A granary precision regulation method based on differential partition grid and three-dimensional interaction

Abstract

The application provides a kind of based on the method for accurate regulation and control of granary with three-dimensional interaction of micro-partition grid, related to the technical field of control system, the method comprises: by executing micro-partition grid division to granary space and constructing the grid pipeline assembly containing sensor and electric controllable valve, form micro-partition unit system;Further construct digital twin model and three-dimensional visualization model, realize state expression variable mapping;Based on state expression variable, identify abnormal micro-partition unit, and generate path optimization results in combination with graph structure model and path cost parameters;According to the path optimization results, selective control and feedback adjustment of electric controllable valve are carried out, to realize directional regulation and control of abnormal area;Finally, corresponding grid pipeline assembly is executed cleaning treatment and cleaning degree detection parameters are judged to determine the cleaning completion state.The application can solve the problem that the existing granary regulation and control process lacks dynamic path optimization and three-dimensional interaction regulation and control mechanism based on the structure of granary space, thereby causing low regulation and control precision.

Description

Technical Field

[0001] This invention relates to the technical field of control systems, and specifically to a method for precise control of grain warehouses based on micro-partitioned grids and three-dimensional interaction. Background Technology

[0002] During grain storage, to ensure the stability of grain quality and storage safety, continuous monitoring and control of internal temperature, humidity, gas environment, and pest and disease conditions are typically required. Existing grain storage control systems usually deploy temperature and humidity sensors, pest monitoring devices, and ventilation ducts inside the grain storage facility. These systems regulate the internal environment through air supply, dehumidification, and pesticide application equipment. When localized high temperatures, high humidity, condensation, or pest and disease spread are detected inside the grain storage facility, the control system activates corresponding control equipment based on the monitoring results. This involves supplying airflow, dehumidifying media, or pesticides to regulate the grain storage environment, thereby reducing the risk of grain mold and inhibiting the spread of pests and diseases. Some existing technologies also incorporate visual monitoring platforms to graphically display the internal status of the grain storage facility, allowing managers to view the storage status of different areas and perform manual control operations.

[0003] However, existing grain storage control methods typically employ fixed conveying structures and coarse-grained regional control modes, lacking dynamic path optimization and three-dimensional interactive control mechanisms based on the grain storage space structure. When an abnormal state occurs in a certain area within the grain storage, existing systems usually cannot automatically generate the optimal conveying path by combining the location of the abnormal area, the length of the conveying path, pipeline connectivity, and the status of the control equipment. Instead, they use large-scale unified conveying or fixed-path conveying methods for control, resulting in a large number of unrelated areas participating in the control process simultaneously. This not only increases energy consumption and control time but also easily leads to insufficient local control accuracy. At the same time, existing technologies struggle to perform dynamic selective control on nodes in the conveying path and cannot adjust valve opening status and conveying direction in real time according to changes in abnormal areas. This results in low control response efficiency and easily leads to conveying conflicts and unreasonable allocation of control resources when multiple abnormal areas occur simultaneously. Summary of the Invention

[0004] This invention provides a precise grain storage control method based on micro-regional grid and three-dimensional interaction, which can solve the problem of low control accuracy caused by the lack of dynamic path optimization and three-dimensional interactive control mechanism based on the grain storage spatial structure in the existing grain storage control process.

[0005] In a first aspect, the present invention provides a method for precise control of grain storage based on micro-regional grids and three-dimensional interaction, the method comprising: The grain storage space is divided into micro-zone grids, and grid pipe components integrating sensors and electrically controllable valves are built in each micro-zone unit to form a micro-zone unit system; Digital twin modeling is performed on the micro-partition unit system and a three-dimensional visualization model is constructed, mapping the state data of each micro-partition unit into state expression variables; Based on the state expression variables, abnormal micro-partition units are identified and a graph structure model is constructed. At the same time, a path cost parameter is introduced to optimize the transport path between the abnormal micro-partition unit and the control equipment to generate path optimization results. Based on the path optimization results, selective opening and closing control is performed on the electrically controllable valves on the grid pipeline assembly, and feedback adjustment is performed in conjunction with the state expression variables to complete directional regulation; After completing the directional control, the mesh pipeline component corresponding to the path optimization result is cleaned and the cleaning completion status is determined based on the cleanliness detection parameters.

[0006] In a second aspect of the invention, a grain storage precision control device based on micro-regional grid and three-dimensional interaction is provided. The device is used to execute a grain storage precision control method based on micro-regional grid and three-dimensional interaction as described above. The device includes an acquisition module, a processing module, and an output module, wherein: The acquisition module is used to perform micro-partitioning and grid division of the grain warehouse space and construct a grid pipeline assembly integrating sensors and electrically controllable valves in each micro-partition unit, thereby forming a micro-partition unit system. The processing module is used to perform digital twin modeling on the micro-partition unit system and construct a three-dimensional visualization model, and to map the state data of each micro-partition unit into state expression variables. The processing module is used to identify abnormal micro-partition units based on the state expression variables and construct a graph structure model. At the same time, it introduces path cost parameters to perform optimization solution on the transmission path between the abnormal micro-partition units and the control equipment to generate path optimization results. The processing module is used to selectively open and close the electrically controllable valves on the grid pipeline assembly according to the path optimization results, and to perform feedback adjustment in conjunction with the state expression variables to complete directional regulation; The output module is used to perform cleaning processing on the grid pipeline component corresponding to the path optimization result after completing the directional control, and determine the cleaning completion status based on the cleanliness detection parameters.

[0007] In a third aspect of the invention, an electronic device is provided, including a processor and a memory, the memory having a computer program stored thereon, wherein the computer program, when executed by the processor, implements a grain storage precision control method based on micro-partitioned grids and three-dimensional interaction as described in any of the preceding claims.

[0008] In a fourth aspect, the present invention provides a non-transitory computer-readable storage medium storing instructions that, when executed, perform a grain warehouse precision control method based on micro-partitioned grid and three-dimensional interaction as described in any of the preceding claims.

[0009] In summary, one or more technical solutions provided in the embodiments of the present invention have at least the following technical effects or advantages: This invention enables each area within a grain silo to possess independent sensing, transport, and controllability capabilities by dividing the space into micro-regions and constructing a micro-regional unit system comprising grid pipe components, electrically controllable valves, and various types of sensors. Simultaneously, by performing digital twin modeling on the micro-regional unit system and constructing a 3D visualization model, the state data of each micro-regional unit is mapped to state expression variables, thereby achieving spatial representation and dynamic interactive display of abnormal states within the grain silo. Furthermore, based on these state expression variables, abnormal micro-regional units are identified, and combined with graph structure models, path cost parameters, and transport path optimization mechanisms, abnormal... The target transport path between the micro-zoning unit and the control equipment is dynamically optimized to ensure that the control medium is transported directionally along a target path with low resistance, low conflict, and low risk of mirror error control. Subsequently, selective opening and closing control is performed on the electrically controllable valves on the grid pipeline assembly, and real-time feedback adjustment is performed in combination with state expression variables to achieve dynamic adaptive adjustment of the transport direction, transport intensity, and control range of the control medium. This avoids the problems of low control accuracy, high control energy consumption, and simultaneous participation of irrelevant areas in control caused by using fixed paths or large-scale uniform transport in the prior art, and improves the precise control capability and control response efficiency of abnormal areas in the grain warehouse. Attached Figure Description

[0010] Figure 1 This is a flowchart illustrating a method for precise control of grain storage based on micro-partitioned grids and three-dimensional interaction, as disclosed in an embodiment of the present invention. Figure 2 This is a schematic diagram of a three-dimensional visualization model disclosed in an embodiment of the present invention; Figure 3 This is a schematic diagram of a graph structure model disclosed in an embodiment of the present invention; Figure 4 This is a schematic diagram of a grain warehouse precision control device based on micro-regional grid and three-dimensional interaction disclosed in an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of the present invention.

[0011] Explanation of reference numerals in the attached drawings: 401, acquisition module; 402, processing module; 403, output module; 501, processor; 502, communication bus; 503, user interface; 504, network interface; 505, memory. Detailed Implementation

[0012] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification 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.

[0013] In the description of the embodiments of the present invention, words such as "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "for example" or "for instance" in the embodiments of the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Rather, the use of words such as "for example" or "for instance" is intended to present the relevant concepts in a specific manner.

[0014] In the description of the embodiments of the present invention, the term "multiple" means two or more. For example, multiple systems means two or more systems, and multiple screen terminals means two or more screen terminals. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0015] While existing grain storage control technologies can perform basic monitoring and regulation of the internal environment of grain storage through sensors, ventilation equipment, and pesticide application equipment, they generally adopt fixed conveying structures and coarse-grained regional control modes. They lack dynamic path optimization mechanisms, three-dimensional interactive control mechanisms, and selective control mechanisms for conveying nodes based on the spatial structure of the grain storage. As a result, when abnormal areas occur, they cannot combine the abnormal location, pipeline connectivity, and control equipment status to generate the optimal conveying path. This can easily lead to problems such as synchronous control of unrelated areas, conveying conflicts, increased energy consumption, insufficient control accuracy, and low control response efficiency.

[0016] This invention discloses a method for precise grain warehouse control based on micro-regional grids and 3D interaction, which is applied to a server. The server includes, but is not limited to, electronic devices such as mobile phones, tablets, wearable devices, and PCs (Personal Computers), and can also be a backend server running the method for precise grain warehouse control based on micro-regional grids and 3D interaction. The server can be implemented using a standalone server or a server cluster composed of multiple servers.

[0017] This embodiment discloses a method for precise control of grain storage based on micro-partitioned grids and three-dimensional interaction, referring to... Figure 1 It includes the following steps: S110 involves dividing the grain storage space into micro-zone grids and constructing grid pipeline components integrating sensors and electrically controllable valves within each micro-zone unit, thereby forming a micro-zone unit system.

[0018] S120 performs digital twin modeling on the micro-partition unit system and constructs a three-dimensional visualization model, mapping the state data of each micro-partition unit into state expression variables.

[0019] S130 identifies abnormal micro-partition units based on state expression variables and constructs a graph structure model. At the same time, it introduces path cost parameters to optimize the transport path between abnormal micro-partition units and control equipment to generate path optimization results.

[0020] S140, based on the path optimization results, performs selective opening and closing control on the electrically controllable valves on the grid pipeline assembly, and performs feedback adjustment in combination with state expression variables to complete directional regulation.

[0021] S150, after completing the directional control, performs cleaning on the grid pipeline components corresponding to the path optimization results and determines the cleaning completion status based on the cleanliness detection parameters.

[0022] In grain silos with a regularized micro-regional grid structure, due to the strong geometric symmetry of each micro-regional unit in terms of spatial layout, pipeline connection method, and node distribution, when multiple abnormal micro-regional units are located in mirror-symmetric positions, the graph structure in the digital twin model easily identifies different regions as structurally equivalent paths with the same connection characteristics and the same transport cost. However, existing path optimization algorithms typically rely mainly on geometric distance, basic connectivity, or static path length to perform path search, lacking the ability to dynamically model local pipeline resistance differences, valve aging status, grain pile compaction degree, directional flow capacity, and historical transport status. This results in the system generating the shortest path or lowest-cost path at the logical level, but the actual path corresponding to that path is not... The physical transport direction may have shifted due to changes in the internal state of the pipeline. In this case, the control medium will flow along a mirror branch with lower actual resistance or faster valve response, causing the air supply, dehumidification, or pesticide application processes to erroneously act on adjacent normal areas, while the actual abnormal areas fail to be effectively controlled. Since this type of miscontrol phenomenon still appears as normal path connectivity, normal valve action, and normal equipment operation in the digital twin model, the system has difficulty in timely identifying that the actual transport target has shifted, thus forming a highly concealed mirror miscontrol problem. This type of deviation is more likely to accumulate and amplify, especially in scenarios such as regular matrix-type grain silos, large-scale symmetrical pipe network structures, and scenarios where long-term operation leads to gradual changes in local resistance.

[0023] In one possible implementation, the grain storage space is divided into micro-regions and grids, and grid pipe assemblies integrating sensors and electrically controllable valves are constructed within each micro-region unit to form a micro-region unit system. Specifically, this includes: establishing a three-dimensional spatial coordinate system for the grain storage space and generating micro-region units with micro-region unit identifiers; constructing grid pipe assemblies with directional connection characteristics within the micro-region units and forming local pipe topologies; configuring electrically controllable valves and multiple types of sensors at the boundary connection nodes of the grid pipe assemblies and generating valve information; performing directional flow calibration on the grid pipe assemblies and generating asymmetric resistance characteristics corresponding to the pipe connection directions; and constructing a micro-region unit system based on the micro-region unit identifiers, local pipe topologies, valve information, and asymmetric resistance characteristics.

[0024] Specifically, a fixed reference point is first selected within the grain silo space, and a three-dimensional spatial coordinate system is established using this reference point. The length, width, and height of the grain silo correspond to three coordinate directions in the three-dimensional spatial coordinate system, respectively. Subsequently, the boundaries of the grain silo's bottom, side walls, and top, the surface contour of the grain pile, and the effective storage height inside the grain pile are collected, and these spatial boundaries are uniformly mapped to the three-dimensional spatial coordinate system. The three-dimensional spatial coordinate system is a unified coordinate reference used to describe any spatial location within the grain silo. It enables subsequent micro-zoning units, grid pipe assemblies, electrically controllable valves, and various types of sensors to be positioned under the same spatial reference, avoiding inconsistencies in spatial location caused by different equipment using different coordinate references.

[0025] After establishing the three-dimensional spatial coordinate system, the grain storage space is divided into multiple micro-zone units with fixed spatial ranges by setting micro-zone division intervals according to the length, width, and height directions of the grain storage. Each micro-zone unit is the smallest spatial control object in the grain storage space that can be independently monitored, modeled, and controlled. Each micro-zone unit records its coordinate boundaries, center coordinates, grain pile depth level, and relationships with adjacent micro-zone units in the three-dimensional spatial coordinate system. To avoid misidentification of geometrically symmetrical positions in a regular matrix-type grain storage space as the same control object, a micro-zone unit identifier is generated for each micro-zone unit. This identifier is obtained by combining spatial coordinate encoding, grain pile depth level encoding, pipeline access direction encoding, and adjacency verification encoding, ensuring that each micro-zone unit remains uniquely identifiable at the data level, even when in a geometrically symmetrical position.

[0026] Within each micro-zone unit, the layout area of ​​the grid pipeline assembly is determined. Horizontal pipeline segments, longitudinal pipeline segments, and vertical branch pipe segments are laid out according to the coordinate boundaries of the micro-zone unit and the grain pile depth levels. The horizontal pipeline segments are used to transport the control medium along the length of the grain silo, the longitudinal pipeline segments are used to transport the control medium along the width of the grain silo, and the vertical branch pipe segments are used to deliver the control medium to different grain pile depth levels along the height of the grain silo. The grid pipeline assembly refers to a multi-directional interconnected pipeline structure formed according to the distribution of micro-zone units within the grain silo space. It can transport air supply media, dehumidification media, and pesticide application media, and can also transport cleaning media after pesticide application. The directional connection characteristics refer to the access direction, output direction, and adjacent connection direction of each pipeline segment in the grid pipeline assembly relative to the micro-zone unit, used to distinguish the actual flow capacity of the same micro-zone unit in different directions.

[0027] When forming a local pipeline topology within a micro-zone unit, horizontal pipeline segments, vertical pipeline segments, vertical branch segments, pipeline intersections, and pipeline boundary connection positions are recorded as topology objects, and the connection relationships between pipeline segments are recorded as topology connection relationships. Local pipeline topology refers to the pipeline connection structure within a micro-zone unit and at its boundaries, used to describe the directions from which the control medium can enter and exit within that micro-zone unit, and through which pipeline nodes it can be transported. For connection positions between adjacent micro-zone units, boundary connection nodes are set. Boundary connection nodes are pipeline nodes that enable pipeline connectivity and valve control between two adjacent micro-zone units. Boundary connection nodes allow the pipeline connection relationships between adjacent micro-zone units to be explicitly recorded, avoiding the need to determine connectivity solely based on geometric proximity during subsequent path optimization.

