information, which reflects the current state of material storage.

[0020] The node feature extraction engine is enabled on the central control platform, and the data processing domain is divided according to the function of the warehouse: the candidate data of the material storage nodes of the cold-chain medical material warehouse focus on low-temperature storage adaptability, and the core representation information includes drug cold-chain temperature zone compliance rate, packaging integrity (seal film damage rate, vacuum degree), inventory turnover rate (frequency of warehouse entry and exit in the past 30 days), and shelf utilization rate (actual storage capacity / rated storage capacity); the candidate data of the material storage nodes of the explosion-proof disaster relief equipment warehouse focus on safety storage compliance, and the core representation information includes equipment explosion-proof level matching degree, metal component corrosion degree, vulnerable component aging index, and stacking stability (center of gravity offset); the candidate data of the material storage nodes of the constant-temperature food storage warehouse focus on preservation storage state, and the core representation information includes packaging airtightness, shelf ventilation coverage, inventory expiration date proportion, and pest control level.

[0021] A unique digital identifier is assigned to each material storage node, and the identification coding rules include warehouse area code, storage location partition, and node serial number, ensuring the spatial positioning and data traceability of the node.

[0022] Step S113: Extract the core characteristic information of the environmental node, which reflects the environmental state of different areas in the park.

[0023] The environmental monitoring unit is divided according to the physical area, and each unit corresponds to an environmental node candidate data set: the core characteristic information of the environmental node of the cold-chain medical material warehouse includes temperature and humidity control accuracy (actual value and set value deviation), air conditioning system operation load rate, cold loss rate, and fresh air ventilation frequency; the core characteristic information of the environmental node of the explosion-proof disaster relief equipment warehouse includes combustible gas concentration alarm threshold compliance rate, ventilation system air pressure distribution, static electricity elimination device operating status, and dust concentration; the core characteristic information of the environmental node of the comprehensive management area includes public area temperature and humidity, lighting system energy consumption, and fire passage smoke concentration.

[0024] The core characteristic information of the environmental node needs to be attached with the monitoring point coordinates and data collection time stamp to form a time and space related feature matrix, for example, "Cold-chain warehouse A area environmental node-202509011000-temperature and humidity gradient: [0-1m layer 2℃ / 50%RH, 1-2m layer 3℃ / 52%RH]".

[0025] Step S114: Extract the core characteristic information of the activity node, which reflects the activity state of personnel in the park.

[0026] Based on the personnel activity monitoring data, a behavior characteristic model is constructed, and the core characteristic information is extracted according to the operation type: the material loading and unloading activity node extracts operation specification (forklift operation path deviation, lifting height compliance), loading and unloading efficiency (unit time loading and unloading amount), and safety protection measures implementation (safety helmet wearing, protective gloves use); the equipment maintenance activity node extracts maintenance qualification matching degree (personnel qualification and equipment type), tool use specification, and maintenance process integrity (power-off-detection-maintenance-acceptance link missing rate); the material inventory activity node extracts inventory accuracy (accounting discrepancy rate), operation time, and warehouse interference degree (personnel flow during inventory).

[0027] In the multi-person collaborative operation scenario, the operation teams are combined into cluster activity nodes, and independent activity nodes are set up for special operations (such as explosion-proof equipment maintenance), and the node identifier is associated with the operation work order number and the identity information of the person in charge.

[0028] Step S115: Analyze the influence relationship between the material storage node and the environment node, and establish the association edge between the material storage node and the environment node according to the state association characteristics of the two in the multi-source monitoring information. The attribute of the association edge is determined by the influence degree of the two.

[0029] Enable association rule mining algorithm to perform time sequence correlation analysis on the feature data of the material storage node and the environment node: in the cold chain medical material warehouse, the "cold chain temperature zone compliance rate" of the material storage node and the "air conditioning system operation load rate" of the environment node have strong positive correlation. When the air conditioning load rate is lower than the threshold, the temperature zone compliance rate decreases by more than a preset proportion; in the explosion-proof disaster relief equipment warehouse, the "metal component corrosion degree" of the material storage node and the "air humidity" of the environment node have moderate positive correlation. When the humidity increases by a certain range, the corrosion degree index increases in the corresponding interval; in the constant temperature food storage warehouse, the "packaging airtightness" of the material storage node and the "air replacement rate" of the environment node have weak negative correlation. Excessive replacement rate will slightly accelerate the aging of the packaging.

[0030] Based on the correlation strength calculation result, define the influence coefficient. Strong influence corresponds to an influence coefficient greater than 0.8, moderate influence corresponds to 0.5-0.8, and weak influence corresponds to 0.2-0.5. The association edge is established between the corresponding nodes, and the edge attribute is marked with the influence coefficient and the association characteristic description, such as "cold chain storage node-environment node: influence coefficient 0.85-air conditioning load influences temperature zone compliance rate".

[0031] Step S116: Analyze the influence relationship between the environment node and the activity node, and establish the association edge between the environment node and the activity node according to the state association characteristics of the two in the multi-source monitoring information. The attribute of the association edge is determined by the influence degree of the two.

[0032] Analyze the interaction relationship between the environment node and the activity node through the behavior-environment coupling model: the "combustible gas concentration" of the explosion-proof warehouse environment node and the "forklift operation frequency" of the loading and unloading activity node have strong correlation. When the gas concentration exceeds the warning value, the spark risk generated by the forklift operation significantly increases; the "temperature fluctuation amplitude" of the cold chain warehouse environment node and the "operation time length" of the inventory activity node have moderate correlation. Temperature fluctuations will prolong the operation adaptation time of the inventory personnel; the "rainfall intensity" of the outdoor environment node and the "transport stability" of the equipment transfer activity node have strong correlation. Rainfall causes the road friction coefficient to decrease, and the transfer equipment overturning risk increases.

[0033] Calculate the influence weight of the environment on the activity. The strong influence weight is set to 1.0, the moderate influence is 0.7, and the weak influence is 0.3. The association edge attribute is marked with the weight value and the risk type, such as "explosion-proof warehouse environment-loading and unloading activity: weight 1.0-combustible gas + forklift operation spark risk".

[0034] Step S117: Analyze the indirect influence relationship between the material storage node and the active node, establish an indirect association edge between the material storage node and the active node through the environment node as an intermediate transmission carrier, and the attribute of the association edge is determined by the degree of indirect influence.

[0035] Identify the indirect influence chain across nodes: the material storage node of the cold chain medical material warehouse (vaccine needs -20℃ storage) indirectly influences the inventory activity node through the environment node (air conditioner temperature setting), the low temperature requirement of the storage node causes the environment node to maintain a low temperature state, which in turn affects the operation endurance and protective equipment configuration of the inventory personnel, forming a transmission path of "storage requirement→environment parameter→activity constraint"; the material storage node of the explosion-proof equipment warehouse (explosion-proof lamp storage) indirectly influences the maintenance activity node through the environment node (static elimination state), the explosion-proof requirement of the storage node determines the static control standard of the environment node, which in turn affects the static protection operation process of the maintenance personnel.

[0036] The degree of indirect influence is calculated by the product of two-level direct influence coefficients, such as storage-environment influence coefficient 0.9, environment-activity influence coefficient 0.8, indirect influence coefficient 0.72, corresponding to medium-strong influence level, the association edge is labeled with indirect influence coefficient and transmission path, such as "vaccine storage node-inventory activity node: 0.72-conducted through cold chain environment node".

[0037] Step S118: Integrate the material storage node, environment node, activity node and each association edge according to the mesh structure to form the safety state link of the emergency material storage park, each node in the safety state link is attached with corresponding characteristic information identifier, and each association edge is attached with corresponding influence attribute identifier.

[0038] In the visualization system of the central control platform, the digital twin scene of the park is constructed based on the BIM model, and the node icons are deployed according to the actual position: the material storage node adopts a cubic icon, the environment node adopts a circular icon, and the activity node adopts a human icon. The association edge is drawn with different line widths according to the influence coefficient, the coefficient greater than 0.8 is 3px line width, 0.5-0.8 is 2px, 0.2-0.5 is 1px, and the indirect association edge adds a dashed texture.

[0039] The node and the association edge are both bound with metadata tags, clicking the node can view the real-time data and historical trend curve of the core characteristic information, and clicking the association edge can view the influence coefficient calculation basis and association rule details. The link data is stored in the attribute graph database, which supports Cypher statement query of node association relationship.

[0040] Step S120: Track the evolution process of the safety state link, record the state information of each node and association edge changing with time, and generate link evolution data.

