AI safe operation situation scheduling method and system for LNG receiving station

Abstract

The application discloses an LNG receiving station-oriented AI safe operation situation scheduling method and system, relates to the technical field of safety scheduling, and comprises the following steps: S1, collecting operation verification data, and performing preprocessing on the operation verification data; S2, establishing an operation broken link matrix, determining a condition conflict evaluation value, and generating an operation situation evidence field; S3, constructing a coupled bipartite graph, marking strong coupling evidence pairs after weighted message passing updates the node activation value, and generating an operation situation scheduling unit; and S4, generating corresponding scheduling tokens based on the operation situation scheduling unit, combining the condition conflict evaluation value to generate operation scheduling instructions, sending the operation scheduling instructions to a corresponding management end, and outputting a safe operation situation scheduling result. The application solves the problems that in the existing LNG receiving station operation, multi-source verification data cannot be continuously bound and dynamically checked, personnel overreach, gas failure, probe alarm and hidden danger information cannot be identified and associated in real time, and the operation safety situation is difficult to accurately judge.

Description

Technical Field

[0001] This invention relates to the field of safety dispatching technology, specifically to an AI-based safety operation status dispatching method and system for LNG receiving terminals. Background Technology

[0002] With the development of next-generation information technologies such as the Industrial Internet, the Internet of Things, artificial intelligence, big data, and cloud computing, the safe scheduling, risk monitoring, and intelligent management of complex industrial systems have become a hot topic in the industry. Domestic and international applications of these technologies in the energy, power, and cloud computing sectors demonstrate that through multi-source data integration, intelligent analysis, and scheduling optimization, it is possible to improve system operating efficiency, quantify risk assessment, and provide real-time decision support.

[0003] For example, application CN117272850B discloses a method for analyzing the flexible space of distribution network safe operation dispatch, which includes the following steps: assessing the reliable regulation capacity of various reliable regulation resources; integrating and aggregating reliable regulation resources, classifying and aggregating resources, and constructing a mapping relationship reflecting the membership degree and aggregation index between reliable regulation resources and cluster centers; creating a county-level power grid dispatch flexible space analysis method that takes into account the interaction between source, grid, load and storage, establishing a load elasticity assessment model, quantifying load elasticity, obtaining the elastic space of reliable regulation resources under different scenarios, and evaluating the elastic space.

[0004] For example, application CN120930132A provides a multi-virtual machine isolation system based on automatic scheduling according to security levels, relating to the fields of cloud computing and virtualization technology. This system generates task security levels through an intelligent security level assessment module combined with multi-dimensional analysis; a security-layered virtual machine resource pool enables graded cluster management and elastic scaling; an intelligent scheduling engine allocates tasks based on security level matching and resource optimization; and a cross-level data exchange control module enables fine-grained data management.

[0005] However, existing LNG receiving terminal operation management technologies still suffer from problems such as the inability to continuously bind multi-source data and insufficient dynamic verification capabilities. This leads to information silos, personnel overstepping boundaries, gas failures, and the inability to correlate probe alarms and potential hazards in real time during operations, making it difficult to determine the continuous compliance of operational conditions. Existing systems typically can only handle single data sources or single operational stages, lacking the ability to conduct cross-stage and cross-departmental risk correlation analysis and dynamic control. This results in delayed responses to operational anomalies, low safety management efficiency, and an inability to meet the real-time monitoring and risk assessment needs of the complex operational environment of modern LNG receiving terminals.

[0006] Therefore, in order to address the above issues, there is an urgent need for AI-based safety operation status scheduling methods and systems for LNG receiving terminals. Summary of the Invention

[0007] Technical problems to be solved

[0008] To address the shortcomings of existing technologies, this invention provides an AI-based safety operation status scheduling method and system for LNG receiving terminals. It solves the problem that multi-source verification data cannot be continuously bound and dynamically verified in existing LNG receiving terminal operations, resulting in the inability to identify and associate personnel crossing boundaries, gas failures, probe alarms, and hidden danger information in real time, making it difficult to accurately judge the operational safety status.

[0009] Technical solution

[0010] To achieve the above objectives, this invention provides the following technical solution: an AI-based safety operation status scheduling method for LNG receiving terminals, comprising: S1, constructing a three-dimensional coordinate system for the terminal area, collecting operation verification data, performing time alignment and coordinate transformation on the operation verification data, and outputting preprocessed operation verification data; S2, establishing an operation chain break matrix based on the preprocessed operation verification data, with personnel crossing boundaries, gas failure, probe alarm, video evidence failure, hidden danger hindrance, and emergency gap as rows and continuous verification periods as columns, and determining condition conflict evaluation values ​​based on the number of records in the node unit and their corresponding weights, generating an operation status evidence field; S3, marking control areas and hazards based on the operation status evidence field, constructing a coupled bipartite graph based on the dominant chain break records and process alarm records, updating node activation values ​​through weighted message passing, marking strongly coupled evidence pairs, and generating an operation status scheduling unit; S4, generating a corresponding scheduling token based on the operation status scheduling unit, generating an operation scheduling instruction based on the condition conflict evaluation value, sending the operation scheduling instruction to the corresponding management terminal, and outputting the safety operation status scheduling result.

[0011] Further, the specific steps for constructing a three-dimensional coordinate system for the station area, collecting operation verification data, and performing time alignment and coordinate transformation on the operation verification data are as follows: Using the southwest corner reference point of the LNG receiving station electronic map as the coordinate origin, the direction from west to east along the east-west road of the station area is set as the positive X-axis, the direction from south to north along the north-south road of the station area is set as the positive Y-axis, and the vertical upward direction is set as the positive Z-axis, thus constructing a three-dimensional coordinate system for the station area; Operation verification data is collected, including data collection timestamp, operation task number, operation type code value, operation level code value, operation plan start timestamp, operation plan end timestamp, operation point center coordinates, operation electronic fence vertex coordinates, personnel positioning coordinates, personnel positioning timestamp, and combustible material... The system collects and processes the following data: gas concentration values, oxygen volume fraction values, hydrogen sulfide concentration values, carbon monoxide concentration values, gas detection timestamps, gas sampling point coordinate values, fixed probe coordinate values, fixed probe alarm duration values, process alarm duration values, video offline duration values, controlled area boundary coordinate values, hazardous source boundary coordinate values, current number of unclosed hidden dangers, number of inspection anomalies, number of available emergency resources, and emergency passage occupancy status values. For the collected operation verification data, a nearest neighbor timestamp matching algorithm is used to perform cross-source record alignment processing. A spatial coordinate transformation algorithm based on rigid coordinate transformation is used to perform station area coordinate unification processing on the operation verification data, converting point data from different sources to a unified station area three-dimensional coordinate system, and outputting preprocessed operation verification data.

[0012] Furthermore, the specific steps for establishing an operation chain break matrix based on preprocessed operation verification data, with personnel crossing boundaries, gas failure, adjacent detector alarms, video evidence failure, hidden dangers, and emergency gaps as rows and continuous verification periods as columns, are as follows: Read the preprocessed operation verification data, extract the operation plan start timestamp, operation plan end timestamp, operation point center coordinates, and operation electronic fence vertex coordinates according to the operation task number, and determine the time interval between the operation plan start timestamp and the operation plan end timestamp as the operation permit time period; Analyze the operation electronic fence vertex coordinates corresponding to the same operation task number according to... Close the connection according to the vertex connection order recorded in the electronic map, and determine the closed boundary and internal area as the work permit plane area; according to the work type code value and work level code value, read the verification period length value, time effective window value, spatial coverage radius value and verification weight value from the work verification rule configuration table; divide the work permit time period into M consecutive verification periods according to the verification period length value; and establish the work chain break matrix with personnel crossing the boundary node, gas failure node, probe adjacent alarm node, video evidence failure node, hidden danger traction node and emergency gap node as matrix rows and consecutive verification periods as matrix columns.

[0013] Further, based on the number of records within a node unit and the corresponding weights, the specific steps for generating the operational status evidence field are as follows: The intersection of each matrix row and column is used as a node unit; within each continuous verification period, records where personnel location coordinates are outside the work permit plane area are written into the node unit corresponding to the personnel boundary crossing node; records where the difference between the gas detection timestamp and the current continuous verification period's end timestamp is greater than the effective time window value, the distance from the gas sampling point coordinates to the work point center coordinates is greater than the spatial coverage radius value, or any of the following values ​​exceeds the corresponding gas safety threshold range: combustible gas concentration, oxygen volume fraction, hydrogen sulfide concentration, and carbon monoxide concentration; records where the distance from the fixed probe coordinates to the work point center coordinates is less than or equal to the spatial coverage radius value and the fixed probe alarm duration is greater than zero are written into the node unit corresponding to the probe neighbor alarm node; the video offline duration... Records with a long value greater than zero are written to the node unit corresponding to the video verification node; records with a current unclosed hazard quantity value greater than zero or an inspection anomaly quantity value greater than zero are written to the node unit corresponding to the hazard traction node; records with an emergency resource availability value of zero or an emergency passage occupancy status value of occupancy are written to the node unit corresponding to the emergency gap node; when multiple records are written to the same node unit, only the record whose data collection timestamp is closest to the end of the current continuous verification period is retained; the number of records retained in each node unit of each matrix row is counted according to the continuous verification period, and the number of records is multiplied by the verification weight value of the corresponding matrix row to obtain the row conflict contribution value of each matrix row; the row conflict contribution values ​​of all matrix rows are added together to obtain the conditional conflict evaluation value corresponding to the current continuous verification period; the broken chain node type corresponding to the matrix row with the largest row conflict contribution value is determined as the dominant broken chain type; the operation broken chain matrix, conditional conflict evaluation value and dominant broken chain type are combined to generate the operation status evidence field.

[0014] Further, the specific steps for marking control areas and hazard sources based on the operational situation evidence field are as follows: Read the operational situation evidence field, extract the coordinate values ​​of the center of the work point, the boundary coordinate values ​​of the control area, and the boundary coordinate values ​​of the hazard source, and calculate the shortest distance from the center of the work point to the boundary of each control area under the XY-axis plane projection, and record it as the area shortest distance; calculate the shortest distance from the center of the work point to the boundary of each major hazard source under the XY-axis plane projection, and record it as the hazard shortest distance; mark the control areas whose area shortest distance is not greater than the control distance threshold as adjacent control areas; mark the hazard sources whose hazard shortest distance is not greater than the hazard distance threshold as adjacent hazard sources.

[0015] Furthermore, the specific steps for constructing a coupled bipartite graph based on the dominant chain break record and process alarm record are as follows: Extract the node unit record corresponding to the dominant chain break type from the operation chain break matrix, and denote it as the main record; extract the spatial coordinates corresponding to the main record from the preprocessed operation verification data, and extract the process alarm record with a process alarm duration value greater than zero within the current continuous verification period; construct a coupled bipartite graph, taking personnel boundary crossing nodes, gas failure nodes, probe adjacent alarm nodes, video evidence failure nodes, hidden danger traction nodes, and emergency gap nodes as the first type of nodes, and taking adjacent control areas and adjacent... Near the hazard source is designated as a second type of node. When the spatial coordinates associated with the master record are located in the region polygon represented by a second type of node, a directed edge is established from the first type of node corresponding to the master record to the second type of node, and the weight of the edge is taken as the row conflict contribution value corresponding to the first type of node. When the alarm point coordinates of the process alarm record fall into the region polygon represented by the second type of node under the XY axis plane projection, a directed edge is established from the second type of node to the first type of node corresponding to the master record, and the weight of the edge is taken as the normalized value obtained by dividing the process alarm duration value by the upper limit of the alarm duration value.

[0016] Further, after updating the node activation value through weighted message passing, marking strongly coupled evidence pairs, and generating the operational status scheduling unit, the specific steps are as follows: The condition conflict evaluation value is used as the initial activation value of the first type of node; the average of the normalized duration values ​​of the process alarms associated with each second type of node is used as the initial activation value of the second type of node; if the second type of node is not associated with any process alarm record, the initial activation value is zero; a graph message propagation algorithm is used to perform N rounds of message passing on the coupled bipartite graph. In each round of propagation, each node sends its current activation value to its neighboring nodes after weighting it according to the weights of the outgoing edges. Each node sums all the incoming edge messages it receives and adds them to its current activation value to obtain the updated activation value; after N rounds of propagation, the final activation value of each node is determined... The coupling activation degree is defined as follows: For a node pair consisting of a dominant broken chain node and each second-type node, if there is at least one directed edge between them, the coupling activation degree of the first-type node and the coupling activation degree of the second-type node are multiplied to obtain the product value. The weights of all edges between the node pairs are added to obtain the weight sum. The coupling strength value is obtained by dividing the product value by the weight sum. Node pairs with coupling strength values ​​greater than the coupling strength threshold are marked as strongly coupled evidence pairs. The broken chain node type corresponding to the first-type node is extracted from the strongly coupled evidence pairs as the coupling broken chain type. The continuous verification period in which the strongly coupled evidence pairs are located is marked as the situation coupling period. The start timestamp, task number, center coordinates of the task point, condition conflict evaluation value, and coupling broken chain type of the situation coupling period are written into the operation situation scheduling unit.

