A gas station perimeter security defense linkage processing method and system

Abstract

A kind of gas station perimeter security defense linkage processing method and system belong to intelligent security and industrial safety production technical field, solve the problem of high false alarm rate of gas station security system, cannot effectively distinguish legal operating personnel and illegal intruder and emergency response lag.This method includes: by constructing the time-space operation filtering mechanism based on work order data: real-time extraction production operation system in the time-space characteristics of operation, with the logic operation of captured invasion target coordinates, filter out the alarm signal caused by legal operation.Combined with the threat level of regional dangerous attribute and visual behavior characteristics, when judged as high-risk threat, automatically link the process safety system through bottom hardware mapping to execute defense action.Security system and production control system are logically integrated for active defense of gas station.The present application is suitable for natural gas gate station, LNG storage and distribution station, pressure regulating station and oil depot, etc., which have very high requirements for fire and explosion prevention and personnel control.

Description

Technical Field

[0001] This application relates to the fields of intelligent security and industrial safety production technology, and in particular to the safety defense of gas stations. Background Technology

[0002] Gas stations, as critical nodes in urban energy supply, have perimeter security that directly impacts public safety and the stability of energy supply. Currently, perimeter intrusion prevention technologies for such high-risk critical infrastructure still face the following significant technical bottlenecks in practical applications:

[0003] Current mainstream solutions largely rely on physical sensor technologies such as infrared beam detectors, tension fences, or distributed fiber optic vibration detection. However, these technologies are typically based on a single signal threshold triggering mechanism, making them poorly adaptable to complex outdoor environments. In the typical open or suburban environments of gas stations, severe weather conditions (such as strong winds, heavy rain, and dense fog) can easily trigger sensor malfunctions. Furthermore, unintentional contact by non-target organisms (such as birds or small mammals) or the natural swaying of surrounding vegetation are often incorrectly identified as intrusion behavior by the system. The massive amount of false alarm data not only reduces the system's reliability but also leads to "alarm desensitization" among monitoring personnel, severely weakening the system's effectiveness in warning of real intrusion events.

[0004] Furthermore, existing technologies typically lack the logical ability to recognize the identity attributes and behavioral intentions of targets. Due to data silos between security systems and production operation and maintenance management systems, traditional security systems cannot verify whether personnel within a protected area possess valid spatiotemporal work permissions. For routine operations such as inspections, equipment maintenance, or infrastructure upkeep performed with legitimate work orders, traditional systems cannot decouple them from unauthorized intrusion activities. This logical flaw leads to the system triggering an alarm whenever someone enters the protected area, resulting in a long-term state of disorder in security management and making it difficult to support the refined management needs of the site.

[0005] On the other hand, existing security video surveillance systems and core process safety systems (SIS) typically employ an independent deployment architecture, lacking a coordinated response mechanism. When the security system detects an intrusion threat in high-risk core areas (such as tank areas and valve chambers), it can only execute local audible and visual deterrence, failing to directly trigger proactive defense mechanisms on the production side. After confirming the alarm, monitoring personnel must manually intervene across systems to execute emergency shutdowns or emergency responses, a process characterized by significant manual operation delays. Considering the extremely rapid evolution of secondary disasters caused by gas leaks, this delayed manual response mode results in a severe mismatch between the safety defense window and the speed of accident evolution, making it difficult to prevent the spread of risk from the physical source in the first instance.

[0006] Therefore, there is an urgent need to develop an intelligent defense system that can combine the operation and maintenance status of the site, automatically filter legitimate behaviors, and achieve deep linkage with the process system based on the threat level. Summary of the Invention

[0007] The purpose of this invention is to solve the technical problems of existing gas station security systems, such as high false alarm rate due to physical environmental interference, inability to effectively distinguish between legitimate workers and illegal intruders, and delayed emergency response caused by the independence of security systems and production process systems. This invention provides a gas station perimeter security defense linkage processing method and system.

[0008] The technical solution adopted by the present invention to solve the above problems is: a method for coordinated handling of perimeter security defense in gas stations, the method comprising:

[0009] Step 1: Receive the abnormal trigger signal from the physical sensing unit and the trigger location information from the visual acquisition unit, and call the visual acquisition unit that has a spatial mapping relationship with the trigger location information to acquire the on-site video image.

[0010] Step 2: Extract the spatiotemporal coordinate information of the triggering target, retrieve the work order data from the gas station production and operation management system through the data interface, and construct spatiotemporal operation filtering conditions that include the dimensions of operation time and operation area;

[0011] The spatiotemporal coordinates of the triggering target are logically matched with the spatiotemporal operation filtering conditions. If the matching result is true, the current abnormal triggering signal is determined to be a legitimate operation disturbance signal, and a first preset strategy including alarm suppression or marking and archiving is executed. If the matching result is not true, it is determined to be an illegal intrusion and step 3 is executed.

[0012] Step 3: Based on the preset electronic map of the gas station, identify the regional attributes and target category of the illegal intrusion target, and determine the intrusion threat level;

[0013] Step 4: Execute differentiated response strategies based on the intrusion threat level; wherein, when the intrusion threat level reaches a preset high-risk threshold, automatically send a linkage command to the gas station's process safety system or programmable logic controller to execute process-side defense actions.

[0014] Furthermore, before step 1, there are also steps of system initialization and spatial calibration, specifically: before the system runs, a spatial mapping relationship between the physical sensing unit and the visual acquisition unit is established in advance, the physical location identifier is recorded by simulated trigger test along the perimeter of the station, and the PTZ parameters of the visual acquisition unit in the corresponding area are calibrated to construct a spatial mapping table of physical location and visual viewpoint.

