Dynamic coordination management system for intelligent chain and supply chain

Abstract

The invention discloses a dynamic coordination management system for an intelligent chain and a supply chain, and relates to the technical field of ventilation systems, which comprises the following steps: S1, respectively setting on-off states and action sequences for fans and protective valves in each region according to three working conditions of clean ventilation, poison filtration ventilation and isolation protection; s2, acquiring plane arrangement and boundary information of the warehouse, converting a corridor, a storage room and a partition wall into line segments under a unified coordinate system, generating a grid point set only located in the ventilated space based on a line segment range, mapping the position of an air quality sensor to a corresponding grid point, and setting the position of the air quality sensor as a grid point set; and S3, collecting concentration data of harmful gas from each air quality sensor according to a fixed time interval, constructing a distance measure only spreading along a ventilated corridor by preprocessing a geometric grid of a warehouse and using a path search algorithm, and then performing interpolation and error optimization under the distance measure. And directly mapping the obtained pollution distribution field into a partition fan operation strategy and a protection valve group control instruction.

Description

Technical Field

[0001] This invention relates to the field of ventilation system technology, specifically to the Smart Supply Chain Dynamic Coordination and Management System. Background Technology

[0002] In the supply chain system, a large number of chemical raw materials and hazardous materials need to be stored in warehouses for extended periods. During storage, these items slowly release harmful gases such as ammonia, hydrogen sulfide, carbon dioxide, and volatile organic compounds. If the warehouse is not properly ventilated, these harmful gases can easily accumulate in localized areas, affecting personnel health and operational safety, potentially leading to a decline in the quality of goods or even causing safety accidents.

[0003] Furthermore, in existing warehouse ventilation monitoring and control systems, the number of gas sensors is limited by cost, deployment conditions, and maintenance complexity. They are typically only deployed in a few key nodes of the air ducts or individual storage rooms, making dense deployment across the entire project impossible. On-duty personnel can only infer the diffusion of harmful gases in corridors and storage rooms based on readings from these discrete monitoring points, making it difficult to accurately determine the boundary location of pollution clouds and the distribution of high-risk areas. This results in a lack of spatial precision in the protective valve group control strategy. Therefore, designing a smart supply chain dynamic coordination management system is essential. Summary of the Invention

[0004] The purpose of this invention is to provide a smart supply chain dynamic coordination and management system to solve the problems mentioned in the background art.

[0005] To address the aforementioned technical problems, this invention provides the following technical solution: a smart supply chain dynamic coordination and management system, comprising a modeling and deployment module, an air quality calculation module, and a collaborative control module. The modeling and deployment module acquires the warehouse's floor plan and 3D information, digitally represents the boundaries of corridors, storage rooms, partitions, and air ducts, and configures the locations of air quality sensors. The air quality calculation module constructs a grid point set for path search based on the warehouse's geometric information, uses a path search algorithm to calculate the propagation distance along the corridor between the sensor and each grid point, and performs near-real-time interpolation estimation of the pollution concentration at each grid point based on this distance metric. The collaborative control module matches the air quality distribution field with a preset warehouse protection and control strategy, generating fan operation commands and group control commands for explosion-proof valves, protective valves, and filter valves for different storage rooms.

[0006] According to the above technical solution, the modeling and deployment module includes: a geometric data acquisition module, which extracts the corridor outline, partition wall outline, and air duct cross-sectional boundary from the building information model, warehouse construction drawings, and on-site survey data, and converts them into a set of line segments under a unified coordinate system; a rasterization processing module, which generates a set of raster points within the boundary line segments output by the geometric data acquisition module, dividing the corridor and storage room interior space into raster points with regular spacing for path search algorithms and interpolation calculations; and a sensor configuration module, which manages the coordinates, types, and numbers of air quality sensors, binds the monitoring points of carbon monoxide sensors, combustible gas sensors, and toxic agent sampling devices to the raster point coordinates, and associates the real-time sensor measurements with the corresponding raster points. The air quality calculation module includes: a distance calculation module, which calculates the shortest path distance from each sensor's grid point to other grid points within the corridor using a path search algorithm based on the grid point set and geometric boundary information, and simultaneously calculates the geometric straight-line distance for heuristic search; an interpolation calculation module, which performs inverse distance-weighted interpolation on the pollution concentration of each grid point based on the distance from the sensor to each grid point output by the distance calculation module and the real-time concentration measurement values ​​of each sensor, to obtain the pollution concentration distribution field within the warehouse area; and a parameter optimization module, which uses cross-validation to evaluate the error of different interpolation parameter combinations, selects the interpolation parameter with the smallest error index as the interpolation parameter at the current moment, and outputs the optimized pollution concentration distribution field. The collaborative control module includes: a risk area identification module, used to compare the pollution concentration distribution field with a preset danger threshold, identify corridor sections and storage rooms that exceed the threshold, and classify and mark them according to their importance level; a fan operation strategy generation module, used to generate start / stop commands, speed adjustment commands, and supply / exhaust air switching strategies for each fan and exhaust fan based on the risk distribution results output by the risk area identification module and the power conditions of the warehouse; and a protective valve group control module, used to determine the combination of protective valves and filter valves that need to be closed and the combination of explosion-proof pressure relief valves that need to be opened based on the location of the risk area and the direction of pollution diffusion, generate a valve action sequence, and send it to the protective valve actuator.