[0017] Furthermore, the specific steps for generating corresponding scheduling tokens based on the operational status scheduling unit and generating job scheduling instructions in conjunction with condition conflict evaluation values ​​are as follows: Read the operational status scheduling unit and pre-processed job verification data; when the coupling chain break type is a personnel boundary crossing node, generate a personnel deviation token; when the coupling chain break type is a gas failure node, generate a gas failure token; when the coupling chain break type is a probe neighbor alarm node, generate a probe neighbor alarm token; when the coupling chain break type is a video evidence failure node, generate a video evidence failure token; when the coupling chain break type is a hazard traction node, generate a hazard traction token; when the process alarm duration value is greater than zero, generate a process disturbance... Activation tokens; when the available quantity of emergency resources is zero or the emergency passage is occupied, a support gap token is generated; each generated token is used to trigger the corresponding operation scheduling instruction; if the condition conflict assessment value is greater than the confidence threshold and the coupling chain break type is personnel crossing the boundary node or gas failure node, an operation suspension scheduling instruction is generated; if the condition conflict assessment value is greater than the confidence threshold and the coupling chain break type is probe neighbor alarm node or video evidence failure node, an on-site review scheduling instruction is generated; if a process disturbance token has been generated and the coupling chain break type is hidden danger traction node, an upgrade control scheduling instruction is generated; if a support gap token has been generated, an emergency support scheduling instruction is generated.

[0018] Further, the specific steps for sending work scheduling instructions to the corresponding management terminals and outputting the safety operation status scheduling results are as follows: Send the work suspension scheduling instruction to the special work management terminal; send the on-site verification scheduling instruction to the mobile video verification terminal and the intelligent inspection terminal; send the upgraded control scheduling instruction to the safety production dashboard; send the emergency support scheduling instruction to the personnel positioning management terminal and the emergency communication terminal; and display the sent scheduling instructions on the 3D electronic map according to the center coordinates of the work point, recording the sending timestamp of each scheduling instruction; read the work suspension confirmation timestamp from the special work management terminal; and read the timestamp from the mobile video verification terminal and... The intelligent inspection terminal reads the on-site verification completion timestamp and takes the earliest completion timestamp as the on-site verification confirmation timestamp; it reads the upgrade control confirmation timestamp from the safety production cockpit; it reads the emergency support arrival timestamp from the personnel positioning management terminal and the emergency communication terminal and takes the earliest arrival timestamp as the emergency support confirmation timestamp; for each confirmation timestamp, the difference is calculated only when the confirmation timestamp is later than the sending timestamp, and the single scheduling duration is obtained by subtracting the sending timestamp from the confirmation timestamp; the minimum value among all single scheduling durations is taken as the scheduling response duration value, and the maximum value is taken as the scheduling closed-loop duration value, and the safety operation status scheduling result is output.

[0019] The second aspect of this invention provides an AI-based safety operation status scheduling system for LNG receiving terminals, comprising: a data acquisition and processing module, a job interruption generation module, a coupled status discrimination module, and a token scheduling output module. The data acquisition and processing module is used to construct a three-dimensional coordinate system for the terminal area, acquire job verification data, perform time alignment and coordinate transformation on the job verification data, and output preprocessed job verification data. The job interruption generation module is used to establish job interruption status based on the preprocessed job verification data, with rows for personnel crossing boundaries, gas failure, adjacent detector alarms, video evidence failure, hidden danger hazard, and emergency gaps, and columns for continuous verification periods. The system generates an operational status evidence field by constructing a chain matrix and determining the conditional conflict assessment value based on the number of records and corresponding weights within each node unit. A coupled status discrimination module is used to mark control areas and hazards based on the operational status evidence field, and constructs a coupled bipartite graph based on dominant chain break records and process alarm records. After updating node activation values ​​through weighted message passing, strongly coupled evidence pairs are marked to generate operational status scheduling units. A token scheduling output module is used to generate corresponding scheduling tokens based on the operational status scheduling units, and generate operational scheduling instructions in conjunction with the conditional conflict assessment value. The operational scheduling instructions are then sent to the corresponding management terminal, outputting the safe operational status scheduling results.

[0020] Beneficial effects

[0021] The present invention has the following beneficial effects:

[0022] (1) AI safety operation status scheduling method and system for LNG receiving terminals: By constructing an operation chain break matrix and quantifying condition conflicts, the information of personnel crossing the boundary, gas failure, probe neighboring alarm, video evidence failure, hidden danger traction and emergency gap is formed into a matrix. Combined with the node weight, the condition conflict evaluation value is calculated to realize the quantitative identification of operation anomalies and the determination of the dominant chain break type, thereby improving the accuracy and reliability of operation status analysis.

[0023] (2) AI safety operation status scheduling method and system for LNG receiving terminals: Through bidirectional coupled bipartite graph analysis, the dominant chain break node is coupled with the adjacent control area and hazard source node. Combined with process alarm records, the node activation value is updated through weighted message transmission, realizing dynamic risk status identification across links and regions, and providing quantitative basis for scheduling decisions.

[0024] (3) AI safety operation status scheduling method and system for LNG receiving terminals, through scheduling token generation and closed-loop instruction control, generates operation scheduling tokens and scheduling instructions based on the operation status scheduling unit, realizes closed-loop monitoring and intelligent response of the entire operation process, and enhances the real-time and traceability of operation safety management.

[0025] (4) AI safety operation status scheduling method and system for LNG receiving terminals: Through continuous status monitoring and abnormal response quantification, real-time tracking of changes in operating conditions, comprehensive judgment of multi-source data anomalies, timely identification of risk points and triggering operation suspension, review or upgrade control, realizing proactive intervention and dynamic optimization of operation safety management. Attached Figure Description

[0026] Figure 1 A flowchart of an AI-based safety operation status quo scheduling method for LNG receiving terminals; Figure 2 A structural diagram of an AI-based safety operation status quo scheduling system for LNG receiving terminals; Figure 3 This is a schematic diagram illustrating the generation and coupling discrimination of the operational situation evidence field; Figure 4 A flowchart is constructed to establish the bipartite graph coupling between the main chain breakage and process alarm. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] Please see Figures 1-4 This invention provides a technical solution: an AI-based safety operation status scheduling method for LNG receiving terminals, comprising: S1, constructing a three-dimensional coordinate system for the terminal area, collecting operation verification data, performing time alignment and coordinate transformation on the operation verification data, and outputting preprocessed operation verification data; S2, establishing an operation chain break matrix based on the preprocessed operation verification data, with personnel crossing boundaries, gas failure, probe alarm, video evidence failure, hidden danger hindrance, and emergency gap as rows and continuous verification period as columns, and determining the condition conflict evaluation value based on the number of records in the node unit and the corresponding weight, generating an operation status evidence field; S3, marking control areas and hazards based on the operation status evidence field, constructing a coupled bipartite graph based on the dominant chain break records and process alarm records, updating the node activation value through weighted message passing, marking strongly coupled evidence pairs, and generating an operation status scheduling unit; S4, generating a corresponding scheduling token based on the operation status scheduling unit, generating an operation scheduling instruction based on the condition conflict evaluation value, sending the operation scheduling instruction to the corresponding management terminal, and outputting the safety operation status scheduling result.

[0029] Specifically, the steps for constructing a three-dimensional coordinate system for the station area, collecting operation verification data, and performing time alignment and coordinate transformation on the operation verification data are as follows: Using the center point of the pre-marked concrete benchmark pile in the southwest corner of the LNG receiving station's electronic map as the origin, the direction from west to east along the main east-west road of the station area is set as the positive X-axis, the direction from south to north along the main north-south road of the station area is set as the positive Y-axis, and the vertical direction perpendicular to the ground plane is set as the positive Z-axis. A right-hand rule-based three-dimensional coordinate system for the station area is constructed, ensuring that any spatial point within the station area can be represented by a unique X, Y, and Z coordinate value. The coordinate values ​​of the operation point center, the apex of the operation electronic fence, the gas sampling point, the fixed probe, the control area boundary, and the hazard source boundary are read from the electronic map database. These coordinate values ​​are three-dimensional coordinate values ​​under Gaussian projection plane coordinates, and each coordinate value consists of X, Y, and Z coordinate components. The system reads data acquisition timestamps, job task numbers, job type codes, job level codes, job plan start timestamps, and job plan end timestamps from the special operation permit management system. Specifically, the data acquisition timestamp is the server time written when the special operation permit management system generates the current job permit record; the job task number is the unique business number in the special operation ticket; the job type code is the type code corresponding to hot work, confined space work, hoisting work, and temporary power supply work in the special operation ticket; the job level code is the level code corresponding to Level 1, Level 2, and Level 3 work in the special operation ticket; and the job plan start timestamps and job plan end timestamps are the approved start and end times in the special operation ticket, respectively. The system receives personnel positioning coordinates and timestamps from the personnel positioning system once per second. The system also receives combustible gas concentration values, oxygen volume fraction values, hydrogen sulfide concentration values, carbon monoxide concentration values, and gas detection timestamps from the gas detection system once per second. The system receives the alarm duration value of the fixed gas probe once per second from the fixed gas probe alarm system. This fixed probe alarm duration value is calculated by the fixed gas probe alarm system based on the alarm start time and the current alarm refresh time of the same fixed probe, and is measured in seconds. The fixed probe alarm duration value is written as a field in the operation verification data; the alarm start time and the current alarm refresh time are only used as internal calculation bases for the fixed probe alarm duration value. When the fixed probe is not in an alarm state within the current sampling period, the fixed probe alarm duration value is recorded as zero seconds. The system also receives the process alarm duration value from the distributed control system once per second. The process alarm duration value represents the time from the first time a process parameter exceeds its limit to the time it returns to normal, and is measured in seconds. Finally, the system receives the video offline duration value from the video management system via an event-triggered method. The video offline duration value represents the time from network interruption to reconnection of the video device, and is measured in seconds.The system receives the current number of unclosed hidden dangers and the number of inspection anomalies from the dual prevention management system on a synchronized basis with each login. It also receives the available quantity of emergency resources and the occupancy status of emergency passages from the emergency resource management system on a synchronized basis with each login. Data collected at different frequencies and in different ways, including timestamps, task numbers, task type codes, task level codes, start and end timestamps of the task plan, coordinates of the task center, coordinates of the electronic fence apex, coordinates of personnel location, timestamps of personnel location, concentrations of combustible gases, oxygen volume fraction, hydrogen sulfide, and carbon monoxide, timestamps of gas detection, coordinates of gas sampling points, coordinates of fixed probes, duration of fixed probe alarms, duration of process alarms, duration of offline video, coordinates of controlled area boundaries, coordinates of hazardous source boundaries, the current number of unclosed hidden dangers, the number of inspection anomalies, the available quantity of emergency resources, and the occupancy status of emergency passages, are then organized into task verification data according to the data collection timestamps and task numbers. For the collected job verification data, a nearest neighbor timestamp matching algorithm is used to perform cross-source record alignment processing on data records from different sources under the same job task number: using the personnel location timestamp as the baseline time axis, for each personnel location record, the data acquisition timestamps corresponding to the gas detection timestamp, the process alarm duration, and the video offline duration are searched within a preset effective time window. The preset effective time window extends five seconds forward and backward from the personnel location timestamp. The five-second value is based on the data reporting cycle of two seconds for personnel location terminals in the LNG receiving station, one second for gas detection terminals, and three seconds for process alarm data. The five-second window can cover two complete reporting cycles of the above three data sources to ensure a high matching success rate. The absolute difference between each candidate timestamp and the personnel location timestamp is calculated. When multiple candidate timestamps have the same and minimum absolute difference from the personnel location timestamp, the record corresponding to the gas detection timestamp is selected first as the matching record. Next, the record corresponding to the data acquisition timestamp of the process alarm duration is selected. Finally, the record corresponding to the data acquisition timestamp of the video offline duration is selected. This priority order is determined based on the importance of the three data sources to the assessment of operational safety: gas concentration data directly reflects the safety of the working environment, process alarm data reflects the equipment operating status, and video data serves as an auxiliary verification method. The record corresponding to the timestamp with the smallest absolute difference (not exceeding five seconds) is selected as the matching record, and the numerical and coordinate fields in the matching record are associated with the personnel location record. If no matching record is found within the five-second window, the personnel location record and all its corresponding unmatched records are discarded.For the time-aligned operation verification data, a spatial coordinate transformation algorithm based on rigid coordinate transformation is used to perform station area coordinate unification processing on the operation verification data: Using the target coordinate values ​​of the station area's three-dimensional coordinate system as a reference, the source coordinate values ​​of each coordinate point are extracted from the operation verification data. The source coordinate values ​​come from coordinate values ​​already included in the operation verification data in electronic map databases, personnel positioning systems, gas detection systems, and fixed gas probe alarm systems, and each uses its own independent local coordinate system or geographic coordinate system. Before performing the rigid coordinate transformation, the scale consistency between the source coordinate values ​​and the target coordinate values ​​is first determined. The ratio of the distance between two points in the source coordinate values ​​to the distance between the corresponding two points in the target coordinate values ​​is calculated for at least three common control point pairs, and the average of all ratios is taken as the scale factor. If the absolute difference between the scale factor and the numerical value is greater than one-thousandth, the source coordinate values ​​are first multiplied by the reciprocal of the scale factor for scale normalization processing to unify the scale of the source coordinate system and the target coordinate system. After scale normalization processing, the rotation matrix and translation vector are calculated. If the absolute difference between the scale factor and the numerical value is less than or equal to one-thousandth, scale normalization is not required, and the rotation matrix and translation vector can be calculated directly. The rotation matrix and translation vector between the source and target coordinate values ​​are calculated. The rotation angle of the rotation matrix ranges from 0 to 360 degrees, and the displacement range of the translation vector is determined based on the actual dimensions of the station area: -500 meters to +500 meters in the X-axis direction, -500 meters to +500 meters in the Y-axis direction, and -50 meters to +50 meters in the Z-axis direction. The rotation matrix and translation vector are calculated by collecting at least three non-collinear common control point pairs. Each common control point pair contains one source coordinate value and one known target coordinate value. The common control points are selected from the corner points of fixed buildings already marked on the electronic map. The target coordinate values ​​of each corner point are pre-stored in the electronic map database, and the source coordinate values ​​are obtained through on-site measurement. The rotation matrix and rotation angle are calculated using the least squares fitting method, ensuring that the root mean square error of all common control points is less than 0.01 meters. The root mean square (RMS) of the transformation error is calculated as follows: For each common control point, calculate the Euclidean distance between the transformed target coordinates and the known target coordinates. Sum the squares of all Euclidean distances, divide by the number of common control points, and then take the square root. When the RMS of the transformation error is greater than 0.01 meters, reselect common control points or increase the number of common control points until the error requirement is met. For each coordinate point to be transformed, subtract the translation vector from the source coordinates and multiply by the inverse of the rotation matrix to obtain the transformed target coordinates. This transforms point data from different sources to the same station area's three-dimensional coordinate system, outputting preprocessed operation verification data. All coordinate values ​​in the preprocessed operation verification data are represented using the X, Y, and Z coordinates in the station area's three-dimensional coordinate system.