[0015] Furthermore, in step 2, the spatiotemporal operation filtering conditions include the operation time period and the operation area range;

[0016] The specific rules for the logical matching operation are as follows: query the database of the production operation and maintenance management system to check whether there is a work order record that meets the following conditions: the planned operation time period of the work order includes the time parameter in the spatiotemporal coordinate information; and the operation area of ​​the work order includes the spatial coordinate parameter in the spatiotemporal coordinate information; if the search result is that it exists, the matching result is true; if the search result is that it does not exist, the matching result is no.

[0017] Furthermore, in step 2, the logical matching operation method specifically includes:

[0018] Step 2.1: Extract the real-time timestamp and physical location coordinates of the current physical sensing unit that triggered the alarm; call the preset spatial mapping table to convert the physical location coordinates into logical operation area labels in the electronic map;

[0019] Step 2.2: The central control processing server initiates a query to the production operation and maintenance management system through the multi-protocol industrial communication gateway. First, it filters out the set of active work orders in the system whose status is "in execution". If the set is empty, then proceed to step 6.

[0020] Step 2.3: Traverse the active work order set and extract the planned start time and planned end time for each work order; determine whether the work order meets the following condition: planned start time ≤ real-time timestamp ≤ planned end time. If it meets the condition, keep the work order and proceed to step 2.4; if it does not meet the condition, remove the work order.

[0021] Step 2.4: Determine whether the currently triggered logical job area label is included within the job area;

[0022] Step 2.5: If, after comparing time and space, there is still at least one work order record that matches completely, the logical matching operation result is output as true, and the current trigger is determined to be a legitimate work disturbance. The central control processing server then extracts the work order number of the work order, writes it as metadata into the system log and the tag of the current silent recording, and terminates the audible and visual alarm process; otherwise, proceed to step 2.6.

[0023] Step 2.6: If the active work order set is empty, or if no work order can pass the comparison in both Step 2.3 and Step 2.4, the logical matching operation result will be output as "no". The system will determine that the current trigger is an unauthorized illegal intrusion and then execute Step 3.

[0024] Furthermore, step 3 specifically includes:

[0025] Step 3.1: Preprocess the on-site video images obtained in Step 1, including dehazing and noise reduction;

[0026] Step 3.2: Input the preprocessed on-site video images into the pre-trained YOLOv11 deep learning model, and output the target category and confidence score;

[0027] Step 3.3: Determine the intrusion threat level of the illegal intrusion target based on the regional attributes, target category, and confidence level of the illegal intrusion target.

[0028] Secondly, the present invention provides a gas station perimeter security defense linkage processing system, the system comprising: a front-end sensing layer, an intermediate control layer and an execution linkage layer;

[0029] The front-end perception layer includes physical perception units and visual acquisition units deployed at the perimeter of the site, used to acquire abnormal trigger signals and on-site video images in real time.

[0030] The central control processing server is configured to extract the spatiotemporal coordinate information of the triggering target, retrieve work order data from the gas station production and operation management system through a data interface, and construct spatiotemporal operation filtering conditions that include operation time and operation area dimensions; perform logical matching operations between the spatiotemporal coordinate information of the triggering target and the spatiotemporal operation filtering conditions; if the matching result is true, the current abnormal triggering signal is determined to be a legitimate operation disturbance signal, and a first preset strategy including alarm suppression or marking and archiving is executed; if the matching result is not true, based on a preset electronic map of the gas station, the regional attributes and target category of the illegal intrusion target are identified, and the intrusion threat level is determined.

[0031] The execution linkage layer is configured to execute a multi-level response strategy based on the intrusion threat level; when the threat level reaches a preset high-risk threshold, a linkage command is sent to the process safety system of the gas station through an industrial communication protocol to execute defensive actions on the process side.

[0032] Furthermore, when the system is in normal operation, the visual acquisition unit is in a low-power preset position polling state;

[0033] When the physical sensing unit detects that the amplitude of the vibration signal exceeds the preset environmental noise threshold, it immediately generates an abnormal trigger signal and, based on the spatial mapping of physical location and visual perspective, prompts the visual acquisition unit in the corresponding area to focus and acquire images.

[0034] Furthermore, the front-end sensing layer consists of a fiber optic signal demodulator and an explosion-proof PTZ camera, which aggregate data through an industrial network switch; the middle control layer includes a central control server equipped with a GPU and a communication gateway that supports bidirectional controlled access; the execution linkage layer is equipped with dual-channel intelligent I / O control modules, which are respectively connected to the explosion-proof supplementary lighting equipment and the non-explosion-proof power supply circuit, and realize the underlying hard linkage with the process safety system SIS through the register address mapping of the Modbus TCP protocol.

[0035] Furthermore, the central control processing server is also configured to, upon receiving an intrusion trigger signal, parse the distance value of the trigger location, and by consulting the spatial mapping table of physical location and visual perspective, send an absolute coordinate positioning command to the corresponding camera, causing the camera to rotate to the target's viewpoint and display the live video image.

[0036] Thirdly, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor runs the computer program stored in the memory, it executes the steps of the gas station perimeter security defense linkage processing method described above.

[0037] The beneficial effects of this invention are:

[0038] This invention introduces a logical verification mechanism between work order data and security signals. By utilizing the matching relationship between work time windows and spatial domains, it can filter out alarm signals caused by legitimate inspections or maintenance operations, thereby reducing false alarms caused by personnel activity. Through real-time data consistency verification between security signals and work order data, the system can automatically identify and filter out personnel activity alarms generated by legitimate operations such as inspections, maintenance, and landscaping, solving the management pain point of gas stations struggling to distinguish between legitimate personnel and unauthorized intrusion targets.

