Granary monitoring and early warning system based on digital twinborn technology

Abstract

The invention discloses a granary monitoring and early warning system based on a digital twin technology, and relates to the technical field of granary monitoring. The system comprises a server, a conduction analysis module and an early warning management module, through complete coverage of sensor data, a mechanism model, numerical simulation, three-dimensional visualization, risk quantification and intelligent response, landing, online acquisition, real-time simulation and dynamic early warning of digital twinning in grain storage management are truly realized, and blind areas and time delay of traditional manual and local sensing are solved.

Description

Technical Field

[0001] This invention relates to the field of grain warehouse monitoring technology, and in particular to a grain warehouse monitoring and early warning system based on digital twin technology. Background Technology

[0002] With global population growth and increasing demand for food security, effective food storage management has become crucial. However, traditional food condition monitoring methods often rely on manual inspections and simple sensor deployments, which suffer from incomplete monitoring, slow response times, and difficulty in acquiring comprehensive data in real time. In recent years, digital twin technology has emerged by using digital models of physical objects to simulate their behavior in real-world environments. It can collect and analyze data in real time, providing in-depth insights into the state of physical objects. While digital twin technology has shown great potential in manufacturing, healthcare, and urban management, its application in food storage management remains relatively limited. Summary of the Invention

[0003] Therefore, it is necessary to provide a grain warehouse monitoring and early warning system based on digital twin technology to address the problems mentioned in the background technology above.

[0004] The objective of this invention can be achieved through the following technical solution: a server that communicates with various sensors in the grain warehouse to obtain the temperature and humidity at various locations within the grain warehouse, and simultaneously builds a dynamic evolution coupled partial differential equation model, and provides initial conditions and boundaries to dynamically simulate the physical state of the grain pile.

[0005] The conduction analysis module, based on the server's coupled partial differential equation model, can obtain the temperature, humidity, and heat at any point inside the grain pile, and then construct a three-dimensional model of temperature, humidity, and heat distribution.

[0006] The early warning management module divides the entire grain pile into several risk clusters based on a three-dimensional model of temperature, humidity, and heat distribution. For each risk cluster, a comprehensive early warning index is obtained through temperature, humidity, and heat risk analysis. This index is then compared with the early warning range. If the comprehensive early warning index is greater than the upper limit of the early warning range, the risk cluster is marked as a high-level alarm cluster, and high-level response measures are implemented. If the comprehensive early warning index is within the early warning range, the risk cluster is marked as a medium-level alarm cluster, and medium-level response measures are implemented. If the comprehensive early warning index is less than the lower limit of the early warning range, continuous monitoring and fixed level one ventilation are sufficient; no other response measures are required.

[0007] In some embodiments, the server-coupled partial differential equation model is constructed as follows:

[0008] Establish the heat conduction-biological respiration coupling equation: ,in Let ρ be the temperature field of the grain pile, and ρ be the density of the grain pile. ρ is the specific heat capacity, k is the thermal conductivity. For respiratory exothermic terms, The convective heat transfer coefficient is... The surface area of ​​the grain pile; The outside air temperature, It is a spatial term, and ▽ is the gradient operator. For convective heat exchange;

[0009] Establish the moisture diffusion-phase transition coupling equation: ,in The moisture content of the grain pile, The water diffusion coefficient is... For phase transition source terms, For ambient humidity, For diffusion term, This is an environmental exchange item.

[0010] In some embodiments, the solution process for the coupled partial differential equation model is as follows:

[0011] 1-1, Spatial Discretization: The entire three-dimensional region of the grain pile is divided into several regular small units, i.e., grid nodes. The spatial step size Δx, Δy, Δz is selected according to the size of the silo, and the nodes are evenly distributed. The initial temperature, moisture content, grain pile density, specific heat capacity parameters corresponding to each node are matched with the sensor measurement points or initial assumed values. The outermost nodes are marked as boundary nodes, and the heat transfer flux and humidity exchange coefficient with the silo wall or air are recorded to provide boundary conditions for subsequent calculations.