[0028] Electrically controlled valves are configured at boundary connection nodes, ensuring that each valve corresponds to a clearly defined boundary connection node, a clearly defined pipeline connection direction, and a clearly defined micro-zone unit identifier. An electrically controlled valve is a pipeline control component capable of opening, closing, or adjusting its opening degree according to control commands, used to control whether the controlled medium can enter or leave the micro-zone unit along the corresponding pipeline connection direction. During configuration, the installation position, installation orientation, valve opening range, response delay, rated flow capacity, and corresponding boundary connection node of the electrically controlled valves are recorded. This allows the subsequent system to accurately control the electrically controlled valves in the target pipeline connection direction based on path optimization results, preventing the erroneous opening of electrically controlled valves in the geometrically mirrored direction.

[0029] Multiple types of sensors are configured within the grid pipe assembly and micro-zone units. These sensors include at least temperature sensors, humidity sensors, insect infestation sensors, pressure sensors, and flow sensors. Temperature sensors collect data on temperature changes in the grain pile within the corresponding micro-zone unit; humidity sensors collect data on humidity changes in the grain pile within the corresponding micro-zone unit; insect infestation sensors collect data on insect density changes in the corresponding micro-zone unit; pressure sensors collect data on pressure changes in the grid pipe assembly along the corresponding pipe connection direction; and flow sensors collect data on flow rate changes in the grid pipe assembly along the corresponding pipe connection direction. Valve information refers to the binding information between the electrically controllable valve and its installation location, boundary connection nodes, pipe connection direction, control status, response capability, and aging status. This valve information ensures that the valve control action is consistent with the actual pipe connection direction.

[0030] After configuring the grid pipe assembly, electrically controllable valves, and various types of sensors, directional flow calibration is performed on the grid pipe assembly. Directional flow calibration involves selecting one pipe connection direction as the target pipe connection direction sequentially according to a preset test order under low-power testing conditions. The electrically controllable valve corresponding to the target pipe connection direction is opened, while the electrically controllable valves corresponding to the other non-test directions are closed. Then, the air supply equipment, dehumidification equipment, or pesticide application equipment is started to perform short-term low-power delivery. Pressure and flow responses in the target pipe connection direction are collected using pressure and flow sensors. This directional testing method allows for the determination of the true flow capacity of each pipe connection direction, rather than assuming that geometrically symmetrical directions have the same flow capacity.

[0031] After obtaining the pressure and flow responses for directional flow calibration, asymmetric resistance characteristics are generated by combining valve information from the electrically controllable valve, the grain pile compaction state, and the local settlement state. Asymmetric resistance characteristics refer to the actual transport resistance differences of the same micro-zone unit in different pipeline connection directions, used to characterize the degree of resistance encountered by the control medium when entering or leaving the micro-zone unit in different directions. Since the degree of grain pile compaction, pipeline bends, valve aging, and local settlement state can all lead to different resistances in geometrically symmetric pipelines, asymmetric resistance characteristics can provide real physical constraints for subsequent path optimization, reducing the probability of mirror-symmetric paths being mistakenly identified as equivalent paths.

[0032] The formula for calculating the asymmetric drag characteristic is: in, Let represent the asymmetric resistance characteristic of the i-th micro-partition unit in the d-th pipe connection direction, where i represents the index of the micro-partition unit and d represents the index of the pipe connection direction. This represents the pressure change value collected by the i-th micro-partition unit during directional flow calibration in the d-th pipe connection direction. The value is obtained based on the difference between the pressure sensor readings before and after the test. This represents the flow response value collected by the i-th micro-partition unit during directional flow calibration in the d-th pipe connection direction. The value is obtained based on the average stable flow rate of the flow sensor during the test period. This represents the stability correction factor, a small constant with a value greater than 0, used to prevent the calculation results from being abnormally amplified due to excessively small flow response values. This represents the weight of the valve aging coefficient on the asymmetric resistance characteristics, with a value ranging from 0 to 1. This represents the weight of the grain pile compaction coefficient on the asymmetric resistance characteristics, with a value ranging from 0 to 1. This represents the weight of the local settlement coefficient on the asymmetric drag characteristics, with a value ranging from 0 to 1. This represents the valve aging coefficient of the i-th micro-partition unit in the d-th pipe connection direction, with a value ranging from 0 to 1. This represents the grain pile compaction coefficient corresponding to the i-th micro-partition unit, with a value ranging from 0 to 1. The value represents the local settlement coefficient corresponding to the i-th micro-partition unit, ranging from 0 to 1. This formula represents the basic flow resistance through the proportional relationship between the pressure change value and the flow response value, and corrects the basic flow resistance through the valve aging coefficient, grain pile compaction coefficient, and local settlement coefficient, so that the geometrically symmetrical pipe connection directions in the same regular grid can form different asymmetric resistance characteristics.

[0033] When constructing the micro-zoning unit system, each micro-zoning unit is recorded as a unit object containing a micro-zoning unit identifier, a three-dimensional spatial coordinate range, relationships between adjacent micro-zoning units, local pipeline topology, boundary connection nodes, pipeline connection directions, valve information, information from multiple types of sensors, directional flow calibration results, and asymmetric resistance characteristics. Furthermore, the boundary connection nodes and pipeline connection directions between adjacent unit objects are linked. The micro-zoning unit system refers to a calculable, controllable, and traceable basic structure for grain storage spatial regulation, composed of all unit objects and their pipeline connection relationships. Subsequent digital twin modeling, anomaly identification, path optimization, valve control, and cleaning processes are all performed based on the micro-zoning unit system.

[0034] As an example, when the grain silo is 20m long, 15m wide, and has an effective grain storage height of 6m, multiple micro-zone units can be generated at intervals of 1×1m / 2×2m. Each micro-zone unit is assigned a micro-zone unit identifier consisting of spatial coordinate codes, grain pile depth level codes, pipeline access direction codes, and adjacency verification codes. Within each micro-zone unit, transverse pipeline sections, longitudinal pipeline sections, and vertical branch pipe sections are laid out. Electric controllable valves are installed at the boundary connection nodes between adjacent micro-zone units. Temperature sensors, humidity sensors, insect infestation sensors, pressure sensors, and flow sensors are also configured. After installation, directional flow calibration is performed sequentially for the east, west, south, north, and vertical pipeline connection directions. Pressure and flow responses in each direction are collected, and the corresponding asymmetric resistance characteristics are calculated. If a geometrically mirrored micro-partition unit has the same spatial layout as the target micro-partition unit, but its westward pipeline connection direction has significantly increased asymmetric resistance due to grain pile compaction, then the two will not be considered as completely equivalent paths during subsequent path optimization, thereby reducing the risk of the control medium being mistakenly sent to the mirror symmetric region.

[0035] In one possible implementation, digital twin modeling is performed on the micro-partitioned unit system and a three-dimensional visualization model is constructed. The state data of each micro-partitioned unit is mapped to state expression variables. Specifically, this includes: writing the structural parameters and directional transport parameters in the micro-partitioned unit system into the digital twin model; constructing three-dimensional voxel objects corresponding to the mesh pipe components based on the micro-partitioned units; performing time synchronization on the sensor data of each micro-partitioned unit and writing it into the corresponding three-dimensional voxel objects, and generating state expression variables in combination with asymmetric drag characteristics; generating mirror-resolved coefficients based on the state expression variables and asymmetric drag characteristics; and mapping the state expression variables, asymmetric drag characteristics, and mirror-resolved coefficients to the three-dimensional visualization model. In this case, directional flow calibration is performed on the mesh pipe components and asymmetric drag characteristics are generated.

[0036] Specifically, refer to Figure 2 The 3D visualization model uses the entire grain silo space as a digital twin object. Under a unified 3D spatial coordinate system, the interior of the grain silo is divided into multiple micro-zones corresponding to different grain pile depth levels, and each micro-zone is constructed as a corresponding 3D voxel object. These 3D voxel objects are spatially connected through mesh pipe components, with electrically controllable valves configured at the boundary connection nodes to allow the control medium to be transported between the micro-zones along different pipe connection directions. The model maps state expression variables through different grayscale levels, thereby displaying the degree of anomaly in each micro-zone unit in real time. Higher state expression variables correspond to deeper 3D voxel objects, indicating more pronounced temperature, humidity, pest infestation, or transport anomalies in the corresponding area. Simultaneously, asymmetric resistance characteristics are mapped to the pipeline lines. The width of the line indicates that the direction of pipe connection with greater resistance is displayed with a thicker line width, thus reflecting the difference in actual conveying capacity in different directions. For micro-zone units in geometric mirror relationship, the display is associated with the mirror relationship connection line and the mirror resolution coefficient. When the mirror resolution coefficient is low, it indicates that the corresponding micro-zone unit is easily misidentified as a structurally equivalent path by the path optimization algorithm, and the direction verification weight needs to be increased. In addition, the 3D visualization model also displays the location and real-time status of temperature sensor, humidity sensor, insect sensor, pressure sensor and flow sensor, and reflects the data synchronization cycle and model update cycle through real-time status timestamps, so that managers can intuitively observe the grain storage status, pipeline conveying status, valve control status and mirror mis-regulation risk status of each micro-zone unit inside the grain warehouse.

[0037] When writing the structural parameters of the micro-zone unit system into the digital twin model, a data object corresponding to the micro-zone unit identifier is first created for each micro-zone unit. Within this data object, the following parameters are written: 3D spatial coordinate range, center coordinates, grain pile depth level, relationship between adjacent micro-zone units, local pipeline topology, boundary connection nodes, grid pipeline component material, grid pipeline component length, grid pipeline component inner diameter, electric controllable valve position, electric controllable valve installation posture, and installation positions of various types of sensors. Structural parameters refer to fixed or slowly changing parameters that describe the spatial morphology, pipeline connection relationships, and equipment installation relationships of the micro-zone units. Their function is to establish a one-to-one correspondence between the virtual objects in the digital twin model and the micro-zone units, grid pipeline components, electric controllable valves, and various types of sensors in the real grain silo, avoiding the situation where the 3D visualization model only presents a geometric grid and fails to reflect the actual pipeline access direction and the actual valve control objects.

[0038] When incorporating the directional transport parameters of the micro-regional unit system into the digital twin model, the directional flow response data, asymmetric resistance characteristics, directional correlation weights, valve response delay, valve aging coefficient, pressure fluctuation coefficient, flow fluctuation coefficient, and historical control counts for each micro-regional unit are recorded according to the pipeline connection direction. Directional transport parameters refer to the flow capacity, resistance differences, and valve response characteristics exhibited when the control medium is transported along different pipeline connection directions. These parameters differ from structural parameters and can reflect the dynamic transport differences caused by grain pile compaction, pipeline dust accumulation, valve aging, and localized settlement after long-term operation. By incorporating both directional transport parameters and structural parameters into the digital twin model, the model gains both spatial geometric representation capabilities and the ability to represent the actual transport state, thus providing a basis for identifying differences between geometrically mirrored micro-regional units.

[0039] The directional correlation weight in the directional conveying parameters can be generated based on the asymmetric resistance characteristics, flow stability, and valve response stability. The expression for the directional correlation weight is: in, This represents the directional association weight of the i-th micro-partition unit in the d-th pipeline connection direction, with a value ranging from 0 to 1; This represents the directional correction factor for the d-th pipeline connection direction, with a value ranging from 0.5 to 1.5. It is used to correct for differences in basic transportation caused by different pipeline layout methods in different directions. This represents the asymmetric resistance characteristic of the i-th micro-partition unit in the d-th pipe connection direction; This represents the stability correction coefficient, with a value ranging from 0.001 to 0.01, used to prevent abnormal weight amplification when the resistance characteristic is close to 0; This represents the flow stability coefficient of the i-th micro-partition unit in the d-th pipe connection direction. The value ranges from 0 to 1, and the larger the value, the smaller the flow fluctuation. This represents the valve response stability coefficient of the i-th micro-partition unit in the d-th pipeline connection direction. The value ranges from 0 to 1, and the larger the value, the more stable the valve opening and closing response. and These represent the influence weights of the flow stability coefficient and the valve response stability coefficient on the directional association weights, respectively, with values ​​ranging from 0 to 1; M represents the number of pipe connection directions corresponding to the i-th micro-partition unit. This expression uses the reciprocal of the asymmetric resistance characteristic to represent directional passability, and increases the association weight of stable flow directions through the flow stability coefficient and the valve response stability coefficient. Then, denominator normalization ensures that the directional association weights of each pipe connection direction within the same micro-partition unit are within a comparable range, facilitating direct use in subsequent path optimization.

[0040] When constructing 3D voxel objects based on micro-partition units, each micro-partition unit is converted into a 3D voxel object with volume boundaries according to the 3D spatial coordinate range. Within each 3D voxel object, the micro-partition unit identifier, 3D spatial coordinate range, grain pile depth level, local pipeline topology, and relationships between adjacent 3D voxel objects are bound. A 3D voxel object is a 3D data unit in a digital twin model used to carry the spatial location, equipment status, and sensor status of a micro-partition unit. Unlike ordinary 3D model blocks, it not only displays spatial boundaries but also stores real-time status data, directional transport parameters, and subsequent control records for the corresponding micro-partition unit. Through 3D voxel objects, each visible unit in the 3D visualization model can be back-located to a micro-partition unit in the actual grain silo.

[0041] When constructing pipeline visualization objects corresponding to mesh pipeline components within a 3D voxel object, horizontal pipeline segments, vertical pipeline segments, vertical branch pipe segments, and boundary connection nodes are embedded into the 3D voxel object according to their actual installation positions. Each pipeline visualization object is then bound to its corresponding pipeline connection direction, pipeline length, pipeline inner diameter, asymmetric resistance characteristics, and directional association weight. Pipeline visualization objects are visualization elements in the 3D visualization model used to present the spatial orientation and transport status of mesh pipeline components. They can display different line widths, transparency, or texture densities based on asymmetric resistance characteristics, allowing managers to directly observe whether there is an abnormal increase in resistance in a particular pipeline connection direction. Boundary connection nodes are used to display the pipeline connection positions between adjacent 3D voxel objects and serve as the installation reference for electrically controllable valve visualization objects.

[0042] When synchronizing sensor data across micro-region units, temperature, humidity, insect infestation, pressure, flow rate, and valve status data are collected according to a unified sampling period, and timestamps are aligned for data from different sampling frequencies. For temperature and humidity data, lower-frequency smoothing sampling can be used; for pressure and flow rate data, higher-frequency sampling and window aggregation can be used; for insect infestation data, insect density can be generated according to a preset statistical period. Time synchronization refers to unifying data generated by different sensors at different sampling times into the same time window, ensuring that the same 3D voxel object has a complete set of state data at the same state update time point, avoiding anomaly identification bias caused by the mixing of state expression variables from data from different time slices.

[0043] After time synchronization, sensor data needs to be written to the corresponding 3D voxel object. During writing, the micro-partition unit identifier is used as an index. Temperature, humidity, insect infestation, pressure, flow rate, and valve status data are written to the state data cache of the same 3D voxel object, respectively. Missing data, abruptly changing data, and delayed upload data are corrected. For short-term missing data, state data from adjacent time windows can be used for smoothing compensation. For abruptly changing data, the state data from adjacent micro-partition units and directional flow response data can be combined to determine if it is a genuine anomaly. For delayed upload data, it is backfilled to the corresponding historical state window according to the timestamp. This writing method ensures that the data used when generating state expression variables maintains a binding relationship with the corresponding micro-partition unit and the corresponding 3D voxel object.