[0039] This invention establishes a communication connection between the security system and the process safety system (SIS). When a high-risk intrusion threat is detected, it can automatically trigger actions such as cutting off non-explosion-proof power or adjusting gas detection thresholds. This solves the problem that traditional security systems cannot directly intervene in production risks and shortens emergency response time.

[0040] Furthermore, this invention employs a triggering mechanism that uses physical sensing signals to wake up the visual acquisition unit, which reduces the system's computational load and data transmission bandwidth usage compared to all-time video analysis. The hierarchical response strategy based on regional attributes ensures flexible handling of minor external interference and rigid blocking of threats to the core area, achieving refined and efficient security management.

[0041] This invention is applicable to critical infrastructure such as natural gas gate stations, LNG storage and distribution stations, pressure regulating stations, and oil depots, which have extremely high requirements for fire prevention, explosion protection, and personnel management. Attached Figure Description

[0042] To more clearly illustrate the technical solution of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 A schematic diagram of the overall architecture of a gas station perimeter security defense linkage system provided for the implementation of the invention;

[0044] Figure 2 Flowchart of the core steps of the gas station perimeter security defense linkage processing method provided in the embodiments of the present invention;

[0045] Figure 3 This is a logical diagram illustrating the threat classification and process system linkage response based on regional attributes in an embodiment of the present invention. Detailed Implementation

[0046] Specific Implementation Method 1: The gas station perimeter security defense linkage processing method described in this implementation method includes the following steps:

[0047] S0: System initialization and spatial calibration steps: Before the system runs, establish the spatial mapping relationship between the physical sensing unit and the visual acquisition unit in advance; record the physical location identifier through simulated trigger test along the perimeter of the station, and calibrate the PTZ parameters of the visual acquisition unit in the corresponding area to construct a mapping table between physical location and visual viewpoint;

[0048] S1: Abnormal signal acquisition steps: Receive the trigger signal sent by the physical sensing unit, analyze the trigger position, and wake up the visual acquisition unit of the corresponding area based on the mapping table to perform focusing and image acquisition;

[0049] S2: Legality verification steps: Extract the spatiotemporal features of the triggering target and compare them with the work order data of the production and operation system in real time. If a matching legal work order exists, the silent recording strategy is executed; otherwise, it is determined to be an illegal intrusion.

[0050] S3: Threat Classification Determination Steps: For targets determined to be illegal intrusions, threat classification is performed based on the danger level attributes of their location and visual behavioral characteristics.

[0051] S4: Multi-level linkage response steps: Based on the threat classification results, the security equipment is linked to carry out audible and visual deterrence, and in the case of high risk, the process safety system (SIS) is linked to execute physical blocking or emergency plan.

[0052] This implementation method only requires video recording when a physical trigger signal, such as vibration, is present, thus avoiding continuous video acquisition and analysis and significantly reducing the system's computational load and data transmission bandwidth usage. Furthermore, by introducing a logical verification mechanism between work order data and security signals, it can filter out alarm signals triggered by legitimate inspections or maintenance work by utilizing the matching relationship between the work time window and spatial domain, thereby reducing false alarms caused by personnel activity. Layered judgment of intrusion signals improves the accuracy of illegal intrusion detection while also reducing the load and accelerating processing speed. It enables deep integration of security data and production operation and maintenance data, providing proactive and hierarchical intelligent defense decision support for gas stations.

[0053] Specific Implementation Method Two: The gas station perimeter security defense linkage method described in this implementation method includes the following steps:

[0054] Abnormal signal acquisition steps: Receive the abnormal trigger signal and trigger location information sent by the front-end perception layer, and call the visual acquisition unit that has a spatial mapping relationship with the trigger location information to acquire the on-site video image;

[0055] Legality verification steps: Extract the spatiotemporal coordinate information of the triggering target; retrieve work order data from the gas station production and operation management system through the data interface; construct spatiotemporal operation filtering conditions including operation time and operation area dimensions; perform logical matching operation between the spatiotemporal coordinate information of the triggering target and the spatiotemporal operation filtering conditions; if the matching result is true, determine that the current physical triggering signal is a legal operation disturbance signal, and execute the first preset strategy including alarm suppression or marking and archiving; if the matching result is not true, determine that it is an illegal intrusion, and execute the second preset strategy to trigger the subsequent threat classification judgment steps.

[0056] Threat classification and determination steps: Based on a preset electronic map of gas stations, identify the regional attributes and target category of the illegal intrusion target, and calculate the intrusion threat level;

[0057] Multi-level linkage response steps: Execute differentiated response strategies according to the intrusion threat level; wherein, when the intrusion threat level reaches a preset high-risk threshold, automatically send linkage instructions to the gas station's process safety system or programmable logic controller to execute process-side defense actions.

[0058] Specifically, before the abnormal signal acquisition step, there is also a system initialization and spatial calibration step: before the system runs, a spatial mapping relationship between the physical sensing unit and the visual acquisition unit is established in advance; the physical location identifier is recorded by simulated trigger test along the perimeter of the station, and the PTZ parameters of the corresponding area visual acquisition unit are calibrated to construct a mapping table between physical location and visual viewpoint.

[0059] Specifically, in the abnormal signal acquisition step, the front-end perception layer includes a physical perception unit and a visual acquisition unit; the acquisition of the on-site video image of the corresponding area specifically includes: the visual acquisition unit is in a low-power standby mode or a preset position polling mode under normal conditions; only when the amplitude of the vibration signal or the duration of the blocking signal detected by the physical perception unit exceeds the preset environmental noise threshold is the abnormal trigger signal generated to wake up the visual acquisition unit to enter the high frame rate recording and focusing state.