[0012] 1-2, Time Discreteness: The total simulation period is divided into several hourly steps Δt. Each step updates a corresponding time level. The current time is represented by an integer n, and the values ​​of all fields at n steps are denoted as... , , The temperature and humidity of all nodes are initialized with sensor measurements when n=0; where the field represents temperature, humidity or heat.

[0013] 1-3, Decoupling Iteration: The temperature obtained in the previous step Humidity (moisture content) Calculate the coupling parameters of the current grid, including thermal conductivity k, diffusion coefficient, breathing source term, and phase transition term. Fix these coupling parameters and solve the heat conduction-breathing equation separately to obtain a new intermediate value for the temperature field. With the updated temperature field, solve the moisture diffusion-phase transition equation separately to obtain a new intermediate value for the humidity field. Recalculate the coupling parameters using the new intermediate values ​​for the temperature and humidity fields as input for the next iteration. By repeatedly updating the parameters, solving for temperature and humidity, and updating the parameters again within a time step, the coupled solution of the two can be gradually approximated.

[0014] 1-4, Linear Solution: In each decoupling iteration, the heat conduction equation and the moisture diffusion equation are discretized into a system of linear algebraic equations in space. The temperature and humidity updates correspond to a sparse linear relationship, which is solved using an efficient sparse matrix algorithm. At the boundary nodes, the contributions of convective heat transfer and humidity exchange are distributed to the coefficients and right-hand side terms of the linear equations. The intermediate values ​​of the temperature field and humidity field obtained from the solution are the intermediate field values ​​of this iteration.

[0015] 1-5, Iterative Judgment: After completing one alternating update of temperature and humidity, it is necessary to check whether the coupled iteration has met the convergence requirements; calculate the maximum difference between the latest solution and the previous solution, which is the maximum temperature difference and the maximum humidity difference, and set a minimum threshold. If the maximum temperature difference and the maximum humidity difference are both less than the corresponding minimum threshold, then the coupled iteration in this time step is considered to have converged; if converged, the iteration in this time step is exited; if not converged, the current new solution is used as input, and the decoupled iteration step is returned to continue updating parameters and linear solution; the heat is calculated based on the temperature and humidity of the converged iteration.

[0016] 1-6, After convergence is determined, the grid node of the current time layer... , , That is, the temperature field, humidity field and heat field at that moment. If it does not converge, repeat the decoupling iteration → linear solution → convergence judgment until the convergence requirement is met. After completing one time step, the time layer n → n+1, and then the next iteration is carried out until the total simulation time ends or the predetermined termination condition is reached.

[0017] In some embodiments, the heat is calculated based on the temperature and humidity that converges iteratively:

[0018] The respiratory heat and phase change heat are calculated using the most recent temperature and humidity. Then, based on the heat conduction exchange between the node and its six neighboring nodes, the temperature gradient terms are calculated in the x, y, and z directions respectively. The difference is obtained by multiplying the gradient terms by the average local thermal conductivity in the corresponding directions. The net thermal power density is obtained by dividing the thermal power difference in each direction by the corresponding grid spacing. The results in the three directions are summed to obtain the heat "lost" by the node due to internal conduction, i.e., the heat outflow.

[0019] For nodes that are in direct contact with the warehouse wall or air, it is also necessary to consider the heat carried away by the convective heat exchange to accurately calculate the convective outflow heat. Read the convective heat transfer coefficient as well as the surface area and volume of the unit, multiply the difference between the node temperature and the ambient air temperature by the convective heat transfer coefficient and the surface area, and then divide by the volume to obtain the convective outflow heat.

[0020] Finally, the heat of respiration and the heat of phase change are added together, and then the heat of conduction and the heat of convection are subtracted to obtain the net heat source intensity. This net heat source intensity is then used as a component in the energy equation. .

[0021] In some embodiments, risk clusters are divided as follows:

[0022] A three-dimensional model of the temperature distribution of grain piles in the grain warehouse at various times is extracted. Temperature intervals are set. If the temperature of any node in the three-dimensional model is greater than the upper limit of the temperature interval, the node is marked in red; if the temperature of any node in the three-dimensional model is within the temperature interval, the node is marked in yellow; if the temperature of any node in the three-dimensional model is less than the lower limit of the temperature interval, the node is marked in blue. On the discrete three-dimensional mesh, nodes of the same risk level are clustered into risk clusters using 6-neighborhood or 26-neighborhood connectivity rules. Thus, the three-dimensional model of temperature distribution can be divided into several clusters.