[0041] Through time series data acquisition and state transition analysis, the dynamic tracking of the safety state link is realized, and the specific process is steps S121 to S126.

[0042] Step S121: Set the tracking period of the safety state link, and continuously collect the change of the characteristic information of each node according to the tracking period, and the change of the characteristic information reflects the change of the node state over time.

[0043] Based on the time sensitivity of the node characteristics, the differentiated tracking period is set: the characteristic change of the material storage node is slow (such as the degree of rust and the aging of the package), the tracking period is set to once every hour; the characteristic fluctuation of the environment node is moderate (such as temperature and humidity, gas concentration), the tracking period is set to once every 15 minutes; the characteristic change of the active node is frequent (such as operation behavior and position coordinates), the tracking period is set to once every 1 minute.

[0044] The monitoring system collects node characteristic information according to the period, and obtains the change amount through difference calculation, such as "cold chain storage node-temperature zone compliance rate: from 98% to 95%, change amount-3%" "explosion-proof warehouse environment node-flammable gas concentration: from 5ppm to 8ppm, change amount+3ppm" "loading and unloading active node-operation path deviation: from 0.5m to 1.2m, change amount+0.7m", and the change data is stored in the time series database with the collection timestamp.

[0045] Step S122: Collect the attribute change of each associated edge, and the attribute change reflects the change of the influence degree between nodes over time.

[0046] Synchronize the change of the influence coefficient of the associated edge, and analyze the influence of the node state change on the association strength: in the cold chain warehouse, when the air conditioner load rate of the environment node decreases from 80% to 60%, the influence coefficient of the associated edge between it and the storage node increases from 0.85 to 0.92; in the explosion-proof warehouse, when the static electricity elimination device of the environment node returns to normal operation, the influence coefficient of the associated edge between it and the maintenance active node decreases from 0.75 to 0.3; in the food warehouse, when the air replacement rate of the environment node increases from 1 time / hour to 3 times / hour, the influence coefficient of the associated edge between it and the storage node increases from 0.3 to 0.55.

[0047] Record the change range, change time and trigger condition of the influence coefficient, and form the evolution log of the associated edge attribute, such as "cold chain storage-environment associated edge: 0.85→0.92-trigger condition: air conditioner load rate decreases to 60%".

[0048] Step S123: According to the change of the node characteristic information, the evolution stage of the node is divided, and different evolution stages correspond to different change characteristics of the node characteristic information.

[0049] A quantitative division model of the evolution stage of the node is established: the steady state maintenance stage refers to that the absolute value of the change amount of the characteristic information is less than a threshold value (such as the storage node change amount <1%, the environment node <5%, and the activity node <10%) in three consecutive periods; the slow degradation stage refers to that the characteristic information changes in the degradation direction in three consecutive periods, and the absolute value of the change amount in a single period is between 1-2 times of the threshold value; the accelerated degradation stage refers to that the absolute value of the change amount in a single period is more than 2 times of the threshold value, and the change amount in two consecutive periods is in an increasing trend; and the recovery and improvement stage refers to that the characteristic information returns to the baseline value from the degradation state, and the change amount in two consecutive periods is a positive improvement value.

[0050] For example, the change amount of the temperature zone compliance rate of the cold chain storage node in three consecutive periods is -0.5%, -0.8%, and -1.0%, and the node is in the slow degradation stage; the combustible gas concentration of the anti-explosion warehouse environment node changes from 3ppm to 7ppm in a single period, and the node is in the accelerated degradation stage; and the change amount of the operation path deviation of the loading and unloading activity node is -0.6m and -0.4m after intervention, and the node is in the recovery and improvement stage.

[0051] Step S124: According to the change of the associated edge attribute, the evolution stage of the associated edge is divided, and different evolution stages correspond to different change characteristics of the associated edge attribute.

[0052] The evolution stage of the associated edge is defined based on the dynamic change of the influence coefficient: the strength stability stage refers to that the absolute value of the change amount of the influence coefficient in three consecutive periods is less than 0.05; the strength enhancement stage refers to that the influence coefficient is in an increasing trend in two consecutive periods, and the total increase is greater than 0.1; the strength attenuation stage refers to that the influence coefficient is in a decreasing trend in two consecutive periods, and the total decrease is greater than 0.1; the strength mutation stage refers to that the absolute value of the change amount of the influence coefficient in a single period is greater than 0.2; and the dormant stage refers to that the influence coefficient is continuously less than 0.2 and has no obvious change.

[0053] For example, the influence coefficient of the cold chain storage-environment associated edge in three consecutive periods is 0.85, 0.86, and 0.84, and the node is in the strength stability stage; the influence coefficient of the anti-explosion environment-loading and unloading activity associated edge changes from 0.4 to 0.65, and the node is in the strength enhancement stage; the influence coefficient of the food storage-environment associated edge changes from 0.5 to 0.3, and the node is in the strength attenuation stage; and when the environment node suddenly leaks gas, the influence coefficient of the associated edge changes from 0.3 to 0.8, and the node is in the strength mutation stage.

[0054] Step S125: The correspondence between the node evolution stage and the associated edge evolution stage in each tracking period is recorded to form a period evolution record, and the period evolution record contains the state snapshot of each node and associated edge in the tracking period.

[0055] At the end of each tracking period, a node-association edge state matrix is generated: the rows represent nodes, the columns represent association edges, and the cells record the evolution stage and association characteristics of the two. At the same time, a state snapshot is generated, which includes the current representation information value of the node, the evolution stage label, the change trend graph, and the influence coefficient of the association edge, the evolution stage label and the trigger condition.

[0056] For example, in a certain period snapshot, the cold chain storage node is labeled “warm area compliance rate 95%-slow degradation”, the association edge is labeled “influence coefficient 0.92-intensity enhancement”, and a feature comparison graph between the previous period and the current period is attached. The period evolution record is archived in the history database in chronological order.

[0057] Step S126: The period evolution records of multiple tracking periods are integrated in chronological order to form link evolution data, which is used to reflect the overall change process of the safety state link in the continuous time dimension, and each evolution record of a period has a connection relationship with the evolution record of the previous period.

[0058] The period evolution record is integrated by time series data fusion technology to build a link evolution timeline: each time node corresponds to a state snapshot of a period, adjacent time nodes are connected through a state transition matrix, and the matrix elements record the evolution stage transition probability of nodes and association edges. For example, from period T1 to T2, the probability of the cold chain storage node transitioning from “steady state maintenance” to “slow degradation” is 0.6, and the probability of the association edge transitioning from “intensity stability” to “intensity enhancement” is 0.8.

[0059] The link evolution data is stored in a time series graph database, which supports querying the evolution trajectory of nodes and association edges by time interval, and can play the dynamic evolution process of the link state through a visualization system.

[0060] Step S130: Adapt the control action set based on the link evolution data, and the control action set has a corresponding relationship with the node state of the safety state link.

[0061] Through multi-dimensional abnormality analysis and action matching, a targeted control action set is generated, and the specific process is steps S131 to S135.

[0062] Step S131: Analyze the node evolution stage in the link evolution data to identify nodes in the non-stable evolution stage, and the non-stable evolution stage node has a representation information change amplitude that exceeds the normal range.

[0063] An abnormal node identification algorithm is enabled to traverse the node evolution stage in the link evolution data: nodes in the slow degradation stage, accelerated degradation stage and intensity mutation stage are determined as non-stable nodes, and nodes in the steady state maintenance stage and recovery improvement stage are determined as stable nodes.

[0064] Identify the nodes in the acceleration deterioration phase and the nodes with abnormal evolution trend: for example, the flammable gas concentration of the explosion-proof warehouse environment node is in the acceleration deterioration phase, which is determined as a high-priority unstable node; the temperature zone compliance rate of the cold-chain warehouse storage node is in the slow deterioration phase, but it has shown a downward trend for 5 consecutive periods, which is determined as a medium-priority unstable node; the operation path deviation of the loading and unloading activity node has shown a sudden mutation in intensity, which is determined as a high-priority unstable node.

[0065] Establish a list of unstable node priorities, including node identification, evolution phase, characteristic change amount, priority level, and influence range.

[0066] Step S132: For each unstable node in the unstable evolution phase, analyze the historical state change rule of the unstable node in the link evolution data to determine the target influencing factor that causes the unstable node to enter the unstable evolution phase.