[0060] Specifically, in the legality verification step, the first preset strategy includes terminating the alarm process, performing silent recording, or attaching a maintenance label for archiving; the second preset strategy includes proceeding to the threat classification determination step.

[0061] Specifically, in the legality verification step, the spatiotemporal job filtering conditions include the job time period and the job area range; the specific rules of the logical matching operation are as follows: query the database of the production operation and maintenance management system to check whether there is a work order record that meets the following conditions: the planned job time period of the work order includes the time parameter in the spatiotemporal coordinate information; and the job area range of the work order includes the spatial coordinate parameter in the spatiotemporal coordinate information; if the search result is that it exists, the matching result is true; if the search result is that it does not exist, the matching result is no.

[0062] Specifically, in the threat classification and determination step, the area attributes include at least the peripheral monitoring area, the general production area, and the core danger area; the calculation rules for the intrusion threat level include: if the target is located in the peripheral monitoring area, it is determined to be a level three threat; if the target is located in the general production area, or the front-end perception layer detects continuous climbing characteristic signals, it is determined to be a level two threat; if the target is located in the core danger area, or the front-end perception layer detects that the target generates open flames or smoke characteristics, it is determined to be a level one threat, i.e., the high-risk threshold.

[0063] Specifically, in the multi-level linkage response steps, when the intrusion threat level reaches level one threat, the process-side defense action includes at least one of the following: cutting off the power circuit of non-intrinsically safe equipment in the intrusion area; forcibly turning on the explosion-proof lighting equipment in the intrusion area for supplemental lighting; sending a warning signal to the control unit associated with the emergency shut-off valve, the warning signal being used to trigger the control unit to adjust the gas leak detection alarm threshold in the associated area to a preset low concentration sensitivity mode, and to set the emergency shut-off valve logic to a ready action state.

[0064] Implementation Method 3: The gas station perimeter security defense linkage system described in this implementation method includes a front-end sensing layer, an intermediate control layer, and an execution linkage layer;

[0065] The front-end perception layer includes physical perception units and visual acquisition units deployed at the perimeter of the site, used to collect intrusion trigger signals and on-site video streams in real time;

[0066] The central control processing server is configured to acquire the spatiotemporal coordinate information of the triggering target and call the work order data of the gas station production and operation management system for legality verification; if the verification is an illegal intrusion, the threat level is calculated by combining the regional attributes and behavioral characteristics of the target.

[0067] The execution linkage layer is configured to execute a multi-level response strategy based on the threat level; when the threat level reaches a preset high-risk threshold, a linkage command is sent to the process safety system (SIS) of the gas station through an industrial communication protocol to execute defensive actions on the process side.

[0068] To ensure the physical safety and real-time response of the system, this invention adopts a physical cascade-based physical hardware architecture: the front-end sensing layer consists of a fiber optic signal demodulator and an explosion-proof PTZ camera, which aggregate data through an industrial network switch; the middle control layer includes a central control server configured with a GPU and a communication gateway that supports bidirectional controlled access; the execution linkage layer is configured with dual-channel intelligent I / O control modules, which are respectively connected to the explosion-proof supplementary lighting equipment (normally open) and the non-explosion-proof power supply circuit (normally closed), and realize the underlying hard linkage with the process safety system (SIS) through the register address mapping of the Modbus TCP protocol.

[0069] When performing a validity check, the central control processing server executes the following steps:

[0070] Get the timestamp of the alarm triggered by the physical sensing unit and position coordinates ;

[0071] Search the production operations and maintenance management database to check if any work orders meet the following conditions: the work order's operation period covers the timestamp. The work order's work area includes location coordinates. And the work order status is "in progress";

[0072] If the query result is yes, the current trigger is determined to be a legitimate operation, and only silent video recording is performed; if the query result is no, the current trigger is determined to be an illegal intrusion.

[0073] When the execution linkage layer performs defensive actions on the process side, it includes:

[0074] Automatically cut off the power circuit of non-intrinsically safe equipment in the intrusion area to eliminate the risk of electrical sparks;

[0075] The high-brightness explosion-proof lighting equipment in the intruded area was forcibly turned on to provide supplemental lighting.

[0076] Send a warning signal to the programmable logic controller (PLC) to put the emergency shut-off valve in the associated area into a ready-to-operate state.

[0077] The front-end perception layer adopts a "physical trigger wake-up" mechanism: the visual acquisition unit is in a low-power standby or fixed-point polling state under normal conditions; only when the physical perception unit detects a vibration or occlusion signal exceeding the environmental noise threshold will the corresponding visual acquisition unit be woken up to perform focusing, capturing, and feature extraction.

[0078] The central control processing server determines the threat level based on a preset electronic map:

[0079] Level 3 threat: The target is located in the monitoring area outside the station and stays for less than the preset time limit, such as loitering reconnaissance, malicious filming, illegal parking of vehicles, or large animals approaching;

[0080] Level 2 threat: The target is located in a general production area or is detected climbing or damaging the fence, such as throwing foreign objects, carrying breaching tools, drone intrusion, tailing, or obstructing cameras;

[0081] Level 1 Threat: The target is located in the core danger zone (LNG storage tank area, pressure regulating and metering area) or open flame or smoke characteristics are detected, such as the use of non-explosion-proof equipment (mobile phone), vehicle collision, personnel falling to the ground (suspected poisoning), liquid leakage, or thermal anomalies.