[0023] In some embodiments, temperature risk cluster characteristic parameters, humidity risk cluster characteristic parameters, and heat risk cluster characteristic parameters are extracted from the risk clusters:

[0024] Extracting temperature cluster characteristic parameters from risk clusters: Calculate the volume of each risk cluster, calculate the nearest distance between the risk cluster and the warehouse wall or outlet, calculate the average temperature of each node in the risk cluster, and select the maximum temperature from each node of the risk cluster. Thus, the temperature risk cluster characteristic parameters can be obtained.

[0025] Similarly, humidity cluster characteristic parameters are extracted from risk clusters: calculate the volume of each risk cluster, calculate the nearest distance between the risk cluster and the warehouse wall or outlet, calculate the average humidity of each node in the risk cluster, and select the node with the highest humidity from the nodes of the risk cluster. Thus, the humidity risk cluster characteristic parameters can be obtained.

[0026] Similarly, heat cluster characteristic parameters are extracted from risk clusters: calculate the volume of each risk cluster, calculate the nearest distance between the risk cluster and the warehouse wall or outlet, calculate the average heat of each node in the risk cluster, and select the node with the maximum heat from each node of the risk cluster. Thus, the heat risk cluster characteristic parameters can be obtained.

[0027] In some embodiments, the comprehensive early warning index is calculated as follows:

[0028] The temperature risk cluster feature parameters, humidity risk cluster feature parameters, and heat risk cluster feature parameters are normalized and weighted fusion calculations to obtain temperature warning values, humidity warning values, and heat warning values; the calculation process for the temperature warning value is as follows:

[0029] The volume V, nearest distance L, average temperature F, and maximum temperature Fmax of the risk cluster are determined using the formula... and The temperature cluster feature parameters are normalized, and the normalized temperature cluster feature parameters are then weighted and fused to obtain the temperature warning value, which is denoted as follows. ,in The temperature representing the temperature state of a risk cluster is obtained by weighting the maximum temperature Fmax and the average temperature F within the risk cluster. The weighting formula is as follows: Vmin and Vmin are the minimum and maximum cluster volumes in the 3D temperature distribution model, respectively; Lmin is the minimum distance between the nearest cluster and the warehouse wall or vent in the 3D temperature distribution model; and Lmax is the maximum minimum distance. warm and T alarm These represent the lower and upper limits of the temperature range;

[0030] The same principle applies to humidity warning values ​​and heat warning values;

[0031] Temperature warning value Humidity warning value Heat warning value According to the formula The comprehensive early warning index (ECR) was calculated.

[0032] In some embodiments, advanced response measures and intermediate response measures:

[0033] The advanced response measures are as follows: extract the location of the advanced alarm cluster, enhance the ventilation of the area to level three, and call the robotic arm to turn over the entire area of ​​the site to break up the high temperature and high humidity clumps. After turning over the clumps, enhance the ventilation in the grain warehouse to level three, maintain it for a fixed time, and then restore it to the normal level one ventilation intensity.

[0034] The intermediate response measures are as follows: extract the location of the intermediate alarm cluster, upgrade the ventilation of this location to level two, and call on a robotic arm to locally turn over the piles in the area to break up the high-temperature and high-humidity clumps. After turning over the piles, maintain level two ventilation for a fixed period of time, and then restore normal level one ventilation.

[0035] Compared with the prior art, the beneficial effects of the present invention are:

[0036] 1. By constructing a partial differential coupled model based on heat conduction-biological respiration and moisture diffusion-phase change, dynamic simulation of the physical state of grain piles was achieved. This completely breaks through the limitations of traditional grain warehouses that can only rely on fixed-point data and static analysis. It not only has powerful data aggregation and modeling capabilities, but also can invert parameters such as thermal conductivity and diffusion coefficient of each node based on data collected by temperature and humidity sensors, realizing dynamic parameter updates and ensuring that the model always maintains a high degree of consistency with the actual grain pile state. At the same time, it considers the complex interaction between multiple physical fields in the warehouse (temperature affects moisture migration, moisture affects biological respiration, which in turn affects heat distribution, etc.), and constructs a scientific and complete coupled response chain, improving the physical realism of monitoring and the foresight of early warning.