[0067] Perform a deep trace analysis on each unstable node, with the process being steps S1321 to S1327:

[0068] Step S1321: Extract the characteristic information change sequence of the unstable node in multiple tracking periods in the link evolution data, which is arranged in chronological order and used to present the change process of the node state of the unstable node.

[0069] Select the characteristic information values of the unstable node for 10 consecutive tracking periods from the time series database and construct a change sequence in chronological order. For example, the change sequence of the explosion-proof warehouse environment node (flammable gas concentration) is [4ppm, 5ppm, 5ppm, 6ppm, 7ppm, 8ppm, 10ppm, 12ppm, 15ppm, 18ppm]; the change sequence of the cold-chain warehouse storage node (temperature zone compliance rate) is [99%, 99%, 98%, 98%, 97%, 96%, 95%, 94%, 93%, 92%]; and the change sequence of the loading and unloading activity node (operation path deviation) is [0.3m, 0.4m, 0.5m, 0.6m, 0.5m, 0.7m, 1.0m, 1.5m, 2.0m, 2.2m], which completely presents the process of the unstable node gradually entering the unstable state from the initial relatively stable state.

[0070] Step S1322: Divide the characteristic information change sequence into stable and unstable change phases to determine the time node at which the unstable node enters the unstable change phase from the stable change phase.

[0071] The sequence segmentation algorithm is called to divide the stages based on the mutation characteristics of the change amount of the characteristic information. For the combustible gas concentration sequence of the explosion-proof warehouse environment node, the change amount of the first 3 periods is less than 1 ppm, and it is determined that it is in a stable change stage; from the 4th period, the change amount increases to more than 1 ppm, and the subsequent trend is increasing, which is determined to be in a non-stable change stage, and the time node is the starting time of the 4th period.

[0072] For the temperature zone compliance rate sequence of the cold chain storage node, the first 2 periods have no change, the change amount of the 3rd-5th period is-1%, and it is at the edge of the stable change stage; from the 6th period, the change amount is maintained at-1% and continuously decreases, which is determined to enter a non-stable change stage, and the time node is the starting time of the 6th period.

[0073] For the operation path deviation sequence of the loading and unloading activity node, the change amount of the first 5 periods is less than 0.2 m, which is in a stable change stage; the change amount of the 6th period increases to 0.3 m, indicating that it enters a non-stable change stage, and the time node is the starting time of the 6th period.

[0074] Step S1323: comparing the characteristic differences of the characteristic information between the stable change stage and the non-stable change stage, extracting the characteristic information features corresponding to the non-stable change stage, and the characteristic information features unique to the non-stable change stage are the key marker features for distinguishing the stable change stage from the non-stable change stage.

[0075] Comparing the characteristics of the explosion-proof warehouse environment node in two stages: the stable stage has a gentle change in combustible gas concentration, a small fluctuation range, and no obvious upward trend; the non-stable stage has a stepwise increase in concentration, a gradual increase in single-period change amount, and a continuously increasing upward trend. The "stepwise increase and increasing change amount" are the marker features of this node.

[0076] Comparing the characteristics of the cold chain storage node: although the stable stage has a decrease in temperature zone compliance rate, the decrease is slow and not persistent; the non-stable stage has a continuous decrease in compliance rate with a fixed amplitude and no rebound, and the "continuous linear decrease and no fluctuation rebound" are the marker features.

[0077] Comparing the characteristics of the loading and unloading activity node: in the stable stage, the operation path deviation fluctuates around the reference value with a small amplitude, and the deviation range is controllable; in the non-stable stage, the deviation exceeds the controllable range, and the fluctuation amplitude shows an expanding trend. The "deviation exceeding the limit and fluctuation amplitude expanding" are the marker features.

[0078] Step S1324: analyzing the attribute changes of the associated edges of the non-stable node before and after the start of the non-stable change stage, identifying the associated nodes whose associated edge attributes have preset changes, and the associated nodes whose associated edge attributes have preset changes are the associated nodes that have an impact on the non-stable node through the associated edges.

[0079] View the associated edge of the explosion-proof warehouse environment node (combustible gas concentration): before the start of the unstable stage, its associated edge with the explosion-proof equipment storage node has an influence coefficient of 0.65 (medium influence); after entering the unstable stage, the influence coefficient of the associated edge rises to 0.85 (strong influence), and the influence coefficient of the associated edge with the loading and unloading activity node rises from 0.5 to 0.9. Thus, the explosion-proof equipment storage node and the loading and unloading activity node are identified as associated nodes, and both nodes have an impact on the unstable node through the change in the associated edge attribute.

[0080] Analyze the associated edge of the cold chain warehouse storage node: before the start of the unstable stage, the influence coefficient of the associated edge with the cold chain environment node is 0.8; after entering the unstable stage, the coefficient rises to 0.88, and the associated edge with the air conditioning equipment running node is activated from the dormant state (coefficient 0.1) to 0.75, so the cold chain environment node and the air conditioning equipment running node are associated nodes.

[0081] Analyze the associated edge of the loading and unloading activity node: before the start of the unstable stage, the influence coefficient of the associated edge with the forklift scheduling node is 0.4; after entering the unstable stage, the coefficient increases to 0.7, and the associated edge coefficient with the warehouse area path planning node rises from 0.3 to 0.6, so the forklift scheduling node and the warehouse area path planning node are associated nodes.

[0082] Step S1325: Analyze the change in external input information related to the unstable node in the multi-source monitoring information before and after the start of the unstable change stage, which is the change in external input information that directly affects the state of the node.

[0083] External input information includes device operating status, personnel operating behavior, external environmental interference, etc. For the explosion-proof warehouse environment node, 10 minutes before the start of the unstable stage, it is monitored that the sealing valve of the explosion-proof equipment storage area changes from "normal" to "half open", which is a device state change that leads to combustible gas leakage, and is direct external input information.

[0084] For the cold chain warehouse storage node, 5 minutes before the start of the unstable stage, it is monitored that the compressor operating power of the air conditioning system decreases from "rated power" to "70% rated power", and the compressor power drop leads to insufficient refrigeration capacity, which is external input information that causes the temperature zone compliance rate to decrease.

[0085] For the loading and unloading activity node, 3 minutes before the start of the unstable stage, it is monitored that the steering system sensor data of the forklift appears abnormal fluctuations, and the temporary passage of the warehouse area is changed due to material stacking, and the two external input information together cause the operation path deviation to increase.

[0086] Step S1326: The unique characterization information features of the unstable phase, the associated nodes that have an impact on the unstable node through the associated edges, and the changes in the external input information are integrated to filter out the factors that play a leading role in the unstable node entering the unstable evolution phase. The factors that play a leading role in the unstable node entering the unstable evolution phase are the target influencing factors that cause the node to enter the unstable evolution phase.

[0087] For the explosion-proof warehouse environment node, it is found through comprehensive analysis that the "seal valve half-open" in the external input information is the direct inducement of the increase in the concentration of flammable gas, and the "increased operation frequency of loading and unloading activity nodes" in the associated nodes accelerates gas diffusion. Both of them jointly cause the node to enter the unstable phase, in which the "abnormal running state of the seal valve" plays a leading role and is determined as the target influencing factor.

[0088] For the cold chain warehouse storage node, "air conditioner compressor power reduction" is the core reason for the continuous decline in the temperature zone compliance rate, and the associated node "cold chain environment node temperature and humidity control precision decline" further aggravates this trend, so "air conditioner compressor abnormal running power" is the target influencing factor.

[0089] For the loading and unloading activity node, "forklift steering system sensor abnormality" directly leads to a decline in operation accuracy, and "temporary change of warehouse area path" increases the operation difficulty, both of which play an important role, so "forklift steering system sensor abnormality" and "temporary change of warehouse area path" are both determined as the target influencing factors.

[0090] Step S1327: The target influencing factors are classified and arranged to determine the association logic between each target influencing factor and the change in the characterization information of the unstable node.

[0091] The target influencing factors are classified by type: device failure class (seal valve abnormality, air conditioner compressor abnormality, forklift sensor abnormality), and environmental change class (warehouse area path change).

[0092] The association logic is clear: the device failure class factors cause the node characterization information to deviate from the benchmark by affecting the normal transmission of materials or energy, such as seal valve abnormality leading to gas leakage → concentration increase, air conditioner compressor abnormality leading to insufficient refrigeration → temperature zone compliance rate decline, and forklift sensor abnormality leading to operation out of control → path deviation increase; the environmental change class factors change the operation constraint conditions and force the node characterization information to adjust, such as warehouse area path change leading to forced deviation of the operation path → deviation increase.

