Plant area dangerous and harmful gas storage tank area internet of things safety interlocking management and control system

Abstract

This invention discloses an IoT-based safety interlocking control system for hazardous gas storage tank areas in a factory, relating to the field of automation control technology. The invention acquires real-time regional data from each monitoring unit through an IoT sensing layer. Based on this data, it determines whether any monitoring unit exhibits any risk anomalies. Based on the anomaly assessment results, a source tracing thread is executed. The results of this thread locate the coordinates of the leaking unit and output a leakage rate curve. Based on the anomaly assessment results, a prediction thread is executed. Based on the prediction thread's results, the expected concentration time series of each monitoring unit is forward-modeled. Based on the expected concentration time series, leaking unit coordinates, and leakage rate curves of each monitoring unit, a hazardous gas spread risk map is constructed. Then, based on the hazardous gas spread risk map, the interlocking action priorities and expected exceedance times are assigned to the affected monitoring units. A priority-ranked list of interlocking units is generated and executed.

Description

Technical Field

[0001] This invention relates to the field of automation control technology, specifically to an Internet of Things (IoT) safety interlocking control system for hazardous gas storage tank areas in industrial plants. Background Technology

[0002] Currently, safety monitoring in hazardous gas storage tank areas generally relies on fixed point-type gas detectors to construct a single-point monitoring network. These devices can only detect concentrations at localized points, exhibiting inherent drawbacks such as large blind spots, weak continuous spatial sensing capabilities, and delayed leak response. They are ill-suited to the monitoring needs of industrial park tank areas with densely packed multiple tanks and complex operating conditions. Existing leak emergency response systems largely rely on single-point concentration threshold-triggered alarms, representing a passive prevention and control model. They generally lack the ability to reverse-locate leak sources and dynamically predict gas spread trends. Safety interlock controls often employ a crude, uniform triggering model, failing to execute differentiated interlock actions based on the timing and severity of risk spread.

[0003] Meanwhile, existing technologies do not fully consider the impact of micro-meteorological conditions in tank areas, the obstruction of structures, and the accumulation effect of low-lying terrain on gas diffusion. Concentration prediction accuracy is insufficient, and interlocking actions are prone to lag or over-response. This makes it difficult to accurately control risks and curb the spread of accidents in the early stages of leakage, and may also disrupt normal production operations due to false interlocks. Therefore, there is an urgent need for a method for the safe management and control of hazardous gases in tank areas that integrates comprehensive perception, intelligent source tracing, trend prediction, and hierarchical interlocking to comprehensively improve the proactive prevention and control capabilities and the accuracy of handling tank area leakage accidents. To this end, an Internet of Things (IoT) safety interlocking management and control system for hazardous gas storage tank areas in industrial plants is provided. Summary of the Invention

[0004] The purpose of this invention is to provide an Internet of Things (IoT) safety interlock control system for hazardous gas storage tank areas in industrial plants, in order to address the shortcomings of the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: An Internet of Things (IoT) safety interlocking control system for hazardous chemical gas storage tank areas in a factory area includes a factory scene perception module, an anomaly tracing module, an anomaly propagation prediction module, and a hierarchical interlocking control module. The plant area scene perception module is used to deploy an Internet of Things (IoT) perception layer in the tank area and delineate a monitoring unit with each gas tank as the center. The IoT perception layer is used to obtain real-time regional data of each monitoring unit and to determine whether there are any risks or anomalies in each monitoring unit based on the real-time regional data. The anomaly tracing module is used to execute a tracing thread based on the anomaly result of the risk judgment, locate the coordinates of the leaking unit based on the execution result of the tracing thread, and output the leakage rate curve. The spread prediction module is used to execute a prediction thread based on the abnormal results of risk judgment, and to perform forward modeling of the expected concentration time series of each monitoring unit based on the execution results of the prediction thread. The hierarchical interlocking control module is used to construct a hazardous gas spread risk map based on the expected concentration time series of each monitoring unit, the coordinates of the leaking unit, and the leakage rate curve. Then, based on the hazardous gas spread risk map, it assigns the priority of interlocking actions and the expected time of exceeding the limit for the affected monitoring units, generates a priority-ranked list of interlocking units, and executes them.

[0006] Furthermore, the process of deploying an IoT sensing layer in the tank area and defining monitoring units centered on each gas storage tank includes: Fixed sensing nodes are deployed in the hazardous gas storage tank area of ​​the target plant. All fixed sensing nodes are numbered sequentially, and a fixed data collection coverage radius is set for each fixed sensing node. The boundary distance between adjacent monitoring units is also adjusted. The target tank area consists of several vertical or horizontal hazardous gas storage tanks. Each tank corresponds to an independent monitoring unit, and each monitoring unit is associated with a fixed sensing node.