[0007] According to the above technical solution, the working method of this system includes: S1. Set the switch status and action sequence of the fans and protective valves in each area according to the three working conditions of clean ventilation, filtration ventilation and isolation protection. S2. Obtain the warehouse's floor plan and boundary information, convert corridors, storage rooms, and partitions into line segments in a unified coordinate system, generate a set of grid points located only within the ventilated space based on the range of the line segments, and map the location of the air quality sensor to the corresponding grid point. S3. Collect concentration data of harmful gases from each air quality sensor at fixed time intervals, perform time synchronization, outlier removal and missing value interpolation on the collected data, and obtain the set of effective concentration data of the sensor corresponding to the current time. S4. Based on the grid point set and geometric boundary information, the path search algorithm is used to calculate the propagation distance from each sensor's grid point to other grid points along the corridor. The path distance is used as the distance metric for interpolation weight. Inverse distance weighted interpolation is performed under this distance metric to obtain the initial distribution field of pollution concentration within the warehouse area at the current time. S5. The leave-one-out cross-validation method is used to calculate the error of different interpolation parameter combinations. The interpolation parameter combination with the smallest error index is selected as the interpolation parameter at the current time, and the pollution concentration distribution field is recalculated based on the interpolation parameter combination. S6. Combine the optimized pollution concentration distribution field with the preset protection threshold and valve control strategy to identify high-risk areas and generate corresponding fan operation commands and protective valve group control commands. The coordinated action of the fan and protective valve is completed through automatic control and human-machine interface linkage.

[0008] According to the above technical solution, S2 specifically refers to: S2-1. Import the warehouse building information model file and floor plan construction drawings into the processing software, extract the corridor outline, partition wall outline and ventilation duct boundary line from the drawings, and unify them into a set of line segments in a plane rectangular coordinate system through coordinate transformation. S2-2. Generate a grid point set within the ventilable area surrounded by the line segment set according to the preset grid spacing. Retain the grid points located inside the corridor and storage room, and delete the grid points located inside the partition wall or outside the project, so that the grid point set only covers the ventilable space inside the warehouse. S2-3. Map the installation location coordinates of the air quality sensor to a grid point set, associate each sensor with a grid point number, establish a one-to-one correspondence between the sensor number and the grid point number, and form a sensor grid point set for subsequent distance calculation.

[0009] According to the above technical solution, S3 specifically refers to: S3-1. During the monitoring period, read the monitoring values ​​of various harmful gases from each air quality sensor at fixed time intervals to form a raw monitoring data table corresponding to the timestamp. S3-2. Preprocess the original monitoring data table. Replace abnormal data that are significantly outside the sensor's measurement range with interpolation between previous and next time points. Supplement short-term missing data with interpolation between adjacent time points to ensure that all sensors have effective concentration data in each time step. S3-3. Normalize the preprocessed concentration data to a uniform unit and numerical range, establish the sensor concentration vector at the current moment, and associate each component in the vector with the sensor grid points established in S2-3 for use in subsequent path distance weighted interpolation.

[0010] According to the above technical solution, S4 specifically refers to: S4-1. For each grid point, check its several adjacent grid points in the horizontal and vertical directions. If the line connecting adjacent grid points falls entirely within the ventilable space, then establish an adjacency relationship between the two grid points. If the line passes through pillars, rock masses, sealed tunnels, and other obstructions, then do not establish an adjacency relationship between the two grid points, forming a grid adjacency graph structure that only contains ventilable passages. This graph structure serves as the input for subsequent path search algorithms, strictly limiting the path search process to the space inside ventilated tunnels and storage rooms. S4-2. For each grid point, use... The path search algorithm calculates the shortest ventilation path distance between the grid point and each sensor grid point: To provide a geometric distance metric for the path search, any two three-dimensional spatial points are defined. and Euclidean distance between ,in for point to vector difference, To represent the Euclidean norm of a vector, the current node The cost of moving from the current position to an adjacent position in one step. Defined as the Euclidean distance between the current position and the next position. To estimate the remaining distance between the current node and the target node, a heuristic cost function is defined. ,in Given the spatial coordinate vector of the target node, a comprehensive evaluation value needs to be given for each candidate node during the search process, which is the comprehensive cost function. Select from the open list at each step Expand the smallest node; S4-3. For any grid point, based on the aforementioned Euclidean distance metric, one-step movement cost function, heuristic cost function, and comprehensive cost function, use... The path search algorithm sequentially expands nodes on the grid adjacency graph structure until the corresponding sensor grid point is removed from the open list. At this point, the shortest path from the grid point to the sensor grid point is obtained. This shortest path consists of a series of consecutive grid nodes. By counting the number of nodes in the path and multiplying it by the grid spacing, the discrete path length value is obtained. This value is used as the path search distance from the grid point to the corresponding sensor grid point. The above process is repeated for each grid point and each sensor grid point to form a set of path search distance data for spatial interpolation. S4-4. Spatial interpolation estimation of pollution concentration using the inverse distance weighting method: for any sensor location With arbitrary interpolation positions Path search distance between Define distance weight values This weight decreases with increasing distance. The distance between the two is the path search distance output by S4-3. These are power coefficients used to describe the rate at which the weights decay with distance, for any interpolation position. Its estimated pollution concentration Calculated by combining the weights of all sensors and their measured values. ,in This represents the number of sensors currently participating in the interpolation calculation, including all sensors with valid concentration readings at the current time step. For the first The measured pollution concentration values ​​of each sensor To be arrive The distance weight value is obtained by substituting the path search distance between them; S4-5. For each grid point, its spatial coordinates are considered as the interpolation position. The estimated pollution concentration of this grid point is calculated based on S4-4. By performing the above interpolation operation on all grid points, a spatial distribution field of pollution concentration inside the entire ventilation system is formed at the current time step.