[0093] Step S133: According to the target influencing factors, the types of control actions that can act on the unstable node are filtered out, and each type of control action corresponds to one or more target influencing factors.

[0094] The action type is filtered through the matching mechanism of the target influencing factor and the control action. The process is steps S1331 to S1337:

[0095] Step S1331: Establish a mapping library of influencing factors and control action types, which pre-stores the control action types corresponding to different influencing factors in the mapping library. Each control action type is labeled with the range of influencing factors it can act on.

[0096] The mapping library stores the mapping relationship according to the type of influencing factors: device failure factors correspond to control action types such as "emergency repair of equipment", "switching of standby equipment", "parameter recalibration", etc. For example, "abnormal sealing valve" corresponds to "emergency valve closure + repair" and "activation of standby valve"; "abnormal air conditioner compressor" corresponds to "compressor power regulation" and "activation of standby compressor"; "abnormal forklift sensor" corresponds to "sensor calibration" and "forklift shutdown for repair".

[0097] Environmental change factors correspond to control action types such as "path replanning", "temporary obstacle removal", and "operation process adjustment". For example, "warehouse area path change" corresponds to "temporary path planning in warehouse area" and "material re-stacking".

[0098] Each control action type is labeled with an action range, such as "emergency valve closure + repair" for all sealing valve failures, and "sensor calibration" for sensor abnormalities of various devices.

[0099] Step S1332: Match the determined target influencing factor with the influencing factors in the mapping library of control action types, and preliminarily filter out the control action types corresponding to the target influencing factor.

[0100] Match the target influencing factor "abnormal running state of sealing valve" of the explosion-proof warehouse environment node with the mapping library, and preliminarily filter out two control action types: "emergency valve closure + repair" and "activation of standby valve".

[0101] Match the target influencing factor "abnormal running power of air conditioner compressor" of the cold chain warehouse storage node, and preliminarily filter out two control action types: "compressor power regulation" and "activation of standby compressor".

[0102] Match the target influencing factors "abnormal sensor of forklift steering system" and "temporary change of warehouse area path" of the loading and unloading activity node, and filter out four control action types: "sensor calibration", "forklift shutdown for repair", "temporary path planning in warehouse area", and "material re-stacking".

[0103] Step S1333: Analyze the application effect of the preliminarily screened management and control action type in the historical link evolution data, and view whether the preliminarily screened management and control action type can effectively guide the node to transit from the non-stable evolution stage to the stable evolution stage when acting on the same type of node in the past.

[0104] The management and control action execution record of "valve emergency shutdown + maintenance" in the historical link evolution data is called. In the past 3 times of application to the same type of sealing valve abnormal scene, it can make the combustible gas concentration decrease within 1 tracking period, guide the node to enter the recovery improvement stage, and the application effect is effective. In 2 applications, 1 time is effective, and 1 time is delayed due to the failure of the standby valve to be debugged in time.

[0105] In 5 applications, "compressor power regulation" successfully restored the air conditioning load rate to 4 times, and the temperature zone compliance rate improved. In 3 applications, "standby compressor start" can quickly restore the refrigerating capacity, and the effect is stable.

[0106] In 4 applications, "sensor calibration" restored the forklift operation accuracy 3 times. "Forklift shutdown for maintenance" has a significant effect in 2 serious fault scenarios. "Temporary path planning in the warehouse area" can effectively reduce the operation deviation in 6 path change scenarios. "Material re-stacking" is relatively delayed in 3 applications due to the long time consumption.

[0107] Step S1334: If any one of the management and control action types can effectively guide the node to transit from the non-stable evolution stage to the stable evolution stage in the historical application, and its target influencing factor matches the target influencing factor of the current non-stable node, the management and control action type is determined as a candidate management and control action type.

[0108] Based on the historical effect analysis, "valve emergency shutdown + maintenance" completely matches the target influencing factor of the explosion-proof warehouse environment node and has effective effect, and is determined as a candidate management and control action type. "Standby compressor start" matches the target influencing factor of the cold chain storage node and has stable effect, and is determined as a candidate management and control action type. "Sensor calibration" and "temporary path planning in the warehouse area" match the target influencing factor of the loading and unloading activity node and have good effect, and are determined as candidate management and control action types.

[0109] Step S1335: If any one of the management and control action types cannot effectively guide the node to transit from the non-stable evolution stage to the stable evolution stage in the historical application, but can adapt to the target influencing factor of the current node by adjusting its action mode, the action mode of the management and control action type is adjusted, and the adjusted management and control action type is included in the candidate management and control action type.

[0110] Analysis of the "backup valve activation" action: In the history application, the effect is delayed due to debugging problems, and the adjustment action is "backup valve activation + immediate pressure test", to ensure that the sealing performance meets the standard after activation. The adjusted action is included in the candidate control action type of the explosion-proof warehouse environment node.

[0111] Analysis of the "material restacking" action: In the history application, the time is longer, and it is adjusted to "regional temporary stacking + priority cleaning main path", to shorten the operation time. The adjusted action is included in the candidate control action type of the loading and unloading activity node.

[0112] Step S1336: If there is no type of the target influencing factor that can adapt to the current unstable node in the preliminary screened control action type, a new control action type is constructed based on the characteristics of the target influencing factor of the current unstable node.

[0113] All preliminary screening actions are checked, and no situation that cannot be adapted is found, so there is no need to construct a new control action type. If there is a special target influencing factor (such as a new type of equipment failure), a new action is designed based on the factor characteristics, such as "equipment manufacturer remote diagnosis + on-site quick maintenance", to clearly define the execution process and required resources of the action.

[0114] Step S1337: The candidate control action type and the newly constructed control action type are integrated to form a control action type set for the unstable node. Each control action type is clearly labeled with its corresponding target influencing factor and action mode.

[0115] After integration: the control action type set of the explosion-proof warehouse environment node is { "valve emergency shutdown + maintenance - corresponding to sealing valve abnormality" "backup valve activation + immediate pressure test - corresponding to sealing valve abnormality"}; the set of the cold chain storage node is { "backup compressor start - corresponding to air conditioner compressor power abnormality"}; and the set of the loading and unloading activity node is { "sensor calibration - corresponding to forklift sensor abnormality" "temporary path planning in warehouse area - corresponding to temporary path change" "material restacking (regional) - corresponding to temporary path change"}.

[0116] Step S134: Analyze the association edge attributes between the unstable node and other nodes. If the unstable state of the unstable node is transmitted to other associated nodes through the association edge, the corresponding control action type is synchronized for the corresponding associated node.

[0117] View the associated edge of the explosion-proof warehouse environment node: its associated edge with the explosion-proof equipment storage node has an influence coefficient of 0.85 (strong influence), and the unstable state has caused the "equipment explosion-proof level matching degree" of the storage node to drop to the critical value, and needs to be synchronized with the screening control action. Based on the target influencing factor "high concentration of flammable gas" of the storage node, two control action types "temporary sealing protection of equipment" and "strengthening ventilation in explosion-proof area" are screened from the mapping library.

[0118] Analyze the associated edge of the cold chain storage node: its indirect associated edge coefficient with the inventory activity node is 0.72, and the unstable state causes the inventory personnel operation endurance to decrease, and the "inventory personnel rotation" and "protection equipment upgrade" control action types are screened synchronously.

[0119] Analyze the associated edge of the loading and unloading activity node: its associated edge coefficient with the material storage node is 0.6, and the unstable state does not cause significant influence on the storage node, and no action needs to be screened synchronously.

[0120] Step S135: According to the evolution urgency of each unstable node in the link evolution data, the execution priority of the control action is determined, the screened control action types are sorted according to the execution priority, and the corresponding nodes and associated edges of each control action type are determined to form a control action set, and each control action in the control action set forms a corresponding relationship with the corresponding node in the safety state link.

[0121] Define the evolution urgency determination standard: the acceleration degradation stage and the influence coefficient > 0.8 are urgent, the slow degradation stage and the influence coefficient 0.5-0.8 are general, and the intensity mutation stage is urgent.

[0122] Determine the urgency of each node: the explosion-proof warehouse environment node (acceleration degradation, coefficient 0.85) is urgent; the cold chain storage node (slow degradation, coefficient 0.88) is general; the explosion-proof equipment storage node (affected by the urgent node) is urgent; the inventory activity node (affected by the general node) is general.