[0044] State representation variables are used to integrate the storage environment status, pest and disease status, pipeline transportation status, and valve control status into a unified variable that can be used for anomaly identification and visualization mapping. When generating state representation variables, temperature data, humidity data, pest data, pressure data, flow rate data, valve status data, and asymmetric resistance characteristics are first normalized separately. Then, corresponding influence weights are set according to grain type, storage stage, and control medium type, enabling the state representation variables to adapt to different grain varieties, different storage cycles, and different control objectives. For example, in high-temperature control scenarios, the temperature deviation coefficient and flow anomaly coefficient have higher influence weights; in pest control scenarios, the pest risk coefficient, residual risk coefficient, and valve status anomaly coefficient have higher influence weights.

[0045] The expression for the state variable is: in, The variable represents the state of the i-th micro-partition unit at time t, and its value ranges from 0 to 1. The larger the value, the more obvious the deviation of the micro-partition unit from the target grain storage state. The normalized temperature deviation coefficient of the i-th micro-partition unit at time t is represented. The value ranges from 0 to 1 and is generated based on the degree of deviation between the real-time temperature and the target temperature threshold range. The normalized humidity deviation coefficient of the i-th micro-partition unit at time t is represented. The value ranges from 0 to 1 and is generated based on the degree of deviation between the real-time humidity and the target humidity threshold range. The normalized pest risk coefficient of the i-th micro-partition unit at time t is represented. The value ranges from 0 to 1 and is generated based on the degree of deviation between the pest density and the pest warning threshold. This represents the normalized flow anomaly coefficient of the i-th micro-partition unit at time t, with a value ranging from 0 to 1, generated based on the attenuation degree of the current flow data relative to the directional flow calibration result; The normalized pressure anomaly coefficient of the i-th micro-partition unit at time t is represented. The value ranges from 0 to 1 and is generated based on the degree of deviation of the current pressure data from the directional flow calibration result. The normalized asymmetric resistance coefficient of the i-th micro-partition unit at time t is represented, with a value ranging from 0 to 1, and is generated based on the differences in asymmetric resistance characteristics in each pipe connection direction; The normalized valve state anomaly coefficient of the i-th micro-partition unit at time t is represented. The value ranges from 0 to 1 and is generated based on the valve opening error, valve response delay and valve aging coefficient. , , , , , and These represent the weights of the corresponding parameters on the state expression variables, with values ​​ranging from 0 to 1, and the sum of each weight is 1. This expression compresses different types of state data into the same comparable variable through a weighted fusion method, enabling the anomaly levels of different micro-partition units to be compared on the same scale, while preserving the asymmetric resistance characteristics and the impact of valve state anomalies on the risk of mirror miscontrol.

[0046] Each normalized deviation coefficient can be generated using a piecewise truncation method. Taking the temperature deviation coefficient as an example, its expression is: in, This represents the normalized temperature deviation coefficient of the i-th micro-partition unit at time t; This represents the real-time temperature value of the i-th micro-partition unit at time t; This represents the lower boundary of the target temperature threshold range corresponding to the i-th micro-partition unit; This represents the upper boundary of the target temperature threshold range corresponding to the i-th micro-partition unit; This indicates the allowable deviation from the scale on the low-temperature side, and a value of 2 to 5 is recommended. This indicates the allowable deviation scale for the high-temperature side, with a recommended value of 2 to 8. The expression sets the deviation coefficient to 0 when the real-time temperature is within the target temperature threshold range. When the real-time temperature is below the lower boundary or above the upper boundary, the deviation coefficient gradually increases according to the deviation magnitude, and an upper limit truncation is used to avoid distortion of the state expression variables due to extreme sensor anomalies. The humidity deviation coefficient, insect infestation risk coefficient, flow anomaly coefficient, and pressure anomaly coefficient can be generated using the same normalization approach, and the allowable deviation scale can be adjusted according to the corresponding threshold range.

[0047] When generating the mirror distinguishability coefficient based on state expression variables and asymmetric resistance characteristics, the first step is to find geometrically mirrored micro-partitions in the digital twin model that are geometrically mirrored to the target micro-partition. Then, the corresponding state expression variables, asymmetric resistance characteristics, directional correlation weights, valve state anomaly coefficients, and flow stability coefficients are read. The mirror distinguishability coefficient is a quantitative indicator used to determine whether two geometrically symmetric micro-partitions can be distinguished in terms of physical transport and storage states. A larger value indicates a more significant difference in the actual state between the two geometrically mirrored micro-partitions; a smaller value indicates that the two geometrically mirrored micro-partitions are easily mistaken for structurally equivalent objects by path optimization algorithms, requiring secondary verification.

[0048] The expression for the mirror resolution coefficient is: in, The mirror resolution coefficient between the i-th micro-partition unit and the g-th geometric mirror micro-partition unit at time t is represented, and its value ranges from 0 to 1. This represents the state expression variable of the i-th micro-partition unit at time t; This represents the state expression variable of the g-th geometric mirror micro-partition at time t; Indicates the number of pipe connection directions involved in the mirror comparison; This represents the normalized asymmetric drag coefficient of the i-th micro-partition unit in the d-th pipe connection direction; This represents the normalized asymmetric drag coefficient of the g-th geometric mirror micro-partition in the direction of the corresponding pipeline connection. This represents the directional association weight of the i-th micro-partition unit in the d-th pipeline connection direction; This represents the directional association weight of the g-th geometric mirror micro-partition unit in the direction of the corresponding pipeline connection in the mirror; Represents the normalized valve state anomaly coefficient of the i-th micro-partition unit at time t; Represents the normalized valve state anomaly coefficient of the g-th geometric mirror micro-partition unit at time t; , , and These represent the influence weights of state expression variable differences, asymmetric resistance differences, directional correlation weight differences, and valve state differences on the mirror discriminability coefficient, respectively. Each weight ranges from 0 to 1, and the sum of all influence weights is 1. This expression determines whether two geometrically mirrored micro-partition units should be considered different control objects by the digital twin model by simultaneously comparing differences in grain storage state, directional resistance, directional conveying capacity, and valve state, thereby reducing the probability of incorrect equivalence of mirrored paths during path optimization.

[0049] When mapping state expression variables in a 3D visualization model, these variables are mapped to the fill grayscale, boundary line type, or transparency of 3D voxel objects, with higher contrast displayed for 3D voxel objects exhibiting higher anomalies. When mapping asymmetric resistance features, the line width, line type density, or directional arrow density of pipeline visualization objects is associated with the asymmetric resistance features, making the blockage state of pipeline connections with higher resistance more obvious in the 3D visualization model. When mapping mirror resolution coefficients, the boundaries of 3D voxel objects with low mirror resolution coefficients are displayed as special line types, and mirror relationship connections are generated between geometric mirror micro-partitions, enabling managers to identify whether current anomalies pose a risk of mirror confusion. The mapping process is not simply displaying sensor values; rather, it binds state expression variables, asymmetric resistance features, and mirror resolution coefficients to 3D voxel objects, pipeline visualization objects, and mirror relationship objects, respectively, allowing the 3D visualization model to simultaneously express grain storage anomalies, pipeline transportation anomalies, and the risk of mirror mis-regulation.

[0050] When using artificial intelligence models to assist in generating state representation variables or mirror-recognizable coefficients, a lightweight spatiotemporal fusion model can be constructed, consisting of a temporal feature extraction module, a spatial neighborhood aggregation module, a directional transport encoding module, and a state mapping module. The temporal feature extraction module receives temperature, humidity, insect infestation, pressure, flow, and valve status data for the same micro-region within a continuous time window, and extracts state change trends through one-dimensional convolutional layers or gated recurrent units. The spatial neighborhood aggregation module receives state data from adjacent micro-regions and aggregates the diffusion effects of adjacent micro-regions on the target micro-region through graph convolutional layers. The directional transport encoding module receives asymmetric resistance features, directional correlation weights, and valve response delays, and generates directional transport representations through fully connected layers. The state mapping module concatenates the temporal features, spatial neighborhood features, and directional transport representations, and outputs predicted values ​​for the state representation variables and mirror-recognizable coefficients. The model is relevant to the application scenario of grain warehouses because abnormal temperature and humidity and abnormal insect infestation have the characteristics of time accumulation, and there is a diffusion effect between adjacent micro-zone units. Furthermore, whether the control medium can accurately reach the target abnormal micro-zone unit is affected by the directional transport parameters. Therefore, each module of the model corresponds to the three key factors of grain warehouse state evolution, neighborhood diffusion, and pipeline transport.

[0051] The training data for the lightweight spatiotemporal fusion model consists of historical operational data, directional flow calibration data, manually verified anomaly records, and post-regulation state change records. Each training sample includes a state data sequence of the target micro-region unit within a continuous time window, state data sequences of adjacent micro-region units, asymmetric resistance characteristics of the target micro-region unit, directional correlation weights, valve state data, and corresponding state expression variable labels and mirror resolvable coefficient labels. During training, the time window can be set to 10 to 60 sampling periods, the learning rate to 0.0001 to 0.001, the batch size to 16 to 128, and the number of training epochs to 50 to 300. The mean squared error loss function is used to constrain the prediction errors of the state expression variables and the mirror resolvable coefficients. For scenarios lacking manual labels, weak labels can be generated first using the above formula. Then, the model can be incrementally updated based on whether the state expression variables of the target micro-region unit decrease after actual regulation and whether the state expression variables of the geometric mirror micro-region unit change abnormally.

[0052] The loss function of the lightweight spatiotemporal fusion model can be expressed as: in, This represents the model training loss; N represents the number of training samples. This represents the predicted value of the state expression variable of the i-th micro-partition unit output by the model; The label represents the state expression variable label corresponding to the i-th micro-partition unit; This represents the predicted mirror resolution coefficient between the i-th micro-partition unit and the g-th geometric mirror micro-partition unit output by the model; This indicates the corresponding image resolution coefficient label; This represents the weighted average of the state expression variables of adjacent micro-partition units within the neighborhood of the i-th micro-partition unit; , and These represent the influence weights of the state representation variable error, the mirror resolvability coefficient error, and the neighborhood smoothing constraint, respectively, all ranging from 0 to 1. This expression uses the first two constraint terms to output labels for the state representation variable and the mirror resolvability coefficient, and the third constraint term to output labels that maintain spatial continuity with the state changes of adjacent micro-partition units, thus avoiding isolated abnormal flickering in the 3D visualization model caused by single-point sensor noise.

[0053] When directional flow calibration is used to generate asymmetric resistance characteristics, the first step is to set one pipe connection direction corresponding to the micro-zone unit to be calibrated as the target pipe connection direction. Then, the electrically controlled valves in the target pipe connection direction are opened, while the electrically controlled valves in other pipe connection directions are closed. The air supply or dehumidification equipment is then controlled to deliver air for a short period at low power. Subsequently, the pressure value before calibration and the stable delivery pressure value are collected using a pressure sensor, and the stable delivery flow rate value is collected using a flow sensor. The basic flow resistance is calculated based on the pressure change and flow response values, and then corrected using the valve aging coefficient, grain pile compaction coefficient, and local settlement coefficient. For each micro-zone unit, directional flow calibration must be performed at least once for each of the east, west, south, north, and vertical pipe connection directions to form a complete set of directional delivery parameters.

[0054] The expression for the asymmetric drag characteristic is: in, This represents the asymmetric resistance characteristic of the i-th micro-partition unit in the d-th pipe connection direction; This represents the pressure change value collected by the i-th micro-partition unit when performing directional flow calibration in the d-th pipe connection direction; This represents the flow response value collected by the i-th micro-partition unit when performing directional flow calibration in the d-th pipe connection direction; This represents the stability correction factor, with a value ranging from 0.001 to 0.01. This represents the valve aging coefficient of the i-th micro-partition unit in the d-th pipe connection direction, with a value ranging from 0 to 1; This represents the grain pile compaction coefficient corresponding to the i-th micro-partition unit, with a value ranging from 0 to 1; This represents the local settlement coefficient corresponding to the i-th micro-partition unit, with a value ranging from 0 to 1; , and These represent the influence weights of the valve aging coefficient, grain pile compaction coefficient, and local settlement coefficient on the asymmetric resistance characteristics, respectively, with values ​​ranging from 0 to 1. This expression first characterizes the basic flow resistance as the ratio between the pressure change value and the flow response value. Then, it uses the valve aging coefficient, grain pile compaction coefficient, and local settlement coefficient to amplify and correct the basic flow resistance, ensuring that the local physical differences that gradually appear during long-term operation are reflected in the asymmetric resistance characteristics.

[0055] As an example, in a grain silo with a length of 20m, a width of 15m, and an effective grain storage height of 6m, the space is divided into micro-partitions of 2m × 2m × 1.5m, and each micro-partition is constructed as a three-dimensional voxel object. For the numbered... The micro-regional unit maps its horizontal pipe segments, vertical pipe segments, and vertical branch pipe segments as pipe visualization objects, and incorporates the asymmetric resistance characteristics of the five pipe connection directions (east, west, south, north, and vertical) into the digital twin model. If the micro-regional unit has a temperature deviation coefficient of 0.70, a humidity deviation coefficient of 0.20, an insect infestation risk coefficient of 0.10, a flow anomaly coefficient of 0.40, a pressure anomaly coefficient of 0.35, an asymmetric resistance coefficient of 0.60, and a valve status anomaly coefficient of 0.30, and sets... , , , , , , If the state expression variable is 0.435, and the state expression variable of its geometrically mirrored micro-partition is 0.210, and the asymmetric resistance characteristics of the two differ significantly in the mirror direction, then the mirror discriminability coefficient increases. The 3D visualization model will display the micro-partition as having a higher degree of anomaly and provide a risk warning regarding its mirror relationship. Subsequent path optimization calls for the state expression variable and the mirror discriminability coefficient to distinguish the real anomaly region from the geometrically mirrored region, preventing the control medium from being incorrectly delivered to the mirror-symmetric region.

[0056] In one possible implementation, abnormal micro-partition units are identified based on state expression variables, and a graph structure model is constructed. Simultaneously, path cost parameters are introduced to optimize the transport path between the abnormal micro-partition units and the control equipment to generate path optimization results. Specifically, this includes: identifying abnormal micro-partition units based on state expression variables and mirror discriminability coefficients, and generating a set of abnormal micro-partition units; performing mirror target verification on the abnormal micro-partition units and geometrically mirrored micro-partition units to reconfirm the true anomaly location; constructing a graph structure model containing directed graph edges based on the connection relationships in the micro-partition unit system, wherein each micro-partition unit, each boundary connection node, each electrically controllable valve, and each control equipment in the micro-partition unit system is mapped as a graph node, and the grid pipeline component connection relationships between adjacent micro-partition units, the access relationships between control equipment and grid pipeline components, and the control relationships between boundary connection nodes and electrically controllable valves are mapped as directed graph edges; filtering available control equipment nodes corresponding to the abnormal micro-partition units based on the anomaly type; calculating comprehensive path cost parameters for the directed graph edges in the candidate transport paths; and performing constraint optimization on the candidate transport paths based on the comprehensive path cost parameters to generate path optimization results.

[0057] Specifically, the state expression variables, mirror distinguishability coefficients, asymmetric resistance characteristics, directional correlation weights, and three-dimensional voxel object positions corresponding to each micro-region unit in the digital twin model are first read, and then the state expression variables are compared with the anomaly judgment threshold. The state expression variables are comprehensive state quantities that integrate temperature deviation, humidity deviation, insect infestation risk, pressure anomaly, flow anomaly, asymmetric resistance anomaly, and valve status anomaly, used to represent the degree to which the micro-region unit deviates from the target grain storage state; the mirror distinguishability coefficient is an indicator used to measure whether the target micro-region unit and the geometrically mirrored micro-region unit can be reliably distinguished, used to prevent geometrically symmetrical positions from being misjudged as the same control object. When the state expression variable of a micro-region unit exceeds the anomaly judgment threshold and its mirror distinguishability coefficient meets the minimum distinguishability condition, the micro-region unit is identified as an abnormal micro-region unit; when the state expression variable exceeds the anomaly judgment threshold but the mirror distinguishability coefficient is lower than the minimum distinguishability condition, the abnormal micro-region unit is not directly identified, but rather marked as an abnormal micro-region unit to be reviewed, so that the mirror target review can be performed later.