[0011] According to the above technical solution, S5 specifically refers to: S5-1. At the current monitoring time, construct a leave-one-out cross-validation structure for error assessment for all sensors participating in the interpolation. Specifically, this includes: Let the sensor set be denoted as... Each element in the set corresponds to an installation location. and ,against For each sensor in the set, temporarily remove that sensor from the interpolation process, retaining only the remaining sensors in the set as the known point set for the current round of interpolation. For the removed sensor index... At its installation location Re-execute based on Inverse distance weighted interpolation calculation using the current power coefficient. This yields the interpolated estimate for that location. By analyzing the set Each sensor sequentially performs the above removal and re-interpolation operations to construct a set of pollutant concentration values ​​measured by the sensors. Interpolated estimates Using a leave-one-out cross-validation structure, the mean absolute percentage error of the interpolation model over the entire sensor ensemble is calculated from the sample. ; S5-2, Obtaining coefficients of different power orders Below Then, by traversing the set of candidate power coefficients and finding the minimum value of the error function, the optimal power coefficient at this point is determined. .

[0012] According to the above technical solution, S6 specifically refers to: S6-1. Compare the pollution concentration distribution field obtained in step S4 with the preset danger threshold, mark the area where the grid point concentration exceeds the threshold as a high-risk grid point set, and identify high-risk corridor sections, high-risk storage rooms and potential pollution source directions based on the spatial distribution of high-risk grid points in corridors and storage rooms. S6-2. Configure the fan operation strategy according to the location and importance level of the high-risk area. Increase the speed of the fan on the upwind side of the pollution and ensure the positive pressure in the key storage room. Adjust the start-stop and speed of the exhaust fan on the downwind side of the pollution according to the preset strategy to guide the pollutants to concentrate in the set discharge channel. At the same time, limit the total power of the fan according to the power supply capacity. Convert the above fan operation strategy into specific start-stop commands and speed setting values. S6-3. Based on the relationship between high-risk areas and warehouse zoning, determine the opening and closing combinations of explosion-proof valves, protective valves, and filter valves for each storage room. Issue a closing command to the protective valves located on the pollution diffusion path, issue an opening command to the filter valves that need to maintain air intake protection, and issue an opening command to the explosion-proof valves that need to release overpressure.

[0013] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: The present invention performs gridding processing on the warehouse layout, introduces a path search algorithm to calculate the shortest path distance propagating inside the corridor, regards the corridor boundary, partition walls and closed sections as insurmountable obstacles, and performs inverse distance weighted interpolation on this basis and automatically optimizes the interpolation parameters through error index, thereby obtaining an air quality distribution field containing details of the interior of the storage room under the condition of a limited number of sensors. Based on the identification of high-risk areas within the pollution distribution field, a coordinated control strategy is generated by adjusting fan speeds, switching between supply and exhaust air, and combining the opening and closing of explosion-proof valves, protective valves, and filter valves according to their importance levels. This strategy can be updated on a minute-by-minute timescale. Through this system, warehouses can achieve precise control of fans and protective valves under limited power conditions in scenarios where harmful gases accumulate, thereby improving the air safety margin in storage rooms and reducing unnecessary energy consumption. Attached Figure Description