[0007] Furthermore, the process of acquiring real-time regional data through the IoT sensing layer and determining whether there are any risk anomalies in each monitoring unit based on the real-time regional data includes: Three levels of concentration thresholds are set for the types of hazardous gas media stored in each monitoring unit: early warning threshold, interlocking preparation threshold, and emergency action threshold. Set a data scanning cycle. After each data scanning cycle, all fixed sensing nodes continuously collect real-time regional data within their coverage area according to the set data scanning cycle. At the same time, perform a full threshold comparison on the real-time regional data of all monitoring units to generate start instructions for the tracing thread and the prediction thread.

[0008] Furthermore, the execution process of the source tracing thread includes: The tracing thread first retrieves historical area data from all monitoring units based on the event trigger timestamp; A unified timestamp benchmark for historical regional data is established, and the concentration peak point and rising edge start point at the event trigger time are marked as constraints for inverse fitting. Identify the rising edge start point in the historical concentration sequence of all monitoring units, sort them according to the order of the rising edge start points of each monitoring unit, and list the 3 to 5 monitoring units with the earliest rising edge time as first-level candidate leakage units, and then list the subsequent units as second-level and third-level candidate leakage units in ascending order of time difference.

[0009] Furthermore, the process of locating the coordinates of the leaking unit and outputting the leakage rate curve based on the execution results of the source tracing thread includes: The peak concentration value and peak occurrence time are extracted from the concentration sequence of each candidate leakage unit. The peak concentration decay gradient and peak time difference between adjacent monitoring units are obtained. Backward diffusion fitting is performed with the decay law of peak concentration with spatial distance and air mass propagation speed as constraints. The reverse fitting process uses the leakage source coordinates and leakage intensity as the parameters to be solved. The measured concentration time series of each candidate leakage unit is used as the verification value. The parameters to be solved are adjusted iteratively and the residual sum of squares is obtained. When the residual sum of squares is lower than the preset convergence threshold or the number of iterations reaches the upper limit, the calculation is stopped and the optimal fitting parameters are output. The optimal leak source coordinates output by the inverse fitting are matched with the coordinates of the monitoring units in the digital twin space of the tank area. The monitoring unit corresponding to the nearest storage tank is determined as the leak unit. The spatial coordinates of the leak source are output simultaneously, along with the sequence of changes in the leak intensity over time obtained by fitting, i.e., the leak rate curve.

[0010] Furthermore, the execution process of the prediction thread includes: The first monitoring unit identified as having an abnormal risk is used as the seed unit for diffusion forward modeling. The source term initial conditions for diffusion forward modeling are calibrated by combining the leak unit coordinates and leak rate curves output in real time by the source tracing thread. The current moment is taken as the prediction starting point, the prediction duration is set, the prediction time step is consistent with the data scanning cycle, and the digital twin space of the tank area is included as a boundary condition in the forward modeling calculation. Gas retention attributes are set for areas with a height lower than the cofferdam. Then, within each time step, the spatial location and diffusion range of the air mass are updated according to the current wind speed vector. The grid concentration at the center location and within the influence range of each monitoring unit is calculated successively at each time node in the next 30 minutes. The area-weighted average of all grid concentrations within the influence range of each monitoring unit is taken as the expected concentration at the corresponding time of the monitoring unit. Finally, a complete expected concentration time series for each monitoring unit is generated.

[0011] Furthermore, the process of constructing the hazardous gas spread risk map includes: Based on the leakage source location parameters and the expected concentration time series, a concentration scalar distribution field is constructed for the current moment and each future moment. The concentration scalar distribution field is then spatially superimposed with the real-time wind speed vector field to synthesize a dynamically updated hazardous gas spread risk map. The hazardous gas spread risk map consists of two core components: an ordered unit chain and a risk chromatogram. The construction of the ordered unit chain starts with the monitoring unit that first exceeds the interlocking preparatory threshold as the seed point, and proceeds along the wind speed vector streamline. Combined with the expected over-limit time sequence of each monitoring unit, an ordered unit chain impacted by gas spread is formed. Simultaneously, starting from the seed unit, the system expands layer by layer to adjacent monitoring units according to the preset overlap of the influence range. Each expansion layer corresponds to a first-level spread level, and the risk chromatogram assigns values ​​based on the predicted peak concentration of each level unit.