[0123] Determine the execution priority: the control action of the urgent node is prior to the general node, and the action with better effect in the same node is prior. The sorting result is: 1. The explosion-proof warehouse environment node "valve emergency closing + maintenance"; 2. The explosion-proof equipment storage node "temporary sealing protection of equipment" and "strengthening ventilation in explosion-proof area"; 3. The explosion-proof warehouse environment node "backup valve activation + immediate pressure test"; 4. The cold chain storage node "backup compressor start"; 5. The inventory activity node "inventory personnel rotation" and "protection equipment upgrade"; 6. The loading and unloading activity node "sensor calibration" "temporary path planning in warehouse area" "material re-stacking (regional)".

[0124] The corresponding node identifier and associated edge information are labeled for each control action, such as "valve emergency shutdown + maintenance - corresponding node: explosion-proof warehouse environment node, associated edge: explosion-proof warehouse environment - explosion-proof equipment storage (coefficient 0.85)", and a complete set of control actions is formed.

[0125] Step S140: sending the control action set to the park control system, driving the execution device to execute the control action, and collecting feedback data of the safety state link after execution.

[0126] Through the cooperative execution of the control system and the execution device, the control action is implemented and feedback data is collected, and the process is steps S141 to S147:

[0127] Step S141: obtaining the function attribute of each execution device in the park control system, the function attribute of the execution device reflecting the type of control action that the execution device can execute, and different execution devices corresponding to different ranges of control action types.

[0128] The execution device of the park control system includes: valve control device (responsible for valve opening and closing, maintenance control), ventilation control device (responsible for ventilation system start and stop, air volume regulation), device switching device (responsible for standby device start, parameter adjustment), personnel scheduling device (responsible for personnel scheduling, instruction issuing), path planning device (responsible for warehouse path design, label update), material handling device (responsible for material stacking, shifting).

[0129] The function attribute of each execution device is stored in the device archive of the control system, such as the function attribute of the valve control device being "supporting remote opening and closing of various industrial valves, fault diagnosis and maintenance process control", which can execute "valve emergency shutdown + maintenance" and "standby valve activation" actions; the function attribute of the ventilation control device is "supporting ventilation equipment start and stop, air volume grading regulation and regional directional ventilation", which can execute "explosion-proof area ventilation intensification" action.

[0130] Step S142: analyzing the type and execution requirement of each control action in the control action set, and assigning each control action to the corresponding execution device according to the function attribute of the execution device.

[0131] The execution requirement of the control action is analyzed: "valve emergency shutdown + maintenance" requires valve closing and starting maintenance process to be completed within 10 minutes; "temporary sealing protection of equipment" requires equipment sealing to be completed within 20 minutes; "standby compressor start" requires standby equipment to be started and adjusted to rated power within 5 minutes.

[0132] According to the function attribute matching execution device: "valve emergency closing + maintenance" "backup valve enable + immediate pressure test" is allocated to the valve control device; "equipment temporary sealing protection" "explosion-proof area ventilation enhancement" is allocated to the ventilation control device + material handling device (cooperative work); "backup compressor start" is allocated to the equipment switching device; "inventory personnel rotation" "protection equipment upgrade" is allocated to the personnel scheduling device; "sensor calibration" is allocated to the equipment switching device (including sensor calibration module); "temporary path planning in the warehouse area" is allocated to the path planning device; "material restacking (by area)" is allocated to the material handling device.

[0133] Step S143: The control action allocation result and the execution sequence of the control action are sent to the park control system, so that the park control system sends execution instructions to each execution device according to the execution sequence, and the execution instructions contain specific execution parameters (such as valve closing pressure threshold, ventilation system air volume level, compressor startup power parameter) and execution time requirements (such as "valve emergency closing + maintenance" needs to be started within 10 minutes, "backup compressor start" needs to be completed within 5 minutes) of the control action.

[0134] After the park control system receives the allocation result and the execution sequence, it generates an instruction queue according to the priority, and sends execution instructions to each execution device through an industrial bus interface. For example, the valve control device is sent an instruction "execute action: valve emergency closing + maintenance; target valve: explosion-proof warehouse A area sealing valve; execution parameter: closing pressure threshold XX; execution time limit: start within 10 minutes"; the equipment switching device is sent an instruction "execute action: backup compressor start; target device: cold chain warehouse B area backup compressor; execution parameter: startup power XX, target temperature XX; execution time limit: complete within 5 minutes".

[0135] After the execution device receives the instruction, it starts the built-in action execution module, executes the control action according to the execution parameter and the time requirement, and feeds back the execution progress to the park control system through the real-time communication interface, such as "valve control device - execution progress: valve has been closed, maintenance process has been started" "equipment switching device - execution progress: backup compressor has been started, power adjustment is in progress".

[0136] Step S144: The running state information of the execution device itself is recorded in real time during the execution process of the control action, and the running state information of the execution device itself reflects the process characteristics of the execution device executing the control action.

[0137] Each execution device is equipped with a state monitoring sensor to collect real-time operation state information: the valve control device records the response time of valve closing, the sealing pressure after closing, and the operation parameters of the maintenance tool; the ventilation control device records the starting current of the fan, the air volume output value, and the wind pressure stabilization time; the equipment switching device records the starting current of the compressor, the running speed, and the output power fluctuation; the personnel scheduling device records the instruction issuing time, the personnel response time, and the number of protective equipment allocation; the path planning device records the path generation time, the identification update progress, and the terminal device synchronization state; the material handling device records the running speed of the handling machinery, the load weight, and the operation path deviation.

[0138] These state information is collected at a frequency of 1 second / time, uploaded to the park control system through the edge computing gateway, and stored in association with the corresponding control action identifier by the system to form an execution process log, for example, "valve control device-action identifier: V001-operation state: response time XX, sealing pressure XX, maintenance tool operation normal".

[0139] Step S145: After the control action is executed, the change of the characteristic information of the corresponding node in the safety state link is collected, which reflects the influence effect of the control action on the node state.

[0140] The node characteristic information collection is completed through the following sub-steps:

[0141] Step S1451: Determine the target node corresponding to the control action, which is the node in the safety state link acted on by the control action.

[0142] According to the type and execution object of the control action, the corresponding target node is determined: the target node corresponding to "valve emergency closing + maintenance" is the explosion-proof warehouse environment node (flammable gas concentration); the target node corresponding to "backup compressor starting" is the cold chain warehouse storage node (temperature zone compliance rate); the target node corresponding to "sensor calibration" is the loading and unloading activity node (operation path deviation); the target node corresponding to "temporary sealing protection of equipment" is the explosion-proof equipment storage node (equipment explosion-proof level matching degree).

[0143] Step S1452: Set the characteristic information collection window, the time range of which covers the period from the completion of the control action execution to the stabilization of the node state.

[0144] The collection window is set according to the state response characteristics of different nodes: the state stabilization time of the environment node and the equipment node is relatively long, and the collection window is set to be within 1 hour after the action is completed; the state stabilization time of the active node is relatively short, and the collection window is set to be within 15 minutes after the action is completed. For example, after the completion of “valve emergency shutdown + maintenance”, the collection window of the explosion-proof warehouse environment node is 1 hour; after the completion of “sensor calibration”, the collection window of the loading and unloading active node is 15 minutes.

[0145] Step S1453: In the characterization information collection window, the characterization information of the target node is continuously collected according to the preset collection interval, and the collected characterization information of the target node is arranged to form a characterization information change sequence in chronological order.

[0146] In the collection window, the environment node collects at an interval of 5 minutes / time, the equipment node collects at an interval of 10 minutes / time, and the active node collects at an interval of 1 minute / time. For example, the combustible gas concentration change sequence of the explosion-proof warehouse environment node in the collection window is [18ppm, 15ppm, 12ppm, 10ppm, 8ppm, 8ppm]; the temperature zone compliance rate change sequence of the cold chain warehouse storage node is [92%, 93%, 95%, 97%, 98%, 98%]; and the operation path deviation change sequence of the loading and unloading active node is [2.2m, 1.8m, 1.2m, 0.8m, 0.6m, 0.5m].

[0147] Step S1454: Comparing the characterization information baseline sequence of the target node before the execution of the control action with the characterization information change sequence after the execution of the control action, extracting the difference features therebetween, and the difference features reflect the influence of the control action on the node state.

[0148] The baseline sequence before the execution of the control action is the characterization information of the previous 10 cycles, for example, the baseline sequence of the explosion-proof warehouse environment node is [4ppm, 5ppm, 5ppm, 6ppm, 7ppm, 8ppm, 10ppm, 12ppm, 15ppm, 18ppm], the change sequence after the execution is [18ppm, 15ppm, 12ppm, 10ppm, 8ppm, 8ppm], and the difference feature is “the concentration continuously decreases from 18ppm to 8ppm and stabilizes, and the decrease amplitude is 10ppm”.