[0014] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the overall modular structure of the present invention. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] Please see Figure 1 This invention provides a technical solution: a smart supply chain dynamic coordination and management system, including a modeling and deployment module, an air quality calculation module, and a collaborative control module. The modeling and deployment module is used to acquire the warehouse's floor plan and 3D information, digitally represent the corridors, storage rooms, partitions, and air duct boundaries, and configure the locations of air quality sensors. The air quality calculation module constructs a grid point set for path search based on the warehouse's geometric information, uses a path search algorithm to calculate the propagation distance along the corridor between the sensors and each grid point, and performs near real-time interpolation estimation of the pollution concentration of each grid point under this distance metric. The collaborative control module matches the air quality distribution field with a preset warehouse protection and control strategy, generating fan operation commands and group control commands for explosion-proof valves, protective valves, and filter valves for different storage rooms. The modeling and deployment module includes: a geometric data acquisition module, which extracts the corridor outline, partition wall outline, and air duct cross-sectional boundary from the building information model, warehouse construction drawings, and on-site survey data, and converts them into a set of line segments in a unified coordinate system; a rasterization module, which generates a set of raster points within the boundary line segments output by the geometric data acquisition module, dividing the corridor and storage room interior space into raster points with regular spacing for path search algorithms and interpolation calculations; and a sensor configuration module, which manages the coordinates, types, and numbers of air quality sensors, binds the monitoring points of carbon monoxide sensors, combustible gas sensors, and toxic agent sampling devices to the raster point coordinates, and associates the real-time sensor measurements with the corresponding raster points. The air quality calculation module includes: a distance calculation module, which calculates the shortest path distance from each sensor's grid point to other grid points within the corridor using a path search algorithm based on the grid point set and geometric boundary information, and simultaneously calculates the geometric straight-line distance for heuristic search; an interpolation calculation module, which performs inverse distance-weighted interpolation on the pollution concentration of each grid point based on the distance from the sensor to each grid point output by the distance calculation module and the real-time concentration measurement values ​​of each sensor, to obtain the pollution concentration distribution field within the warehouse area; and a parameter optimization module, which uses cross-validation to evaluate the error of different interpolation parameter combinations, selects the interpolation parameter with the smallest error index as the interpolation parameter for the current moment, and outputs the optimized pollution concentration distribution field. The collaborative control module includes: a risk area identification module, which compares the pollution concentration distribution field with preset danger thresholds, identifies corridor sections and storage rooms that exceed the thresholds, and classifies and marks them according to their importance level; a fan operation strategy generation module, which generates start / stop commands, speed adjustment commands, and supply / exhaust air switching strategies for each fan and exhaust fan based on the risk distribution results output by the risk area identification module and the power conditions of the warehouse; and a protective valve group control module, which determines the combination of protective valves and filter valves that need to be closed and the combination of explosion-proof pressure relief valves that need to be opened based on the location of the risk area and the direction of pollution diffusion, generates the valve action sequence, and sends it to the protective valve actuator.

[0017] The system's operating method includes: S1, setting the on / off status and action sequence of the fans and protective valves in each area according to three operating conditions: clean ventilation, filtration ventilation, and isolation protection; S2. Obtain the warehouse's floor plan and boundary information, convert corridors, storage rooms, and partitions into line segments in a unified coordinate system, generate a set of grid points located only within the ventilated space based on the range of the line segments, and map the location of the air quality sensor to the corresponding grid point. S3. Collect concentration data of harmful gases from each air quality sensor at fixed time intervals, perform time synchronization, outlier removal and missing value interpolation on the collected data, and obtain the set of effective concentration data of the sensor corresponding to the current time. S4. Based on the grid point set and geometric boundary information, the path search algorithm is used to calculate the propagation distance from each sensor's grid point to other grid points along the corridor. The path distance is used as the distance metric for interpolation weight. Inverse distance weighted interpolation is performed under this distance metric to obtain the initial distribution field of pollution concentration within the warehouse area at the current time. S5. The leave-one-out cross-validation method is used to calculate the error of different interpolation parameter combinations. The interpolation parameter combination with the smallest error index is selected as the interpolation parameter at the current time, and the pollution concentration distribution field is recalculated based on the interpolation parameter combination. S6. Combine the optimized pollution concentration distribution field with the preset protection threshold and valve control strategy to identify high-risk areas and generate corresponding fan operation instructions and protective valve group control instructions. The coordinated action of the fan and protective valve is completed through automatic control and human-machine interface linkage. S2 specifically refers to: S2-1. Import the warehouse building information model file and floor plan construction drawings into the processing software, extract the corridor outline, partition wall outline and ventilation duct boundary line from the drawings, and unify them into a set of line segments in a plane rectangular coordinate system through coordinate transformation. S2-2. Generate a grid point set within the ventilable area surrounded by the line segment set according to the preset grid spacing. Retain the grid points located inside the corridor and storage room, and delete the grid points located inside the partition wall or outside the project, so that the grid point set only covers the ventilable space inside the warehouse. S2-3. Map the installation location coordinates of the air quality sensor to a grid point set, associate each sensor with a grid point number, establish a one-to-one correspondence between the sensor number and the grid point number, and form a sensor grid point set for subsequent distance calculation.