[0012] Furthermore, the process of assigning interlock action priorities and expected exceedance times to affected monitoring units based on the hazardous gas spread risk map, and generating a priority-ranked list of interlock units includes: Based on the risk map of hazardous gas spread, the priority of interlock actions and the expected over-limit time are comprehensively calibrated for all affected monitoring units; According to the interlocking unit list, hierarchical interlocking instructions are issued to the interlocking execution equipment of the corresponding monitoring unit in order of priority. Different risk levels correspond to different combinations of interlocking actions. After the tiered interlocking command is issued, feedback signals from the interlocking execution equipment are collected in real time and the operational status of the equipment is confirmed. If equipment failure is detected, the risk priority of the corresponding monitoring unit is immediately increased, and the tiered interlocking command is regenerated. At the same time, alarm information is pushed to on-site operation and maintenance and safety management personnel. After each data scanning cycle, the hazardous gas spread risk map and interlocking unit list are updated based on the latest measured concentration data and the corrected prediction results. The priority and interlocking actions of each unit are dynamically adjusted until the real-time regional data of all monitoring units continuously fall below the warning threshold, at which point the safety interlocking event is lifted.

[0013] The technical effects and advantages provided by the present invention in the above technical solution are as follows: 1. This invention constructs an overlapping and covered gridded monitoring unit system, combined with multiple types of explosion-proof IoT sensing devices, to achieve full-area monitoring of the tank area without blind spots, eliminating the sensing dead zones of traditional point monitoring. At the same time, through the inverse fitting algorithm of concentration rise time difference and peak decay gradient, the coordinates of the leakage unit are effectively located and a continuous leakage rate curve is output, which improves the accuracy and response efficiency of leakage source location and provides precise source term support for subsequent interlocking treatment. 2. This invention achieves forward modeling prediction of gas diffusion through localized micro-meteorological fields and structural constraints. Combined with the scalar distribution of concentration across the entire region, it generates a risk map of hazardous gas spread. This can accurately determine the priority of interlocking actions and the expected time of exceeding limits for each affected unit, enabling differentiated interlocking control based on grade and time sequence. To a certain extent, this avoids the over-response problem of traditional synchronous and unified interlocking control across the entire region. It prioritizes the efficiency of handling high-risk areas while reducing the interference of interlocking on normal production, thus comprehensively improving the proactive prevention and control level of tank farm leakage accidents. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0015] Figure 1 This is a system block diagram of the present invention.

[0016] Figure 2 This is a flowchart of the source tracing thread described in this invention.

[0017] Figure 3 This is a flowchart of the prediction thread described in this invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0019] Please see Figure 1 As shown, an IoT-based safety interlocking control system for hazardous gas storage tank areas in a factory includes a factory scene perception module, an anomaly tracing module, an anomaly propagation prediction module, and a hierarchical interlocking control module. The plant area scene perception module is used to deploy an Internet of Things (IoT) perception layer in the tank area and delineate a monitoring unit with each gas tank as the center. The IoT perception layer is used to obtain real-time regional data of each monitoring unit, and then the real-time regional data is used to determine whether there are any risks or anomalies in each monitoring unit. The anomaly tracing module is used to execute a tracing thread based on the anomaly result of the risk judgment, locate the coordinates of the leaking unit based on the execution result of the tracing thread, and output the leakage rate curve. The spread prediction module is used to execute a prediction thread based on the abnormal results of risk judgment, and to perform forward modeling of the expected concentration time series of each monitoring unit based on the execution results of the prediction thread. The hierarchical interlocking control module is used to construct a hazardous gas spread risk map based on the expected concentration time series of each monitoring unit, the coordinates of the leaking unit, and the leakage rate curve. Then, based on the hazardous gas spread risk map, it assigns the priority of interlocking actions and the expected time of exceeding the limit for the affected monitoring units, generates a priority-ranked list of interlocking units, and executes them.