[0149] The baseline sequence of the cold chain warehouse storage node is [99%, 99%, 98%, 98%, 97%, 96%, 95%, 94%, 93%, 92%], the change sequence after the execution is [92%, 93%, 95%, 97%, 98%, 98%], and the difference feature is “the compliance rate increases from 92% to 98% and stabilizes, and the increase amplitude is 6%”.

[0150] Step S1455: Analyze the change direction of the difference feature. If the difference feature shows that the node representation information changes towards the stable direction, it indicates that the control action has a positive impact effect. If the difference feature shows that the node representation information still changes towards the unstable direction, it indicates that the impact effect of the control action does not meet the expectation.

[0151] The concentration of the explosion-proof warehouse environment node changes from an upward trend to a downward and stable trend, which changes towards the stable direction, indicating that the "valve emergency shutdown + maintenance" has a positive impact effect. The compliance rate of the cold chain warehouse storage node changes from a downward trend to an upward and stable trend, indicating that the "backup compressor startup" has a positive impact effect. If the representation information of the target node continues to deteriorate (such as the concentration continues to rise) after the execution of a certain control action, it is determined that the effect does not meet the expectation.

[0152] Step S1456: Integrate the representation information change sequence, difference feature, and impact effect judgment result to form a record of the representation information change of the target node.

[0153] The record content includes: target node identification, control action identification, sequence comparison before and after execution, difference feature description, and effect judgment result. For example, "target node: explosion-proof warehouse environment node - action identification: V001 - execution sequence before: [4ppm...18ppm] - execution sequence after: [18ppm...8ppm] - difference feature: 10ppm decrease and stable - effect: positive".

[0154] Step S146: Collect the associated edge attribute change of the corresponding node in the safety state link, which reflects the impact effect of the control action on the influence relationship between nodes.

[0155] After the execution of the control action is completed, the influence coefficient changes of the target node associated edge: the influence coefficient of the explosion-proof warehouse environment node (combustible gas concentration) associated edge with the explosion-proof device storage node decreases from 0.85 to 0.6, and the influence coefficient with the loading and unloading activity node decreases from 0.9 to 0.5; the influence coefficient of the cold chain warehouse storage node associated edge with the cold chain environment node decreases from 0.88 to 0.8, and the influence coefficient with the air conditioning device running node decreases from 0.75 to 0.3.

[0156] Record the attribute change of the associated edge, including associated edge identification, target node identification, influence coefficient before and after execution, and change amplitude, such as "associated edge: explosion-proof warehouse environment - explosion-proof device storage - execution coefficient before: 0.85 - execution coefficient after: 0.6 - change amplitude: -0.25", to form a record of the attribute change of the associated edge.

[0157] Step S147: Integrate the running state information of the execution device, the change of the characteristic information of the corresponding node in the security state link, and the change of the associated edge attribute associated with the corresponding node in the security state link to form feedback data of the execution security state link; upload the feedback data of the execution security state link to the park control system in real time, and the feedback data of the execution security state link is attached with the corresponding control action identifier and execution time identifier.

[0158] Integrate the three types of information according to the control action identifier. Each feedback data includes: control action identifier, execution time, execution device running state log, target node characteristic information change record, and associated edge attribute change record. For example, “action identifier: V001 - execution time: 202509011430 - execution device state: valve control device running normally - target node change: explosion-proof warehouse environment concentration decreased by 10 ppm - associated edge change: coefficient decreased by 0.25”.

[0159] Upload the feedback data to the park control system in real time through the 5G industrial module, store it in the feedback database, and establish an associated index with the link evolution data.

[0160] Step S150: Calibrate the evolution trend of the security state link according to the feedback data.

[0161] Adjust the evolution benchmark of the security state link through matching analysis of the feedback data and the evolution rule. The specific process is steps S151 to S157.

[0162] Step S151: Establish the corresponding relationship between the feedback data and the security state link. The collected post-execution feedback data is associated to the material storage node, environment node, activity node and associated edge in the security state link according to the corresponding logic of control action, target node and associated edge, forming an action link feedback mapping table.

[0163] Analyze the control action identifier, target node identifier and associated edge identifier in the feedback data, and establish a mapping relationship according to the logic of “control action→target node→associated edge”. For example, the feedback data “action identifier V001 (valve emergency shutdown + maintenance)→target node: explosion-proof warehouse environment node→associated edge: explosion-proof warehouse environment - explosion-proof equipment storage, explosion-proof warehouse environment - loading and unloading activity” records the above corresponding relationship in the mapping table.

[0164] The mapping table includes four fields of action ID, target node ID, associated edge ID and feedback data index. Through the index, the corresponding execution device state, node change and associated edge change data can be quickly queried to realize the precise association of feedback data and link nodes.

[0165] Step S152: Call the basic evolution rule library of the security state link, which pre-stores node stable state characteristics and associated edge influence relationship benchmarks.

[0166] The basic evolution rule library stores rules classified by node type and associated edge type: node stable state characteristics include "explosion-proof warehouse environment node: combustible gas concentration <10 ppm and fluctuation <2 ppm" "cold chain warehouse storage node: temperature zone compliance rate >95% and no decline for 3 consecutive periods" "loading and unloading activity node: operation path deviation <1 m and fluctuation <0.3 m"; associated edge influence relationship benchmarks include "explosion-proof warehouse environment-explosion-proof equipment storage: influence coefficient <0.7 for stable association" "cold chain warehouse storage-cold chain environment: influence coefficient <0.8 for stable association".

[0167] These rules are based on historical data and industry standards and serve as benchmark criteria for determining whether the state of nodes and associated edges is stable.

[0168] Step S153: Match the actual node state and actual associated edge influence relationship in the action link feedback mapping table with the benchmark characteristics in the rule library to identify the to-be-updated nodes and to-be-updated associated edges that have differences in actual state and benchmark characteristics.

[0169] Compare the actual node state with the benchmark characteristics: the concentration of the explosion-proof warehouse environment node stabilizes at 8 ppm after the control action, which meets the benchmark characteristics (<10 ppm) and does not need to be updated; if the compliance rate of a certain cold chain warehouse storage node stabilizes at 94% after the action, which is lower than the benchmark characteristics (>95%), then the node is a to-be-updated node.

[0170] Compare the actual associated edge influence relationship with the benchmark: the actual coefficient of the explosion-proof warehouse environment-explosion-proof equipment storage is 0.6, which meets the benchmark (<0.7) and does not need to be updated; if the actual coefficient of a certain associated edge is 0.85, which is higher than the benchmark (<0.7), then the associated edge is a to-be-updated associated edge.

[0171] Record the IDs and difference contents of the to-be-updated nodes and associated edges, such as "to-be-updated node ID: L001-difference: temperature zone compliance rate 94% < benchmark 95%" "to-be-updated associated edge ID: E002-difference: coefficient 0.85 > benchmark 0.7".

[0172] Step S154: Based on the actual state of the to-be-updated nodes and to-be-updated associated edges, extract their new stable characteristics after the execution of the control action, and supplement the new stable characteristics to the evolution rule library and replace the original benchmark characteristics.

[0173] For the cold chain storage node with a compliance rate of 94%, analyze its actual stable state: the compliance rate has been maintained at 94%±0.5% for 5 consecutive periods, and the associated edge influence coefficient is stable at 0.75. Determine that this state is a new stable state, extract the feature "cold chain storage node: temperature zone compliance rate 94%±0.5% and associated edge coefficient <0.8", and supplement it to the evolution rule library and replace the original benchmark.

[0174] For the associated edge to be updated (coefficient 0.85), analyze its actual influence relationship: the coefficient has been stable at 0.85±0.03 for 3 consecutive periods, and has not caused the degradation of the target node. Extract the new benchmark "influence coefficient <0.9 is stable association" and update the corresponding benchmark in the rule library.

[0175] Step S155: Based on the updated evolution rule library, determine the stable direction and change direction of each node under the new benchmark, and the transmission path of the associated edge under the new influence logic, to form the calibrated evolution trend benchmark.