[0030] In this implementation plan, a three-dimensional coordinate system is constructed with the southwest corner of the station area as the origin. This system unifies the coordinates of the work point center, the top of the work electronic fence, the gas sampling point, the fixed probe, the control area boundary, and the hazard source boundary from the electronic map into the same coordinate frame. Using the personnel positioning timestamp as a reference, cross-source alignment of the gas detection timestamp, process alarm timestamp, and video frame acquisition timestamp is completed within a five-second window. Then, the data from different sources are transformed to the target coordinate system through rotation matrices and translation vectors. This ensures that the personnel positioning coordinates, gas sampling point coordinates, fixed probe coordinates, control area boundary coordinates, and hazard source boundary coordinates have a directly comparable spatial inclusion relationship and temporal sequence relationship. This provides spatiotemporally consistent preprocessed data for writing and statistically analyzing the node unit records in the work chain break matrix.

[0031] Specifically, the steps for establishing an operation chain break matrix based on preprocessed operation verification data, with personnel boundary crossing nodes, gas failure nodes, probe alarm nodes, video evidence failure nodes, hidden danger leading nodes, and emergency gap nodes as matrix rows and continuous verification periods as matrix columns, are as follows: Read the preprocessed operation verification data. All coordinate values ​​in the preprocessed operation verification data are represented using X, Y, and Z coordinates in the station's three-dimensional coordinate system. Extract the operation plan start timestamp, operation plan end timestamp, operation point center coordinates, and operation electronic fence vertex coordinates according to the operation task number. The operation point center coordinates consist of X-axis, Y-axis, and Z-axis components. The operation electronic fence vertex coordinates are a sequence of polygon vertex coordinates arranged clockwise. Determine the time interval between the operation plan start timestamp and the operation plan end timestamp as the operation permission time period, which represents the time range within which the operation task is allowed to be executed. Concatenate the operation electronic fence vertex coordinates corresponding to the same operation task number sequentially according to the vertex index order pre-stored in the electronic map database. The electronic map database is calibrated by the surveying unit based on the actual terrain of the station area during system initialization. The vertex coordinates of each work electronic fence are stored in a clockwise direction, with the first vertex indexed as 1, subsequent vertices incremented, and the last vertex indexed as N. The index order is determined by the surveying unit's clockwise movement along the fence boundary during on-site data collection. Connections begin with the vertex at index 1, sequentially connecting to vertices at indices 2, 3, and up to index N. Finally, the vertex at index N is connected to the vertex at index 1 to form a closed polygon. If the vertices of a work electronic fence in the electronic map database are not stored in a clockwise direction, a vertex order reordering algorithm is automatically executed during system initialization: the directed area of ​​the polygon is calculated; if the directed area is negative, it indicates that the vertices are in counter-clockwise order, so the entire vertex sequence is reversed to become clockwise, and the reordered vertex sequence is rewritten back to the electronic map database. The boundary segments of this closed polygon and the internal area enclosed by the boundary segments are collectively defined as the work permit plane area. The work permit area is a two-dimensional polygon region projected onto the XY-axis plane, used to subsequently determine whether the personnel's location coordinates are outside the work permit area. Based on the work type code and work level code, the corresponding verification period length, time validity window value, spatial coverage radius value, and verification weight value are read from the work verification rule configuration table.The operation verification rule configuration table is pre-stored in the database, using the operation type code value and operation level code value as a combined primary key. Each configuration record contains the following fields: the verification period length value is in minutes, ranging from 10 minutes to 60 minutes, with a default of 30 minutes; the time validity window value is in seconds, ranging from 30 seconds to 120 seconds, with a default of 60 seconds; the spatial coverage radius value is in meters, ranging from 3 meters to 10 meters, with a default of 5 meters; the control distance threshold value is in meters, ranging from 5 meters to 50 meters, with a default of 10 meters; the danger distance threshold value is in meters, ranging from 10 meters to 100 meters, with a default of 30 meters; the verification weight value is a floating-point vector of length 6, corresponding to the weight coefficients of personnel crossing the boundary node, gas failure node, probe adjacent alarm node, video evidence failure node, hidden danger leading node, and emergency gap node, respectively. The value range of each weight coefficient is from 0 to 100, with default values ​​of 2.5, 5.0, 7.5, 10.0, 12.5, and 15.0, respectively. The operation verification rule configuration table is manually entered by safety management personnel during system initialization, based on the "Safety Specifications for Special Operations in Chemical Production Units" and the company's internal safety management regulations. The verification period length is entered based on the operation level code: 30 minutes for Level 1 operations, 40 minutes for Level 2 operations, and 50 minutes for Level 3 operations. The effective time window is entered based on the average transmission delay of the on-site communication network, defaulting to 60 seconds to cover three standard data reporting cycles. The spatial coverage radius is entered based on the effective detection range of the fixed gas probe, defaulting to 5 meters corresponding to the standard coverage radius of a single probe. The control distance threshold and danger distance threshold are entered based on the safety distance calculation results in the LNG receiving station safety assessment report; the control distance threshold defaults to 10 meters, and the danger distance threshold defaults to 30 meters. The verification weight is entered based on the ranking of the impact of each chain break node type on operation safety determined by the analytic hierarchy process (AHP), with hazard-driven nodes having the highest weight, followed by emergency gap nodes, and so on down. The configuration table is updated using a combination of manual review and periodic audits: a configuration table review meeting is held quarterly by safety management personnel to adjust the values ​​of each field based on the statistical analysis results of the previous quarter's work safety data. Adjustments are performed through the configuration management interface, and each modification is recorded with the operator's account, modification timestamp, and field values ​​before and after the modification, forming a complete change log. The work permit time period is divided into M consecutive verification periods based on the verification period length value, where M is a positive integer greater than 1. The division method is as follows: starting from the start timestamp of the work plan, the verification period length value is incremented sequentially to generate the start and end timestamps of each period. The end timestamp of the last period is the end timestamp of the work plan. If the actual length of the last period is less than one verification period length value, it is still considered an independent consecutive verification period. Each consecutive verification period is uniquely identified by its start and end timestamps, with period numbers increasing from 1 to M.The matrix uses six rows as rows for nodes indicating personnel crossing boundaries, gas failures, adjacent alarm points from detectors, video evidence failures, hidden dangers, and emergency gaps. It also uses M consecutive verification periods (M being a positive integer greater than 1) as columns, with M columns equal to M. Node cells are defined at the intersections of matrix rows and columns, each storing a record set. Each record in the set must contain at least a data acquisition timestamp and judgment data corresponding to that node type. Initially, all node cell record sets are empty. The defined rows, columns, and node cells are combined to create a task chain break matrix with six rows and M columns. This matrix is ​​used in subsequent steps to write task verification records that meet the conditions according to consecutive verification periods and node types.

[0032] In this implementation plan, the work permit time period is divided into a number of consecutive verification time periods with a length greater than 1, and a work chain break matrix is ​​established with six fixed rows and consecutive verification time periods as columns, based on the nodes of personnel crossing boundaries, gas failure, probe alarm, video evidence failure, hidden danger, and emergency gap. This ensures that the statistics of the number of records in subsequent node units and the calculation of row conflict contribution values ​​have a unified time slice benchmark and a fixed matrix dimension, thus ensuring the comparability of chain break node types between different consecutive verification time periods under the same work task.