[0020] The working principle of the present invention is illustrated below through examples: Deploy n sets of fixed sensing nodes and supporting industrial IoT transmission gateways in the hazardous gas storage tank area of ​​the target plant to build a fully covered IoT sensing layer, where n is the total number of storage tanks, and each fixed sensing node is connected to the plant scene sensing module through the industrial IoT transmission gateway. The fixed sensing node integrates an explosion-proof laser methane telemetry instrument, an electrochemical toxic gas detector, a three-dimensional ultrasonic anemometer, an atmospheric stability meter, and an integrated temperature, humidity, and pressure sensor. The laser methane telemetry instrument has a detection range of 0-100m and a concentration detection accuracy of no less than ±2%FS, used for non-contact, large-area gas concentration telemetry. The electrochemical toxic gas detector has a response time of no more than 30s and a detection range covering 0-100% of immediately hazardous concentrations, and can be matched with corresponding detection probes for different hazardous media such as methane, hydrogen sulfide, and chlorine. The three-dimensional ultrasonic anemometer has a sampling frequency of no less than 1Hz and can output three-dimensional wind speed components and wind direction data. The atmospheric stability meter is used to monitor the near-surface atmospheric stability level of the tank area in real time, providing meteorological parameters for subsequent diffusion calculations. All fixed sensing nodes are sequentially numbered S1, S2, ..., S... n At the same time, a fixed data collection coverage radius is set for each fixed sensing node. The data collection coverage radii of sensing nodes in adjacent spatial locations may partially overlap, with an overlap ratio of no less than 15%, thus eliminating sensing blind spots in the tank area. After the factory scene perception module completes the calibration of the influence range of each monitoring unit, it adjusts the boundary distance between adjacent monitoring units to control the proportion of the overlapping area of ​​the influence range of adjacent monitoring units to the total influence area of ​​a single unit to be between 10% and 20%. The tank area is divided into 1m×1m grid points. Each grid point is checked to see if it is covered by the influence range of at least one monitoring unit. After confirming that there are no blind spots in the entire area, a digital twin space of the tank area is established based on the boundaries, numbers and corresponding tank information of all monitoring units, and the unit space index mapping relationship is marked. The target tank area consists of several vertical or horizontal hazardous gas storage tanks. Each tank corresponds to an independent monitoring unit, and each monitoring unit is associated with a fixed sensing node. The influence range of each monitoring unit is determined by taking the geometric center of the corresponding tank as the center and combining the data collection coverage radius of the fixed sensing node.

[0021] For each type of hazardous gas medium stored in the monitoring unit, three concentration thresholds are set with reference to occupational exposure limits for hazardous factors in the workplace and emergency response standards for hazardous chemical leaks: warning threshold, interlocking preparedness threshold, and emergency action threshold. The warning threshold is set at 50% of the occupational exposure limit of the corresponding medium and is used to trigger daily inspection reminders; the interlock preparation threshold is set at the occupational exposure limit of the corresponding medium and is used to initiate the interlock preparation process and the tracing and prediction threads; the emergency action threshold is set at 30% of the immediate hazardous concentration of the corresponding medium and is used to trigger the highest level of emergency interlock action. The factory area scene perception module is set to have a data scanning cycle, with a data scanning cycle duration of 5 to 10 seconds. During each data scanning cycle, all fixed sensing nodes continuously collect real-time data such as gas concentration, wind speed and direction, atmospheric stability, temperature, humidity, and air pressure within their coverage area according to the set data scanning cycle. The monitoring unit packages this data into real-time regional data and performs a full threshold comparison on the real-time regional data of all monitoring units. If it is determined that the real-time regional concentration of any monitoring unit exceeds the interlocking preparation threshold or emergency action threshold, the event trigger timestamp is immediately locked and a safety interlocking event is triggered. The plant area scene perception module simultaneously sends the start instructions for the source tracing thread and prediction thread to the anomaly tracing module and the anomaly propagation prediction module. Alternatively, if it is determined that the real-time regional concentration only exceeds the warning threshold, a manual inspection event is generated and issued to the corresponding staff for execution. Otherwise, the next data scanning cycle begins directly.