[0176] Based on the updated rule library, determine the stable direction of the node: the stable direction of the cold chain storage node to be updated is "maintain a compliance rate of 94%±0.5%", and the change direction needs to avoid falling below 93.5%; the transmission path of the associated edge to be updated needs to be adjusted to "when the coefficient is 0.85, the influence range is limited to the directly associated nodes, and does not transmit to the indirect nodes".

[0177] Integrate the stable direction, change direction and transmission path of all nodes and associated edges to form the calibrated evolution trend benchmark, which is more consistent with the actual link state after management.

[0178] Step S156: Call the historical feedback database to compare the actual evolution in the current action link feedback mapping table with the link evolution corresponding to the same type of management action in history, extract the evolution commonness rule in the same type of scene and integrate it into the calibrated evolution trend benchmark to form the final calibrated evolution trend.

[0179] Query the evolution record corresponding to the "valve emergency shutdown + maintenance" action in the historical feedback database, and find that in the same type of scene, the concentration of the explosion-proof warehouse environment node stabilizes at 8-10 ppm within 1 hour after the action, and the associated edge coefficient decreases by 0.2-0.3. Extract the commonness rule "this action can reduce the concentration by 8-10 ppm and the coefficient by 0.2-0.3".

[0180] Integrate this rule into the calibrated evolution trend benchmark to determine that "after executing valve emergency shutdown + maintenance, the evolution trend of the explosion-proof warehouse environment node is to decrease to 8-10 ppm within 1 hour and stabilize, and the associated edge coefficient decreases by 0.2-0.3", forming the final calibrated evolution trend.

[0181] Step S157: synchronize the calibrated evolution trend to the link evolution tracking link to adjust the next round of monitoring focus, and to the control action set adaptation link as a link state reference standard for subsequent screening control actions, and then preset the collection dimension of the next round of feedback data based on the calibrated evolution trend.

[0182] The calibrated evolution trend is sent to the link evolution tracking module to adjust the monitoring focus: for the explosion-proof warehouse environment node, the next round of focus is to monitor whether the concentration is maintained at 8-10 ppm and the correlation coefficient is stable at about 0.6; for the cold chain warehouse to be updated node, the focus is to monitor whether the compliance rate is maintained at 94%±0.5%.

[0183] Synchronize to the control action set adaptation module as a reference for action screening: when the concentration of the explosion-proof warehouse environment node exceeds 10 ppm, preferentially select the "valve emergency closing + maintenance" action; when the compliance rate of the cold chain warehouse storage node is lower than 93.5%, the action needs to be reselected and adapted.

[0184] Based on the calibrated trend, the next round of collection dimensions are preset: the "concentration drop rate" collection dimension of the explosion-proof warehouse environment node is increased, and the "compliance rate fluctuation frequency" collection dimension of the cold chain warehouse node is increased, to ensure that the next round of feedback data can more accurately reflect the evolution trend.

[0185] Figure 2 A schematic diagram of exemplary hardware and software components of the safety control system 100 for emergency material storage parks that can implement the idea of the present application provided by some embodiments of the present application is shown. For example, the processor 120 can be used in the safety control system 100 for emergency material storage parks and used to execute the functions in the present application.

[0186] The safety control system 100 for emergency material storage parks can be a general server or a special-purpose server, both of which can be used to implement the safety control method for emergency material storage parks of the present application. Although only one server is shown in the present application, for the sake of convenience, the functions described in the present application can be implemented in a distributed manner on multiple similar platforms to balance the processing load.

[0187] For example, the security management system 100 applied to the emergency material storage park can include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and different forms of storage media 140, such as a disk, a ROM, or a RAM, or any combination thereof. Exemplarily, the security management system 100 applied to the emergency material storage park can also include program instructions stored in a ROM, a RAM, or other types of non-transitory storage media, or any combination thereof. The method of the present application can be implemented according to these program instructions. The security management system 100 applied to the emergency material storage park also includes an I / O interface 150 between the computer and other input / output devices.

[0188] For ease of illustration, only one processor is described in the security management system 100 applied to the emergency material storage park. However, it should be noted that the security management system 100 applied to the emergency material storage park in the present application can also include multiple processors, so the steps performed by one processor described in the present application can also be jointly performed or separately performed by multiple processors. For example, if the processor of the security management system 100 applied to the emergency material storage park performs steps A and B, it should be understood that steps A and B can also be jointly performed by two different processors or separately performed in one processor. For example, a first processor performs step A, a second processor performs step B, or the first processor and the second processor jointly perform steps A and B.

[0189] In addition, the present application also provides a readable storage medium, wherein computer executable instructions are pre-set in the readable storage medium, and when a processor executes the computer executable instructions, the security management method applied to the emergency material storage park is implemented.

[0190] It should be noted that, in order to simplify the description of the present application and help understand one or more embodiments of the present application, in the foregoing description of the embodiments of the present application, various features are sometimes combined into one embodiment, figure or description thereof.

Claims

1. A safety management and control method applied to emergency material storage parks, characterized in that, The method includes: A safety status link is constructed for the emergency supplies storage park. This safety status link is based on multi-source monitoring information and includes material storage nodes, environmental nodes, and activity nodes. These nodes are connected via interconnecting edges to form a mesh structure. The construction of this safety status link for the emergency supplies storage park includes: Collect multi-source monitoring information from emergency material storage parks, including information related to material storage, park environment, and personnel activities. The core characterization information of the material storage nodes is extracted from multi-source monitoring information, and the core characterization information reflects the current status of material storage; Extract the core characterization information of environmental nodes, which reflects the environmental status of different areas of the park; Extract the core representation information of the activity nodes, which reflects the activity status of people in the park; Analyze the influence relationship between material storage nodes and environmental nodes. Based on the state correlation characteristics of the two in multi-source monitoring information, establish the association edge between material storage nodes and environmental nodes. The attributes of the association edge are determined by the degree of influence between the two. Analyze the influence relationship between environmental nodes and activity nodes. Based on the state correlation characteristics of the two in multi-source monitoring information, establish the association edge between environmental nodes and activity nodes. The attributes of the association edge are determined by the degree of influence between the two. Analyze the indirect influence relationship between material storage nodes and activity nodes. Using environmental nodes as intermediate transmission carriers, establish indirect association edges between material storage nodes and activity nodes. The attributes of the association edges are determined by the degree of indirect influence. The material storage nodes, environmental nodes, activity nodes, and all associated edges are integrated in a network structure to form a safety status link for the emergency material storage park. Each node in the safety status link is accompanied by a corresponding characterization information identifier, and each associated edge is accompanied by a corresponding impact attribute identifier. The evolution process of the security state link is tracked, and the state information of each node and associated edge changes over time is recorded to generate link evolution data; Based on the link evolution data, a set of control and management actions is adapted, and the set of control and management actions corresponds to the node status of the security status link. The set of control actions is sent to the park control system, which then drives the execution device to perform the control actions and collects feedback data from the security status link after execution. The evolution trend of the security state link is calibrated based on the feedback data.

2. The safety management and control method for emergency material storage parks according to claim 1, characterized in that, The process of tracking the evolution of the secure state link involves recording the state information of each node and associated edge over time, generating link evolution data, including: Set the tracking period for the security status link, and continuously collect the changes in the representation information of each node according to the tracking period. The changes in the representation information reflect the changes in the node status over time. Collect the attribute changes of each associated edge, and the attribute changes reflect the change in the degree of influence between nodes over time; Based on the changes in node representation information, the evolution stages of nodes are divided, and different evolution stages correspond to different changes in node representation information. Based on the changes in the attributes of the associated edges, the evolution stages of the associated edges are divided, and different evolution stages correspond to different changes in the attributes of the associated edges. Record the corresponding relationship between the evolution stages of nodes and the evolution stages of associated edges within each tracking period to form a periodic evolution record. The periodic evolution record contains a snapshot of the state of each node and associated edge within the tracking period. The cycle evolution records of multiple tracking cycles are integrated in chronological order to form link evolution data. The link evolution data is used to reflect the overall change process of the safe state link in the continuous time dimension, and the evolution record of each cycle is connected with the evolution record of the previous cycle.