[0033] Specifically, the steps for determining the conditional conflict evaluation value and generating the operational status evidence field based on the number of records and corresponding weights within a node unit are as follows: The intersection of each matrix row and column is defined as a node unit. Each node unit stores the operational verification record corresponding to that node type within a single continuous verification period. Each node unit maintains a record list, and each record in the list contains at least a data acquisition timestamp field and a numerical or coordinate field related to the node type determination. Within each continuous verification period, the following six types of node unit record writing operations are performed: For personnel boundary crossing nodes, the personnel positioning coordinate value is extracted from the preprocessed operational verification data. Under XY-axis plane projection, it is determined whether the personnel positioning coordinate value is located outside the two-dimensional polygon area enclosed by the boundary line segment of the operational permit plane area. If so, the current personnel positioning coordinate value and the corresponding personnel positioning timestamp are written as a record into the node unit corresponding to the personnel boundary crossing node. The personnel boundary crossing node is used to mark the behavior of operational personnel leaving the operational permit area without authorization. For gas failure nodes, three conditions are simultaneously determined: The first condition is that if the difference between the gas detection timestamp and the end timestamp of the current continuous verification period is greater than the effective time window value, the gas detection records within the current continuous verification period are determined to have exceeded the effective time limit; the second condition is to calculate the Euclidean distance between the coordinates of the gas sampling point and the coordinates of the operation point center under the XY axis plane projection, and determine whether the Euclidean distance is greater than the spatial coverage radius value; the third condition is to sequentially determine whether the combustible gas concentration value is greater than the upper limit of the corresponding gas safety threshold, whether the oxygen volume fraction value is less than the lower limit of the corresponding gas safety threshold or greater than the upper limit of the corresponding gas safety threshold, whether the hydrogen sulfide concentration value is greater than the upper limit of the corresponding gas safety threshold, and whether the carbon monoxide concentration value is greater than the upper limit of the corresponding gas safety threshold. The upper limit of the gas safety threshold for combustible gas concentration is taken as 10% of the lower explosive limit; the lower limit of the gas safety threshold for oxygen volume fraction is taken as 19.5% by volume; the upper limit of the gas safety threshold for oxygen volume fraction is taken as 23% by volume; the upper limit of the gas safety threshold for hydrogen sulfide concentration is taken as 10 parts per million by volume; and the upper limit of the gas safety threshold for carbon monoxide concentration is taken as 24 parts per million by volume. When any one of these three conditions is met, the current gas detection timestamp and the corresponding gas concentration value are written as a record to the node unit corresponding to the gas failure node. The gas failure node is used to mark the state where the gas environment around the work site does not meet the safety requirements for operation.For the probe neighbor alarm node, the Euclidean distance between the fixed probe coordinates and the center coordinates of the work point is calculated under the XY-axis plane projection. It is then determined whether this Euclidean distance is less than or equal to the spatial coverage radius. Simultaneously, it is determined whether the fixed probe alarm duration is greater than zero seconds. If both conditions are met, the current fixed probe coordinates and alarm duration are recorded as a single record and written to the corresponding node unit of the probe neighbor alarm node. The probe neighbor alarm node is used to mark the continuous alarm status of fixed gas probes around the work point. For the video offline node, it is determined whether the video offline duration is greater than zero seconds. If it is, the current video offline duration is recorded as a single record and written to the corresponding node unit of the video offline node. The video offline node is used to mark the status of monitoring offline due to video equipment malfunction around the work point. For the hazard-driven node, it is determined whether the current number of unclosed hazard instances or the number of inspection anomalies is greater than zero. If both are greater than zero, the current number of unclosed hazard instances and the number of inspection anomalies are recorded as a single record and written to the corresponding node unit of the hazard-driven node. The hazard-driven node is used to mark the existence of unresolved hazards or inspection anomalies. For emergency gap nodes, it is determined whether the available emergency resources value is zero or the emergency passage occupancy status value is occupied. If the conditions are met, the available emergency resources value and the emergency passage occupancy status value are written as a record to the node unit corresponding to the emergency gap node. The emergency gap node is used to mark the state of insufficient emergency resources or occupied emergency passages. When multiple records are written to the same node unit, the time distance between the data collection timestamp of each record and the end timestamp of the current continuous verification period is compared in turn. The time distance is calculated by subtracting the record's data collection timestamp from the end timestamp of the current continuous verification period and taking the absolute value. The record with the smallest time distance is retained, and all other records in the node unit are deleted. When multiple records have the same time distance from the end timestamp of the current continuous verification period and are all the minimum, different retention rules are executed according to the node type. For personnel boundary crossing nodes, the record with the personnel positioning coordinates farthest from the boundary of the work permit plane area is retained. The farthest distance indicates the most serious degree of personnel boundary crossing. For gas failure nodes, retain the record with the largest exceedance of combustible gas concentration, hydrogen sulfide concentration, or carbon monoxide concentration. If multiple gas concentrations exceed the limit simultaneously, prioritize retaining the record with the largest exceedance of combustible gas concentration. For detector alarm nodes, retain the record with the longest fixed detector alarm duration. For video evidence failure nodes, retain the record with the longest offline video duration. For hazard-related nodes, prioritize retaining records with a current number of unclosed hazards greater than zero. If both records meet this criterion, retain the record with the higher hazard level, which is retrieved from the dual prevention management system.For emergency gap nodes, priority is given to retaining records where the available emergency resources are zero and the emergency passage is occupied. If both conditions are met, the record with the most recent emergency resource verification timestamp further from the end timestamp of the current continuous verification period is retained. If the above rules still cannot distinguish between them, the record with the larger data acquisition timestamp (i.e., the later one) is retained. All unretained records are deleted from the node unit. The data acquisition timestamp for each record is taken from the corresponding source based on the node type: personnel boundary crossing nodes take the personnel location timestamp, gas failure nodes take the gas detection timestamp, probe alarm nodes take the fixed probe alarm timestamp, video evidence failure nodes take the video frame acquisition timestamp, hazard-induced nodes take the later of the most recent hazard registration timestamp or the most recent inspection completion timestamp, and emergency gap nodes take the most recent emergency resource verification timestamp. Each column of the operation chain break matrix is ​​traversed according to the continuous verification period. For the six node units in each column, the number of records retained in each node unit is counted, with the record count being either zero or one. The row conflict contribution value for each node is obtained by multiplying the number of records in each matrix unit by the verification weight value of the corresponding matrix row during the current continuous verification period. The verification weight values ​​are pre-read from the operation verification rule configuration table. The verification weight value corresponding to the personnel boundary crossing node quantifies the impact of personnel boundary crossing behavior on operation safety; the verification weight value corresponding to the gas failure node quantifies the impact of gas environment failure on operation safety; the verification weight value corresponding to the probe neighbor alarm node quantifies the impact of probe neighbor alarm status on operation safety; the verification weight value corresponding to the video evidence failure node quantifies the impact of missing video surveillance on operation safety; the verification weight value corresponding to the hidden danger leading to the node quantifies the impact of unclosed hidden dangers on operation safety; and the verification weight value corresponding to the emergency gap node quantifies the impact of emergency resource gaps on operation safety. The row conflict contribution values ​​of the six matrix rows are summed, and the sum is used as the conditional conflict assessment value for the current continuous verification period. The conditional conflict assessment value is used to quantify the overall severity of the operation chain disruption conditions within the current period. Find the maximum value among the row conflict contribution values ​​of the six matrix rows, and determine the broken chain node type of the matrix row corresponding to the maximum value as the dominant broken chain type. The dominant broken chain type is used to indicate the most severe broken chain condition category in the current time period. Combine the operation broken chain matrix, condition conflict evaluation value, and dominant broken chain type to generate an operation status evidence field. The operation status evidence field is used as input data for the subsequent coupling discrimination step. The operation broken chain matrix provides the record distribution of each node unit in each time period, the condition conflict evaluation value provides the comprehensive evaluation score for each time period, and the dominant broken chain type provides the most critical broken chain category in each time period.

[0034] In this implementation plan, by associating independent judgment conditions and weighting coefficients with personnel boundary crossing nodes, gas failure nodes, probe alarm nodes, video evidence failure nodes, hidden danger leading nodes, and emergency gap nodes, the recording and single record filtering of six types of node units are completed in parallel within the same continuous verification period. This ensures that the number of records retained in each node unit is uniquely determined to be 0 or 1, avoiding multiple counting interferences from multiple records within the same node unit. Based on this, the condition conflict assessment value is obtained by statistically counting the number of records according to the continuous verification period and weighted summing. The condition conflict assessment value is a normalized value that comprehensively reflects the overall severity of the six types of chain break conditions in the current period. At the same time, the dominant chain break type is located by the maximum value of the row conflict contribution value. This ensures that the operational situation evidence field output includes both the quantitative assessment score of each period and the most critical chain break category in the period, providing structurally and content-clear evidence data for subsequent coupling judgment.

[0035] In this embodiment, sixteen consecutive verification periods are used as samples to illustrate the generation process of the operational situation evidence field and the discrimination process of situation coupling periods in this invention. Each consecutive verification period corresponds to a set of node unit records, and records the number of records for personnel boundary crossing nodes, gas failure nodes, probe adjacent alarm nodes, video evidence failure nodes, hidden danger leading nodes, and emergency gap nodes. At the same time, the condition conflict assessment value is calculated and the dominant chain break type is determined. Under the judgment rules of this embodiment, the number of records retained in each node unit is first counted (only the record closest to the end of the period is retained in the same node unit, so the maximum value is 1). Then, the number of records is multiplied by the corresponding verification weight value to obtain the conflict contribution value of each row, and the sum is used to obtain the condition conflict assessment value. Finally, the node type with the largest row conflict contribution value is determined as the dominant chain break type. When the condition conflict assessment value is high and the dominant chain break type is a hidden danger leading node, the period is marked as a situation coupling period.

[0036] Table 1. Example data of node records and confidence values ​​for each time period.

[0037]

[0038] Table 1 lists exemplary data for sixteen consecutive verification periods, recording the number of records for personnel crossing the boundary, gas failure, probe alarm, video evidence failure, hazard-related nodes, and emergency gap nodes in chronological order, along with the condition conflict assessment value and dominant chain break type. Specifically, in periods 1 and 2, all nodes had 0 records, a condition conflict assessment value of 0, and no dominant chain break type; in period 3, only the personnel crossing the boundary node had 1 record, a condition conflict assessment value of 2.5, and the dominant chain break type was personnel crossing the boundary node; in period 4, both personnel crossing the boundary node and gas failure node had 1 record, a condition conflict assessment value of 7.5, and the dominant chain break type was gas failure node; in periods 5 and 6, the personnel crossing the boundary node, gas failure node, and probe alarm node all had 1 record, a condition conflict assessment value of 15.0, and the dominant chain break type was probe alarm node; in periods 7 to 11, the personnel crossing the boundary node, gas failure node, probe alarm node, and hazard-related node all had 1 record. The condition conflict assessment value was 27.5, and the dominant chain break type was the hidden danger-driven node. Among them, time period 8 was marked as the situation coupling time period. In time period 12, the number of records for personnel crossing the boundary node, gas failure node and hidden danger-driven node was 1, the condition conflict assessment value was 20.0, and the dominant chain break type was the hidden danger-driven node. In time period 13, the number of records for personnel crossing the boundary node and hidden danger-driven node was 1, the condition conflict assessment value was 15.0, and the dominant chain break type was the hidden danger-driven node. In time periods 14 and 15, only the number of records for personnel crossing the boundary node was 1, the condition conflict assessment value was 2.5, and the dominant chain break type was personnel crossing the boundary node. In time period 16, the number of records for all nodes was 0, the condition conflict assessment value was 0, and there was no dominant chain break type.

[0039] like Figure 3As shown in the figure, the horizontal axis represents the sequence number of continuous verification periods (period 1 to period 16), the left side of the vertical axis represents the six node types of the operation chain break matrix (from top to bottom: personnel crossing the boundary node, gas failure node, probe adjacent alarm node, video evidence failure node, hidden danger leading node, and emergency gap node), and the right side of the vertical axis represents the condition conflict assessment value. The main part uses a heat map, where each cell is color-coded to indicate whether the node unit retains a record within the corresponding period. Dark cells are marked with the number "1" to indicate that the number of records is 1. The blue broken line above connects the condition conflict assessment values ​​of each period. The value is obtained by multiplying the number of records of each node unit by the corresponding verification weight value and summing them. In the figure, period 1 to period 2 is 0, period 3 rises to 2.5, period 4 to period 6 gradually climbs, period 7 to period 11 reaches the highest value of 27.5, and then gradually falls back to 0 in period 16. The red arrows mark the dominant chain break types for each time period: Time period 3 is the personnel boundary crossing node, Time period 4 is the gas failure node, Time periods 5 to 6 are the probe neighbor alarm node, Time periods 7 to 13 are the hidden danger-driven node, and Time periods 14 to 15 are again the personnel boundary crossing node, clearly showing the evolution trend of the dominant chain break types. Time period 8 is highlighted in red and marked as "Strong Coupling Time Period 8" in the figure, indicating that this time period is judged to be the situation coupling period where the evidence pair is strongly coupled. At this time, the condition conflict assessment value is high and the dominant chain break type is the hidden danger-driven node, and the system generates corresponding scheduling instructions accordingly. Figure 3 It fully covers the technical path from establishing the operation chain break matrix, recording and statistically analyzing node units, calculating condition conflict assessment values, determining the dominant chain break type, to identifying the situation coupling period.

[0040] Specifically, the steps for marking adjacent control areas and adjacent hazard sources based on the operational situation evidence field are as follows: First, read the center coordinates of the work point from the operational situation evidence field. These center coordinates consist of X-axis, Y-axis, and Z-axis components. Second, read the boundary coordinates of the control area and the hazard source from the pre-processed operational verification data. The control area boundary coordinates are a sequence of polygonal vertex coordinates arranged clockwise, with each vertex consisting of X-axis, Y-axis, and Z-axis components. This sequence defines the boundary of a two-dimensional closed area in the station's three-dimensional coordinate system. The hazard source boundary coordinates are a sequence of outer edge point coordinates of major hazard source storage facilities. Each outer edge point coordinate consists of X-axis, Y-axis, and Z-axis components. This sequence defines the projected boundary of the major hazard source facilities on the XY-axis plane in the station's three-dimensional coordinate system. The hazard source boundary coordinates and the control area boundary coordinates share the same source but have different uses: the control area is used for delineating the scope of daily operational risk management, while the hazard source is used for assessing the safety distance to legally defined major hazard sources. Under XY-axis plane projection, the X-axis and Y-axis components of the work point center coordinates are extracted to form the work point projection point. For the control area boundary coordinate value sequence, the X-axis and Y-axis components of each vertex coordinate value are extracted to form the control area projection polygon. The perpendicular distance from the work point projection point to each boundary line segment of the control area projection polygon is calculated, and the minimum value among all perpendicular distances is taken as the shortest distance from the work point center to the control area boundary. This shortest distance is denoted as the area shortest distance, which is used to quantify the spatial proximity between the work point and the control area. In the case of multiple control areas, the corresponding area shortest distance is calculated for each control area. Under XY-axis plane projection, the X-axis and Y-axis components of the work point center coordinates are extracted to form the work point projection point. For the hazard source boundary coordinate value sequence, the X-axis and Y-axis components of each boundary point coordinate value are extracted to form the hazard source projection polygon. Calculate the perpendicular distance from each boundary segment of the polygon projected from the work point to the hazard source. Take the minimum of all perpendicular distances as the shortest distance from the center of the work point to the boundary of the major hazard source. This shortest distance is denoted as the hazard shortest distance, which quantifies the spatial proximity between the work point and the major hazard source. For cases with multiple major hazard sources, calculate the corresponding hazard shortest distance for each major hazard source sequentially. Read the control distance threshold and hazard distance threshold from the work verification rule configuration table. The control distance threshold is a length value in meters, ranging from five to fifty meters, with a default of ten meters, used to determine whether the work point and the control area are adjacent. The hazard distance threshold is a length value in meters, ranging from ten to one hundred meters, with a default of thirty meters, used to determine whether the work point and the major hazard source are adjacent.A control distance threshold lower than a hazard distance threshold indicates that the sensitivity to proximity of the controlled area during operation is higher than the sensitivity to proximity of major hazard sources. Control areas whose shortest distance is not greater than the control distance threshold are marked as adjacent control areas. Adjacent control areas indicate that the spatial distance between the work point and the control area under XY-axis projection is less than a preset safety control value. Subsequent coupling judgments will prioritize the association between this area and operation chain break records. Major hazard sources whose shortest hazard distance is not greater than the hazard distance threshold are marked as adjacent hazard sources. Adjacent hazard sources indicate that the spatial distance between the work point and the major hazard source under XY-axis projection is less than a preset safety distance value. Subsequent coupling judgments will prioritize the association between this hazard source and process alarm records. For control areas whose shortest distance is greater than the control distance threshold, no proximity marking is performed, and the association between this control area and operation chain break records is no longer considered in subsequent coupling judgments. For major hazard sources whose shortest hazard distance is greater than the hazard distance threshold, no proximity marking is performed, and the association between this major hazard source and process alarm records is no longer considered in subsequent coupling judgments.