[0022] For further details, please refer to Figure 2 As shown, when the anomaly tracing module receives the tracing thread start instruction from the factory scene perception module, it starts the tracing thread. The tracing thread first retrieves historical regional data of all monitoring units over a period of time T based on the event trigger timestamp. The value of T is set according to the tank area size and medium diffusion characteristics, ranging from 5 minutes to 30 minutes, and the data time granularity is consistent with the data scanning cycle. The retrieved historical regional data includes historical concentration time series of each monitoring unit, wind speed and direction data of each sensing node during the same period, atmospheric stability data, and spatial coordinates and influence range parameters of all monitoring units; A unified timestamp benchmark for historical regional data is established, and the concentration peak point and rising edge start point at the event trigger time are marked as constraints for inverse fitting. Identify the rising edge start point in the historical concentration sequence of all monitoring units, that is, the first moment when the concentration data shows a continuous rise for three consecutive update cycles starting from the stable baseline level, and mark the rising edge trigger time t for each monitoring unit in sequence. i (i is the monitoring unit number, i is a positive integer less than or equal to n); Based on the spatiotemporal transmission patterns of hazardous gas cloud diffusion, the concentration in the monitoring unit where the leak source is located starts to rise first, and the rising edge is triggered earliest; the farther away the unit is from the leak source, the later the gas cloud arrives, and the later the rising edge is triggered. Based on the order of the rising edge trigger time of each monitoring unit, the 3 to 5 monitoring units with the earliest rising edge time are listed as first-level candidate leakage units, and the subsequent units are listed as second-level and third-level candidate leakage units in order of increasing time difference. At the same time, based on the basic characteristics of gas diffusion downwind, candidate leakage units in the downwind direction are excluded, taking into account the wind speed, wind direction, atmospheric stability, temperature, humidity and air pressure data in the minute before the event is triggered. For the selected candidate leakage unit range, the peak concentration value and peak occurrence time in the concentration sequence of each candidate leakage unit are extracted. The peak concentration decay gradient and peak time difference between adjacent monitoring units are calculated. Backward diffusion fitting is performed with the decay law of peak concentration with spatial distance and air mass propagation speed as constraints. The inverse fitting process uses the leak source coordinates and leak intensity as the parameters to be solved. The measured concentration time series of each candidate leak unit is used as the verification value. The parameters to be solved are adjusted iteratively, and the sum of squared residuals between the output concentration and the measured concentration is calculated. During the iteration process, the tank area structure correction coefficient is introduced. For the leeward area blocked by buildings and equipment, the tank area structure correction coefficient is adjusted to 0.7 times that of the open area to correct the gas diffusion path and attenuation rate. When the sum of squared residuals is lower than the preset convergence threshold or the number of iterations reaches the upper limit, the calculation stops and the optimal fitting parameters are output. The output concentration is calculated by using the coordinates and intensity of the leak source to be solved as the source term input, and simultaneously importing the micro-meteorological field parameters (wind speed vector, atmospheric stability level) and hazardous gas physical property parameters (molar mass, air density ratio) for the corresponding time period to establish the boundary and initial conditions for diffusion calculation. Then, the spatial relative distance and azimuth between the leak source and the center of each candidate leak unit are calculated, and the corresponding horizontal and vertical diffusion coefficients are matched according to the atmospheric stability level. For leeward areas with structures blocking the wind, the turbulent diffusion coefficient is lowered according to the preset correction coefficient to correct the gas diffusion attenuation rate. Then, the theoretical concentration value of the candidate leak unit location at a single time step is calculated. Finally, the calculation is recursively performed according to the time step of the data update cycle, covering all historical backtracking periods, to form an output concentration time series sequence that is completely aligned with the time dimension of the measured concentration time series. The optimal leak source coordinates output by the inverse fitting are matched with the coordinates of the monitoring units in the digital twin space of the tank area. The monitoring unit corresponding to the nearest storage tank is determined to be the leak unit. The spatial coordinates of the leak source are output simultaneously, and the sequence of changes in the leakage intensity over time obtained by fitting is also output, i.e. the leakage rate curve. The time resolution of the curve is consistent with the data scanning cycle, covering the entire time period from the start of the leak to the current event trigger time. It can intuitively reflect whether the leak aperture has expanded and whether the leak status is stable. The anomaly tracing module stores the leak unit number, leak source coordinates, leak rate curve, fitting residual, and other results into the interlocking event database and pushes them synchronously to the anomaly propagation prediction module.

[0023] For further details, please refer to Figure 3 As shown, when the anomaly propagation prediction module receives the prediction thread start instruction from the factory scene perception module, the prediction thread is started. After the prediction thread is started, it first retrieves the real-time monitoring data from the three-dimensional ultrasonic anemometer and atmospheric stability meter, and then combines the digital twin space of the tank area to construct a micro-meteorological field covering all monitoring units. The micro-meteorological field is divided into equidistant grids with a grid side length of no more than 5m. Each grid point corresponds to a wind speed vector, where the wind direction is represented by the vector direction and the wind speed is represented by the vector magnitude. To address flow field distortions caused by structures such as storage tanks, fire dikes, and pump rooms within the tank farm, the magnitude and direction of the wind speed vector are adjusted for the wind speed rise zone on the windward side of the buildings and the vortex stagnation zone on the leeward side. The ventilation coefficient is reduced for semi-enclosed areas such as dikes and trenches. The corrected micro-meteorological field must deviate from the measured wind speed data by less than 10%. The micro-meteorological field is updated synchronously with the data scanning cycle; the overall wind speed vector is updated once after each data acquisition cycle to ensure the real-time nature of the prediction calculation. The first monitoring unit identified as having an abnormal risk is used as the seed unit for diffusion forward modeling. Combined with the leak unit coordinates and leak rate curves output in real time by the source tracing thread, the initial conditions of the source term for diffusion forward modeling are calibrated, including the three-dimensional spatial coordinates of the leak source, the leak rate time series, and physical property parameters such as the molar mass, density, and toxicity threshold of the hazardous gas. Using the current moment as the prediction starting point, the prediction duration is set to 0-30 minutes, and the prediction time step is consistent with the data scanning cycle. At the same time, the digital twin space of the tank area is included as a boundary condition in the forward modeling calculation. Gas retention attributes are set for areas with a height lower than the cofferdam, and diffusion barrier boundaries are set for enclosed plants and closed pipe corridors to improve the fit between the forward modeling results and the actual scenario. Furthermore, within each time step, the spatial location and diffusion range of the air mass are updated based on the current wind speed vector, while the diffusion coefficient is adjusted in conjunction with the real-time atmospheric stability level: under unstable atmospheric conditions, the horizontal and vertical diffusion coefficients are increased to accelerate gas dilution; under stable atmospheric conditions, the diffusion coefficient is decreased to prolong gas residence time. The grid concentration at the center of each monitoring unit and within its influence range is calculated sequentially for each time point within the next 30 minutes. The area-weighted average of all grid concentrations within the influence range of each monitoring unit is taken as the expected concentration for that monitoring unit at the corresponding time. Finally, a complete expected concentration time series for each monitoring unit is generated. The time series is 30 minutes long, and the time resolution is consistent with the data scanning cycle. Based on the digital twin space identification of tank farms, low-lying areas within the tank farms are identified, including underground pipe trenches, inner corners of cofferdams, equipment foundation depressions, underground tanks, and other areas prone to the accumulation of heavy hazardous gases. For hazardous gas media with a density greater than air, the actual concentration in these areas will be significantly higher than the forward modeling results of open areas, so accumulation correction is required. Based on the depth, opening area, and ventilation conditions of the low-lying areas, a concentration accumulation correction factor of 1.2 to 2.0 is set. The greater the depth and the worse the ventilation conditions, the higher the correction factor. The forward concentration value at the corresponding location is multiplied by the correction factor to obtain the expected concentration time series of the low-lying accumulation area. All low-lying accumulation areas are listed as key risk points and included in subsequent risk mapping and interlocking priority determination to avoid missed risk assessments due to local gas accumulation.