[0018] S3 specifically refers to: S3-1. During the monitoring period, read the monitoring values ​​of various harmful gases from each air quality sensor at fixed time intervals to form a raw monitoring data table corresponding to the timestamp. S3-2. Preprocess the original monitoring data table. Replace abnormal data that are significantly outside the sensor's measurement range with interpolation between previous and next time points. Supplement short-term missing data with interpolation between adjacent time points to ensure that all sensors have effective concentration data in each time step. S3-3. Normalize the preprocessed concentration data to a uniform unit and numerical range, establish the sensor concentration vector at the current moment, and associate each component in the vector with the sensor grid points established in S2-3 for use in subsequent path distance weighted interpolation.

[0019] S4 specifically refers to: S4-1. For each grid point, check its several adjacent grid points in the horizontal and vertical directions. If the line connecting adjacent grid points falls entirely within the ventilable space, then establish an adjacency relationship between the two grid points. If the line passes through pillars, rock masses, sealed tunnels, and other obstructions, then do not establish an adjacency relationship between the two grid points, forming a grid adjacency graph structure that only contains ventilable passages. This graph structure serves as the input for subsequent path search algorithms, strictly limiting the path search process to the space inside ventilated tunnels and storage rooms. S4-2. For each grid point, use... The path search algorithm calculates the shortest ventilation path distance between the grid point and each sensor grid point: To provide a geometric distance metric for the path search, any two three-dimensional spatial points are defined. and Euclidean distance between ,in for point to vector difference, To represent the Euclidean norm of a vector, the current node The cost of moving from the current position to an adjacent position in one step. Defined as the Euclidean distance between the current position and the next position. To estimate the remaining distance between the current node and the target node, a heuristic cost function is defined. ,in Given the spatial coordinate vector of the target node, a comprehensive evaluation value needs to be given for each candidate node during the search process, which is the comprehensive cost function. Select from the open list at each step Expand the smallest node; S4-3. For any grid point, based on the aforementioned Euclidean distance metric, one-step movement cost function, heuristic cost function, and comprehensive cost function, use... The path search algorithm sequentially expands nodes on the grid adjacency graph structure until the corresponding sensor grid point is removed from the open list. At this point, the shortest path from the grid point to the sensor grid point is obtained. This shortest path consists of a series of consecutive grid nodes. By counting the number of nodes in the path and multiplying it by the grid spacing, the discrete path length value is obtained. This value is used as the path search distance from the grid point to the corresponding sensor grid point. The above process is repeated for each grid point and each sensor grid point to form a set of path search distance data for spatial interpolation. S4-4. Spatial interpolation estimation of pollution concentration using the inverse distance weighting method: for any sensor location With arbitrary interpolation positions Path search distance between Define distance weight values This weight decreases with increasing distance. The distance between the two is the path search distance output by S4-3. These are power coefficients used to describe the rate at which the weights decay with distance, for any interpolation position. Its estimated pollution concentration Calculated by combining the weights of all sensors and their measured values. ,in This represents the number of sensors currently participating in the interpolation calculation, including all sensors with valid concentration readings at the current time step. For the first The measured pollution concentration values ​​of each sensor To be arrive The distance weight value is obtained by substituting the path search distance between them; S4-5. For each grid point, its spatial coordinates are considered as the interpolation position. The estimated pollution concentration of this grid point is calculated based on S4-4. By performing the above interpolation operation on all grid points, a spatial distribution field of pollution concentration inside the entire ventilation system is formed at the current time step.

[0020] Conventional systems typically use ventilation zoning and threshold judgment. This solution first filters out the grid adjacency graph of areas that can only be ventilated; then, path search is used to obtain the actual propagation distance that bypasses walls; finally, this distance is used for weighted interpolation. A near real-time digital twin of the pollution field is constructed within the ventilation system. The warehouse is treated as a graph structure, and path search methods like A* are used to calculate the actual propagation distance that allows airflow. IDW interpolation is then performed under this distance metric. The physical constraint that toxic gases / smoke can only travel along ventilation corridors is explicitly encoded into the interpolation weights through ventilation path distances, ensuring that the spatial shape of the interpolation result closely matches the actual propagation path, rather than a simple circular diffusion. This solves the problem of having sparse sensor locations but needing to understand the pollution distribution of the entire project; it maps the complex geometry of corridors / storage rooms and the state of wall closures into a mathematical model of graph + distance metric. S5 specifically refers to: S5-1. At the current monitoring time, construct a leave-one-out cross-validation structure for error assessment for all sensors participating in the interpolation. Specifically, this includes: Let the sensor set be denoted as... Each element in the set corresponds to an installation location. and ,against For each sensor in the set, temporarily remove that sensor from the interpolation process, retaining only the remaining sensors in the set as the known point set for the current round of interpolation. For the removed sensor index... At its installation location Re-execute based on Inverse distance weighted interpolation calculation using the current power coefficient. This yields the interpolated estimate for that location. By analyzing the set Each sensor sequentially performs the above removal and re-interpolation operations to construct a set of pollutant concentration values ​​measured by the sensors. Interpolated estimates Using a leave-one-out cross-validation structure, the mean absolute percentage error of the interpolation model over the entire sensor ensemble is calculated from the sample. ; S5-2, Obtaining coefficients of different power orders Below Then, by traversing the set of candidate power coefficients and finding the minimum value of the error function, the optimal power coefficient at this point is determined. .