[0082] This implementation constructs a spatiotemporal operation filtering mechanism based on work order data: it extracts the spatiotemporal characteristics of operations in the production and maintenance system in real time, performs logical operations with the coordinates of intrusion targets captured by the front-end perception layer, and thus filters out alarm signals caused by legitimate operations. Furthermore, the system calculates the threat level by combining regional hazard attributes and visual behavioral characteristics. When a high-risk threat is identified, it automatically triggers the process safety system through underlying hardware mapping to execute defensive actions such as cutting off non-explosion-proof power supplies, activating explosion-proof supplementary lighting, or adjusting gas detection thresholds. This achieves logical integration between the security system and the production control system, enhancing the proactive defense level of the gas station.

[0083] Example 1: A method for coordinated perimeter security defense at gas stations. The system architecture corresponding to this method is as follows: Figure 1-3 As shown, the gas station perimeter security defense linkage system in this embodiment adopts a layered physical hardware architecture, and the specific configuration of each layer is as follows:

[0084] 1. Front-end Sensing Layer: This layer mainly consists of a physical intrusion detection component (in this embodiment, a distributed vibration sensing component) and a visual acquisition component. The distributed vibration sensing component includes distributed vibration sensing optical cables laid along the perimeter fence of the site, with their ends connected to an optical fiber signal demodulator. This demodulator integrates a hardware bandpass filter circuit (preferably with a frequency band of 50Hz-500Hz) to filter out low-frequency environmental noise and is equipped with a physical alarm output port. The visual acquisition component uses explosion-proof PTZ cameras deployed in critical areas such as tank areas and valve chambers. These cameras meet Ex d IIC T4Gb or higher explosion-proof standards, and the PTZ motors are capable of responding to high-speed rotation commands.

[0085] 2. Middle Control Layer: This layer comprises a central control processing server and a multi-protocol industrial communication gateway. The central control processing server utilizes industrial-grade equipment equipped with a high-performance graphics processing unit (GPU) to run real-time inference models and pre-stores 3D or 2D vector maps of the site containing area hazard attribute labels (such as "core hazard area" and "general area"). The multi-protocol industrial communication gateway is configured with bidirectional isolated physical network interfaces, connecting the security network and the production control network respectively to physically block broadcast storms and supports the conversion between protocols such as OPC UA and Modbus TCP.

[0086] 3. Execution Linkage Layer: This layer is crucial for achieving physical blocking and includes the hard-linkage interface between the intelligent I / O control module and the process safety system (SIS). The output of the intelligent I / O control module uses a relay dry contact method, directly connected in series to the power supply circuit of the explosion-proof lighting equipment on site, to cut off the power in response to central control commands. Simultaneously, the multi-protocol industrial communication gateway is physically connected to the PLC controller of the SIS system via a shielded cable, using a dedicated data interaction address reserved within the PLC (specifically, a Modbus protocol holding register in this embodiment, such as address 40001) to receive warning status codes.

[0087] Based on the system of this embodiment, a gas station perimeter security defense linkage processing method specifically includes:

[0088] Step S0: System initialization and spatial calibration

[0089] Physical trigger test: Perform artificial intrusion triggering at preset intervals (e.g., 10 meters) along the site's physical fence, and record the physical location identifier output by the physical sensing unit (e.g., fiber optic distance calibration value "245 meters" or zone ID "Zone-03").

[0090] Visual parameter calibration: For each physical trigger point, the corresponding visual acquisition unit (PTZ camera) is manually operated to align its field of view center with the trigger position, and the zoom is adjusted to obtain a clear image. The system records the PTZ parameters of the camera at this time (horizontal azimuth angle Pan, pitch angle Tilt, zoom).

[0091] Building a mapping library: The central control processing server associates the above data and generates and stores a key-value pair mapping table (look-up table) in the database, which consists of "physical location identifier - camera ID - PTZ parameters". This mapping table serves as the data foundation for achieving precise wake-up in the subsequent step S1.

[0092] Step S1: Abnormal Signal Acquisition Step

[0093] When the system is running normally, the visual acquisition unit is in a low-power preset position polling state.

[0094] When the physical sensing unit detects that the amplitude of the vibration signal exceeds the preset environmental noise threshold (e.g., signal-to-noise ratio > 15dB after excluding wind and rain interference), it immediately generates a trigger signal.

[0095] After receiving the trigger signal, the central control server parses the distance value of the trigger position and, by consulting the mapping table established in step S0, directly sends an absolute coordinate positioning command to the camera with the corresponding ID. Within 1 second, the camera rotates to the target's field of view via a servo motor and starts high-frame-rate (e.g., 60fps) recording and snapshot capture.

[0096] Step S2: Spatiotemporal coordinate mapping and validity verification

[0097] To address the false alarm issue, the central control processing server executes the following dual verification logic:

[0098] Spatial semantic transformation: Map the physical coordinates (GPS or relative coordinates) of the trigger location to the logical functional area in the station's electronic map (e.g., transform "fiber optic 200-220 meters" into the semantic label "No. 1 pressure regulating valve chamber area").

[0099] Work order database comparison: A structured query (SQL query) is initiated to the production operation and maintenance management system through the OPC UA interface of the multi-protocol industrial communication gateway:

[0100] Work order data logical comparison: The central control processing server initiates a logical query request containing "time" and "space" dimensions to the production operation and maintenance management system through the data interface of the multi-protocol industrial communication gateway. The query conditions are configured as follows: retrieve records whose work area field matches the spatial semantic label of the current trigger location, whose planned work time period covers the current trigger timestamp, and whose work order status is marked as "in execution" or an equivalent active status. Judgment logic: If the system returns a valid work order record index, the current behavior is determined to be consistent with the pre-reported legitimate work, the system automatically executes the alarm suppression strategy, and archives the triggered event after associating it with the work order number; if the query result is empty or returns an abnormal status code, it is determined to be an unauthorized illegal intrusion behavior, and the subsequent threat classification judgment process is triggered.