[0041] In this implementation scheme, by calculating the shortest distance from the center coordinates of the work point to the boundary coordinates of the controlled area and the boundary coordinates of the hazard source under the XY-axis plane projection, the spatial proximity judgment is reduced from three-dimensional space to two-dimensional plane. This eliminates the influence of height differences in the Z-axis direction on distance calculation, ensuring that the shortest distance in the area and the shortest distance in the hazard only reflect the actual horizontal distance between the work point and the controlled area and the hazard source. Furthermore, by comparing the shortest distance in the area with the control distance threshold and the shortest distance in the hazard with the hazard distance threshold, adjacent controlled areas with a horizontal distance less than a preset safety value are selected from all controlled areas, and adjacent hazard sources with a horizontal distance less than a preset safety value are selected from all major hazard sources. This provides clear screening results for the subsequent construction of spatial association edges between the first type of nodes and the second type of nodes in the coupled bipartite graph.

[0042] Specifically, the steps for constructing a coupled bipartite graph based on the dominant chain break record and process alarm record are as follows: Figure 4As shown, node unit records corresponding to the dominant chain break type are extracted from the operation chain break matrix and designated as master records. The dominant chain break type is the chain break node type corresponding to the maximum value determined in the previous step by comparing the row conflict contribution values ​​of the six matrix rows. The master record contains the data acquisition timestamp and the judgment data corresponding to the node type. The master record is used to establish the spatial association between chain break nodes and controlled areas or hazard sources in subsequent steps. The spatial coordinates corresponding to the master record are extracted from the preprocessed operation verification data. The method for extracting spatial coordinates is determined based on the dominant chain break type: when the dominant chain break type is a personnel boundary crossing node, the personnel positioning coordinates are extracted from the preprocessed operation verification data as spatial coordinates; when the dominant chain break type is a gas failure node, the gas sampling point coordinates are extracted as spatial coordinates; when the dominant chain break type is a probe adjacent alarm node, the fixed probe coordinates are extracted as spatial coordinates; when the dominant chain break type is a video evidence failure node, the video equipment coordinates are extracted as spatial coordinates; when the dominant chain break type is a hazard-driven node, the hazard location coordinates are extracted as spatial coordinates; when the dominant chain break type is an emergency gap node, the emergency resource coordinates are extracted as spatial coordinates. The spatial coordinate sources for each dominant chain break type are different, but all have undergone coordinate transformation in the station's three-dimensional coordinate system within the preprocessed operation verification data. Process alarm records with a duration greater than zero seconds within the current continuous verification period are extracted from the preprocessed operation verification data. Each process alarm record includes at least the process alarm duration, alarm point coordinates, alarm-associated equipment identifier, and alarm-associated area identifier. The duration of the process alarm is measured in seconds, representing the time it takes for the process parameters to return to normal after the initial exceedance. The alarm point coordinates consist of X-axis, Y-axis, and Z-axis components, used in subsequent steps to determine the spatial inclusion relationship between the alarm point and the controlled area or hazard source. A coupled bipartite graph is constructed. This coupled bipartite graph is a bidirectional directed graph containing a first-type set of nodes and a second-type set of nodes, with edges between the two sets established according to spatial inclusion relationships. Nodes indicating personnel exceeding boundaries, gas failure, adjacent alarm probes, video evidence failure, hazard-related issues, and emergency gaps are designated as first-type nodes, with a fixed total of six. Adjacent controlled areas and adjacent hazard sources, marked in previous steps, are designated as second-type nodes, with the number of second-type nodes equal to the sum of the number of adjacent controlled areas and adjacent hazard sources. No edges are established between first-type and second-type nodes, with no edges pointing from one first-type node to another. When the spatial coordinates corresponding to the master record are located inside the region polygon represented by a second type node under the projection of the XY axis plane, a directed edge is established from the first type node corresponding to the master record to the second type node.The method for determining whether spatial coordinates are located inside the region polygon is as follows: A ray is emitted from the projection point of the spatial coordinates in any direction. The number of intersections between this ray and each boundary segment of the region polygon is calculated. If the number of intersections is odd, it is determined to be inside the polygon; if the number of intersections is even, it is determined to be outside the polygon. The direction of this directed edge indicates that the location of the chain break event spatially overlaps with a specific control area or hazard source, meaning that the chain break behavior during the operation is directly related to this area. The weight of this directed edge is taken as the row conflict contribution value of the first type of node in the current continuous verification period. The row conflict contribution value reflects the severity of the chain break node and is used as the weight of the directed edge for the weighted transmission of activation values ​​in subsequent message propagation. When the alarm point coordinates of the process alarm record fall inside the region polygon represented by a second type of node under the XY axis plane projection, a directed edge is established from the second type of node to the first type of node corresponding to the main record. The method for determining whether alarm point coordinates fall into the region polygon is the same as the aforementioned method for determining whether spatial coordinates fall into the polygon, both using the ray method. The direction of the directed edge indicates that the location of the process alarm event spatially overlaps with a specific controlled area or hazard source, meaning the process alarm is directly related to that area. The weight of the directed edge is the normalized value obtained by dividing the process alarm duration by the upper limit of alarm duration. The upper limit of alarm duration is read from the operation verification rule configuration table, with a range of 60 to 360 seconds, and a default value of 180 seconds. The normalized value ranges from zero to one. The closer the process alarm duration is to the upper limit of alarm duration, the closer the normalized value is to one, indicating a more severe duration of the process alarm. This normalized value is used as the weight of the directed edge for the weighted transmission of activation values ​​in subsequent message propagation. For the same process alarm record, if the alarm point coordinates fall into the polygon represented by multiple second-type nodes, a directed edge is created for each second-type node pointing to the first-type node corresponding to the main record, with each edge having the same normalized weight. For the same second-type node, if both the conditions for the main record's spatial coordinates and the process alarm point's coordinates to fall within the specified area are met, two directed edges in opposite directions are established, forming a bidirectional connection. If the spatial coordinates of the main record do not fall within the polygonal region represented by the second-type node, no positive edge is established from the first-type node corresponding to the main record to that second-type node; similarly, if the alarm point coordinates of the process alarm record do not fall within the polygonal region represented by the second-type node, no negative edge is established from that second-type node to the first-type node corresponding to the main record. After the coupled bipartite graph is constructed, it is used for weighted message passing calculations in subsequent steps.

[0043] In this implementation scheme, a bidirectional weighted coupled bipartite graph is constructed between the main record spatial coordinates corresponding to the dominant chain break type and adjacent control areas and adjacent hazards. This is achieved by performing point-to-surface spatial inclusion judgment on the main record spatial coordinates corresponding to the dominant chain break type and adjacent control areas and hazards. Simultaneously, the alarm point coordinates of the process alarm record are also performed with the same group of adjacent control areas and adjacent hazards. The row conflict contribution value is used as the weight of the directed edge from the first type of node to the second type of node, and the normalized value of the process alarm duration is used as the weight of the directed edge from the second type of node to the first type of node. This bipartite graph establishes a quantifiable association between the chain break node and the process alarm node in the same spatial region, providing a clear propagation path and weight basis for the bidirectional update of activation values ​​between the two types of nodes in subsequent weighted message transmission.