[0058] The anomaly intensity of an anomaly micro-partition can be generated jointly by the state expression variable and each normalized anomaly component. The formula for calculating the anomaly intensity is: in, This represents the anomaly intensity of the i-th micro-partition unit, with a value ranging from 0 to 1. The larger the value, the higher the degree of anomaly. The variable representing the state of the i-th micro-partition unit takes values ​​from 0 to 1. The normalized temperature deviation coefficient of the i-th micro-partition unit is represented, and its value ranges from 0 to 1. The normalized humidity deviation coefficient of the i-th micro-partition unit is represented, and its value ranges from 0 to 1. The normalized insect infestation risk coefficient represents the i-th micro-partition unit, and its value ranges from 0 to 1; The normalized flow anomaly coefficient of the i-th micro-partition unit is represented, and its value ranges from 0 to 1. The normalized pressure anomaly coefficient of the i-th micro-partition unit is represented, and its value ranges from 0 to 1. The normalized asymmetric drag coefficient of the i-th micro-partition unit is represented, and its value ranges from 0 to 1. The normalized valve state anomaly coefficient represents the i-th micro-partition unit, and its value ranges from 0 to 1. , , , , , , and These represent the weights of the corresponding parameters on the anomaly intensity, with values ​​ranging from 0 to 1, and the sum of each weight is 1. This formula uses a weighted fusion method to uniformly convert anomalies in grain storage environment, pest infestation, pipeline transportation, and valve control into anomaly intensity, avoiding the need to determine the anomaly micro-zone unit based solely on a single indicator such as temperature or humidity.

[0059] The anomaly detection threshold can be dynamically set based on grain type, storage stage, and historical stable state. When the grain silo is in a long-term safe storage stage, the anomaly detection threshold can be set to 0.35 to 0.55 to identify risks such as slow temperature rise, high humidity accumulation, or early pest infestation in advance. When the grain silo is in a short-term turnover storage stage, the anomaly detection threshold can be set to 0.50 to 0.70 to reduce frequent start-ups and shutdowns of control equipment. The minimum resolvable condition corresponding to the mirror resolvable coefficient threshold can be set to 0.15 to 0.35. When the mirror resolvable coefficient is below this range, it indicates that the target micro-partition unit and the geometric mirror micro-partition unit are not sufficiently different in terms of state expression variables, directional conveying state, or valve state, and are easily identified as structurally equivalent paths by the path optimization algorithm, requiring mirror target verification. The above threshold settings are not fixed empirical values, but are adjusted based on the grain silo safety level, grain type sensitivity, control medium risk level, and historical miscontrol records, so that anomaly identification can maintain sensitivity while reducing the probability of mirror misidentification.

[0060] When performing mirror target verification on abnormal micro-zone units and geometrically mirrored micro-zone units, the corresponding geometrically mirrored micro-zone units are first determined in the three-dimensional spatial coordinate system based on the grain silo center plane, the length direction symmetry plane, the width direction symmetry plane, or the height hierarchy symmetry relationship. Simultaneously, the state expression variables, asymmetric resistance characteristics, directional flow response data, valve information, and state expression variables of adjacent micro-zone units are read. Geometrically mirrored micro-zone units refer to micro-zone units in a regular matrix-type grain silo that are symmetrically positioned with the abnormal micro-zone unit. They may have the same connection form in terms of geometric structure, but may differ in actual pipeline resistance, valve aging degree, and grain pile compaction degree. The verification process not only compares the spatial positions of the two units but also compares their pressure response, flow response, and asymmetric resistance characteristics in the corresponding mirrored pipeline connection directions, enabling the true abnormal location to be distinguished from the geometrically symmetrical location.

[0061] Mirroring the target verification can generate the target confidence score. The formula for calculating the target confidence score is: in, This represents the target confidence level of the i-th anomalous micro-partition unit. The value range can be normalized to 0 to 1. The larger the value, the more likely the i-th anomalous micro-partition unit is to be a real anomalous location. This represents the anomalous intensity of the i-th anomalous micro-partition unit; The mirror resolution coefficient between the i-th anomalous micro-partition unit and the g-th geometric mirror micro-partition unit is represented; M represents the number of pipe connection directions involved in the mirror comparison. This represents the normalized asymmetric resistance coefficient of the i-th anomalous micro-partition unit in the d-th pipe connection direction; This represents the normalized asymmetric drag coefficient of the g-th geometric mirror micro-partition in the direction of the corresponding pipeline connection. The normalized directional flow response coefficient of the i-th abnormal micro-partition unit in the d-th pipe connection direction is represented, and its value ranges from 0 to 1. The normalized directional flow response coefficient of the g-th geometric mirror micro-partition unit in the direction of the mirrored pipe connection is represented, and its value ranges from 0 to 1. The state synchronization change coefficient between the i-th abnormal micro-partition unit and the g-th geometric mirror micro-partition unit is represented by a value ranging from 0 to 1. The larger the value, the more similar the state changes of the two units. , , , and These represent the weights of the corresponding parameters on the target confidence level, with values ​​ranging from 0 to 1. This formula enhances the contributions of anomaly intensity, mirror discernibility difference, asymmetric drag difference, and directional flow response difference to the target confidence level, and suppresses the misidentification of geometric mirror regions as true anomaly locations through the state synchronization change coefficient, thereby achieving reconfirmation of the true anomaly location.

[0062] When the target confidence level is higher than the target confirmation threshold, the anomalous micro-region is confirmed as a true anomaly location, and its micro-region identifier, anomaly type, anomaly intensity, target confidence level, and geometrically mirrored micro-region identifier are written into the anomalous micro-region set. When the target confidence level is lower than the target confirmation threshold, a short-term verification flow is performed. This involves short-term air supply or short-term exhaust of the grid pipe components corresponding to the direction of the anomalous micro-region under low power conditions, and observing whether the pressure and flow responses of the anomalous micro-region and the geometrically mirrored micro-region conform to the target pipe connection direction. The target confirmation threshold can be set to 0.55 to 0.80. When the risk of the controlled medium is high, the target confirmation threshold should be set to 0.70 to 0.80. When only air supply control is performed, the target confirmation threshold can be set to 0.55 to 0.70 to improve the control response speed. In this way, all objects in the anomalous micro-region set have true anomaly location attributes verified by mirroring, and subsequent graph structure modeling and path optimization solutions no longer directly rely on geometric coordinate results.

[0063] Reference Figure 3 When constructing the graph structure model, each micro-region unit in the micro-region unit system is mapped to a micro-region unit graph node, each boundary connection node is mapped to a boundary connection graph node, each electrically controllable valve is mapped to a valve graph node, and air supply equipment, dehumidification equipment, chemical application equipment, and cleaning equipment are mapped to control equipment graph nodes. Graph nodes are the basic objects in the graph structure model that can be retrieved and connected by path optimization algorithms. Different types of graph nodes carry different attributes: micro-region unit graph nodes carry state expression variables and anomaly intensity; boundary connection graph nodes carry pipeline connection directions and adjacency relationships; valve graph nodes carry valve information and valve status; and control equipment graph nodes carry equipment type, equipment load, output capacity, and current occupancy status. By mapping real physical objects to graph nodes, path optimization algorithms can simultaneously access spatial location, pipeline connections, valve controllability, and equipment capabilities.

[0064] When mapping connectivity to directed graph edges, the connectivity between adjacent micro-partition units and their corresponding grid pipe components is mapped to directed graph edges between micro-partition unit graph nodes and boundary connection graph nodes. Similarly, the access relationship between control equipment and grid pipe components is mapped to directed graph edges between control equipment graph nodes and boundary connection graph nodes. Furthermore, the control relationship between boundary connection nodes and electrically controllable valves is mapped to directed graph edges between boundary connection graph nodes and valve graph nodes. A directed graph edge is a graph connection object with a defined delivery direction. It not only represents a connection between two graph nodes but also indicates that the control medium can flow according to a specific pipe connection direction. The directed graph edge records pipe length, pipe connection direction, asymmetric resistance characteristics, direction association weight, valve opening state, valve response delay, cleaning state, residual risk, and pipe occupancy state, enabling the graph structure model to reflect the reachable path and delivery cost of the control medium within the actual grid pipe components.

[0065] The graph structure model can be represented as: Wherein, G represents the graph structure model; V represents the set of graph nodes, including at least micro-partition unit graph nodes, boundary connection graph nodes, valve graph nodes, and control device graph nodes; E represents the set of directed graph edges, including at least directed graph edges corresponding to the connection relationships of mesh pipeline components, the connection relationships of control devices, and the control relationships of valves; A represents the adjacency matrix, used to record whether there are directed connections between graph nodes; and W represents the edge weight matrix, used to record the comprehensive cost, asymmetric resistance characteristics, directional association weights, and valve switching costs corresponding to the directed graph edges. This expression transforms the micro-partition unit system into a graph structure that can be retrieved by the algorithm, enabling the path optimization process to perform constrained search between graph nodes and directed graph edges.

[0066] The elements in the adjacency matrix can be represented as: in, This represents the connection state between graph node u and graph node v in the adjacency matrix; u and v represent any two graph nodes in the graph structure model. This expression records traversable connections in a binary manner. When the corresponding electrically controllable valve fails, pipeline cleaning is incomplete, or the pipeline is occupied by other control tasks, even if a physical connection exists, the corresponding adjacency matrix element will be set to 0, so that the path optimization algorithm will not select an unavailable transport path.

[0067] When filtering available controllable device nodes based on anomaly type, the anomaly type corresponding to each anomaly micro-zone unit in the anomaly micro-zone unit set is first read. Anomaly types can include high temperature anomaly, high humidity anomaly, condensation risk anomaly, insect infestation anomaly, mold risk anomaly, and flow anomaly. Then, the device type and control medium type of the controllable device node are matched according to the anomaly type. For example, high temperature anomaly matches air supply device nodes, high humidity anomaly matches dehumidification device nodes, insect infestation anomaly matches pesticide application device nodes, mold risk anomaly can match both air supply device nodes and dehumidification device nodes, and flow anomaly matches cleaning device nodes or short-term verification flow device nodes. Available controllable device nodes refer to controllable device nodes whose device type corresponds to the anomaly type, whose equipment is in an operational state, whose equipment load does not exceed the limit, and whose access path is not occupied.

[0068] The selection criteria for available controllable device nodes can be expressed as follows: in, This indicates the availability status of the r-th control device node relative to the i-th abnormal micro-partition unit, and its value is 0 or 1; This indicates a conditional indicator function, which takes the value 1 if the condition within the parentheses is true, and 0 otherwise. This indicates the device control type of the r-th control device node; This indicates the required control type for the anomaly type corresponding to the i-th abnormal micro-partition unit; This represents the current load level of the r-th control device node, with a value ranging from 0 to 1; This indicates the maximum allowable device load; a value between 0.70 and 0.90 is recommended. This indicates the occupancy status of the r-th control device node, where 0 indicates unoccupied and 1 indicates occupied; This represents the running status of the r-th control device node, where 1 indicates it is operational and 0 indicates it is not. This expression filters available control device nodes by combining device type matching, load limits, occupancy status, and running status, avoiding path optimization results connecting to unavailable or incompatible control device nodes.

[0069] After identifying available controllable equipment nodes, candidate transport paths are generated in the graph structure model, starting from these nodes and ending at the corresponding micro-partition unit graph nodes. A candidate transport path is a reachable path composed of multiple graph nodes and directed graph edges connected in directional relationships. It represents the possible route for the control medium to be transported from the controllable equipment node through the grid pipeline assembly, electrically controllable valves, and boundary connection nodes to the abnormal micro-partition unit. When generating candidate transport paths, directed graph edges with valve malfunctions, pipeline occupancy, incomplete cleaning, excessively high residual risk, mirror confusion risk exceeding limits, and excessively low directional association weights need to be removed to ensure that the candidate transport paths meet the basic feasibility conditions before entering cost calculation.

[0070] When calculating the comprehensive path cost parameter for directed graph edges in candidate transport paths, the local cost of each directed graph edge is calculated first, and then the costs of all directed graph edges in the candidate transport paths are summed. The comprehensive path cost parameter is a comprehensive indicator used to evaluate whether a candidate transport path is suitable for directional control. The smaller the value, the better the candidate transport path is in terms of transport distance, flow resistance, valve switching, mirror confusion, equipment load, residual risk, and cleaning status. Unlike using only the shortest distance, the comprehensive path cost parameter incorporates the actual physical transport capacity, control execution reliability, and contamination residual risk into path optimization, making the controlled medium more likely to be transported along a truly low-resistance, low-risk, and low-conflict path.

[0071] The formula for calculating the comprehensive path cost parameter is as follows: in, This represents the comprehensive path cost parameter of the k-th candidate transport path; the smaller the value, the better the candidate transport path. Let represent the k-th candidate transport path; e represents the directed graph edge in the candidate transport path. This represents the normalized pipe length of edge e in the directed graph, with a value ranging from 0 to 1; This represents the normalized asymmetric resistance characteristic of edge e in a directed graph, with a value ranging from 0 to 1. This represents the directional association weight of edge e in the directed graph, with a value ranging from 0 to 1; This represents the normalized valve switching cost of edge e in the directed graph, with a value ranging from 0 to 1. This represents the normalized valve response delay of edge e in the directed graph, with a value ranging from 0 to 1. This represents the normalized control medium residual risk of edge e in the directed graph, with a value ranging from 0 to 1. This represents the normalized mirror confusion risk of edge e in the directed graph, with a value ranging from 0 to 1. This represents the normalized device load level of edge e in the directed graph, with a value ranging from 0 to 1. This represents the normalized pipe occupancy level of edge e in a directed graph, with a value ranging from 0 to 1. This represents the normalized clean state risk of edge e in the directed graph, with a value ranging from 0 to 1. This represents the normalized directional reachability coefficient of the k-th candidate transport path, with a value ranging from 0 to 1; , , , , , , , , , and These represent the weights of the corresponding parameters on the overall path cost parameter, with values ​​ranging from 0 to 1. This formula obtains the total path cost by accumulating the local risks and local costs of all directed graph edges in the candidate transport path, and reduces the total cost of the truly reachable path by using a directional reachability coefficient, thus prioritizing candidate transport paths with low resistance, stability, low mirror confusion, and low residual risk.

[0072] The weights of each influence can be dynamically set according to the type of control medium. When the control medium is the supply air medium, , and It can be set to a higher value, such as 0.10 to 0.25, to emphasize path length, flow resistance, and directional accessibility; when the control medium is a dehumidifying medium, , and It can be set to a higher value, such as 0.12 to 0.25, to emphasize pipeline resistance, valve response, and cleaning status; when the control medium is a chemical injection medium, , and It can be set to a higher value, such as 0.15 to 0.30, to reduce the risks of pesticide residue, accidental application, and insufficient cleaning. The weighting is based on the different sources of risk for different control media: the air supply process mainly focuses on delivery efficiency, the dehumidification process mainly focuses on dehumidification smoothness and condensation risk, and the pesticide application process mainly focuses on target accuracy and residue safety.

[0073] When performing constrained optimization based on the comprehensive path cost parameter, unusable directed graph edges in the graph structure model are first deleted or set to an impassable state. Then, a minimum cost search is performed based on candidate transport paths between available control equipment nodes and anomalous micro-region units. Constrained optimization refers to finding the candidate transport path with the minimum comprehensive path cost parameter under the premise of satisfying constraints such as valve availability, pipeline accessibility, equipment availability, acceptable mirror confusion risk, acceptable residual risk, and no pipeline occupation conflict. For a single anomalous micro-region unit, a constrained Dijkstra's search or A* search can be used; for multiple anomalous micro-region units, a control priority can be generated first based on the anomalous intensity and diffusion risk, and then the candidate transport path with the minimum conflict can be assigned one by one.