[0024] Furthermore, the leakage source location parameters output by the hierarchical interlocking control module and the concentration calculation results of the entire area output by the abnormal spread prediction module are integrated to construct the concentration scalar distribution field at the current moment and at each future moment. The concentration scalar distribution field is spatially superimposed with the real-time wind speed vector field to synthesize a dynamically updated hazardous gas spread risk map. The hazardous gas spread risk map consists of two core components: ordered unit chains and risk chromatography. The ordered unit chains are used to characterize the sequential order of gas spread, while the risk chromatography is used to visually characterize the risk level of different regions. The construction of the ordered unit chain begins with the monitoring unit that first exceeds the interlocking preparatory threshold as the seed point. It proceeds along the streamline direction of the wind vector driving force field, combining the expected sequence of exceedance times for each monitoring unit to form an ordered unit chain impacted by gas spread. Simultaneously, starting from the seed unit, it expands layer by layer to adjacent monitoring units with a preset overlap of 10%~20% in the influence range. Each expansion layer corresponds to a first-level spread layer. Risk chromatography assigns grades based on the predicted peak concentration of each unit level: areas where the predicted peak concentration reaches the emergency action threshold are marked as dark areas, corresponding to an emergency risk level; areas where the predicted peak concentration reaches the interlocking preparatory threshold but not the emergency action threshold are marked as medium-colored areas, corresponding to a warning risk level; and areas where the predicted peak concentration reaches the early warning threshold but not the interlocking preparatory threshold are marked as light-colored areas, corresponding to a caution risk level. Based on the risk map of hazardous gas spread, the interlock action priority and expected over-limit time are comprehensively calibrated for all affected monitoring units. The expected over-limit time is calculated as follows: extract the expected concentration time series of each monitoring unit, locate the time point when the concentration first exceeds the corresponding risk level threshold, and use the difference between this time point and the current timestamp as the expected over-limit duration, and convert it to the expected over-limit time accurate to the second. The interlocking action priorities include: the first priority is for monitoring units that have already exceeded the emergency action threshold, which must immediately execute the highest level interlocking action; the second priority is for units that are expected to exceed the emergency action threshold within 5 minutes, as well as units that have already exceeded the interlocking preparation threshold; the third priority is for units that are expected to exceed the interlocking preparation threshold within 5 to 15 minutes; and the fourth priority is for units that are expected to reach the warning threshold within 15 to 30 minutes. All affected monitoring units are sorted from high to low according to the interlocking action priority to generate an interlocking unit list. The interlocking unit list includes the unit number, spatial location, risk level, expected time of exceeding the limit, and suggested interlocking action type. The hierarchical interlocking control module issues hierarchical interlocking instructions to the interlocking execution equipment of the corresponding monitoring unit according to the interlocking unit list and priority order. Different risk levels correspond to different combinations of interlocking actions. The interlocking actions include: triggering the emergency shut-off valve of the storage tank to close, cutting off the material conveying of inlet and outlet pipelines, starting the fixed fire sprinkler and water curtain system, opening the emergency forced ventilation device, and triggering the on-site full-area audible and visual alarm; the warning level interlocking actions include: starting the backup sprinkler system, shutting off non-explosion-proof electrical equipment in the area, and broadcasting a notification to on-site personnel to evacuate to a safe area; the attention level interlocking actions include: triggering the area warning indicator light, dispatching inspection personnel to the site for verification, and starting the backup monitoring equipment to collect data more intensively. After the tiered interlock command is issued, the tiered interlock control module collects feedback signals from the interlock execution equipment in real time through the plant area scene perception module to confirm the action status of equipment such as valves, sprinklers, and alarms. If equipment execution failure is detected, the risk priority of the corresponding monitoring unit is immediately increased, and the tiered interlock command is regenerated. At the same time, alarm information is pushed to on-site operation and maintenance and safety management personnel. After each data scanning cycle, the tiered interlock control module updates the hazardous gas spread risk map and interlock unit list based on the latest measured concentration data and corrected prediction results, and dynamically adjusts the priority and interlock action of each unit until the real-time regional data of all monitoring units continuously fall below the warning threshold, at which point the safety interlock event is released.