[0044] Specifically, the steps for marking strongly coupled evidence pairs and generating operational status scheduling units after updating node activation values ​​through weighted message transmission are as follows: The condition conflict assessment value is used as the initial activation value for the first type of nodes. The condition conflict assessment value is a value calculated in the previous steps, reflecting the comprehensive severity of the six types of chain-breaking conditions within the current continuous verification period. There are six first-type nodes: personnel boundary crossing node, gas failure node, probe adjacent alarm node, video evidence failure node, hidden danger traction node, and emergency gap node. Each first-type node receives the same initial activation value, which is equal to the condition conflict assessment value. The condition conflict assessment value serves as the initial activation value and is used as the basis for iterative updates of node activation values ​​in subsequent message propagation. The average of the normalized duration values ​​of the process alarms associated with each second-type node is used as the initial activation value for the second-type nodes. The second-type nodes include all adjacent control areas and all adjacent hazard sources marked in the previous steps. The set of process alarm records associated with each second-type node is obtained in the previous steps by determining whether the process alarm point coordinates fall within the polygon area of ​​the second-type node. For each second-type node, the normalized duration values ​​of all associated process alarm records are extracted, and the arithmetic mean of these normalized values ​​is calculated. If a second-type node is associated with K process alarm records, the K normalized values ​​are added together and divided by K to obtain the average value. This average value serves as the initial activation value for the second-type node, used as the basis for iterative updates of node activation values ​​in subsequent message propagation. If a second-type node is not associated with any process alarm records, the initial activation value is zero. A graph message propagation algorithm is used to perform N rounds of message passing on the coupled bipartite graph. N is a preset positive integer, ranging from two to ten, with a default value of three. Each round of message passing includes two phases: a message sending phase and a message receiving and updating phase. During the message sending phase, each node traverses all its outgoing edges and identifies the terminal node of each outgoing edge as its adjacent node connected through that outgoing edge. Each node weights its current activation value according to the weight of the corresponding outgoing edge to obtain the incoming edge message to be sent to that adjacent node. The weighting calculation method is to multiply the current activation value by the weight of the outgoing edge to obtain the weighted message corresponding to that outgoing edge, and then send the weighted message along the outgoing edge to the terminal node of that outgoing edge, that is, to the adjacent node corresponding to that outgoing edge. Each node can send messages to multiple adjacent nodes simultaneously, and the weighted message corresponding to each outgoing edge is calculated separately according to the weight of its respective outgoing edge. The sending order does not affect the final result. During the message receiving and updating phase, each node collects the weighted messages transmitted from all incoming edges, adds all the values ​​of these weighted messages to obtain the incoming edge message sum, adds the incoming edge message sum to its own current activation value, and uses the result as the updated activation value of that node.In each round of message passing, all nodes synchronously complete message sending and message receiving updates. That is, all nodes send messages using their activation values ​​from before the start of the round. After all nodes have sent their messages, they uniformly update their activation values ​​by adding the incoming edge message value to their existing activation values. At the end of each round of message passing, the activation values ​​of all nodes are updated once. To prevent the activation values ​​from growing indefinitely over multiple rounds of propagation, after updating the node's activation value in each round of message receiving update, a truncation process is performed on the node's updated activation value. The truncation process is as follows: it checks whether the updated activation value is greater than a preset activation value upper limit, which is 100.0. If the updated activation value is greater than 100.0, the node's activation value is set to 100.0; if the updated activation value is less than or equal to 100.0, the original value is retained. The upper limit of the activation value is based on the theoretical maximum value of the condition conflict assessment value. When the number of records in all six node units is 1 and each is multiplied by the maximum verification weight value of 15.0, the maximum condition conflict assessment value is 2.5 + 5.0 + 7.5 + 10.0 + 12.5 + 15.0 = 52.5. After weighted summation through message propagation, the upper limit of the activation value is set to 100.0, which covers twice the theoretical maximum value and prevents infinite growth. The final activation value of each node after N rounds of propagation is determined as the coupling activation degree. The coupling activation degree reflects the sum of the comprehensive impact received by each node from its neighboring nodes and the initial activation value of that node after N rounds of message propagation. Truncation ensures the boundedness of the activation value. The coupling activation degree of the first type of node reflects the final state of the broken-chain node after being negatively amplified by the process alarm information of the adjacent second type of node. The coupling activation degree of the second type of node reflects the final state of the controlled area or hazard source after being positively amplified by the broken-chain information of the adjacent first type of node. The coupling activation degree is used in subsequent steps to calculate the coupling strength between node pairs. For each node pair consisting of a dominant broken-chain node and each second-type node, the coupling strength is calculated. The dominant broken-chain node is the only first-type node determined in the previous steps. Each second-type node is traversed, and the dominant broken-chain node is paired with it to form a node pair. First, it is determined whether there is at least one directed edge between the node pair. The existence of a directed edge is determined by whether a directed edge from the dominant broken-chain node to the second-type node or from the second-type node to the dominant broken-chain node was established in the previous steps. If there is no directed edge between the node pairs, the node pair is skipped, and no further calculation is performed. If there is at least one directed edge between the node pairs, the coupling activation value of the dominant broken-chain node and the coupling activation value of the second-type node are extracted, and the two coupling activation values ​​are multiplied to obtain the product. Simultaneously, the weights of all directed edges between the node pairs are extracted, and the weights of all directed edges are added together to obtain the weight sum.When calculating the coupling strength value, it is first confirmed that there is at least one directed edge between the node pairs, and the node pairs with directed edges are taken as the node pairs to be calculated. For the node pairs to be calculated, the weights of all directed edges between the node pairs are extracted, and the weights of all directed edges are added together to obtain the weight sum. Since the weight of the directed edge from the dominant broken chain node to the second type of node is taken as the row conflict contribution value corresponding to the first type of node, and the dominant broken chain node is determined by the matrix row with the largest row conflict contribution value in the current continuous verification period, the row conflict contribution value corresponding to the dominant broken chain node entering the coupling strength calculation is a valid value with a value greater than zero. Since the weight of the directed edge from the second type of node to the dominant broken chain node is taken as the normalized value obtained by dividing the process alarm duration value by the alarm duration upper limit value, and the process alarm record is limited to the configuration value that the process alarm duration value is greater than zero and the alarm duration upper limit value is greater than zero when extracted, the weight sum corresponding to the node pairs entering the coupling strength calculation is a valid value with a value greater than zero. Multiply the coupling activation degree of the first type of node by the coupling activation degree of the second type of node to obtain the product value. Divide the product value by the weighted sum to obtain the coupling strength value of the node pair. The coupling strength value is used to quantify the tightness of the association between the node pair's broken chain node and the controlled area or hazard source. Read the coupling strength threshold from the operation verification rule configuration table. The coupling strength threshold ranges from 0.1 to 1.0, with a default value of 0.5. Node pairs with coupling strength values ​​greater than the coupling strength threshold are marked as strong coupling evidence pairs. Strong coupling evidence pairs indicate that within the current continuous verification period, there is a strong correlation between the dominant broken chain node and the second type of node, and the broken chain event and process alarm form verifiable coupling evidence in this spatial area. Extract the broken chain node type corresponding to the first type of node from the strong coupling evidence pair as the coupling broken chain type. The coupling broken chain type is the same as the dominant broken chain type, which is one of the following: personnel boundary crossing node, gas failure node, probe adjacent alarm node, video evidence failure node, hidden danger traction node, or emergency gap node, used to indicate the broken chain category that plays a dominant role in the strong coupling evidence pair. The continuous verification period containing the strongly coupled evidence pair is marked as the situation coupling period. A situation coupling period indicates that within this period, the dominant broken-chain node and at least one second-type node form a strongly coupled evidence pair. The security situation during this period needs to be processed through the subsequent scheduling token generation step. The start timestamp, job task number, job point center coordinates, condition conflict assessment value, and coupling broken-chain type of the situation coupling period are written into the operational situation scheduling unit. The operational situation scheduling unit is a data storage structure used to read and generate the corresponding scheduling token and job scheduling instruction in the subsequent scheduling token generation step. The start timestamp of the situation coupling period is used to locate the time point of the scheduling instruction. The job task number is used to associate the corresponding job task. The job point center coordinates are used to display the scheduling location on the 3D electronic map. The condition conflict assessment value is used to determine the triggering condition of the scheduling instruction. The coupling broken-chain type is used to generate the corresponding broken-chain token.After the operational status scheduling unit completes the writing, it serves as the input data for the subsequent scheduling token generation step.

[0045] In this implementation scheme, the condition conflict evaluation value is used as the initial activation value of the first type of node, and the average value of the normalized duration of the process alarm is used as the initial activation value of the second type of node. A graph message propagation algorithm is used to perform multiple rounds of weighted message passing on the coupled bipartite graph, so that the activation values ​​between the first type of node and the second type of node are updated bidirectionally along the directed edges, and finally the coupling activation degree of each node is obtained. Then, the coupling strength value is calculated by dividing the product of the coupling activation degrees by the weight and the sum. Node pairs with coupling strength values ​​greater than the coupling strength threshold are marked as strong coupling evidence pairs. The coupling break type is extracted from the strong coupling evidence pairs and the situation coupling period is located. This ensures that the operation situation scheduling unit contains only key data that has been coupled and verified, providing a denoised and reliable input for the subsequent generation of scheduling tokens.

[0046] Specifically, the steps for generating corresponding scheduling tokens based on the operational status scheduling unit and generating job scheduling instructions in conjunction with condition conflict assessment values ​​are as follows: Read the coupling chain break type, condition conflict assessment value, and job task number from the operational status scheduling unit. Read the process alarm duration, available emergency resource quantity, and emergency channel occupancy status from the pre-processed job verification data. The coupling chain break type is the type of chain break node extracted from strongly coupled evidence pairs in the previous steps, and its value is one of the following: personnel boundary crossing node, gas failure node, probe adjacent alarm node, video evidence failure node, hidden danger traction node, or emergency gap node. The condition conflict assessment value is the comprehensive severity value of the six types of chain break conditions within the current continuous verification period calculated in the previous steps. When the coupling chain break type is a personnel boundary crossing node, a personnel deviation token is generated. Each token is a structured data object containing the following fields: token type identifier, job task number, start timestamp of the status coupling period, center coordinates of the job point, condition conflict assessment value, handling priority value, and instruction routing information. The token type identifier of the personnel deviation token is 1. Personnel departure tokens are used to mark personnel who have left their permitted work area without authorization during the current continuous verification period, and this behavior forms strong evidence of coupling with adjacent controlled areas or adjacent hazards. A gas failure token is generated when the coupling chain type is a gas failure node. The token type identifier for the gas failure token is 2. A gas failure token is also used to mark when the gas environment around the work site does not meet safe operating requirements during the current continuous verification period, and this gas failure state forms strong evidence of coupling with adjacent controlled areas or adjacent hazards. A probe alarm node is generated when the coupling chain type is a probe alarm node. The token type identifier for the probe alarm node is 3. A probe alarm node is used to mark when fixed gas probes around the work site continuously alarm during the current continuous verification period, and this alarm state forms strong evidence of coupling with adjacent controlled areas or adjacent hazards. A video interruption token is generated when the coupling chain type is a video interruption node. The token type identifier for the video interruption token is 4. A video interruption token is used to mark when video equipment around the work site goes offline during the current continuous verification period, causing monitoring interruption, and this interruption state forms strong evidence of coupling with adjacent controlled areas or adjacent hazards. When the coupling break type is a hazard-driven node, a hazard-driven token is generated. The token type identifier of the hazard-driven token is 5. The hazard-driven token is used to mark the existence of an unclosed hazard or inspection anomaly within the current continuous verification period, and the hazard status forms strong coupling evidence with the adjacent controlled area or adjacent hazard source. When the duration of a process alarm is greater than zero seconds, a process disturbance token is generated. The token type identifier of the process disturbance token is 6. The duration of the process alarm is obtained from the distributed control system, in seconds, representing the time from the first time the process parameter exceeds the limit to the time it returns to normal. A process alarm duration greater than zero seconds indicates that there is at least one process alarm record within the current continuous verification period.Process disturbance tokens are used to mark process parameters that exceed limits during the current continuous verification period, and these alarms may be causally related to chain break events. A support gap token is generated when the available emergency resources are zero and the emergency passage is occupied. The token type identifier for the support gap token is 7. The available emergency resources are obtained from the emergency resource management system, and the emergency passage occupancy status is obtained from the video analytics system or access control system. The support gap token is used to mark that emergency resources are insufficient and the emergency passage is occupied during the current continuous verification period, and this gap status cannot support normal emergency response needs. When each token is generated, the job task number field is directly copied from the job task number in the operational status scheduling unit. The situation coupling period start timestamp field is directly copied from the situation coupling period start timestamp in the operational status scheduling unit. The job point center coordinate value field is copied from the job point center coordinate value in the operational status scheduling unit. The condition conflict assessment value field is copied from the condition conflict assessment value in the operational status scheduling unit. The priority value field is determined based on the token type identifier: personnel deviation tokens and gas failure tokens have a priority value of 1, indicating the highest priority; probe alarm tokens and video evidence failure tokens have a priority value of 2; hazard-induced tokens and process disturbance tokens have a priority value of 3; and support gap tokens have a priority value of 4. The instruction routing information field contains a list of network addresses for the target management end. Personnel deviation tokens and gas failure tokens are routed to the special operations management end; probe alarm tokens and video evidence failure tokens are routed to the mobile video verification end and intelligent inspection end; hazard-induced tokens and process disturbance tokens are routed to the safety production dashboard; and support gap tokens are routed to the personnel positioning management end and emergency communication end. After tokens are generated, they are temporarily stored in the message queue, awaiting retrieval in subsequent dispatch instruction generation steps. The generated personnel deviation token, gas failure token, probe neighbor alarm token, video evidence failure token, hazard entrapment token, process disturbance token, and support gap token are used to trigger corresponding operation scheduling instructions. Specifically, the personnel deviation token and gas failure token trigger operation suspension scheduling instructions; the probe neighbor alarm token and video evidence failure token trigger on-site verification scheduling instructions; the combination of the hazard entrapment token and process disturbance token triggers escalation control scheduling instructions; and the support gap token triggers emergency support scheduling instructions. A confidence threshold is read from the operation verification rule configuration table. The confidence threshold ranges from zero to one hundred, with a default value of ten.0. The confidence threshold is used to determine whether the condition conflict assessment value reaches the severity level required to trigger a scheduling instruction. The upper limit of the condition conflict assessment value is also read from the operation verification rule configuration table. This upper limit is used for normalization but does not directly participate in condition judgment. When generating an operation scheduling instruction, the following four rules are judged sequentially: Rule 1: If the condition conflict assessment value is greater than the confidence threshold, and the coupling chain break type is a personnel out-of-bounds node or a gas failure node, then an operation suspension scheduling instruction is generated.A condition conflict assessment value greater than the confidence threshold indicates that the overall severity of the chain break during the current continuous verification period exceeds the preset safety tolerance limit. Coupled chain break types of "personnel out-of-bounds node" or "gas failure node" indicate that the most severe chain break type is unauthorized departure of personnel from the work area or gas environment failure. The work suspension dispatch instruction is used to command the operators to immediately stop their current work activities and evacuate the work area. Rule 2: If the condition conflict assessment value is greater than the confidence threshold, and the coupled chain break type is "probe adjacent alarm node" or "video evidence failure node," then a field review dispatch instruction is generated. A coupled chain break type of "probe adjacent alarm node" indicates that the most severe chain break type is continuous alarm from a fixed gas probe; a coupled chain break type of "video evidence failure node" indicates that the most severe chain break type is video equipment offline causing monitoring evidence failure. The field review dispatch instruction is used to command on-site monitoring personnel to go to the work area to conduct an on-site review of the gas environment status or video monitoring status. Rule 3: If a process disturbance token has been generated, and the coupled chain break type is "hazard traction node," then an upgraded control dispatch instruction is generated. A generated process disturbance token indicates that there is a process parameter out-of-bounds alarm during the current continuous verification period. The "coupled chain break type" indicates the most serious type of chain break, signifying the existence of unclosed loop hazards or inspection anomalies. The upgraded control and dispatch instruction is used to raise the operational safety control level by one level, increasing control resource investment. Rule 4: If a support gap token has been generated, an emergency support dispatch instruction is generated. A generated support gap token indicates that the number of available emergency resources is zero and the emergency channel is occupied during the current continuous verification period. The emergency support dispatch instruction commands the emergency management department to immediately replenish emergency resources and clear the emergency channel. These four rules are parallel and are judged sequentially according to the above order. Each rule independently generates a corresponding dispatch instruction, and multiple dispatch instructions may be generated simultaneously within a continuous verification period. All generated dispatch tokens and dispatch instructions are associated with the current operational task number and the start timestamp of the current continuous verification period.