[0101] Judgment Result: If the query result is not empty, it is judged as a "legitimate operation," and the system automatically suppresses the alarm, only writing the work order number into the metadata of the video stream for archiving; if the query result is empty, it is judged as an "illegal intrusion," and proceed to step S3. In this step, the work order database of the production operation and maintenance management system is configured as a "spatiotemporal signal filter." By converting the work order information (time window, spatial domain) of the business dimension into machine-readable spatiotemporal operation filtering conditions, and performing a logical "AND" operation with the alarm signal output by the physical sensing unit, the system can automatically filter out disturbance signals generated by legitimate operations based on data consistency. This method of verifying real-time security signals based on offline operation and maintenance data effectively reduces the false alarm rate caused by maintenance or inspection operations.

[0102] Table 1. Work Order Data Content and Data Structure

[0103] Field Name Data types Data content and description Its role in legality verification Work order number String A unique global identifier for this task. If the operation is determined to be legitimate, it will be recorded in the metadata of the video stream for archiving. Homework type String Record the specific job categories, such as equipment inspection, routine maintenance, and plant area greening. It helps determine the nature of the task and can be used for more refined feature matching later. Planned start time Timestamp Pre-reported operations are permitted to enter the designated time slot. The lower limit of the time parameter constitutes the filtering conditions for spatiotemporal operations. Planned end time Timestamp The evacuation time must be reported in advance for any work. The upper limit of the time parameter in the spatiotemporal task filtering conditions is that the current trigger timestamp must fall between the start and end timestamps. Work area String It includes spatial semantic labels (such as "No. 1 pressure regulating valve chamber area") or specific two-dimensional / three-dimensional spatial coordinate system boundary sets. The spatial coordinate parameters that constitute the spatiotemporal operation filtering conditions must include the spatiotemporal coordinate information of the triggering target. Work order status Enumeration value Indicates the lifecycle of the current work order, such as 0 - not started, 1 - in progress, 2 - completed, 3 - cancelled. For logical queries, the key condition is that the status identifier must be "in execution" or an equivalent active status to be considered a valid match. Authorized personnel array Record the IDs or signatures of authorized personnel allowed to enter the area. Provides data interface extensions for multi-dimensional identity comparison.

[0104] To achieve accurate false alarm filtering, upon receiving an abnormal trigger signal, the central control processing server, based on the aforementioned work order data structure, executes the following work order logic comparison and judgment sub-process:

[0105] Step S2-1 Feature Extraction and Conversion: Extract the real-time timestamp and physical location coordinates of the current physical sensing unit that triggered the alarm. Use a preset spatial mapping table to convert the physical location coordinates into logical work area labels on the electronic map.

[0106] Step S2-2 Preliminary Data Retrieval: The central control processing server initiates a query to the production operation and maintenance management system through the multi-protocol industrial communication gateway. First, it filters out the set of active work orders in the system whose status field value is "1-Executing". If the set is empty, it skips directly to step S2-6.

[0107] Step S2-3 Time Dimension Comparison: Traverse the active work order set and extract the planned start time and planned end time for each work order. The judgment condition is: planned start time ≤ real-time timestamp ≤ planned end time. If the condition is met, the work order is retained and proceeds to the next level of comparison; otherwise, it is discarded.

[0108] Step S2-4 Spatial Dimension Comparison: For work orders that pass the time comparison, extract their work area field. The judgment condition is: whether the currently triggered logical work area label is included within the scope of the work area.

[0109] Step S2-5 Legality Determination Output: If, after comparing both time and space, there is still at least one completely matching work order record, the logical matching operation result is output as "true," and the system determines that the current trigger is a legitimate work disturbance. The central control server then extracts the work order number, writes it as metadata into the system log and the current silent recording tag, and terminates the audible and visual alarm process.

[0110] Step S2-6 Illegal Intrusion Judgment Output: If the active work order set is empty, or no work order can pass the comparison in both Step S2-3 and Step S2-4, the logical matching operation result output is "No". The system determines that the current trigger is an unauthorized illegal intrusion and then triggers the threat classification judgment process in Step S3.

[0111] Step S3: Threat classification based on regional attributes and visual features

[0112] The system combines spatial attributes with visual AI analysis results to calculate the threat level. :

[0113] Image feature extraction: The system preprocesses the image captured in step S1 (dehazing, noise reduction) and inputs it into the pre-trained YOLOv11 deep learning model. This model is trained on a sample set containing "flames, smoke, people falling to the ground, and climbing actions" and outputs the target category and confidence score.

[0114] The specific classification and response logic is as follows:

[0115] First, when a target is located in the "peripheral monitoring area" and identified as a "pedestrian" by the AI ​​model, and no climbing action is detected, the system classifies it as a Level 3 threat (early warning level). For this type of threat, the system implements a flexible response strategy, only issuing a pop-up notification in the monitoring backend without triggering physical alarms.

[0116] Secondly, when the target is located in a "general production area," or when the AI ​​model detects "climbing" or "vandalism" and the confidence level is higher than a preset threshold (e.g., 85%), the system classifies it as a level two threat (alarm level). At this time, the central control processing server will close the relay through the intelligent I / O control module to physically activate the on-site audible and visual alarms to drive it away.

[0117] Finally, when a target enters the "core danger zone" (i.e., the tank area or valve well area marked in red on the electronic map), or when the AI ​​model detects "flame" or "smoke" features with extremely high confidence (e.g., >90%), the system classifies it as a Level 1 threat (high-risk level). At this point, the system will immediately trigger subsequent multi-level linkage response steps.