[0025] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An Internet of Things (IoT) safety interlock control system for hazardous chemical gas storage tank areas in a factory, characterized in that, It includes a factory area scene perception module, an anomaly tracing module, an anomaly propagation prediction module, and a hierarchical interlocking control module; The plant area scene perception module is used to deploy an Internet of Things (IoT) perception layer in the tank area and delineate a monitoring unit with each gas tank as the center. The IoT perception layer is used to obtain real-time regional data of each monitoring unit and to determine whether there are any risks or anomalies in each monitoring unit based on the real-time regional data. The anomaly tracing module is used to execute a tracing thread based on the anomaly result of the risk judgment, locate the coordinates of the leaking unit based on the execution result of the tracing thread, and output the leakage rate curve. The spread prediction module is used to execute a prediction thread based on the abnormal results of risk judgment, and to perform forward modeling of the expected concentration time series of each monitoring unit based on the execution results of the prediction thread. The hierarchical interlocking control module is used to construct a hazardous gas spread risk map based on the expected concentration time series of each monitoring unit, the coordinates of the leaking unit, and the leakage rate curve. Then, based on the hazardous gas spread risk map, it assigns the priority of interlocking actions and the expected time of exceeding the limit for the affected monitoring units, generates a priority-ranked list of interlocking units, and executes them.

2. The IoT-based safety interlocking control system for hazardous gas storage tank areas in a factory area according to claim 1, characterized in that, The process of deploying an IoT sensing layer in the tank farm and defining monitoring units centered on each gas storage tank includes: Fixed sensing nodes are deployed in the hazardous gas storage tank area of ​​the target plant. All fixed sensing nodes are numbered sequentially, and a fixed data collection coverage radius is set for each fixed sensing node. The boundary distance between adjacent monitoring units is also adjusted. The target tank area consists of several vertical or horizontal hazardous gas storage tanks. Each tank corresponds to an independent monitoring unit, and each monitoring unit is associated with a fixed sensing node.

3. The IoT-based safety interlocking control system for hazardous chemical gas storage tank areas in a factory, as described in claim 2, is characterized in that... The process of acquiring real-time regional data through the IoT sensing layer and determining whether there are any risk anomalies in each monitoring unit based on the real-time regional data includes: Three levels of concentration thresholds are set for the types of hazardous gas media stored in each monitoring unit: early warning threshold, interlocking preparation threshold, and emergency action threshold. Set a data scanning cycle. After each data scanning cycle, all fixed sensing nodes continuously collect real-time regional data within their coverage area according to the set data scanning cycle. At the same time, perform a full threshold comparison on the real-time regional data of all monitoring units to generate start instructions for the tracing thread and the prediction thread.

4. The IoT-based safety interlocking control system for hazardous chemical gas storage tank areas in a factory, as described in claim 3, is characterized in that... The execution process of the source tracing thread includes: The tracing thread first retrieves historical area data from all monitoring units based on the event trigger timestamp; A unified timestamp benchmark for historical regional data is established, and the concentration peak point and rising edge start point at the event trigger time are marked as constraints for inverse fitting. Identify the rising edge start point in the historical concentration sequence of all monitoring units, sort them according to the order of the rising edge start points of each monitoring unit, and list the 3 to 5 monitoring units with the earliest rising edge time as first-level candidate leakage units, and then list the subsequent units as second-level and third-level candidate leakage units in ascending order of time difference.