[0037] 2. By numerically solving the partial differential equation model constructed on the server, the system outputs three-dimensional temperature, humidity, and heat field distribution data inside the grain pile. Utilizing various computational methods such as spatial discretization, temporal discretization, decoupled iteration, and sparse linear equation solving, it achieves efficient and stable simulation of the grain pile evolution process. An iterative convergence mechanism ensures computational accuracy, and by performing heat conservation analysis (conduction, convection, phase change, and respiration heat) on each node, it comprehensively understands the energy balance of each micro-unit, ensuring the model reflects the real physical processes. Special consideration is given to the boundary condition changes brought about by the grain silo structure (such as silo walls and ventilation openings), as well as the computational challenges posed by the structural complexity of the grain pile itself (such as vertical stratification and local clumping). Through layered iterative and refined solutions, the accuracy of local anomaly detection is greatly improved. Its core contribution lies in successfully transforming the theoretical results of partial differential modeling into a high-precision dynamic computing engine that is feasible and executable within the digital twin system, thereby supporting the refined modeling and high-reliability early warning of the entire system.

[0038] 3. Based on a three-dimensional temperature, humidity, and heat distribution model, and employing techniques such as regional clustering, threshold determination, cluster feature extraction, and comprehensive early warning index calculation, a precise, hierarchical, and responsive intelligent early warning system has been achieved. This system can not only extract high temperature, high humidity, and high heat risk points from a large number of nodes, but also form risk clusters using 6-neighbor or 26-neighbor connectivity clusters. For each risk cluster, characteristic parameters such as volume, average value, maximum value, and shortest distance are calculated to quantify the risk intensity and urgency of response. The ECR comprehensive early warning index, generated by weighted fusion of temperature, humidity, and heat risk warning values, can further subdivide risk clusters into low, medium, and high levels, and precisely trigger response mechanisms. This achieves risk control through hazard mitigation and stability maintenance through proactive measures. The advanced response scheme emphasizes system self-healing, rapidly intervening in high-risk areas through automated data turnover and strong ventilation, while the medium-level response reflects energy conservation and strategic approach, addressing only localized areas. This improves the accuracy, efficiency, and timeliness of early warning, truly achieving the goals of early detection, rapid response, and minimal losses.

[0039] In summary, this invention provides a complete coverage from sensor data to mechanism modeling, numerical simulation, 3D visualization, risk quantification, and intelligent response, truly realizing the application of digital twins in grain storage management. Online data acquisition, real-time simulation, and dynamic early warning solve the blind spots and time delays of traditional manual and localized sensing. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of this application or the conventional technology, the drawings used in the description of the embodiments or the conventional technology will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 This is a schematic diagram of the principle of the present invention;

[0042] Figure 2 This is a cable distribution diagram for monitoring grain conditions in a flat warehouse, as per the present invention.

[0043] Figure 3 This is a distribution diagram of the shallow circular grain storage monitoring cable of the present invention;

[0044] Figure 4 This is a schematic diagram of the iterative solution process of the present invention. Detailed Implementation

[0045] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of the present invention. However, the present invention can be practiced in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0046] like Figure 1 As shown, a grain warehouse monitoring and early warning system based on digital twin technology includes: a server, a transmission and analysis module, and an early warning management module;

[0047] The specific application scenarios of this invention are flat warehouses and shallow round warehouses, and the distribution diagram of their grain condition monitoring cables is shown below. Figures 2-3 As shown, the cable distribution in the flat warehouse is as follows: multiple monitoring cables are suspended at equal intervals along the X (depth) and Y (lateral) directions below the warehouse roof. The distances between the cables and from the cables to the walls or beams form a regular rectangular (or square) grid on the horizontal plane. Several probes are evenly suspended from the main cables. Multiple sensing units are arranged on each probe at height intervals. The length of the probe can be customized according to the height of the highest point of the grain pile to ensure that the entire pile is covered from top to bottom.