[0074] The objective of path optimization can be expressed as: in, This represents the final path optimization result; This represents the set of candidate transport paths corresponding to the i-th abnormal micro-partition unit; This represents the comprehensive path cost parameter for the k-th candidate transport path; This indicates the risk of confusion in the normalized mirror image of the directed graph edge e; This indicates the maximum permissible risk of image obfuscation; a value of 0.25 to 0.45 is recommended. This indicates the risk of residual medium in the normalized control of edge e in the directed graph; This indicates the upper limit of the allowable residual risk of the control medium, and a value of 0.20 to 0.50 is recommended, with a lower value preferred in application scenarios; This indicates the occupancy status of the pipe on edge e in the directed graph, with 0 indicating that it is not occupied; This indicates the availability status of valve graph nodes in the candidate delivery path; 1 indicates availability. This indicates the availability status of the control device node relative to the abnormal micro-partition unit, with 1 indicating availability. The objective function above selects the optimal candidate transport path by minimizing the comprehensive path cost parameter, and eliminates candidate transport paths with excessive mirror confusion risk, excessive residual risk, pipeline occupancy, valve unavailability, and equipment unavailability through constraints.

[0075] When multiple abnormal micro-partition units exist, the control priority can be calculated first, and then the execution path can be allocated according to the control priority from high to low. The formula for calculating the control priority is:

[0076] in, This indicates the control priority of the i-th abnormal micro-partition unit; the larger the value, the higher the priority of control. This represents the anomalous intensity of the i-th anomalous micro-partition unit; represents the normalized anomaly duration of the i-th anomaly micro-partition unit, with a value ranging from 0 to 1; This represents the normalized anomaly propagation risk of the i-th anomalous micro-partition unit, with a value ranging from 0 to 1; This represents the normalized clustering degree of abnormal micro-partition units within the neighborhood of the i-th abnormal micro-partition unit, with a value ranging from 0 to 1; This represents the normalized image obfuscation risk of the i-th anomalous micro-partition unit, with a value ranging from 0 to 1; , , , and These represent the weights of the corresponding parameters on the control priority, with values ​​ranging from 0 to 1, and the sum of each weight is 1. This formula increases the priority of anomalous micro-partitions with high anomaly intensity, long duration, high diffusion risk, and high neighborhood clustering, while decreasing the priority of anomalous micro-partitions with high mirror confusion risk. This allows path optimization in multi-anomaly scenarios to prioritize high-risk areas while avoiding premature application of high-risk control media to uncertain mirror areas.

[0077] When generating path optimization results, the optimal candidate delivery path is converted into a path control object. This path control object records the target control equipment node, target abnormal micro-zone unit, target pipeline node sequence, target electrically controlled valve sequence, target pipeline connection direction, comprehensive path cost parameters, mirror confusion risk, expected control medium delivery intensity, expected control duration, and backup candidate delivery paths. The path control object serves as the direct control basis for the subsequent selective opening and closing of electrically controlled valves. It records not only which graph nodes and directed graph edges the path passes through, but also the opening sequence of each target electrically controlled valve and the corresponding pipeline connection direction. By writing the path optimization results back to the 3D visualization model in the digital twin model, managers can see the target delivery path of the control medium from the control equipment node to the abnormal micro-zone unit in the 3D interface and can identify whether the path passes through grid pipeline components with low mirror confusion risk and low residual risk.

[0078] As an example, in a grain silo with a length of 20m, a width of 15m, and an effective grain storage height of 6m, micro-zoning units are used. The state expression variable is 0.68, the anomaly detection threshold is 0.50, the mirror resolution coefficient is 0.42, and the minimum resolution condition is 0.30. Therefore, the micro-partition unit is identified as an abnormal micro-partition unit. After verification with the mirror target, its geometrically mirrored micro-partition unit... In comparison, the asymmetric resistance characteristics and directional flow response data of the micro-regional unit in the eastward pipeline connection direction show significant differences, with a target confidence level of 0.76, which is higher than the target confirmation threshold of 0.65. Therefore, it is confirmed. This represents the actual location of the anomaly. If the anomaly type is a high-temperature anomaly, then the air supply equipment node is selected as an available controllable equipment node, and multiple paths from the air supply equipment node to the controllable equipment node are generated in the graph structure model. The system calculates the comprehensive path cost parameters for each candidate transport path. One path, although shorter in length, has a higher risk of mirror confusion and asymmetric resistance, resulting in a comprehensive path cost parameter of 0.61. Another path, slightly longer, has a higher directional correlation weight and a lower risk of mirror confusion, resulting in a comprehensive path cost parameter of 0.43. The latter path is ultimately selected as the path optimization result, and a path control object is generated, containing the target air supply equipment node, the target pipeline node sequence, the target electrically controllable valve sequence, and the target pipeline connection direction.

[0079] In one possible implementation, selective opening and closing control is performed on the electrically controllable valves on the grid pipeline assembly based on the path optimization results, and feedback adjustment is performed in conjunction with state expression variables to complete directional regulation. Specifically, this includes: extracting target control device nodes, target abnormal micro-partition units, target pipeline node sequences, target electrically controllable valve sequences, target pipeline connection directions, and expected control medium delivery intensity based on the path optimization results; generating a corresponding valve control topology based on the target pipeline node sequences, ensuring that each target electrically controllable valve in the valve control topology maintains a mapping relationship with its corresponding target pipeline connection direction and target abnormal micro-partition unit; and controlling the target electrically controllable valves... Each target electrically controllable valve in the valve sequence undergoes pre-verification of opening and checks the consistency of the target pipeline connection direction; the target electrically controllable valves are controlled in stages according to the target pipeline node sequence, and non-target electrically controllable valves are locked; the target control equipment node is activated and the expected control medium delivery intensity is dynamically adjusted based on the state expression variables; the target abnormal micro-zone unit and adjacent micro-zone units are monitored for state expression variables and the risk of mirror miscontrol is checked. When there is a risk of mirror miscontrol, the graph structure model and path optimization results are called again to perform secondary path optimization; after the target abnormal micro-zone unit recovers to the target grain storage state threshold range, reverse stage closure control is performed.

[0080] Specifically, the target control device node, target anomaly micro-zone unit, target pipeline node sequence, target electrically controllable valve sequence, target pipeline connection direction, and expected controllable medium delivery intensity are first read from the path control object. The target control device node refers to the node corresponding to the control equipment selected in the graph structure model for outputting air supply medium, dehumidification medium, or chemical injection medium. The target anomaly micro-zone unit refers to the micro-zone unit confirmed to require control after state expression variable identification and mirror target verification. The target pipeline node sequence refers to the arrangement order of boundary connection nodes and pipeline intersection nodes traversed by the controllable medium from the target control device node to the target anomaly micro-zone unit. The target electrically controllable valve sequence refers to the arrangement order of electrically controllable valves corresponding to the target pipeline node sequence that need to be opened or maintained at a specific opening degree. The target pipeline connection direction refers to the actual flow direction of the controllable medium within each segment of the grid pipeline assembly. The expected controllable medium delivery intensity refers to the output intensity of the control equipment pre-determined by the path optimization results based on the anomaly type, anomaly intensity, and path delivery capacity, used as the initial value for subsequent dynamic adjustment.

[0081] After reading the path optimization results, the system first verifies whether the equipment type of the target control device node matches the anomaly type of the target anomaly micro-partition unit. For example, high temperature anomaly corresponds to air supply device node, high humidity anomaly corresponds to dehumidification device node, and insect infestation anomaly corresponds to pesticide application device node. Next, it verifies whether the target pipeline node sequence is continuously connected in the graph structure model and confirms that each target electrically controllable valve in the target electrically controllable valve sequence is located at the boundary connection node between the corresponding target pipeline nodes. This verification process ensures that subsequent valve control actions do not deviate from the path optimization results and avoids errors in writing the path control object, which could cause the control medium to output from the wrong control device node or enter a non-target pipeline connection direction.

[0082] When generating the valve control topology based on the target pipeline node sequence, each target pipeline node in the sequence is used as a control node of the valve control topology. The connection direction between adjacent target pipeline nodes is used as the control edge of the valve control topology, and the corresponding target electrically controllable valve is bound to the control edge. The valve control topology refers to the local control network used to execute valve opening and closing control. It is derived from the path optimization results, but it is more execution-oriented than the graph structure model. It mainly records which target electrically controllable valves need to be opened, in what order they are opened, to what opening degree they are opened, and which non-target electrically controllable valves need to be closed and locked.

[0083] When forming the valve control topology, each target electrically controlled valve needs to be bound to the corresponding target pipeline connection direction, target abnormal micro-zone unit, target control device node, and target pipeline node sequence position. The mapping relationship ensures that the opening action of a target electrically controlled valve can be traced back to a specific control purpose, i.e., whether the target electrically controlled valve is intended to allow the control medium to enter which target abnormal micro-zone unit along which target pipeline connection direction. For grain silos with similar valve layouts in geometrically mirrored positions, the mapping relationship is also used to distinguish between target electrically controlled valves and mirrored electrically controlled valves, preventing the control system from incorrectly opening electrically controlled valves in the mirrored direction based solely on geometric similarity.

[0084] When performing pre-opening verification on the target electrically controllable valve sequence, the valve information of each target electrically controllable valve is read one by one. The valve information includes the valve installation position, boundary connection node identifier, valve physical installation posture, current valve opening degree, valve response delay, valve aging status, most recent opening and closing record, and directional flow response data. Pre-opening verification refers to confirming whether the target electrically controllable valve can reliably operate according to the target pipeline connection direction before officially outputting the control medium. This avoids directly activating the target control equipment node in cases of valve jamming, abnormal valve feedback, incorrect installation direction record, or excessive response delay.

[0085] When verifying the consistency of the target pipeline connection direction, the actual installation direction in the valve information, the actual flow direction in the directional flow response data, and the target pipeline connection direction in the path optimization results are compared. If the three are consistent, the corresponding target electrically controllable valve is marked as an executable valve. If the valve's physical installation orientation is inconsistent with the target pipeline connection direction, or if the directional flow response data indicates that the actual flow direction of the control medium deviates from the target pipeline connection direction, the target electrically controllable valve is marked as a directionally abnormal valve, and the execution of the current valve control topology is suspended. For directionally abnormal valves, low-power flow verification can be performed first, or the graph structure model can be called to regenerate an alternative target pipeline node sequence, thereby preventing the control medium from entering the geometrically mirrored branch or adjacent normal micro-partition units.

[0086] After successful pre-verification, the target electrically controlled valves are opened in stages according to the connection sequence of the target pipeline nodes. Staged opening control means not opening all target electrically controlled valves at once, but opening them segment by segment from the target control equipment node to the target abnormal micro-zone unit, allowing the control medium to form a stable, directional flow within the grid pipeline assembly. First, the target electrically controlled valves closest to the target control equipment node are opened, allowing the control medium to enter the initial segment of the target path. After the pressure and flow sensors confirm stable flow in the initial segment, the next segment of the target electrically controlled valves is opened, until the final target electrically controlled valve near the target abnormal micro-zone unit is opened. This method reduces instantaneous pressure surges and lowers the risk of the control medium entering mirror branches.

[0087] While performing phased opening control, non-target electrically controllable valves not included in the target electrically controllable valve sequence are subject to closing and locking control. Closing and locking control means keeping the non-target electrically controllable valves in the closed state and temporarily interlocking their control commands, preventing them from being opened by other ordinary control tasks before the current directional control task is completed. For mirror branch valves, bypass branch valves, and cross-connected valves adjacent to the target path, a higher level of locking can be set, requiring their valve feedback signals to continuously indicate that they are closed. Through closing and locking control, the control medium can be ensured to flow only along the target pipeline connection direction defined by the valve control topology, preventing unrelated micro-zone units from being synchronously controlled.

[0088] After the target electrically controllable valve completes its opening phase and establishes a stable target path, the target control equipment node is activated, outputting the corresponding control medium according to the expected control medium delivery intensity. For air supply control, the target control equipment node outputs cold air or low-temperature air. For dehumidification control, the target control equipment node outputs dehumidification airflow or extraction power. For pesticide application control, the target control equipment node outputs a predetermined dosage of the application medium. During startup, a short-term trial run is conducted using an initial output intensity lower than the expected control medium delivery intensity. Pressure and flow data from the target pipeline node sequence are used to confirm that continuous flow has been established in the target path before gradually increasing the output intensity to the expected control medium delivery intensity.

[0089] When dynamically adjusting the predicted control medium delivery intensity based on state expression variables, the trend of state expression variable changes in the target abnormal micro-zone unit is continuously read. The operating power of the target control equipment node and the opening degree of the target electrically controllable valve are adjusted based on the anomaly type of the target abnormal micro-zone unit. When the state expression variable decreases slowly and the target path pressure and flow data are stable, the control medium delivery intensity can be appropriately increased. When the state expression variable decreases rapidly, or when adjacent micro-zone units show a trend of being affected, the control medium delivery intensity can be decreased. When pressure data abnormally increases or flow data abnormally decreases, it indicates that there may be local blockage in the target path or insufficient valve opening. In this case, the opening degree of the target electrically controllable valve should be adjusted first, rather than directly increasing the operating power of the target control equipment node.

[0090] When performing state expression variable feedback monitoring on the target abnormal micro-zone unit and adjacent micro-zone units, the temperature, humidity, insect infestation, pressure, flow rate, and valve status data of the target abnormal micro-zone unit are synchronously read according to the set monitoring cycle, and the state expression variables of the target abnormal micro-zone unit are regenerated. Simultaneously, the state expression variables of micro-zone units adjacent to the target abnormal micro-zone unit are read to determine whether the control effect is concentrated on the target abnormal micro-zone unit. Feedback monitoring does not simply observe whether the target abnormal micro-zone unit's temperature drops; it also needs to observe whether unreasonable synchronous changes occur in adjacent micro-zone units to determine whether the control medium has experienced lateral diffusion, mirror crossflow, or bypass leakage.

[0091] When verifying the risk of mis-regulation in mirror imaging, the trend of the state expression variables of the target abnormal micro-zone unit and its geometric mirror micro-zone unit is compared. This is combined with pressure data, flow data, and feedback status of the target electrically controllable valve in the target pipeline connection direction to determine whether the controlled medium accurately reaches the target abnormal micro-zone unit. If the state expression variables of the target abnormal micro-zone unit continuously decrease, while the state expression variables of the geometric mirror micro-zone unit and non-target adjacent micro-zone units do not undergo synchronous abnormal changes, the current directional regulation is deemed effective. If the state expression variables of the target abnormal micro-zone unit do not decrease significantly, while the state expression variables of the geometric mirror micro-zone unit decrease significantly, or if the target path pressure data and flow data are inconsistent with the directional flow calibration results, a risk of mis-regulation in mirror imaging is identified.

[0092] When there is a risk of mirror miscontrol, first reduce the operating power of the target control device node and suspend the opening of any new target electrically controllable valves. Then, close the target electrically controllable valves or bypass electrically controllable valves related to the mirror direction branch, and reread the state expression variables of the target abnormal micro-partition unit, the geometric mirror micro-partition unit, and adjacent micro-partition units. After confirming that the current path has a miscontrol trend, re-invoke the graph structure model, temporarily mark the directed graph edges that have experienced mirror miscontrol risk as high-risk edges, and increase the comprehensive path cost parameter of the corresponding directed graph edges in the new path optimization result, thereby generating a secondary path optimization result. The secondary path optimization result needs to re-form the valve control topology and re-execute the opening pre-verification and phased opening control.

[0093] During targeted regulation, it is necessary to continuously assess whether the target abnormal micro-region unit has recovered to the target grain storage state threshold range. The target grain storage state threshold range refers to the target state boundary dynamically generated based on grain type, storage stage, grain pile depth, historical stable state, and regulation medium type, and is not a fixed single temperature and humidity threshold. Only when the state expression variables of the target abnormal micro-region unit continuously fall within the target grain storage state threshold range, and the retention time meets the recovery and retention requirements of the corresponding regulation medium type, and no reverse abnormal diffusion trend is observed in adjacent micro-region units, is the target abnormal micro-region unit determined to have recovered to the target grain storage state threshold range.