[0047] In this implementation scheme, coupling chain break types are mapped to personnel deviation tokens, gas failure tokens, probe neighbor alarm tokens, video evidence failure tokens, and hidden danger traction tokens, respectively. These tokens, along with process disturbance tokens generated from process alarm duration values ​​and support gap tokens generated from emergency resource availability values ​​and emergency channel occupancy status values, constitute a set of scheduling tokens. The operation suspension scheduling command or on-site review scheduling command is generated by combining the condition conflict assessment value with the confidence threshold and the coupling chain break type. The combination of process disturbance tokens and coupling chain break types generates escalation control scheduling commands, and the support gap token generates emergency support scheduling commands. This ensures that the triggering conditions for operation scheduling commands depend simultaneously on the determination of three dimensions: quantitative assessment score, chain break type classification, and token existence status, avoiding the one-sidedness of relying solely on a single threshold or single type of event to trigger scheduling.

[0048] Specifically, the steps for sending the operation scheduling instruction to the corresponding management terminal and outputting the safety operation status scheduling result are as follows: The operation suspension scheduling instruction is sent to the special operation management terminal via a network interface. The special operation management terminal is an independent business terminal deployed in the central control room of the LNG receiving station, used to receive and process scheduling instructions related to operation suspension. The operation suspension scheduling instruction includes the operation task number, the center coordinates of the operation point, the condition conflict assessment value, and the coupling chain break type. After receiving the operation suspension scheduling instruction, the special operation management terminal displays a confirmation window on the operation interface. The shift supervisor clicks to confirm, and an operation suspension confirmation timestamp is generated. The on-site verification scheduling instruction is sent to the mobile video verification terminal and the intelligent inspection terminal via a network interface. The mobile video verification terminal is installed on the handheld terminal of the on-site monitoring personnel, and the intelligent inspection terminal is installed on the explosion-proof handheld terminal of the inspection personnel. The on-site verification scheduling instruction includes the operation task number, the center coordinates of the operation point, the coupling chain break type, and a list of specific items to be verified. When the coupling chain break type is a probe adjacent alarm node, the list of verification items includes the fixed probe location and the current gas concentration value. When the coupling chain break type is a video verification node, the review item list includes the offline video device number and on-site photo requirements. Upon receiving the instruction, the mobile video verification terminal sends on-site monitoring personnel with a mobile camera to the work site to capture and upload the video. Upon receiving the instruction, the intelligent inspection terminal sends inspection personnel with a gas detector to the work site to detect gas concentration and upload the detection data. After each independently completing its review task, the mobile video verification terminal and the intelligent inspection terminal generate their own on-site review completion timestamps, which may differ. The upgraded control and dispatch instruction is sent to the safety production dashboard via a network interface. The safety production dashboard is a comprehensive display terminal deployed on a large display screen in the LNG receiving station dispatch center. The upgraded control and dispatch instruction includes the work task number, work site center coordinates, condition conflict assessment value, coupling chain break type, and the upgraded control level. The upgraded control level is one level higher than the original work level. Upon receiving the upgraded control and dispatch instruction, the safety production dashboard highlights the work task on its large screen, and the dispatch center supervisor clicks to confirm, generating an upgraded control confirmation timestamp. Emergency support dispatch instructions are sent via network interface to the personnel location management terminal and the emergency communication terminal. The personnel location management terminal is deployed on the workstation of safety management personnel, and the emergency communication terminal is deployed in the emergency command center. The emergency support dispatch instructions include the task number, the center coordinates of the task location, the type of emergency resource shortage, and the support requirements. Emergency resource shortage types include zero available emergency resources and occupied emergency access routes. The support requirements are to replenish emergency resources or clear emergency access routes. Upon receiving the instructions, the personnel location management terminal displays the location of occupied emergency access routes on an electronic map and notifies security personnel to clear them. Upon receiving the instructions, the emergency communication terminal calls for emergency material transport personnel to replenish resources via the emergency broadcast system.After each independent completion of its support task, the personnel positioning management terminal and the emergency communication terminal generate their own emergency support arrival timestamps, which may differ. Sent dispatch instructions are displayed on a 3D electronic map based on the center coordinates of the work point. The 3D electronic map is pre-loaded into the system and supports rotation and scaling along the X, Y, and Z axes. The display method is as follows: a circular icon is overlaid on the map position corresponding to the center coordinates of the work point. The icon color corresponds to the dispatch instruction type: a red circular icon for work pause, a yellow circular icon for on-site verification, an orange circular icon for escalation control, and a blue circular icon for emergency support. The dispatch instruction type and sending timestamp are displayed floating above the icon. After each dispatch instruction is successfully sent, its corresponding sending timestamp is recorded. The sending timestamp is taken from the system clock with millisecond precision. A unified network time protocol synchronization service is deployed on the special operation management terminal, mobile video verification terminal, intelligent inspection terminal, safety production dashboard, personnel positioning management terminal, and emergency communication terminal. A master time server is set up within the station area, acquiring standard time through a BeiDou satellite receiver with millisecond precision. The master time server broadcasts a time synchronization message to all management terminals every second. Upon receiving the synchronization message, each management terminal calibrates its local system clock. When the connection between the master time server and the BeiDou satellite receiver is interrupted, the master time server switches to its local high-precision crystal oscillator to maintain time output. The daily deviation of the crystal oscillator is less than one-thousandth of a second, maintaining time synchronization accuracy for at least 72 hours. When generating confirmation timestamps, each management terminal reads the time value from its local synchronized and calibrated system clock to ensure comparability between confirmation timestamps generated by different management terminals, with a difference calculation error of less than ten milliseconds. The operation pause confirmation timestamp is read from the special operation management terminal. This timestamp is generated by the shift supervisor when confirming the operation pause on the special operation management terminal, with millisecond precision. If no operation pause scheduling instruction is generated within the continuous verification period, this reading step is not executed. The on-site verification completion timestamps are read from the mobile video verification terminal and the intelligent inspection terminal, respectively. The on-site verification completion timestamp has two sources: one generated after the mobile video verification terminal completes video recording and uploading, and another generated after the intelligent inspection terminal completes gas detection and uploading. The values ​​of these two timestamps are compared, and the earlier timestamp is taken as the on-site verification confirmation timestamp. If no on-site verification dispatch instruction is generated within the continuous verification period, this step of the reading operation is not performed. The upgrade control confirmation timestamp is read from the safety production dashboard. The upgrade control confirmation timestamp is generated by the dispatch center supervisor when confirming the upgrade control operation in the safety production dashboard, with a precision of milliseconds. If no upgrade control dispatch instruction is generated within the continuous verification period, this step of the reading operation is not performed. The emergency support arrival timestamps are read from the personnel positioning management terminal and the emergency communication terminal, respectively.The emergency support arrival timestamp has two sources: one timestamp is generated after the personnel location management terminal confirms the emergency channel is clear, and another timestamp is generated after the emergency communication terminal confirms the emergency resources are replenished. The values ​​of these two timestamps are compared, and the earlier timestamp is taken as the emergency support confirmation timestamp. If no emergency support dispatch instruction is generated within the continuous verification period, this step of the reading operation is not performed. For each confirmation timestamp, the sending timestamp of the dispatch instruction corresponding to that confirmation timestamp is first obtained. The confirmation timestamp is compared with the sending timestamp, and the difference is calculated only if the value of the confirmation timestamp is strictly greater than the value of the sending timestamp. The calculation method is to subtract the sending timestamp from the confirmation timestamp, and the difference is in seconds, rounded to three decimal places, as the single dispatch duration corresponding to the dispatch instruction. If the value of the confirmation timestamp is less than or equal to the sending timestamp, the difference is not calculated, and the single dispatch duration corresponding to the dispatch instruction is recorded as an invalid value. Invalid single dispatch durations are not included in subsequent statistical calculations. All valid single dispatch durations are collected, and the minimum value is taken as the dispatch response duration value. The dispatch response time is defined as the time interval from the issuance of a dispatch command to the receipt of the first acknowledgment. The first acknowledgment corresponds to the earliest completed on-site handling action. This metric is used to evaluate the response speed to dispatch commands and the efficiency of on-site handling initiation. The maximum value among all valid single dispatch durations is taken as the dispatch closed-loop time. The dispatch closed-loop time is defined as the time interval from the issuance of a dispatch command to the receipt of the last acknowledgment. The last acknowledgment corresponds to the latest completed on-site handling action. This metric is used to evaluate the complete execution cycle of all on-site handling tasks required by the dispatch command. These two metrics quantify the handling effect of dispatch commands from the dimensions of response speed and execution completeness, respectively. The dispatch response time and dispatch closed-loop time are output as the safety operation status dispatch results. The safety operation status dispatch results are written to a log file in JSON format and simultaneously displayed as text on the central control room monitor.

[0049] In this implementation plan, by sending work suspension dispatch instructions to the special work management terminal, on-site verification dispatch instructions to the mobile video verification terminal and intelligent inspection terminal, upgrade control dispatch instructions to the safety production cockpit, and emergency support dispatch instructions to the personnel positioning management terminal and emergency communication terminal, and displaying the sent dispatch instructions on a 3D electronic map according to the center coordinates of the work point, different types of dispatch instructions are distributed to the corresponding execution terminals, thus achieving functional matching between dispatch instructions and execution terminals. Furthermore, by reading the work suspension confirmation timestamp, on-site verification completion timestamp, upgrade control confirmation timestamp, and emergency support arrival timestamp respectively, and calculating the duration of a single dispatch based on the premise that the confirmation timestamp is later than the sending timestamp, the minimum value among all single dispatch durations is used as the dispatch response duration value, and the maximum value is used as the dispatch closed-loop duration value, so that the safety operation status dispatch result simultaneously includes quantitative indicators of two dimensions: instruction reception response speed and instruction complete execution cycle.

[0050] like Figure 2 As shown, the second aspect of this invention provides an AI-based safety operation status scheduling system for LNG receiving terminals, comprising: a data acquisition and processing module, a job disconnection generation module, a coupled status discrimination module, and a token scheduling output module. The data acquisition and processing module is used to construct a three-dimensional coordinate system for the terminal area, acquire job verification data, perform time alignment and coordinate transformation on the job verification data, and output preprocessed job verification data. The job disconnection generation module is used to establish job status based on the preprocessed job verification data, with rows for personnel crossing boundaries, gas failure, adjacent detector alarms, video evidence failure, hidden danger hazard, and emergency gaps, and columns for continuous verification periods. The system generates an operational status evidence field by establishing a chain break matrix and determining the condition conflict assessment value based on the number of records and corresponding weights within each node unit. A coupled status discrimination module is used to mark control areas and hazards based on the operational status evidence field, and constructs a coupled bipartite graph based on dominant chain break records and process alarm records. After updating node activation values ​​through weighted message passing, strongly coupled evidence pairs are marked, generating an operational status scheduling unit. A token scheduling output module is used to generate corresponding scheduling tokens based on the operational status scheduling unit, and generate operational scheduling instructions in conjunction with the condition conflict assessment value. These instructions are then sent to the corresponding management terminal, outputting the safe operational status scheduling result.

[0051] In this implementation plan, the data acquisition and processing module outputs preprocessed operation verification data, the operation chain break generation module outputs an operation status evidence field consisting of an operation chain break matrix, condition conflict assessment value, and dominant chain break type, the coupled status discrimination module outputs an operation status scheduling unit containing the start timestamp of the status coupling period, operation task number, operation point center coordinates, condition conflict assessment value, and coupled chain break type, and finally the token scheduling output module outputs operation pause scheduling instructions, on-site review scheduling instructions, upgraded control scheduling instructions, emergency support scheduling instructions, and scheduling response duration and scheduling closed loop duration values. This process transforms the original operation verification data into an executable safe operation status scheduling result after passing through three progressive processing stages: chain break quantification, coupling discrimination, and token scheduling. This achieves a complete closed loop from data acquisition to scheduling instruction generation.