[0021] When using IDW interpolation, the power factor is typically set empirically (e.g., fixed at 2); or several values ​​are manually tried during the data analysis phase, and a satisfactory one is chosen and left unchanged during operation. It's rare to see a systematic parameter search and cross-validation for each monitoring moment in a real-time control system. This solution upgrades the power factor parameter of IDW interpolation from an empirical value to a data-driven, adaptive value. The optimal power factor may differ depending on the time and pollution state: a larger power factor is needed to highlight the nearest point when concentration changes drastically; a smaller power factor can be used to avoid overfitting when the field is relatively smooth. Through leave-one-out cross-validation, the system can fully utilize existing point data to calibrate the model without adding new sensors, improving the confidence of the pollution field estimation. In engineering systems like ventilation control, a specially designed leave-one-out cross-validation structure automatically selects the optimal power factor for IDW, transforming pollution field interpolation from an empirical parameter to data-adaptive interpolation. This is a clever design that deeply embeds data science methods into traditional ventilation control.

[0022] S6 specifically refers to: S6-1. Compare the pollution concentration distribution field obtained in step S4 with the preset danger threshold, mark the area where the grid point concentration exceeds the threshold as a high-risk grid point set, and identify high-risk corridor sections, high-risk storage rooms and potential pollution source directions based on the spatial distribution of high-risk grid points in corridors and storage rooms. S6-2. Configure the fan operation strategy according to the location and importance level of the high-risk area. Increase the speed of the fan on the upwind side of the pollution and ensure the positive pressure in the key storage room. Adjust the start-stop and speed of the exhaust fan on the downwind side of the pollution according to the preset strategy to guide the pollutants to concentrate in the set discharge channel. At the same time, limit the total power of the fan according to the power supply capacity. Convert the above fan operation strategy into specific start-stop commands and speed setting values. S6-3. Based on the relationship between high-risk areas and warehouse zoning, determine the opening and closing combinations of explosion-proof valves, protective valves, and filter valves for each storage room. Issue a closing command to the protective valves located on the pollution diffusion path, issue an opening command to the filter valves that need to maintain air intake protection, and issue an opening command to the explosion-proof valves that need to release overpressure.

[0023] This step puts the previously obtained pollution distribution field into practical use, achieving refined spatial collaborative control. Traditional ventilation system control logic is almost entirely based on pre-configured rule tables; this solution, however, first calculates a high-dimensional state pollution field, and then generates control actions based on the state field and strategy rules. By first achieving spatial accuracy, it then uses this to drive the control strategy, no longer relying on single-point alarms to trigger fixed linkage tables, but automatically generating a collaborative strategy for fans and protective valves based on the real-time pollution field distribution.

[0024] The warehouse is transformed into a ventilation path map, interpolated along the actual propagation path, and adaptively tuned through leave-one-out cross-validation; then, based on this, fan / valve collaborative control commands that take into account both importance level and power constraints are generated; this significantly improves the accuracy of pollution field estimation and the precision of ventilation control under limited sensor conditions.

[0025] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0026] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A dynamic coordination management system for a smart chain supply chain, characterized by: The system comprises a modeling and layout module, an air quality calculation module, and a collaborative control module, the modeling and layout module is used to obtain the plan of the warehouse and three-dimensional information, and to digitally express the corridors, storage rooms, partition walls, and air duct boundaries and configure the positions of air quality sensors; the air quality calculation module is used to construct a grid point set for path search based on the geometric information of the warehouse, to calculate the propagation distance along the corridor between the sensors and each grid point by using a path search algorithm, and to perform near-real-time interpolation estimation on the pollution concentration of each grid point under the distance measurement; the collaborative control module is used to match the air quality distribution field with the preset warehouse protection control strategy, and to generate fan operation instructions for different storage rooms and group control instructions for explosion-proof valves, protection valves, and filter valves.