3. The safety management and control method for emergency material storage parks according to claim 1, characterized in that, The set of adaptive control actions based on the link evolution data includes: The node evolution stages in the link evolution data are analyzed to identify nodes in the unstable evolution stage, whose characteristic information changes beyond the normal range. For each unstable node in an unstable evolution stage, analyze the historical state change patterns of the unstable node in the link evolution data, and determine the target influencing factors that cause the unstable node to enter the unstable evolution stage; Based on the target influencing factors, select the types of control actions that can be applied to the unstable node. Each type of control action corresponds to one or more target influencing factors. Analyze the attributes of the associated edges between the unstable node and other nodes. If the unstable state of the unstable node is transmitted to other associated nodes through the associated edges, then the corresponding control action type is selected for the corresponding associated nodes. Based on the urgency of the evolution of each unstable node in the link evolution data, the execution priority of the control action is determined. The selected control action types are sorted according to their execution priority. At the same time, the node and associated edge corresponding to each control action type are determined to form a control action set. Each control action in the control action set is associated with a corresponding node in the safe state link.

4. The safety management and control method for emergency material storage parks according to claim 3, characterized in that, For each unstable node in an unstable evolution stage, the historical state change patterns of that unstable node in the link evolution data are analyzed to determine the target influencing factors that cause the unstable node to enter the unstable evolution stage, including: Extract the sequence of changes in the characterization information of the unstable node in the link evolution data over multiple tracking periods. The sequence of changes in characterization information is arranged in chronological order to present the process of changes in the node state of the unstable node. The sequence of changes in the characterization information is divided into stages, distinguishing between stable and unstable change stages, and the time point at which the unstable node transitions from the stable change stage to the unstable change stage is determined. By comparing the differences in representational information features between the stable change stage and the unstable change stage, the representational information features corresponding to the unstable change stage are extracted. The unique representational information features of the unstable change stage are the key marker features that distinguish the stable change stage from the unstable change stage. Analyze the attribute changes of the associated edges of the unstable node before and after the start of the unstable change phase, identify the associated nodes whose associated edge attributes have undergone a preset change, and the associated nodes whose associated edge attributes have undergone a preset change are the associated nodes that have an impact on the unstable node through the associated edges. The analysis examines the changes in external input information related to the unstable node in multi-source monitoring information before and after the start of the unstable change phase. These changes in external input information directly affect the node's state. By combining the unique representational information characteristics of the unstable stage, the associated nodes that affect the unstable node through the associated edges, and the changes in the external input information, the factors that play a dominant role in the unstable node's entry into the unstable evolution stage are selected. The factors that play a dominant role in the unstable node's entry into the unstable evolution stage are the target influencing factors that lead the node into the unstable evolution stage. The target influencing factors are classified and organized to determine the logical relationship between each target influencing factor and the changes in the characterization information of unstable nodes.

5. The safety management and control method for emergency material storage parks according to claim 3, characterized in that, The selection of control action types that can be applied to the unstable node based on the target influencing factors includes: Establish a mapping library between influencing factors and control action types. The mapping library pre-stores the control action types corresponding to different influencing factors, and each control action type is marked with the range of influencing factors it can affect. The identified target influencing factors are matched with the influencing factors in the mapping library of the control action types to initially screen out the control action types corresponding to the target influencing factors; Analyze the application effect of the initially screened control action types in historical link evolution data, and see whether the initially screened control action types can effectively guide nodes from unstable evolution stages to stable evolution stages when applied to similar nodes in the past. If any control action type can effectively guide a node from an unstable evolution stage to a stable evolution stage in historical applications, and its target influencing factors are completely matched with the target influencing factors of the current unstable node, then the control action type is determined as a candidate control action type. If any control action type cannot effectively guide a node from an unstable evolution stage to a stable evolution stage in historical applications, but can adapt to the target influencing factors of the current node after adjusting its mode of action, then the mode of action of the control action type will be adjusted, and the adjusted control action type will be included in the candidate control action types. If none of the control action types initially selected are suitable for the target influencing factors of the current unstable node, then a new control action type is constructed based on the characteristics of the target influencing factors of the current unstable node. The candidate control action types and the newly constructed control action types are integrated to form a set of control action types for the unstable node. Each control action type is clearly marked with its corresponding target influencing factors and its mode of action.

6. The safety management and control method for emergency material storage parks according to claim 1, characterized in that, The process involves sending the set of control actions to the park management system, which then drives the execution device to perform the control actions. Simultaneously, feedback data from the security status link is collected after execution, including: Obtain the functional attributes of each execution device in the park management and control system. The functional attributes of the execution device reflect the types of management and control actions that the execution device can perform. Different execution devices correspond to different ranges of management and control action types. The type and execution requirements of each control action in the control action set are analyzed, and each control action is assigned to the corresponding execution device according to the functional attributes of the execution device. The control action allocation results and the execution order of the control actions are sent to the park control system, so that the park control system sends execution instructions to each execution device according to the execution order. The execution instructions contain the specific execution parameters and execution time requirements of the control actions. After receiving the execution instructions, the execution devices start the control action execution process according to the execution parameters and execution time requirements. During the execution of the control actions, the execution device's own operating status information is recorded in real time. The execution device's own operating status information reflects the process characteristics of the execution device in executing the control actions. After the control action is completed, the change of the representation information of the corresponding node in the security status link is collected. The change of the representation information of the corresponding node in the security status link reflects the effect of the control action on the node status. The changes in the attributes of the associated edges in the security status link associated with the corresponding node are collected. These changes reflect the impact of control actions on the relationships between nodes. The operating status information of the execution device, the changes in the representation information of the corresponding nodes in the safety status link, and the changes in the attributes of the associated edges associated with the corresponding nodes in the safety status link are integrated to form the feedback data of the safety status link after execution. The feedback data of the post-execution safety status link is uploaded to the park management and control system in real time. The feedback data of the post-execution safety status link includes the corresponding management and control action identifier and execution time identifier.

7. The safety management and control method for emergency material storage parks according to claim 6, characterized in that, After the control action is completed, the process of collecting changes in the representation information of the corresponding nodes in the security status link includes: The target node corresponding to the control action is determined, and the target node corresponding to the control action is a node in the security state link where the control action is applied; A representation information collection window is set, and the time range of the representation information collection window covers the period from the completion of the control action to the stabilization of the node status. Within the representation information acquisition window, the representation information of the target node is continuously acquired according to the preset acquisition interval. The acquired representation information of the target node is then organized and a sequence of representation information changes is formed according to the acquisition time. By comparing the baseline sequence of the target node's representation information before the execution of the control action with the sequence of changes in the representation information after the execution of the control action, the difference features between the two are extracted. These difference features reflect the impact of the control action on the node's state. Analyze the direction of change of the differential characteristics. If the differential characteristics show that the node representation information is changing in a stable direction, it indicates that the control action has a positive effect. If the difference features show that the node representation information is still changing in an unstable direction, it indicates that the impact of the control measures has not met expectations. The sequence of changes in representational information, differential characteristics, and the results of the impact assessment are integrated to form a record of the changes in representational information of the target node.

8. The safety management and control method for emergency material storage parks according to claim 1, characterized in that, The step of calibrating the evolution trend of the security state link based on the feedback data includes: Establish the correspondence between feedback data and safety status links, and orient the collected post-execution feedback data to the material storage nodes, environment nodes, activity nodes and related edges in the safety status links according to the corresponding logic of control actions, target nodes and related edges, forming an action link feedback mapping table; Call the basic evolution rule base of the safe state link, which pre-stores the stable state characteristics of nodes and the benchmark of the influence relationship of associated edges; The actual node status and actual associated edge influence relationship in the action link feedback mapping table are matched with the benchmark features in the rule base to identify nodes and associated edges to be updated that have differences between the actual status and the benchmark features. Based on the actual state of the node to be updated and the associated edge to be updated, extract its new stable features after the execution of the control action, supplement the evolution rule base with the new stable features and replace the original baseline features; Based on the updated evolution rule base, the stable direction and change direction of each node under the new benchmark, and the transmission path of the associated edge under the new influence logic are clarified, forming a calibrated evolution trend benchmark; The historical feedback database is called up, and the actual evolution of the current action link feedback mapping table is compared with the link evolution of the same control action in the past. The common evolution patterns under the same scenario are extracted and integrated into the calibrated evolution trend benchmark to form the final calibrated evolution trend. The calibrated evolution trend is synchronized to the link evolution tracking stage to adjust the monitoring focus for the next round, and simultaneously synchronized to the control action set adaptation stage as a link status reference standard for subsequent screening of control actions. Then, the collection dimensions of the next round of feedback data are preset based on the calibrated evolution trend.

9. A safety management and control system for emergency material storage parks, characterized in that, The device includes a processor and a memory, the memory and the processor being connected. The memory is used to store programs, instructions or code, and the processor is used to execute the programs, instructions or code in the memory to implement the safety management and control method for emergency material storage parks as described in any one of claims 1-8.

Images