[0052] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0053] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. An AI-based safety operation status scheduling method for LNG receiving terminals, characterized in that, Includes the following steps: S1, construct a three-dimensional coordinate system for the station area, collect operation verification data, perform time alignment and coordinate transformation on the operation verification data, and output the preprocessed operation verification data; S2, based on the preprocessed operation verification data, establish an operation chain break matrix with personnel crossing the boundary, gas failure, probe neighbor alarm, video evidence failure, hidden danger pull, and emergency gap as rows and continuous verification period as columns, and determine the condition conflict evaluation value based on the number of records in the node unit and the corresponding weight to generate the operation situation evidence field. The specific steps for generating the operational status evidence field by determining the conditional conflict evaluation value based on the number of records within a node unit and its corresponding weights are as follows: The intersection of each matrix row and each matrix column is used as a node element; During each consecutive verification period, records of personnel whose location coordinates are outside the work permit plane area are written into the node unit corresponding to the personnel boundary crossing node; Records where the difference between the gas detection timestamp and the current continuous verification period's end timestamp is greater than the effective time window value, the distance from the gas sampling point's coordinates to the work point's center coordinates is greater than the spatial coverage radius, or any of the following values ​​exceed the corresponding gas safety threshold range: combustible gas concentration, oxygen volume fraction, hydrogen sulfide concentration, and carbon monoxide concentration, are written to the node unit corresponding to the gas failure node. Records where the distance from the fixed probe's coordinates to the work point's center coordinates is less than or equal to the spatial coverage radius and the fixed probe's alarm duration is greater than zero are written to the node unit corresponding to the probe's adjacent alarm node. Records where the video offline duration is greater than zero are written to the node unit corresponding to the video verification node. Records where the current number of unclosed loop hazards is greater than zero or the number of inspection anomalies is greater than zero are written to the node unit corresponding to the hazard traction node. Records with an emergency resource availability value of zero or an emergency channel occupancy status value of "occupancy" are written to the node unit corresponding to the emergency gap node; when multiple records are written to the same node unit, only the record with the data collection timestamp closest to the end of the current continuous verification period is retained; The number of records retained in each node cell within each matrix row is counted according to the continuous verification period, and the number of records is multiplied by the verification weight value of the corresponding matrix row to obtain the row conflict contribution value of each matrix row. Add up the row conflict contribution values ​​of all matrix rows to obtain the condition conflict assessment value corresponding to the current continuous verification period; The type of broken chain node corresponding to the matrix row with the largest row conflict contribution value is determined as the dominant broken chain type; The operational break-chain matrix, condition conflict assessment value, and dominant break-chain type are combined to generate an operational situation evidence field. S3, based on the operational status evidence field, marks the control area and hazard source, and constructs a coupled bipartite graph based on the dominant chain break record and process alarm record. After updating the node activation value through weighted message transmission, it marks the strongly coupled evidence pair and generates the operational status scheduling unit. The specific steps for constructing a coupled bipartite graph based on the dominant chain break record and process alarm record are as follows: Extract node unit records corresponding to the dominant chain break type from the operation chain break matrix and designate them as master records. Extract the spatial coordinates corresponding to the master records from the preprocessed operation verification data and extract process alarm records with process alarm duration values ​​greater than zero within the current continuous verification period. Construct a coupled bipartite graph, classifying personnel boundary crossing nodes, gas failure nodes, probe adjacent alarm nodes, video evidence failure nodes, hidden danger traction nodes, and emergency gap nodes as first-class nodes, and adjacent control areas and adjacent hazard sources as second-class nodes. When the spatial coordinates associated with the master record are located in the area polygon represented by a second-class node, establish a directed edge from the first-class node corresponding to the master record to the second-class node, with the edge weight taken as the row conflict contribution value corresponding to the first-class node. When the alarm point coordinates of the process alarm record fall into the area polygon represented by the second-class node under the XY axis plane projection, establish a directed edge from the second-class node to the first-class node corresponding to the master record, with the edge weight taken as the normalized value obtained by dividing the process alarm duration value by the upper limit of the alarm duration. S4 generates a corresponding scheduling token based on the operational status scheduling unit, and generates a job scheduling instruction by combining the condition conflict evaluation value. The job scheduling instruction is sent to the corresponding management terminal, and the safe operational status scheduling result is output.

2. The AI-based safety operation status scheduling method for LNG receiving terminals according to claim 1, characterized in that: The specific steps for constructing a three-dimensional coordinate system for the station area, collecting operation verification data, and performing time alignment and coordinate transformation on the operation verification data are as follows: Using the southwest corner of the LNG receiving terminal electronic map as the origin of the coordinate system, the direction from west to east along the east-west road of the terminal is set as the positive X-axis, the direction from south to north along the north-south road of the terminal is set as the positive Y-axis, and the vertical upward direction is set as the positive Z-axis, thus constructing a three-dimensional coordinate system for the terminal area. Collect operation verification data, which includes data collection timestamp, operation task number, operation type code value, operation level code value, operation plan start timestamp, operation plan end timestamp, operation point center coordinate value, operation electronic fence vertex coordinate value, personnel positioning coordinate value, personnel positioning timestamp, combustible gas concentration value, oxygen volume fraction value, hydrogen sulfide concentration value, carbon monoxide concentration value, gas detection timestamp, gas sampling point coordinate value, fixed probe coordinate value, fixed probe alarm duration value, process alarm duration value, video offline duration value, control area boundary coordinate value, hazard source boundary coordinate value, current number of unclosed hidden dangers, number of inspection abnormalities, number of available emergency resources, and emergency passage occupancy status value. For the collected operation verification data, the nearest neighbor timestamp matching algorithm is used to perform cross-source record alignment processing on the operation verification data; the spatial coordinate transformation algorithm based on rigid coordinate transformation is used to perform station area coordinate unification processing on the operation verification data, so that the point data from different sources are transformed into a unified station area three-dimensional coordinate system, and the preprocessed operation verification data is output.

3. The AI-based safety operation status scheduling method for LNG receiving terminals according to claim 2, characterized in that: The specific steps for establishing an operation chain break matrix based on preprocessed operation verification data, with personnel crossing boundaries, gas failure, adjacent alarms from detectors, video evidence failure, hidden dangers leading to problems, and emergency gaps as rows and continuous verification periods as columns, are as follows: Read the preprocessed job verification data, extract the job plan start timestamp, job plan end timestamp, job point center coordinates, and job electronic fence vertex coordinates according to the job task number, and determine the time interval between the job plan start timestamp and the job plan end timestamp as the job permission time period; connect the job electronic fence vertex coordinates corresponding to the same job task number in the closed connection order recorded in the electronic map, and determine the closed boundary and internal area as the job permission plane area; Based on the operation type code value and operation level code value, the verification period length value, time effective window value, spatial coverage radius value and verification weight value are read from the operation verification rule configuration table; the operation permit time period is divided into M consecutive verification periods according to the verification period length value; and an operation chain break matrix is ​​established with personnel boundary crossing node, gas failure node, probe alarm node, video evidence failure node, hidden danger traction node and emergency gap node as matrix rows and consecutive verification periods as matrix columns.

4. The AI-based safety operation status scheduling method for LNG receiving terminals according to claim 3, characterized in that: The specific steps for marking control areas and hazard sources based on operational situation evidence are as follows: Read the operational situation evidence field, extract the coordinates of the center of the operational point, the boundary coordinates of the controlled area, and the boundary coordinates of the hazard source. Calculate the shortest distance from the center of the operational point to the boundary of each controlled area under the XY-axis plane projection, and record it as the regional shortest distance. Calculate the shortest distance from the center of the operational point to the boundary of each major hazard source under the XY-axis plane projection, and record it as the hazard shortest distance. Mark the controlled areas whose regional shortest distance is not greater than the control distance threshold as adjacent controlled areas. Mark the hazard sources whose hazard shortest distance is not greater than the hazard distance threshold as adjacent hazard sources.

5. The AI-based safety operation status scheduling method for LNG receiving terminals according to claim 4, characterized in that: The specific steps for generating the operational status scheduling unit by updating the node activation value through weighted message passing, marking strongly coupled evidence pairs, and then generating the operational status scheduling unit are as follows: The condition conflict assessment value is used as the initial activation value of the first type of node, and the average value of the normalized duration of the process alarm associated with each second type of node is used as the initial activation value of the second type of node. If the second type of node is not associated with any process alarm record, the initial activation value is zero. A graph message propagation algorithm is used to perform N rounds of message passing on the coupled bipartite graph. In each round of propagation, each node sends its current activation value to its neighboring nodes after weighting it according to the weight of the outgoing edges. Each node sums all the incoming edge messages it receives and adds them to its current activation value to obtain the updated activation value. The final activation value of each node after N rounds of propagation is determined as the coupling activation degree. For a node pair consisting of a dominant broken-chain node and each second-type node, if there is at least one directed edge between them, multiply the coupling activation of the first-type node and the coupling activation of the second-type node to obtain the product value, add the weights of all edges between the node pair to obtain the weight sum, and divide the product value by the weight sum to obtain the coupling strength value; mark the node pair with the coupling strength value greater than the coupling strength threshold as a strong coupling evidence pair. Extract the broken-chain node type corresponding to the first type of node from the strongly coupled evidence pair as the coupled broken-chain type; The continuous verification period containing strongly coupled evidence pairs is marked as the situational coupling period. Write the start timestamp of the situation coupling period, the task number, the center coordinates of the task point, the condition conflict assessment value, and the coupling break type into the operation situation scheduling unit.

6. The AI-based safety operation status scheduling method for LNG receiving terminals according to claim 5, characterized in that: The specific steps for generating the corresponding scheduling token based on the operational status scheduling unit and generating the job scheduling instruction in combination with the condition conflict evaluation value are as follows: Read the operation status scheduling unit and the pre-processed operation verification data. When the coupling chain break type is a personnel outbound node, generate a personnel deviation token; when the coupling chain break type is a gas failure node, generate a gas failure token. When the coupling break type is a probe neighbor alarm node, a probe neighbor alarm token is generated; when the coupling break type is a video evidence break node, a video evidence break token is generated. When the coupling chain break type is a hidden danger traction node, a hidden danger traction token is generated; when the duration of the process alarm is greater than zero, a process disturbance token is generated. When the available quantity of emergency resources is zero or the emergency channel is occupied, a support gap token is generated. The generated tokens are used to trigger the corresponding job scheduling instructions; If the condition conflict assessment value is greater than the confidence threshold and the coupling chain break type is a personnel boundary crossing node or a gas failure node, a work suspension scheduling instruction is generated; if the condition conflict assessment value is greater than the confidence threshold and the coupling chain break type is a probe neighbor alarm node or a video evidence failure node, a field review scheduling instruction is generated; if a process disturbance token has been generated and the coupling chain break type is a hidden danger traction node, an upgraded control scheduling instruction is generated. If a support gap token has been generated, an emergency support dispatch instruction will be generated.

7. The AI-based safety operation status scheduling method for LNG receiving terminals according to claim 6, characterized in that: The specific steps for sending the job scheduling instruction to the corresponding management terminal and outputting the safety operation status scheduling result are as follows: The system sends work stoppage dispatch instructions to the special work management terminal; sends on-site verification dispatch instructions to the mobile video verification terminal and intelligent inspection terminal; sends upgraded control dispatch instructions to the safety production cockpit; sends emergency support dispatch instructions to the personnel positioning management terminal and emergency communication terminal; and displays the sent dispatch instructions on the 3D electronic map according to the center coordinates of the work point, and records the sending timestamp of each dispatch instruction. Read the job pause confirmation timestamp from the special job management terminal; Read the on-site verification completion timestamps from the mobile video verification terminal and the intelligent inspection terminal, and take the earliest completion timestamp as the on-site verification confirmation timestamp; Read the upgrade control confirmation timestamp from the safety production cockpit; Read the emergency support arrival timestamps from the personnel location management terminal and the emergency communication terminal, and take the earliest arrival timestamp as the emergency support confirmation timestamp; For each confirmation timestamp, the difference is calculated only if the confirmation timestamp is later than the sending timestamp. The duration of a single scheduling session is obtained by subtracting the sending timestamp from the confirmation timestamp. Take the minimum value among all single scheduling durations as the scheduling response duration value, and take the maximum value as the scheduling closed-loop duration value, and output the safe operation status scheduling result.

8. An AI-based safety operation status quo scheduling system for LNG receiving terminals, employing the AI-based safety operation status quo scheduling method for LNG receiving terminals as described in any one of claims 1-7, characterized in that, include: The module comprises a data acquisition and processing module, a job disconnection generation module, a coupling situation determination module, and a token scheduling output module, among which: The data acquisition and processing module is used to construct a three-dimensional coordinate system for the station area, acquire operation verification data, perform time alignment and coordinate transformation on the operation verification data, and output preprocessed operation verification data. The operation chain break generation module is used to establish an operation chain break matrix based on preprocessed operation verification data, with personnel crossing the boundary, gas failure, probe neighbor alarm, video evidence failure, hidden danger pull, and emergency gap as rows and continuous verification period as columns, and to determine the condition conflict evaluation value based on the number of records in the node unit and the corresponding weight, and generate an operation situation evidence field. The coupled situation discrimination module is used to mark the control area and hazard source based on the operational situation evidence field, and to construct a coupled bipartite graph based on the dominant chain break record and process alarm record. After updating the node activation value through weighted message transmission, it marks the strongly coupled evidence pair and generates the operational situation scheduling unit. The token scheduling output module is used to generate a corresponding scheduling token based on the operation status scheduling unit, and generate a job scheduling instruction in combination with the condition conflict evaluation value. The job scheduling instruction is then sent to the corresponding management terminal, and the safe operation status scheduling result is output.

Images