5. The IoT-based safety interlocking control system for hazardous gas storage tank areas in a factory area according to claim 4, characterized in that, The process of locating the coordinates of the leaking unit and outputting the leakage rate curve based on the execution results of the source tracing thread includes: The peak concentration value and peak occurrence time are extracted from the concentration sequence of each candidate leakage unit. The peak concentration decay gradient and peak time difference between adjacent monitoring units are obtained. Backward diffusion fitting is performed with the decay law of peak concentration with spatial distance and air mass propagation speed as constraints. The reverse fitting process uses the leakage source coordinates and leakage intensity as the parameters to be solved. The measured concentration time series of each candidate leakage unit is used as the verification value. The parameters to be solved are adjusted iteratively and the residual sum of squares is obtained. When the residual sum of squares is lower than the preset convergence threshold or the number of iterations reaches the upper limit, the calculation is stopped and the optimal fitting parameters are output. The optimal leak source coordinates output by the inverse fitting are matched with the coordinates of the monitoring units in the digital twin space of the tank area. The monitoring unit corresponding to the nearest storage tank is determined as the leak unit. The spatial coordinates of the leak source are output simultaneously, along with the sequence of changes in the leak intensity over time obtained by fitting, i.e., the leak rate curve.

6. The IoT-based safety interlocking control system for hazardous chemical gas storage tank areas in a factory area according to claim 5, characterized in that, The execution process of the prediction thread includes: A micro-meteorological field is constructed based on real-time regional data and the digital twin space of the tank area. The micro-meteorological field is divided into equidistant grids, and each grid point corresponds to a wind speed vector, where the wind direction is represented by the vector direction and the wind speed is represented by the vector magnitude. The first monitoring unit identified as having an abnormal risk is used as the seed unit for diffusion forward modeling. The source term initial conditions for diffusion forward modeling are calibrated by combining the leak unit coordinates and leak rate curves output in real time by the source tracing thread. The current moment is taken as the prediction starting point, the prediction duration is set, the prediction time step is consistent with the data scanning cycle, and the digital twin space of the tank area is included as a boundary condition in the forward modeling calculation. Gas retention attributes are set for areas with a height lower than the cofferdam. Then, within each time step, the spatial location and diffusion range of the air mass are updated according to the current wind speed vector. The grid concentration at the center location and within the influence range of each monitoring unit is calculated successively at each time node in the next 30 minutes. The area-weighted average of all grid concentrations within the influence range of each monitoring unit is taken as the expected concentration at the corresponding time of the monitoring unit. Finally, a complete expected concentration time series for each monitoring unit is generated.

7. The IoT-based safety interlocking control system for hazardous chemical gas storage tank areas in a factory, as described in claim 6, is characterized in that... The process of constructing a hazardous gas spread risk map includes: Based on the leakage source location parameters and the expected concentration time series, a scalar distribution field of concentration at the current moment and at each future moment is constructed. The concentration scalar distribution field is then spatially superimposed with the real-time wind speed vector field to construct a hazardous gas spread risk map. The hazardous gas spread risk map consists of two core components: an ordered unit chain and a risk chromatogram. The construction of the ordered unit chain starts with the monitoring unit that first exceeds the interlocking preparatory threshold as the seed point, and proceeds along the wind speed vector direction. Combined with the expected over-limit time sequence of each monitoring unit, an ordered unit chain impacted by gas spread is formed. Simultaneously, starting from the seed unit, the system expands layer by layer to adjacent monitoring units according to the preset overlap of the influence range. Each expansion layer corresponds to a first-level spread level, and the risk chromatogram assigns values ​​based on the predicted peak concentration of each level unit.

8. The IoT-based safety interlocking control system for hazardous chemical gas storage tank areas in a factory, as described in claim 7, is characterized in that... The process of determining the priority of interlock actions and the expected time of exceeding limits for affected monitoring units based on the hazardous gas spread risk map, and generating a priority-ranked list of interlock units includes: Based on the risk map of hazardous chemical gas spread, the interlock action priority and expected over-limit time are comprehensively calibrated for all affected monitoring units. All affected monitoring units are sorted from high to low according to the interlock action priority to generate an interlock unit list. According to the interlocking unit list, hierarchical interlocking instructions are issued to the interlocking execution equipment of the corresponding monitoring unit in order of priority. Different risk levels correspond to different combinations of interlocking actions. After the tiered interlocking command is issued, the feedback signals of the interlocking execution equipment are collected in real time and the operation status of the equipment is confirmed. If the equipment execution failure is detected, the risk priority of the corresponding monitoring unit is immediately increased, and the tiered interlocking command is regenerated. The safety interlocking event is released only after the real-time regional data of all monitoring units have continuously fallen below the warning threshold.

Images