[0094] After the target abnormal micro-zone unit recovers to the target grain storage state threshold range, reverse stage shutdown control is executed. Reverse stage shutdown control refers to closing the target electrically controllable valves segment by segment in the order from the node furthest from the target control equipment node to the node closest to the target control equipment node. This allows residual control media in the target path to continue to be discharged or flow back to the designated discharge node along the predetermined target pipeline connection direction, rather than being confined within a local pipeline segment. For pesticide application control scenarios, reverse stage shutdown control also needs to reserve a discharge window to prevent residual pesticide media from remaining in the target pipeline node sequence. For dehumidification control scenarios, reverse stage shutdown control needs to prevent humid gas from flowing back to the target abnormal micro-zone unit. For air supply control scenarios, reverse stage shutdown control needs to avoid rapid closure causing local pressure rebound.

[0095] After completing the reverse-phase shutdown control, the actual opening and closing sequence of the target electrically controllable valves, valve feedback records, target control equipment node operation records, control medium delivery process, target abnormal micro-zone unit state expression variable change trajectory, adjacent micro-zone unit state expression variable change trajectory, and mirror miscontrol risk verification results are written back to the digital twin model. By writing back the above process data, the subsequent graph structure model can update the directional association weights, asymmetric resistance characteristics, mirror confusion risk, and comprehensive path cost parameters of the corresponding directed graph edges, enabling the next path optimization to utilize the actual feedback results of this directional control.

[0096] As an example, when the micro-partition unit After being identified as a micro-zone unit with high temperature anomalies, the path optimization results select the air supply equipment node. As the target control device node, and generate by , , The target pipeline node sequence is determined, along with the corresponding target electrically controllable valve sequence and target pipeline connection direction. The control system first generates a valve control topology based on the target pipeline node sequence, then performs an opening pre-verification on each target electrically controllable valve to confirm that the valve installation posture, directional flow response data, and target pipeline connection direction are consistent. Subsequently, the control system proceeds according to the sequence from the air supply equipment node to... The system sequentially opens the target electrically controlled valves in stages and locks other non-target electrically controlled valves. After the air supply equipment node starts, the system... The state expression variable shows a downward trend, which dynamically adjusts the air supply intensity while monitoring its geometric mirror micro-zone unit. The state expression variable. If The state expression variable continued to decrease and If no synchronous decline occurs, the current targeted control measures will continue. If... If a synchronous decline occurs, it indicates a risk of misalignment in mirroring and triggers a secondary path optimization. When After continuously meeting the target grain storage state threshold range, the system closes the target electrically controllable valve segment by segment according to the reverse stage closure control, and writes the entire control process back to the digital twin model.

[0097] In one possible implementation, before selectively opening and closing the electrically controllable valves on the grid pipeline assembly based on path optimization results, and performing feedback adjustment in conjunction with state expression variables to complete directional regulation, the method further includes: generating a basic grain storage state threshold range based on the grain type, storage stage, and grain pile depth level of the target abnormal micro-region; calculating the local stable baseline of the target abnormal micro-region based on historical state expression variables, and calculating the neighborhood diffusion baseline based on the state expression variables of adjacent micro-region units; and generating a dynamic boundary of the target grain storage state threshold range based on the basic grain storage state threshold range, the local stable baseline, and the neighborhood diffusion baseline.

[0098] Specifically, the process begins by retrieving the grain type, storage stage, and grain pile depth level corresponding to the target abnormal micro-zone unit from the digital twin model, and then matching this information with a pre-defined grain storage safety rule base. The grain type refers to the variety of grain currently stored within the target abnormal micro-zone unit, such as wheat, rice, corn, or soybeans. Different grain types have varying sensitivities to temperature, humidity, and pest risks. The storage stage refers to different management stages of the grain, such as the stabilization period after storage, long-term storage period, ventilation and cooling period, pre-exit turnover period, or pest control period. Different storage stages correspond to different safety boundaries and control tolerances. The grain pile depth level refers to the spatial hierarchy of the target abnormal micro-zone unit in the grain pile height direction, which can be divided into bottom, middle, and top layers. The bottom layer is more susceptible to ground moisture and ventilation resistance, the middle layer is more prone to heat accumulation, and the top layer is more susceptible to external temperature and humidity disturbances. By using these three types of information to jointly determine the basic grain storage state threshold range, it is possible to avoid over- or under-regulation caused by using a uniform fixed threshold for all micro-zone units.

[0099] The basic grain storage state threshold range includes at least the basic temperature threshold range, basic humidity threshold range, basic insect infestation threshold range, and basic state expression variable threshold range. During generation, the initial temperature and humidity boundaries required for grain quality and safety are first determined based on the grain type. Then, these initial temperature and humidity boundaries are adjusted according to the storage stage. Finally, the adjusted temperature and humidity boundaries are adjusted based on the depth of the grain pile. For example, for wheat stored for long periods, stricter upper temperature and humidity boundaries can be set, while the upper boundaries of state expression variables can be appropriately relaxed during the pre-extraction turnover period. Higher-level micro-zoning units, being more significantly affected by the external environment, can have a higher tolerance for short-term fluctuations. Middle-level micro-zoning units, with slower heat diffusion, can have a lower upper temperature boundary to trigger regulation earlier. The resulting basic grain storage state threshold range reflects the correlation between grain type, stage, and spatial level.

[0100] The boundary correction expression for the basic grain storage state threshold range is: in, This represents the lower boundary of the basic grain storage state threshold range corresponding to the i-th target anomaly micro-partition unit. This represents the upper boundary of the basic grain storage state threshold range corresponding to the i-th target anomaly micro-partition unit. This represents the initial safety lower boundary determined by the grain type. This represents the initial safety upper boundary determined by the grain type. This represents the stage correction coefficient corresponding to the grain storage stage of the i-th target anomaly micro-partition unit. The value ranges from -1 to 1, with a positive value indicating a relaxed boundary and a negative value indicating a tightened boundary. This represents the depth correction coefficient for the grain pile depth level corresponding to the i-th target anomaly micro-partition unit. Its value ranges from -1 to 1, and the direction of the value is set according to the disturbance characteristics of the bottom, middle, and top layers. and These represent the influence weights of the grain storage stage on the lower and upper boundaries, respectively, with values ​​ranging from 0 to 1. and These represent the influence weights of the grain pile depth level on the lower and upper boundaries, respectively, with values ​​ranging from 0 to 1. This expression first provides a basic safety boundary based on the grain type, and then superimposes correction terms for the grain storage stage and grain pile depth level, so that the basic grain storage state threshold range can reflect the grain storage conditions of the target abnormal micro-zone unit, rather than using a uniform threshold for the entire warehouse.

[0101] When the basic grain storage state threshold range is used as a state expression variable... It can be set to 0 to 0.20. It can be set to 0.30 to 0.60; the corresponding value for long-term storage. It can be set to -0.30 to -0.10 to tighten the target grain storage status threshold range; the corresponding turnover period before delivery. It can be set to 0.10 to 0.30 to improve short-term volatility tolerance; the underlying value corresponds to... It can be set to -0.20 to 0.10, corresponding to the middle layer. It can be set to -0.30 to -0.10, corresponding to high-rise buildings. It can be set to 0 to 0.20. The basis for setting the above range is that the higher the value of the state expression variable, the higher the degree of abnormality. Long-term storage and the middle heat accumulation zone require a lower upper boundary for early regulation, while the short-term turnover stage and the upper short-term disturbance zone can allow for higher instantaneous fluctuations. If the basic grain storage state threshold range is used for single indicators such as temperature, humidity or insect infestation, the same boundary correction logic is used, only the initial safety boundary is replaced with the safety boundary of the corresponding single indicator.

[0102] When calculating the local stability baseline based on historical state expression variables, the historical state expression variable sequence of the target anomalous micro-region before the occurrence of the anomaly is first read, and historical data already marked as anomaly control periods, application periods, cleaning periods, and sensor failure periods are removed. The local stability baseline refers to the stability level of the state expression variables of the target anomalous micro-region under normal grain storage conditions, and is used to represent the normal fluctuation characteristics of the target anomalous micro-region. Different micro-regions may have different normal state fluctuations even when storing the same type of grain due to differences in location, pipeline layout, grain pile compaction degree, and local ventilation conditions. Therefore, it is necessary to establish a separate local stability baseline for each target anomalous micro-region, rather than simply using the average state of the entire warehouse as a reference.

[0103] Historical state expression variable sequences can be filtered using a sliding time window, which can be set to the most recent 24 hours to 30 days, and the window length can be adjusted according to the storage stage. For the stable period after storage, state changes are relatively rapid, and the sliding time window can be set to 24 to 72 hours; for long-term storage, state changes are relatively slow, and the sliding time window can be set to 7 to 30 days; for the pest control period, data from the recovery phase after pesticide application should be excluded to avoid the control process affecting the local stability baseline. The sliding time window allows the local stability baseline to reflect both the historical stable state of the target abnormal micro-region and to adapt to seasonal changes and changes in the storage stage.

[0104] The expression for calculating the local stable baseline is: in, This represents the local stable baseline of the i-th target anomaly micro-partition at time t. This represents the i-th target anomaly micro-partition unit at a historical time. The state expression variable is denoted by W, which represents the length of the sliding time window and can range from 24 hours to 30 days for the number of sampling points. This represents the time decay factor, ranging from 0.90 to 0.999. A higher value indicates a higher weighting for retaining earlier historical data. This represents the i-th target anomaly micro-partition unit at a historical time. Whether it belongs to valid and stable data, the value is 0 or 1. 1 indicates that the historical moment was not in an abnormal control period, application period, cleaning period, or sensor failure period, while 0 indicates that the data for that historical moment was removed. This represents the stability correction coefficient, with a value ranging from 0.001 to 0.01. This expression calculates the normal state baseline of the target anomalous micro-partition unit using a time-decay weighted method, ensuring that stable data closer to the current time has a greater impact on the local stable baseline, and using effective stable data identifiers to remove anomalous data unsuitable for baseline calculation.

[0105] When calculating the neighborhood diffusion baseline based on the state expression variables of adjacent micro-regional units, the neighborhood range of the target anomalous micro-regional unit is first determined according to the relationship between adjacent micro-regional units in the micro-regional unit system. The neighborhood range can include first-level adjacent micro-regional units that directly share boundary connection nodes with the target anomalous micro-regional unit, and can also be extended to second-level adjacent micro-regional units that are indirectly connected through an intermediate micro-regional unit. The neighborhood diffusion baseline refers to the level of diffusion influence that the state expression variables of the surrounding area of ​​the target anomalous micro-regional unit may have on the target anomalous micro-regional unit. It is used to determine whether the current anomalousness of the target anomalous micro-regional unit is an isolated anomalousness or may be affected by the spread of surrounding high temperature, high humidity, or pests. Since the temperature, humidity, and pests inside the grain pile all have spatial diffusion characteristics, the neighborhood diffusion baseline can provide a spatial environmental correction basis for the target stored grain state threshold range.

[0106] When calculating the neighborhood diffusion baseline, it is necessary to consider the spatial distance between adjacent micro-regional units and the target anomalous micro-regional unit, the pipeline direction correlation weight, the asymmetric resistance characteristics, and the anomalousness of adjacent micro-regional units. The closer the distance, the higher the direction correlation weight, the lower the asymmetric resistance, and the higher the state expression variables of adjacent micro-regional units, the stronger the diffusion influence of the adjacent micro-regional unit on the target anomalous micro-regional unit. Through this processing, the neighborhood diffusion baseline is not just a simple average of the state expression variables of adjacent micro-regional units, but a weighted baseline that reflects the combined influence of actual pipeline connectivity and grain pile spatial diffusion.

[0107] The expression for calculating the neighborhood diffusion baseline is: in, This represents the neighborhood diffusion baseline of the i-th target anomaly micro-partition at time t. Let represent the set of neighboring micro-partitions corresponding to the i-th target anomaly micro-partition, and j represent any neighboring micro-partition in the set of neighboring micro-partitions. This represents the directional association weight between the i-th target anomaly micro-partition and the j-th adjacent micro-partition, with a value ranging from 0 to 1. This represents the spatial distance between the i-th target anomaly micro-partition and the j-th adjacent micro-partition. This represents the spatial diffusion attenuation scale, and its value can be set to the interval between 1 and 5 micro-partitions. This represents the spatial distance attenuation term, used to reduce the impact of distant adjacent micro-partition units. This represents the asymmetric resistance characteristic of the pipeline connection direction between the i-th target anomaly micro-partition unit and the j-th adjacent micro-partition unit. This represents the state expression variable of the j-th adjacent micro-partition unit at time t. This represents the stability correction coefficient, with a value ranging from 0.001 to 0.01. This expression uses spatial distance, directional transport capacity, and asymmetric drag as weights for neighborhood influence, ensuring that the neighborhood diffusion baseline reflects the true diffusion impact of surrounding micro-regions on the target anomalous micro-region.

[0108] When generating the dynamic boundary of the target grain storage state threshold range based on the basic grain storage state threshold range, the local stability baseline, and the neighborhood diffusion baseline, the basic grain storage state threshold range is first used as a safety boundary. Then, the normal fluctuations of the target abnormal micro-region unit itself are corrected by the local stability baseline, and the diffusion impact of the surrounding area on the target abnormal micro-region unit is corrected by the neighborhood diffusion baseline. The target grain storage state threshold range is a dynamic threshold range used to determine whether targeted regulation has reached the stopping condition. It is closer to the actual state of the target abnormal micro-region unit than the basic grain storage state threshold range. The dynamic boundary includes a dynamic lower boundary and a dynamic upper boundary. The dynamic lower boundary is used to avoid excessive cooling, excessive dehumidification, or excessive pesticide application that could affect grain quality, while the dynamic upper boundary is used to determine whether the abnormality has been sufficiently suppressed.

[0109] The dynamic boundary expression for the target grain storage state threshold range is: in, This represents the dynamic lower boundary of the target abnormal micro-partition unit at time t, corresponding to the target grain storage state threshold range. This represents the dynamic upper boundary of the target grain storage state threshold range corresponding to the i-th target anomaly micro-partition unit at time t. This represents the lower boundary of the basic grain storage state threshold range corresponding to the i-th target anomaly micro-partition unit. This represents the upper boundary of the basic grain storage state threshold range corresponding to the i-th target anomaly micro-partition unit. This represents the local stable baseline of the i-th target anomaly micro-partition at time t. This represents the neighborhood diffusion baseline of the i-th target anomaly micro-partition at time t. This represents the anomaly intensity of the i-th target anomaly micro-partition at time t, with a value ranging from 0 to 1. and These represent the influence weights of the local stable baseline on the dynamic lower boundary and the dynamic upper boundary, respectively, with values ​​ranging from 0 to 1. and These represent the influence weights of the neighborhood diffusion baseline on the dynamic lower boundary and dynamic upper boundary, respectively, with values ​​ranging from 0 to 1. and These represent the tightening weights of the anomaly intensity on the dynamic lower and upper boundaries, respectively, with values ​​ranging from 0 to 1. This expression uses the basic grain storage state threshold range as the initial boundary, pulls the boundary towards the historical normal state of the target anomaly micro-region unit through a local stability baseline, reflects the surrounding diffusion pressure through the neighborhood diffusion baseline, and tightens the dynamic boundary when the anomaly intensity is high, ensuring that targeted regulation does not prematurely stop before the anomaly has stabilized and subsided.