[0118] Step S4: Multi-level linkage response execution

[0119] To address Level 1 threats, the system performs deep physical linkage via the industrial bus:

[0120] Intrinsically safe processing: The central control server sends a power-off command to the intelligent I / O control module to physically disconnect the contactor coils of non-explosion-proof lighting and non-essential equipment in the intrusion area, preventing electric sparks from being generated by human sabotage.

[0121] Auxiliary lighting linkage: Force the activation of the independent explosion-proof searchlight in this area to provide the high-illuminance environment required for AI analysis.

[0122] Process system linkage (ESD early warning): An early warning code is written to a specific register address in the SIS system via the Modbus TCP protocol (e.g., write address 40001 with a value of 0xFF). Upon detecting this signal, the logic controller within the SIS system automatically sets the emergency shut-off valve (ESD) in the associated area to a "ready-to-trigger state" and adjusts the alarm threshold of the gas leak detector (e.g., lowering it from 20% LEL to 10% LEL) for rapid response. At this time, the system sends an early warning code to the SIS to put it into a "ready-to-trigger state" and adjusts the alarm threshold of the gas leak detector (e.g., lowering it from 20% LEL to 10% LEL), instead of directly triggering the shut-off action. This dual-modal cascaded triggering mechanism avoids unnecessary station-wide emergency shut-off (ESD) and production stoppage losses caused by a single visual false alarm. Furthermore, because the system is already in a highly sensitive state, once the gas detector confirms a leak, zero-delay physical blocking can be achieved without waiting for concentration accumulation.

[0123] Example 2: A gas station perimeter security defense linkage system. The system adopts a physical cascade-based architecture. The system includes: a front-end sensing layer, consisting of physical intrusion detection components deployed along the perimeter of the gas station and explosion-proof visual acquisition components deployed within the explosion-proof area of ​​the station; the physical intrusion detection components are equipped with alarm signal output ports, and the explosion-proof visual acquisition components are connected to the system through a network interface;

[0124] The intermediate control layer includes a central control processing server and a multi-protocol industrial communication gateway; the central control processing server establishes a communication connection with the front-end sensing layer;

[0125] The execution linkage layer includes an intelligent I / O control module and the original process safety system controller of the gas station. Specifically, the cascaded structure of the system includes: the physical intrusion detection component transmitting a trigger signal to the central control processing server; the intelligent I / O control module is configured with multiple control channels: a first control channel is connected in series to the explosion-proof lighting equipment circuit and is configured to close the circuit to provide supplementary lighting upon receiving an alarm command; a second control channel is connected in series to the non-explosion-proof equipment power supply circuit and is configured to open the circuit to cut off the power supply upon receiving an alarm command; the multi-protocol industrial communication gateway is configured with a secure isolation interface, one end of which is connected to the central control processing server, and the other end is physically connected to the communication port of the process safety system controller; the process safety system controller has a pre-set specific data interaction address, which is configured to receive the warning status code written from the multi-protocol industrial communication gateway and connect to the control circuit of the on-site emergency shut-off valve.

[0126] The physical intrusion detection component includes a distributed vibration sensing optical cable and an optical fiber signal demodulator; the distributed vibration sensing optical cable is connected to the laser transmitting and receiving module of the optical fiber signal demodulator, and the demodulator integrates a hardware bandpass filter circuit for wind and rain noise; the explosion-proof PTZ camera has an explosion-proof rating of Ex d IIC T4 Gb or higher, and the PTZ motor has the ability to rotate at high speed in response to physical wake-up commands.

[0127] The central control processing server is equipped with a high-performance graphics processing unit (GPU), and the memory stores a three-dimensional point cloud map or a two-dimensional vector map of the gas station. The map data includes regional hazard attribute labels corresponding to physical spatial coordinates.

[0128] The multi-protocol industrial communication gateway supports OPC UA, Modbus TCP, or MQTT industrial communication protocols. The physical connection cable between the gateway and the process safety system (SIS) controller uses shielded twisted pair or optical fiber, and an industrial-grade firewall or bidirectional security isolation gateway is configured at the physical interface. The firewall or gateway is configured to only allow whitelist-based work order query requests and status feedback signals to pass through, and to block other unauthorized network access.

Claims

1. A method for coordinated security defense at the perimeter of a gas station, characterized in that, The method includes: Step 1: Receive the abnormal trigger signal from the physical sensing unit and the trigger location information from the visual acquisition unit, and call the visual acquisition unit that has a spatial mapping relationship with the trigger location information to acquire the on-site video image. Step 2: Extract the spatiotemporal coordinate information of the triggering target, retrieve the work order data from the gas station production and operation management system through the data interface, and construct spatiotemporal operation filtering conditions that include the dimensions of operation time and operation area; The spatiotemporal coordinates of the triggering target are logically matched with the spatiotemporal operation filtering conditions. If the matching result is true, the current abnormal triggering signal is determined to be a legitimate operation disturbance signal, and a first preset strategy including alarm suppression or marking and archiving is executed. If the matching result is not true, it is determined to be an illegal intrusion and step 3 is executed. Step 3: Based on the preset electronic map of the gas station, identify the regional attributes and target category of the illegal intrusion target, and determine the intrusion threat level; Step 4: Execute differentiated response strategies based on the intrusion threat level; wherein, when the intrusion threat level reaches a preset high-risk threshold, automatically send a linkage command to the gas station's process safety system or programmable logic controller to execute process-side defense actions.