[0048] Distribution of monitoring cables for grain conditions in shallow circular silos: Several concentric circles are drawn on the bottom plane, and multiple sensor cable supports are arranged at equal angles on each circle. Figure 3At locations A and B, during the construction of the shallow circular silo, a pre-installed steel wire cable is placed at the entrance for pest trapping. A vertical support rod is usually also placed at the center to detect information at the deepest part of the core. Cables are led outward from the center along several radial channels to connect probes on different rings, forming a two-way grid of "radial + circumferential". Probes are hung at regular intervals (vertically) on the cables arranged in a planar manner. Each probe is equipped with multiple sensors, covering different heights from the top to the bottom of the silo, realizing multi-layer monitoring of the vertical profile.

[0049] The server communicates with various sensors within the grain silo to obtain temperature and humidity information at different locations within the silo, and simultaneously constructs a dynamically evolving coupled partial differential equation model; specifically:

[0050] Establish the heat conduction-biological respiration coupling equation: ,in Let ρ be the temperature field of the grain pile, and ρ be the density of the grain pile. ρ is the specific heat capacity, k is the thermal conductivity. For respiratory exothermic terms, The convective heat transfer coefficient is... The surface area of ​​the grain pile; Let represent the ambient air temperature. It should be noted that this equation is a partial differential equation describing temperature changes within biological tissues, and is commonly used in studies of biological heat transfer. This represents the rate of change of temperature over time. ∠ is a spatial term, representing the temperature change caused by heat conduction, and ∠ is the gradient operator, representing the rate of change in space; The term represents the convective heat exchange, indicating the temperature change caused by convective heat exchange with the environment. The entire formula considers the influence of three main factors: heat conduction, biological heat sources, and convective heat exchange.

[0051] Establish the moisture diffusion-phase transition coupling equation: ,in The moisture content of the grain pile, The water diffusion coefficient is... This is a phase change source term, representing the humidity change caused by evaporation (or a similar process). This refers to ambient humidity; it should be noted that... Let be the partial derivative of humidity with respect to time t, representing the rate of change of humidity over time. The diffusion term represents the flow of humidity caused by uneven distribution of humidity in space. This is an environmental exchange term, representing the exchange between humidity and the environment caused by humidity differences;

[0052] Temperature affects saturated water content and respiration rate; humidity changes alter heat dissipation and respiration parameters. The coupling can be approximated linearly. ,in Based on the thermal conductivity, The coupling coefficient between thermal conductivity and water content is... This is the coupling coefficient between thermal conductivity and temperature. As the reference diffusion coefficient, As the baseline moisture content, The reference temperature is used. It should be noted that, in order to take into account the mutual influence of temperature and humidity, a coupled expression for thermal conductivity and diffusion coefficient is introduced. This expression shows that thermal conductivity and diffusion coefficient are dynamically adjusted as temperature and humidity change.

[0053] Boundary and initial conditions: Temperature and humidity sensors are installed at various locations inside the grain silo to collect temperature and humidity data at various times.

[0054] By constructing a partial differential coupled model based on heat conduction-biological respiration and moisture diffusion-phase transition, dynamic simulation of the physical state of grain piles was achieved. This completely breaks through the limitations of traditional grain warehouses that can only rely on fixed-point data and static analysis. It not only has powerful data aggregation and modeling capabilities, but also can invert parameters such as thermal conductivity and diffusion coefficient of each node based on data collected by temperature and humidity sensors, realizing dynamic parameter updates and ensuring that the model always maintains a high degree of consistency with the actual state of the grain pile. At the same time, it considers the complex interaction between multiple physical fields in the warehouse (temperature affects moisture migration, moisture affects biological respiration, which in turn affects heat distribution, etc.), and constructs a scientific and complete coupled response chain, improving the physical realism of monitoring and the foresight of early warning.