[0110] After the dynamic boundaries are generated, boundary rationality constraints need to be applied to the dynamic lower and upper boundaries to ensure that the target grain storage state threshold range does not exceed the allowable range for grain quality safety. For state expression variable scenarios, the dynamic lower boundary is usually constrained between 0 and 0.30, and the dynamic upper boundary is usually constrained between 0.25 and 0.65, and the dynamic upper boundary is required to be greater than the dynamic lower boundary. For temperature scenarios, the dynamic boundaries should not exceed the safe temperature boundary for the corresponding grain type. For humidity scenarios, the dynamic boundaries should not exceed the mold risk control boundary. If the neighborhood diffusion baseline is high, it indicates that there is still strong diffusion pressure around the target abnormal micro-region unit, so the dynamic upper boundary should be appropriately lowered to require the target abnormal micro-region unit to reach a more stable state before ending the regulation. If the local stability baseline is low for a long period of time, it indicates that the target abnormal micro-region unit is usually in a better state, so the dynamic upper boundary should also be appropriately lowered to avoid the recovery standard being too lenient.

[0111] In practical applications, the target grain storage state threshold range is used not only to determine whether directional control has ended, but also to guide the dynamic adjustment of the expected control medium delivery intensity. When the state expression variable of the target abnormal micro-zone unit is still higher than the dynamic upper boundary, it indicates that the control has not yet reached the target state, and it is necessary to maintain or increase the control medium delivery intensity. When the state expression variable falls between the dynamic lower boundary and the dynamic upper boundary, and the duration reaches the recovery and maintenance requirement, it indicates that the target abnormal micro-zone unit has recovered to an acceptable state and can enter the reverse phase to close the control. When the state expression variable is lower than the dynamic lower boundary, it indicates that there may be an over-control risk, and it is necessary to reduce the control medium delivery intensity or enter the slow-release control state in advance. Through this application method, the target grain storage state threshold range is correlated with selective valve control and feedback regulation.

[0112] As an example, during the long-term storage of wheat, the target abnormal micro-region unit is located in the middle layer of the grain pile, with a state expression variable of 0.68. The initial threshold range of the basic state expression variable is set to 0.10 to 0.45. Due to the high stability required for long-term storage, the grain storage stage correction coefficient is set to -0.20; because the middle layer of the grain pile is prone to heat accumulation, the grain pile depth level correction coefficient is set to -0.15, tightening the upper boundary of the basic grain storage state threshold range to approximately 0.40. Subsequently, the historical state expression variables of the target abnormal micro-region unit within the last 7 days, excluding the control period and sensor failure period, are read, and the local stability baseline is calculated to be 0.26. The state expression variables of the first-level and second-level adjacent micro-region units are read, and combined with directional correlation weights, spatial distance, and asymmetric resistance characteristics, the neighborhood diffusion baseline is calculated to be 0.34. Since the neighborhood diffusion baseline is higher than the local stability baseline, it indicates that there is still some diffusion pressure in the surrounding area; therefore, the dynamic upper boundary is further tightened to approximately 0.36. During subsequent targeted regulation, the target abnormal micro-partition unit is only deemed to have recovered to the target grain storage state threshold range when the state expression variable of the target abnormal micro-partition unit continuously decreases and remains stable between the dynamic lower boundary and 0.36, and when there is no reverse abnormal diffusion trend in adjacent micro-partition units, and the reverse phase shutdown control is allowed to be executed.

[0113] In one possible implementation, after completing the directional control, the grid pipeline component corresponding to the path optimization result is cleaned, and the cleaning completion status is determined based on the cleanliness detection parameters. Specifically, this includes: generating a corresponding cleaning control object based on the path optimization result and the target control medium type, wherein the cleaning control object includes the target cleaning path; performing staged cleaning opening control on each target electrically controllable valve in the target cleaning path, and gradually increasing the valve opening of the corresponding target electrically controllable valve in the order from farthest from the target discharge node to closest to the target discharge node, so that the target cleaning medium forms a segmented advancing flow state inside the grid pipeline component; starting the cleaning equipment node and collecting the cleaning status data inside the target cleaning path; calculating the cleanliness detection parameters based on the cleaning status data and determining the cleaning completion status of the grid pipeline component; and dynamically adjusting the target cleaning medium type and target cleaning intensity when the cleanliness detection parameters do not reach the preset cleanliness threshold.

[0114] Specifically, after completing the targeted control, the path control object corresponding to the path optimization result is read first, and the target control equipment node, target abnormal micro-zone unit, target pipeline node sequence, target electrically controllable valve sequence, target pipeline connection direction, control duration, control medium delivery intensity, and control medium residual risk are extracted from the path control object. The path control object is a record of the path that has been actually executed during the targeted control process, used to determine which grid pipeline components participated in the control medium delivery; the target control medium type refers to the type of medium used in this targeted control, which may include air supply medium, dehumidification medium, pesticide application medium, bio-inhibition medium, or pre-drainage medium before cleaning. A cleaning control object is only generated when the target control medium type may cause residue, moisture accumulation, particle adhesion, or microbial carryover risk inside the grid pipeline components; for ordinary short-term air supply control, only no-load flow detection can be performed, while for pesticide application control, high humidity emission control, and pest and disease control control, a complete cleaning control object needs to be generated.

[0115] The cleaning control object is used to receive the path optimization results and convert them into the execution structure required for cleaning treatment. It includes at least the target cleaning path, target cleaning medium type, target cleaning intensity, target cleaning duration, target discharge node, target cleaning valve sequence, and cleaning risk identifier. The target cleaning path refers to the path of the grid pipeline component that needs cleaning treatment, typically generated from the target pipeline node sequence and target electrically controllable valve sequence actually traversed in this directional control. The target cleaning medium type refers to the category of cleaning medium used to remove residual control media, insect egg particles, mold particles, dust particles, or humid gases, and may include high-speed dry air, pulsed airflow, low-humidity air, or inert purge gas. The target discharge node refers to the designated pipeline node where the target cleaning medium, carrying residues, is discharged; priority is given to boundary connection nodes near drain outlets, vents, or cleaning recovery devices. Through the cleaning control object, subsequent cleaning treatments can maintain consistency with the actual path of this directional control, avoiding ineffective cleaning of grid pipeline components that did not participate in the control.

[0116] The target cleaning path can be generated based on the residual risk of the control medium and the actual frequency of path passage. The higher the residual risk, the longer the control duration, and the greater the intensity of the control medium transport, the more comprehensively the target pipeline node sequence and adjacent buffer pipeline sections should be covered by the target cleaning path. The expression for the target cleaning path coverage coefficient is:

[0117] in, This represents the coverage coefficient of the target cleaning path corresponding to the k-th path control object. The value ranges from 0 to 1, and the larger the value, the larger the coverage of the target cleaning path. This represents the residual risk of the normalized control medium corresponding to the control object of the kth path, with a value ranging from 0 to 1; This represents the normalized control duration corresponding to the k-th path control object, with a value ranging from 0 to 1; This represents the normalized control medium transport intensity corresponding to the kth path control object, with a value range of 0 to 1; This represents the normalized asymmetric resistance average value corresponding to the control object of the k-th path, with a value range from 0 to 1; , , and These represent the weights of the influence of residual risk of the control medium, control duration, control medium delivery intensity, and average asymmetric resistance on the target cleaning path coverage coefficient, respectively, with values ​​ranging from 0 to 1. This expression determines the cleaning coverage area through residual risk, control duration, delivery intensity, and pipeline resistance. Higher residual risk or greater pipeline resistance indicates that residues are more likely to adhere to the inner wall of the target path or remain in local pipeline sections; therefore, the target cleaning path needs to cover a more complete pipeline area.

[0118] When implementing phased cleaning opening control for each target electrically controlled valve in the target cleaning path, the target cleaning path is first divided into multiple cleaning pipeline segments according to the direction from the target cleaning medium inlet to the target discharge node. The target electrically controlled valves corresponding to each cleaning pipeline segment are then arranged into a target cleaning valve sequence. Phased cleaning opening control means not simultaneously opening all target electrically controlled valves completely, but rather opening them segment by segment according to the progression of the target cleaning path, gradually increasing the valve opening degree, so that the target cleaning medium is pushed forward in a controlled manner within the grid pipeline assembly. This method can prevent residual control medium from entering the mirror branch due to instantaneous high pressure, and can also prevent the reverse diffusion of residues due to pressure changes in local pipeline segments.

[0119] When progressively increasing the valve opening of the corresponding target electrically controlled valves in order from furthest from the target emission node to closest to it, the valve furthest from the target emission node and closest to the target cleaning medium inlet is opened first, allowing the target cleaning medium to enter the initial section of the target cleaning path. After the pressure and flow rate in the initial section stabilize, the valve opening of the next level of target electrically controlled valves is increased, until the target electrically controlled valves closest to the target emission node are gradually opened. This sequence allows the target cleaning medium to progressively push residual control medium, dust particles, and microbial particles from the inlet end towards the target emission node, forming a progressive flow state. This progressive flow state means that the target cleaning medium does not diffuse simultaneously throughout the entire grid piping assembly, but rather forms a flow front with a clear direction of propulsion along the target cleaning path, thereby improving residue removal efficiency and reducing the risk of cross-contamination.

[0120] The opening degree of the target electrically controllable valve during the cleaning stage can be dynamically set according to residual pressure, asymmetric resistance, residual concentration, and mirror crossflow risk. The expression for the opening degree during the cleaning stage is: in, This represents the opening degree of the nth target electrically controllable valve during the cleaning stage, with a value range of... to ; This indicates the minimum opening during the cleaning phase. A value of 15% to 30% is recommended to ensure initial flow while avoiding sudden pressure changes. This indicates the maximum opening during the cleaning phase. A value of 70% to 100% is recommended to ensure that the target cleaning medium can carry away the residue. This represents the normalized pressure residual coefficient of the cleaning pipeline section corresponding to the nth target electrically controllable valve, with a value ranging from 0 to 1; The normalized asymmetric resistance coefficient of the cleaning pipeline section corresponding to the nth target electrically controllable valve is represented, with a value ranging from 0 to 1. This represents the normalized residual control medium concentration coefficient of the cleaning pipeline section corresponding to the nth target electrically controllable valve, with a value range of 0 to 1; The normalized mirror crossflow risk coefficient represents the cleaning pipeline section corresponding to the nth target electrically controllable valve, with a value ranging from 0 to 1; , , and These represent the weights of residual pressure, asymmetric resistance, residual control medium concentration, and mirror crossflow risk on the opening degree during the cleaning stage, with values ​​ranging from 0 to 1. This indicates boundary truncation, ensuring that the calculated valve opening is neither lower than the minimum opening nor higher than the maximum opening. This expression enhances the local cleaning capability by increasing the target electrically controllable valve opening through residual pressure, asymmetric resistance, and residual control medium concentration. Simultaneously, it reduces the corresponding valve opening by mimicking crossflow risk to prevent the target cleaning medium from entering the mimic branch.

[0121] Before starting the cleaning equipment node, first verify whether the target cleaning path has formed a one-way cleaning loop according to the cleaning control object, and confirm that the target discharge node is in the discharge open state and the non-target electrically controllable valves are in the closed and locked state. The cleaning equipment node refers to the equipment node that can output the target cleaning medium to the target cleaning path, and may include a high-speed fan, a low-humidity air generator, a pulse airflow generator, or a cleaning recovery device. During startup, a low-intensity pre-rinse method is used, allowing the target cleaning medium to enter the target cleaning path at a low flow rate. After confirming continuous flow through the target cleaning path using pressure and flow sensors, the intensity is gradually increased to the target cleaning intensity. This process avoids backflow of residue caused by direct high-intensity rinsing when there are closed valves or partial blockages in the target cleaning path.

[0122] The cleaning status data consists of pressure, flow rate, volatile concentration, particle concentration, humidity, cleaning medium temperature, and valve feedback data within the target cleaning path. Pressure data is used to determine if there is localized blockage or residual pressure within the target cleaning path; flow rate data is used to determine if the target cleaning medium is flowing stably through the target cleaning path; volatile concentration data reflects whether there are still volatile residues in the applied pesticide or bio-inhibitory medium; particle concentration data reflects whether insect egg particles, mold particles, dust particles, or grain debris are continuously carried out; humidity data reflects whether there is still moisture residue on the inner wall of the pipe after dehumidification; and valve feedback data confirms whether the target electrically controllable valves are executing the staged cleaning opening control. These cleaning status data collectively reflect the residual contamination status and cleaning medium flow status within the mesh pipeline assembly.

[0123] When calculating cleanliness detection parameters based on cleaning status data, the volatile concentration data, particle concentration data, humidity data, pressure fluctuation data, flow fluctuation data, and cleaning stability data are first normalized separately, and then fused according to the target control medium type with corresponding weights. Cleanliness detection parameters are comprehensive indicators used to determine whether the target cleaning path has reached a cleanliness completion state. Higher values ​​indicate more severe residual contamination or flow instability, while lower values ​​indicate that the target cleaning path is closer to a clean state. For cleaning after pesticide application, residual volatile concentration and residual particle concentration have higher weights; for cleaning after high-humidity discharge, residual humidity and flow fluctuation have higher weights; for cleaning after pest and disease control, residual particle concentration and microbial particle-related indicators have higher weights.

[0124] The expression for the cleanliness test parameters is: in, This represents the cleanliness detection parameter corresponding to the kth target cleaning path; the smaller the value, the more thorough the cleaning. This represents the normalized volatile concentration residual coefficient in the k-th target cleaning path, with a value ranging from 0 to 1; This represents the normalized particle residual concentration coefficient in the k-th target cleaning path, with a value ranging from 0 to 1; This represents the normalized residual humidity coefficient in the k-th target cleaning path, with a value ranging from 0 to 1; This represents the normalized pressure fluctuation residual coefficient in the k-th target cleaning path, with a value ranging from 0 to 1; This represents the normalized flow fluctuation residual coefficient in the k-th target cleaning path, with a value ranging from 0 to 1; This represents the normalized cleaning stability coefficient in the k-th target cleaning path, with a value ranging from 0 to 1; , , , , and These represent the weights of the corresponding parameters on the cleanliness detection parameters, with values ​​ranging from 0 to 1, and the sum of each weight is 1. This expression represents the degree of residual contamination in the target cleaning path through residual volatile concentration, residual particles, residual humidity, residual pressure fluctuation, and residual flow fluctuation. It also uses a cleaning stability coefficient to deduct misjudgments caused by instantaneous fluctuations, enabling the cleanliness detection parameters to simultaneously reflect residual contamination and flow stability.

[0125] The preset cleanliness threshold can be set according to the type of target control medium and the food safety level. For the target cleaning path after pesticide application, the preset cleanliness threshold can be set to 0.10 to 0.25 to improve residue control requirements; for the target cleaning path after high humidity discharge, the preset cleanliness threshold can be set to 0.20 to 0.35 to focus on controlling moisture residue and condensation risks; for routine cleaning of ordinary air supply paths, the preset cleanliness threshold can be set to 0.30 to 0.45 to reduce unnecessary cleaning time. When the cleanliness detection parameter is less than or equal to the preset cleanliness threshold and remains stable for multiple consecutive detection cycles, the grid pipeline assembly is determined to have reached the cleaning completion state; when the cleanliness detection parameter is higher than the preset cleanliness threshold, or although it is briefly lower than the preset cleanliness threshold but the pressure fluctuation and particle concentration still show an upward trend, the grid pipeline assembly is determined to have not reached the cleaning completion state.

[0126] When the cleanliness detection parameters do not reach the preset cleanliness threshold, first identify the dominant residue type causing the high cleanliness detection parameters, and then dynamically adjust the target cleaning medium type and target cleaning intensity according to the dominant residue type. If the volatile concentration residual coefficient is high, it indicates that the pesticide or bio-inhibitory medium is still present on the inner wall of the pipeline or in local dead corners. The target cleaning medium type can be switched to high-flow-rate dry air, and the continuous purging time can be increased. If the particle residual concentration coefficient is high, it indicates that insect egg particles, mold particles, or dust particles are still being continuously carried out in the target cleaning path. The target cleaning medium type can be switched to pulsed airflow, and the pulse frequency can be increased. If the humidity residual coefficient is high, it indicates that there is still moisture residue on the inner wall of the pipeline. The target cleaning medium type can be switched to low-humidity air, and the medium temperature can be appropriately increased or the relative humidity can be decreased. If the pressure fluctuation residual coefficient and flow fluctuation residual coefficient are high, it indicates that there may be local blockage or insufficient valve opening in the target cleaning path. The opening of the target electrically controllable valve in the cleaning stage needs to be readjusted.