2. The gas station perimeter security defense linkage method according to claim 1, characterized in that: Before step 1, there are also steps of system initialization and spatial calibration, specifically: before the system runs, a spatial mapping relationship between the physical sensing unit and the visual acquisition unit is established in advance, the physical location identifier is recorded by simulated trigger test along the perimeter of the station, and the PTZ parameters of the visual acquisition unit in the corresponding area are calibrated to construct a spatial mapping table of physical location and visual viewpoint.

3. The gas station perimeter security defense linkage method according to claim 1, characterized in that: In step 2, the spatiotemporal operation filtering conditions include the operation time period and the operation area range; The specific rules for the logical matching operation are as follows: query the database of the production operation and maintenance management system to check whether there is a work order record that meets the following conditions: the planned operation time period of the work order includes the time parameter in the spatiotemporal coordinate information; and the operation area of ​​the work order includes the spatial coordinate parameter in the spatiotemporal coordinate information; if the search result is that it exists, the matching result is true; if the search result is that it does not exist, the matching result is no.

4. The gas station perimeter security defense linkage method according to claim 1, characterized in that: In step 2, the logical matching operation method specifically includes: Step 2.1: Extract the real-time timestamp and physical location coordinates of the current physical sensing unit that triggered the alarm; call the preset spatial mapping table to convert the physical location coordinates into logical operation area labels in the electronic map; Step 2.2: The central control processing server initiates a query to the production operation and maintenance management system through the multi-protocol industrial communication gateway. First, it filters out the set of active work orders in the system whose status is "in execution". If the set is empty, then proceed to step 6. Step 2.3: Traverse the active work order set and extract the planned start time and planned end time for each work order; determine whether the work order meets the following condition: planned start time ≤ real-time timestamp ≤ planned end time. If it meets the condition, keep the work order and proceed to step 2.4; if it does not meet the condition, remove the work order. Step 2.4: Determine whether the currently triggered logical job area label is included within the job area; Step 2.5: If, after comparing time and space, there is still at least one work order record that matches completely, the logical matching operation result is output as true, and the current trigger is determined to be a legitimate work disturbance. The central control processing server then extracts the work order number of the work order, writes it as metadata into the system log and the tag of the current silent recording, and terminates the audible and visual alarm process; otherwise, proceed to step 2.

6. Step 2.6: If the active work order set is empty, or if no work order can pass the comparison in both Step 2.3 and Step 2.4, the logical matching operation result will be output as "no". The system will determine that the current trigger is an unauthorized illegal intrusion and then execute Step 3.

5. The gas station perimeter security defense linkage method according to claim 1, characterized in that: Step 3 specifically includes: Step 3.1: Preprocess the on-site video images obtained in Step 1, including dehazing and noise reduction; Step 3.2: Input the preprocessed on-site video images into the pre-trained YOLOv11 deep learning model, and output the target category and confidence score; Step 3.3: Determine the intrusion threat level of the illegal intrusion target based on the regional attributes, target category, and confidence level of the illegal intrusion target.

6. A gas station perimeter security defense linkage system, characterized in that: The system includes: a front-end perception layer, a middle control layer, and an execution linkage layer; The front-end perception layer includes physical perception units and visual acquisition units deployed at the perimeter of the site, used to acquire abnormal trigger signals and on-site video images in real time. The central control processing server is configured to extract the spatiotemporal coordinate information of the triggering target, retrieve work order data from the gas station production and operation management system through a data interface, and construct spatiotemporal operation filtering conditions that include operation time and operation area dimensions; perform logical matching operations between the spatiotemporal coordinate information of the triggering target and the spatiotemporal operation filtering conditions; if the matching result is true, the current abnormal triggering signal is determined to be a legitimate operation disturbance signal, and a first preset strategy including alarm suppression or marking and archiving is executed; if the matching result is not true, based on a preset electronic map of the gas station, the regional attributes and target category of the illegal intrusion target are identified, and the intrusion threat level is determined. The execution linkage layer is configured to execute a multi-level response strategy based on the intrusion threat level; when the threat level reaches a preset high-risk threshold, a linkage command is sent to the process safety system of the gas station through an industrial communication protocol to execute defensive actions on the process side.

7. A gas station perimeter security defense linkage system according to claim 6, characterized in that: When the system is running normally, the vision acquisition unit is in a low-power preset position polling state; When the physical sensing unit detects that the amplitude of the vibration signal exceeds the preset environmental noise threshold, it immediately generates an abnormal trigger signal and, based on the spatial mapping of physical location and visual perspective, prompts the visual acquisition unit in the corresponding area to focus and acquire images.

8. A gas station perimeter security defense linkage system according to claim 6, characterized in that: The front-end sensing layer consists of a fiber optic signal demodulator and an explosion-proof PTZ camera, which aggregate data through an industrial network switch; the middle control layer includes a central control server equipped with a GPU and a communication gateway that supports bidirectional controlled access; the execution linkage layer is equipped with dual-channel intelligent I / O control modules, which are respectively connected to the explosion-proof supplementary lighting equipment and the non-explosion-proof power supply circuit, and realize the underlying hard linkage with the process safety system SIS through the register address mapping of the Modbus TCP protocol.

9. A gas station perimeter security defense linkage system according to claim 6, characterized in that: The central control processing server is also configured to, upon receiving an intrusion trigger signal, parse the distance value of the trigger location, and by consulting the spatial mapping table between physical location and visual perspective, send an absolute coordinate positioning command to the corresponding camera, which then rotates to the target's viewpoint to display the live video image.

10. A computer device, characterized in that: It includes a memory and a processor, wherein the memory stores a computer program, and when the processor runs the computer program stored in the memory, it performs the steps of the gas station perimeter security defense linkage processing method according to any one of claims 1-5.

Images