2. The intelligent chain supply chain dynamic coordination management system according to claim 1, characterized in that: the modeling and layout module comprises a geometric data acquisition module that extracts the corridor profile, partition wall profile, and air duct cross-section boundary of the warehouse from the building information model, warehouse construction drawings, and field surveying and mapping data, and converts them into a line segment set in a unified coordinate system; a gridding processing module that generates a grid point set within the boundary line segments output by the geometric data acquisition module, and divides the interior space of the corridor and storage room into regularly spaced grid points for use in path search algorithms and interpolation calculation; and a sensor configuration module that manages the coordinates, types, and numbers of air quality sensors, binds the monitoring points of carbon monoxide sensors, combustible gas sensors, and toxic agent sampling devices to the grid point coordinates, and associates the real-time measurement values of the sensors with the corresponding grid points; the air quality calculation module comprises a distance calculation module that calculates the shortest path distance within the corridor between the grid point where each sensor is located and other grid points by using a path search algorithm based on the grid point set and geometric boundary information, and simultaneously calculates the geometric straight-line distance for heuristic search; an interpolation calculation module that inversely distance-weighted interpolates the pollution concentration of each grid point according to the distance between the sensors and each grid point output by the distance calculation module and the real-time concentration measurement values of each sensor, to obtain the pollution concentration distribution field within the warehouse; and a parameter optimization module that performs error evaluation on different interpolation parameter combinations in a cross-validation manner, selects the interpolation parameter with the smallest error index as the interpolation parameter at the current time, and outputs the optimized pollution concentration distribution field; the collaborative control module comprises a risk area identification module that compares the pollution concentration distribution field with a preset danger threshold, identifies the corridor sections and storage rooms that exceed the threshold, and marks them according to the importance level; a fan operation strategy generation module that generates start-stop instructions, speed gear adjustment instructions, and air supply and exhaust switching strategies for each fan and exhaust fan according to the risk distribution results output by the risk area identification module and the power supply conditions of the warehouse; and a protection valve group control module that determines the combination of protection valves and filter valves that need to be closed and the combination of explosion-proof relief valves that need to be opened according to the location and pollution diffusion direction of the risk area, generates a valve action sequence, and issues it to the protection valve actuator. 3.The dynamic coordination management system of the chain supply chain according to claim 2, characterized in that: The working method of the system comprises: S1, set the switch state and action sequence of the fan and protection valve for each region according to the three working conditions of clean ventilation, filter ventilation and isolation protection; S2, obtain the planar layout and boundary information of the warehouse, convert the corridors, storage rooms and partition walls into line segments in a unified coordinate system, generate a grid point set located only in the ventilable space based on the line segment range, and map the air quality sensor positions to the corresponding grid points; S3, collect the concentration data of harmful gases from each air quality sensor at fixed time intervals, perform time synchronization, outlier rejection and missing value interpolation on the collected data, and obtain a set of effective concentration data corresponding to the current time; S4, based on the grid point set and geometric boundary information, use a path search algorithm to calculate the propagation distance of each sensor grid point to other grid points along the corridor, use the path distance as the distance measure of the interpolation weight, and perform inverse distance weighted interpolation under the distance measure to obtain the initial pollution concentration distribution field in the warehouse range at the current time; S5, use leave-one-cross validation to calculate the error of different interpolation parameter combinations, select the interpolation parameter combination with the smallest error index as the interpolation parameter at the current time, and recalculate the pollution concentration distribution field based on the interpolation parameter combination; S6, combine the optimized pollution concentration distribution field with the preset protection threshold and valve control strategy, identify high-risk areas and generate corresponding fan operation instructions and protection valve group control instructions, and complete the coordinated action of the fan and the protection valve through automatic control and human-machine interface linkage. 4.The dynamic coordination management system of the chain supply chain according to claim 3, characterized in that: The S2 is specifically: S2-1, import the building information model file and plan construction drawing of the warehouse into the processing software, extract the corridor contour line, partition wall contour line and air duct boundary line from the drawing, and convert them into a line segment set in the planar rectangular coordinate system through coordinate conversion; S2-2, generate a grid point set in the ventilable area surrounded by the line segment set according to the preset grid spacing, retain the grid points located in the corridor and the interior space of the storage room, and delete the grid points falling inside the partition wall or outside the project, so that the grid point set only covers the ventilable space inside the warehouse; S2-3, map the installation position coordinates of the air quality sensor to the grid point set, associate a grid point number with each sensor, establish a one-to-one correspondence between the sensor number and the grid point number, and form a sensor grid point set for subsequent distance calculation. 5.The dynamic coordination management system of the chain supply chain according to claim 4, characterized in that: The S3 is specifically: S3-1, read the monitoring values of various harmful gases from each air quality sensor at fixed time intervals within the monitoring period, and form a raw monitoring data table corresponding to the time stamp; S3-2, preprocess the raw monitoring data table, replace the abnormal data that obviously exceeds the sensor range with the interpolation method of the previous and next time, and supplement the short-time missing data with the interpolation method of the adjacent time, so that all sensors have effective concentration data at each time step; S3-3, normalize the pre-processed concentration data to a unified unit and numerical range, establish the sensor concentration vector at the current time, and correspond each component in the vector to the sensor grid point established in S2-3 for subsequent path distance weighted interpolation. 6.The dynamic coordination management system of the chain supply chain according to claim 5, characterized in that: The S4 is specifically: S4-1, for each grid point, check its several adjacent grid points in the horizontal and vertical directions, if the adjacent grid point line falls completely within the ventilable space, then establish an adjacency relationship between the two grid points; if the line passes through the column, rock mass, closed roadway and other blocking bodies, then do not establish an adjacency relationship between the two grid points, form a grid adjacency graph structure containing only the ventilable path, which is used as the input of the subsequent path search algorithm, so that the path search process is strictly limited within the ventilation roadway and the internal space of the storage room; S4-2. For each grid point, use... The path search algorithm calculates the shortest ventilation path distance between the grid point and each sensor grid point: To provide a geometric distance metric for the path search, any two three-dimensional spatial points are defined. and Euclidean distance between ,in for point to vector difference, To represent the Euclidean norm of a vector, the current node The cost of moving from the current position to an adjacent position in one step. Defined as the Euclidean distance between the current position and the next position. To estimate the remaining distance between the current node and the target node, a heuristic cost function is defined. ,in Given the spatial coordinate vector of the target node, a comprehensive evaluation value needs to be given for each candidate node during the search process, which is the comprehensive cost function. Select from the open list at each step Expand the smallest node; S4-3, for any grid point, based on the above-mentioned Euclidean distance metric, one-step movement cost function, heuristic cost function and comprehensive cost function, using The path search algorithm expands nodes on the grid adjacency graph structure in turn until the corresponding sensor grid point is taken out from the open list, at which time the shortest path from the grid point to the sensor grid point is obtained, which is composed of a series of consecutive grid nodes. By counting the number of nodes in the path and multiplying by the grid spacing, the discrete path length value is obtained. This value is taken as the path search distance of the grid point to the corresponding sensor grid point. The above process is repeated for each grid point and each sensor grid point to form a set of path search distance data for spatial interpolation. S4-4, spatial interpolation of pollution concentration using inverse distance weighting: for any interpolation location , the path search distance between the interpolation location and any sensor location , define a distance weight value , which decays with distance, , where d is the path search distance between the interpolation location and any sensor location , the output of S4-3, , and p is the power coefficient describing the decay rate of the weight with distance. For any interpolation location , its pollution concentration estimate is calculated by combining the weight and the sensor measurement of all sensors, , where is the number of sensors participating in the interpolation calculation, including all sensors with valid concentration readings at the current time step, is the measured pollution concentration value of the th sensor, is the distance weight value obtained by substituting the path search distance between the interpolation location and the th sensor into the distance weight function; S4-5, for each grid point, take its spatial coordinates as the interpolation position , calculate the pollution concentration estimation value of the grid point according to S4-4 , form the pollution concentration spatial distribution field inside the entire ventilation system at the current time step by performing the above interpolation operation on all grid points.