[0127] The dynamic adjustment of the target cleaning intensity can be determined by the deviation between the cleanliness detection parameters and the preset cleanliness threshold. The expression for the target cleaning intensity adjustment coefficient is as follows: in, This represents the target cleaning intensity adjustment coefficient corresponding to the k-th target cleaning path; This represents the minimum cleaning intensity coefficient, with a recommended value range of 0.20 to 0.40, to ensure the basic cleaning flow rate; This indicates the maximum cleaning intensity coefficient, with a recommended value range of 0.80 to 1.00, for handling high residue conditions; This represents the cleanliness detection parameter corresponding to the k-th target cleaning path; This indicates the preset cleanliness threshold; This indicates the upper limit reference value for cleanliness testing parameters, which is usually set to 1; This represents the stability correction coefficient, with a value ranging from 0.001 to 0.01. This expression gradually increases the cleaning intensity based on the degree to which the cleanliness detection parameters exceed the preset cleanliness threshold. When the residue level is low, a lower cleaning intensity is used to reduce energy consumption and pipeline impact; when the residue level is high, the intensity is increased to near the maximum to enhance cleaning capability.

[0128] After dynamically adjusting the target cleaning medium type and target cleaning intensity, the phased cleaning start control is re-executed, and cleaning status data within the target cleaning path continues to be collected until the cleanliness detection parameters meet the preset cleanliness threshold. If the cleanliness detection parameters still cannot decrease after multiple adjustments, the relevant cleaning pipe segments in the target cleaning path are marked as residual abnormal pipe segments, and these residual abnormal pipe segments are written back to the digital twin model and graph structure model to improve their subsequent comprehensive path cost parameters, preventing the next application control or high-humidity emission control from preferentially passing through these residual abnormal pipe segments. Through this feedback and write-back mechanism, the cleaning process not only completes a single residue removal but also corrects the judgment of the grid pipe component status in subsequent path optimization.

[0129] As an example, after the target abnormal micro-zone unit completes the insect pest control, the path optimization results show that the pesticide medium reaches the target abnormal micro-zone unit through the pesticide delivery equipment node, three boundary connection nodes, and four target electrically controllable valves. Therefore, the cleaning control object determines this actual path as the target cleaning path and sets the boundary connection node closest to the discharge outlet as the target discharge node. The system first closes all non-target electrically controllable valves, and then, in the order from furthest from the target discharge node to closest to the target discharge node, sequentially increases the opening of the four target electrically controllable valves to the cleaning stage openings of 30%, 45%, 60%, and 80%, allowing high-speed dry air to advance segment by segment in the target cleaning path. After the cleaning equipment node is started, the volatile concentration sensor detects a high level of residual pesticide medium, and the particle concentration sensor detects that insect egg particles and dust particles are still being discharged. The system calculates the cleanliness detection parameter to be 0.38, which is higher than the preset cleanliness threshold of 0.20 under the pesticide control scenario. Therefore, the cleaning is not determined to be complete. The system then switched the target cleaning medium type from continuous dry air to pulsed dry airflow and increased the target cleaning intensity to 0.85. After several more cleaning cycles, the cleanliness detection parameter dropped to 0.16 and remained stable, indicating that the corresponding grid pipe assembly had reached the cleaning completion state.

[0130] This embodiment also discloses a grain storage precision control device based on micro-regional grids and three-dimensional interaction, referring to... Figure 4 The device includes an acquisition module 401, a processing module 402, and an output module 403. It is used to execute any of the above-described methods for precise grain storage control based on micro-partitioned grids and three-dimensional interaction, wherein: The acquisition module 401 is used to perform micro-regional grid division on the grain warehouse space and build a grid pipeline assembly integrating sensors and electrically controllable valves in each micro-regional unit, thereby forming a micro-regional unit system; Processing module 402 is used to perform digital twin modeling on the micro-partition unit system and construct a three-dimensional visualization model, mapping the state data of each micro-partition unit to state expression variables; Processing module 402 is used to identify abnormal micro-partition units based on state expression variables and construct a graph structure model. At the same time, it introduces path cost parameters to perform optimization solution on the transmission path between abnormal micro-partition units and control equipment to generate path optimization results. The processing module 402 is used to perform selective opening and closing control on the electrically controllable valves on the grid pipeline assembly based on the path optimization results, and to perform feedback adjustment in combination with the state expression variables to complete directional regulation; The output module 403 is used to perform cleaning processing on the grid pipe components corresponding to the path optimization results after the directional control is completed, and to determine the cleaning completion status based on the cleanliness detection parameters.

[0131] It should be noted that the above embodiments of the apparatus are only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0132] This embodiment also discloses an electronic device, as shown in the reference. Figure 5 The electronic device may include: at least one processor 501, at least one communication bus 502, user interface 503, network interface 504, and at least one memory 505.

[0133] The communication bus 502 is used to enable communication between these components.

[0134] The user interface 503 may include a display screen and a camera. Optionally, the user interface 503 may also include a standard wired interface and a wireless interface.

[0135] The network interface 504 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0136] The processor 501 may include one or more processing cores. The processor 501 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in memory 505, and by calling data stored in memory 505. Optionally, the processor 501 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 501 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content to be displayed on the screen; and the modem handles wireless communication. It is understood that the modem may also not be integrated into the processor 501 and may be implemented as a separate chip.

[0137] The memory 505 may include random access memory (RAM) or read-only memory. Optionally, the memory may include a non-transitory computer-readable storage medium. The memory 505 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 505 may include a program storage area and a data storage area. The program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch functionality, sound playback functionality, image playback functionality, etc.), instructions for implementing the various method embodiments described above, etc.; the data storage area may store data involved in the various method embodiments described above, etc. Optionally, the memory 505 may also be at least one storage device located remotely from the aforementioned processor 501. As a computer storage medium, the memory 505 may include an operating system, a network communication module, a user interface 503 module, and an application program for a grain warehouse precision control method based on micro-partitioned grids and three-dimensional interaction.

[0138] exist Figure 5In the electronic device shown, the user interface 503 is mainly used to provide an input interface for the user and to obtain the user input data; while the processor 501 can be used to call the application stored in the memory 505, which is a method for precise control of grain warehouses based on micro-partitioned grid and three-dimensional interaction. When executed by one or more processors 501, the electronic device executes one or more methods as described in the above embodiments.

[0139] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, as some steps can be performed in other orders or simultaneously according to the present invention. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0140] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0141] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some service interface; the indirect coupling or communication connection between apparatuses or units may be electrical or other forms.

[0142] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0143] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0144] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory 505 and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned memory 505 includes various media capable of storing program code, such as a USB flash drive, external hard drive, magnetic disk, or optical disk.

[0145] The present invention also discloses a non-transitory computer-readable storage medium storing instructions. When executed by one or more processors 501, these instructions cause an electronic device to perform one or more methods as described in the above embodiments.

[0146] The above are merely exemplary embodiments of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Those skilled in the art will readily conceive of other embodiments of this disclosure upon considering the specification and the disclosure of practical truths. This invention is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A method for precise control of grain storage based on micro-regional grids and three-dimensional interaction, characterized in that, The method includes: The grain storage space is divided into micro-zone grids, and grid pipe components integrating sensors and electrically controllable valves are built in each micro-zone unit to form a micro-zone unit system; Digital twin modeling is performed on the micro-partition unit system and a three-dimensional visualization model is constructed, mapping the state data of each micro-partition unit into state expression variables; Based on the state expression variables, abnormal micro-partition units are identified and a graph structure model is constructed. At the same time, a path cost parameter is introduced to optimize the transport path between the abnormal micro-partition units and the control equipment to generate path optimization results. Based on the path optimization results, selective opening and closing control is performed on the electrically controllable valves on the grid pipeline assembly, and feedback adjustment is performed in conjunction with the state expression variables to complete directional regulation; After completing the directional control, the mesh pipeline component corresponding to the path optimization result is cleaned and the cleaning completion status is determined based on the cleanliness detection parameters.

2. The method for precise control of grain storage based on micro-regional grid and three-dimensional interaction according to claim 1, characterized in that, The process of dividing the grain storage space into micro-zones and constructing integrated sensor and electrically controllable valve grid pipeline components within each micro-zone unit to form a micro-zone unit system specifically includes: A three-dimensional spatial coordinate system is established for the grain storage space, and micro-partition units with micro-partition unit identifiers are generated; Within the micro-partition unit, a grid pipe assembly with directional connectivity features is constructed to form a local pipe topology; The electrically controllable valve and multiple types of sensors are configured at the boundary connection nodes of the grid pipe assembly, and valve information is generated. Perform directional flow calibration on the mesh pipe assembly and generate asymmetric resistance characteristics for the corresponding pipe connection direction; The micro-partition unit system is constructed based on the micro-partition unit identifier, the local pipeline topology, the valve information, and the asymmetric resistance characteristics.

3. The method for precise control of grain storage based on micro-regional grid and three-dimensional interaction according to claim 1, characterized in that, The step of performing digital twin modeling on the micro-partition unit system and constructing a three-dimensional visualization model, mapping the state data of each micro-partition unit to state expression variables, specifically includes: The structural parameters and directional transport parameters of the micro-partition unit system are written into the digital twin model; Construct a three-dimensional voxel object corresponding to the mesh pipeline component based on the micro-partition unit; The sensor data of each micro-partition unit is synchronized in time and written into the corresponding three-dimensional voxel object, and the state expression variable is generated by combining the asymmetric drag characteristics. Based on the state expression variables and the asymmetric resistance characteristics, a mirror-recognizable coefficient is generated; The state expression variables, asymmetric resistance features, and mirror resolvability coefficients are mapped to the three-dimensional visualization model, wherein directional flow calibration is performed on the grid pipe assembly and the asymmetric resistance features are generated.

4. The method for precise control of grain storage based on micro-regional grid and three-dimensional interaction according to claim 1, characterized in that, The process of identifying anomalous micro-partition units based on the state expression variables and constructing a graph structure model, while simultaneously introducing path cost parameters to optimize the transport path between the anomalous micro-partition units and the control equipment to generate path optimization results, specifically includes: Based on the state expression variables and mirror distinguishability coefficients, abnormal micro-partition units are identified and a set of abnormal micro-partition units is generated. Perform mirror target verification on the abnormal micro-partition unit and the geometric mirror micro-partition unit and reconfirm the true abnormal location; Based on the connection relationships in the micro-partition unit system, a graph structure model containing directed graph edges is constructed. Each micro-partition unit, each boundary connection node, each electrically controllable valve, and each control device in the micro-partition unit system is mapped as a graph node. The grid pipeline component connection relationship between adjacent micro-partition units, the access relationship between the control device and the grid pipeline component, and the control relationship between the boundary connection node and the electrically controllable valve are mapped as directed graph edges. Based on the anomaly type, select the available control device nodes corresponding to the anomaly micro-partition unit; Calculate the comprehensive path cost parameters for the directed graph edges in the candidate transport paths; Based on the comprehensive path cost parameters, constraint optimization is performed on the candidate transport path to generate the path optimization result.

5. The method for precise control of grain storage based on micro-regional grid and three-dimensional interaction according to claim 1, characterized in that, The selective opening and closing control of the electrically controllable valves on the grid pipeline assembly based on the path optimization results, and the feedback adjustment based on the state expression variables to complete directional regulation, specifically includes: Based on the path optimization results, the target control equipment nodes, target abnormal micro-partition units, target pipeline node sequences, target electrically controllable valve sequences, target pipeline connection directions, and expected control medium transport intensity are extracted. Based on the target pipeline node sequence, a corresponding valve control topology is generated, so that each target electrically controllable valve in the valve control topology maintains a mapping relationship with the corresponding target pipeline connection direction and the target abnormal micro-partition unit. Perform an opening pre-verification on each target electrically controllable valve in the target electrically controllable valve sequence, and verify the consistency of the target pipeline connection direction; Perform phased opening control on the target electrically controllable valve according to the target pipeline node sequence and lock the non-target electrically controllable valve; The target control device node is activated, and the expected control medium delivery intensity is dynamically adjusted based on the state expression variables. The state expression variable feedback monitoring and verification of the target abnormal micro-partition unit and its adjacent micro-partition units are performed, and the risk of mirror mis-regulation is verified. When the risk of mirror mis-regulation exists, the graph structure model and path optimization results are called again to perform secondary path optimization. After the target abnormal micro-partition unit recovers to the target grain storage state threshold range, reverse phase shutdown control is executed.

6. The method for precise control of grain storage based on micro-regional grid and three-dimensional interaction according to claim 5, characterized in that, Before performing selective opening and closing control on the electrically controllable valves on the grid pipeline assembly based on the path optimization results, and performing feedback adjustment in conjunction with the state expression variables to complete directional regulation, the method further includes: The basic grain storage state threshold range is generated based on the grain type, storage stage, and grain pile depth level of the target abnormal micro-partition unit. The local stability baseline of the target abnormal micro-partition unit is calculated based on the historical state expression variables, and the neighborhood diffusion baseline is calculated based on the state expression variables of the adjacent micro-partition units. The dynamic boundary of the target grain storage state threshold range is generated based on the basic grain storage state threshold range, the local stable baseline, and the neighborhood diffusion baseline.

7. The method for precise control of grain storage based on micro-regional grid and three-dimensional interaction according to claim 1, characterized in that, The step of performing cleaning on the grid pipeline components corresponding to the path optimization result after completing the directional control and determining the cleaning completion status based on the cleanliness detection parameters specifically includes: Based on the path optimization results and the target control medium type, a corresponding cleaning control object is generated, wherein the cleaning control object includes the target cleaning path; A staged cleaning opening control is performed on each target electrically controllable valve in the target cleaning path, and the valve opening degree of the corresponding target electrically controllable valve is gradually increased in the order from farthest from the target discharge node to closest to the target discharge node, so that the target cleaning medium forms a segmented advancing flow state inside the grid pipeline assembly. Start the cleaning equipment node and collect cleaning status data within the target cleaning path; Based on the cleaning status data, cleanliness detection parameters are calculated and the cleaning completion status of the mesh pipe assembly is determined. When the cleanliness detection parameters do not reach the preset cleanliness threshold, the target cleaning medium type and target cleaning intensity are dynamically adjusted.

8. A grain storage precision control device based on micro-regional grid and three-dimensional interaction, characterized in that, The device is used to execute a grain storage precision control method based on micro-regional grid and three-dimensional interaction as described in any one of claims 1-7. The device includes an acquisition module, a processing module, and an output module, wherein: The acquisition module is used to perform micro-regional grid division on the grain warehouse space and construct a grid pipeline assembly integrating sensors and electrically controllable valves in each micro-regional unit, thereby forming a micro-regional unit system; The processing module is used to perform digital twin modeling on the micro-partition unit system and construct a three-dimensional visualization model, and to map the state data of each micro-partition unit into state expression variables. The processing module is used to identify abnormal micro-partition units based on the state expression variables and construct a graph structure model. At the same time, it introduces path cost parameters to perform optimization solution on the transmission path between the abnormal micro-partition units and the control equipment to generate path optimization results. The processing module is used to selectively open and close the electrically controllable valves on the grid pipeline assembly according to the path optimization results, and to perform feedback adjustment in conjunction with the state expression variables to complete directional regulation; The output module is used to perform cleaning processing on the grid pipeline component corresponding to the path optimization result after completing the directional control, and determine the cleaning completion status based on the cleanliness detection parameters.

9. An electronic device, characterized in that, It includes a processor and a memory, the memory storing a computer program, wherein the computer program, when executed by the processor, implements a grain warehouse precision control method based on micro-partitioned grid and three-dimensional interaction as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed, perform a grain storage precision control method based on micro-partitioned grids and three-dimensional interaction as described in any one of claims 1-7.

Images