7. The dynamic coordination management system of the chain supply chain according to claim 6, characterized in that: The S5 is specifically: S5-1. At the current monitoring time, construct a leave-one-out cross-validation structure for error assessment for all sensors participating in the interpolation. Specifically, this includes: Let the sensor set be denoted as... Each element in the set corresponds to an installation location. and ,against For each sensor in the set, temporarily remove that sensor from the interpolation process, retaining only the remaining sensors in the set as the known point set for the current round of interpolation. For the removed sensor index... At its installation location Re-execute based on Inverse distance weighted interpolation calculation using the current power coefficient. This yields the interpolated estimate for that location. By analyzing the set Each sensor sequentially performs the above removal and re-interpolation operations to construct a set of pollutant concentration values ​​measured by the sensors. Interpolated estimates Using a leave-one-out cross-validation structure, the mean absolute percentage error of the interpolation model over the entire sensor ensemble is calculated from the sample. ; S5-2, obtaining different power coefficients under different power exponents After that, the optimal power coefficient at this time is determined by traversing the candidate power coefficient set and finding the minimum value of the error function . 8.The dynamic coordination management system of the chain supply chain according to claim 7, characterized in that: The S6 is specifically: S6-1, compare the pollution concentration distribution field obtained in step S4 with the preset danger threshold, mark the area where the grid point concentration exceeds the threshold as a high-risk grid point set, and identify the high-risk corridor section, high-risk storage room and potential pollution source direction according to the spatial distribution of the high-risk grid points in the corridor and the storage room; S6-2, according to the position of the high-risk area and the important level, configure the fan operation strategy, increase the speed of the fan on the upwind side of the pollution and ensure the positive pressure in the key storage room, adjust the start-stop and speed of the exhaust fan on the downwind side of the pollution according to the preset strategy to guide the pollution to the set exhaust channel, and limit the total power of the fan according to the power capacity; convert the above fan operation strategy into specific start-stop instructions and speed setting values; S6-3, according to the relationship between the high-risk area and the warehouse partition, determine the opening and closing combination of the explosion-proof valve, the protective valve and the filter valve for each storage room, issue a closing instruction to the protective valve located on the pollution diffusion path, issue an opening instruction to the filter valve that needs to maintain air intake protection, and issue an opening instruction to the explosion-proof valve that needs to release overpressure.

Images