[0055] like Figure 4 As shown, the conduction analysis module, based on the server's coupled partial differential equation model, can obtain the temperature, humidity, and heat at any point inside the grain pile, and thus construct a three-dimensional model of the temperature, humidity, and heat inside the grain pile; the specific solution process is as follows:

[0056] 1-1 Spatial Discretization: The entire three-dimensional region of the grain pile is divided into several regular small units (grids). The spatial step size Δx, Δy, and Δz are selected according to the size of the silo, and the nodes are evenly distributed. The initial temperature, moisture content, density, specific heat capacity, and other parameters corresponding to each node are mapped one-to-one with the sensor measurement points or initial assumed values. The outermost nodes are marked as boundary nodes, and the heat transfer flux and humidity exchange coefficient with the silo wall or air are recorded to provide boundary conditions for subsequent calculations. Through spatial discretization, the entire continuous medium is transformed into a discrete, point-by-point three-dimensional node matrix.

[0057] 1-2, Time Discretization: The total simulation period is divided into several hourly intervals Δt (time grid). Each update corresponds to a time level n→n+1. The integer n represents the current time. The values ​​of all fields (temperature, humidity, heat) at step n are denoted as... , , The temperature and humidity of all nodes are initialized with sensor measurements at n=0 (i.e., initially); the time discretization allows the model to progressively describe the evolution of the system in a step-by-step manner.

[0058] 1-3, Decoupling Iteration: In the coupled equations, the temperature field and humidity field influence each other. To simplify the solution, an iterative decoupling strategy is adopted. The temperature obtained in the previous step... Humidity (moisture content) Calculate the thermal conductivity k, diffusion coefficient, and coupling parameters such as the breathing source term (heat) and phase transition term for the current grid. Fix these parameters and solve the heat conduction-breathing equation separately to obtain a new intermediate value for the temperature field. With the updated temperature field (parameters can be recalculated if necessary), solve the moisture diffusion-phase transition equation separately to obtain a new intermediate value for the humidity field. Recalculate the coupling parameters using the new intermediate values ​​for the temperature and humidity fields as input for the next iteration. By repeatedly updating the parameters, solving the temperature and humidity problems, and updating the parameters again within a time step, the coupled solution of the two can be gradually approximated.

[0059] 1-4, Linear Solution: In each decoupling iteration, the heat conduction equation and the moisture diffusion equation are discretized into a system of linear algebraic equations in space. The temperature and humidity updates correspond to a large-scale sparse linear relationship, which is solved using efficient sparse matrix algorithms (such as conjugate gradient, Gauss-Seidel iteration, or multigrid). At the boundary nodes, the contributions of convective heat transfer and humidity exchange are distributed to the coefficients and right-hand side terms of the linear equations. The intermediate values ​​of the temperature field and humidity field obtained from the solution are the intermediate field values ​​of this iteration. The linear solution stage is crucial and determines the accuracy and convergence speed of each iteration.

[0060] 1-5. Iterative Judgment: After completing one alternating update of temperature and humidity, it is necessary to check whether the coupled iteration has met the convergence requirements. Calculate the maximum difference between the latest solution and the previous solution, which is the maximum temperature difference and the maximum humidity difference. Set a minimum threshold. If the maximum temperature difference and the maximum humidity difference are both less than the corresponding minimum threshold, the coupled iteration in this time step is considered to have converged. If converged, the iteration in this time step is exited. If not converged, the current new solution is used as input, and the decoupled iteration step is returned to continue updating parameters and linear solution.

[0061] 1-5.1, Heat Calculation: The respiratory heat and phase change heat are calculated using the most recent temperature and humidity. Then, based on the heat conduction exchange between the node and its six neighboring nodes (this heat may flow out or in), the temperature gradient terms are calculated in the x, y, and z directions respectively. The difference is multiplied by the average local thermal conductivity in the corresponding direction to obtain the flux difference. The flux difference in each direction is divided by the corresponding grid spacing to obtain the net thermal power density. The results in the three directions are summed to obtain the heat "lost" by the node due to internal conduction, i.e., the heat outflow of thermal conduction.

[0062] For nodes that are in direct contact with the warehouse wall or air, it is also necessary to consider the heat carried away by the convective heat exchange to accurately calculate the convective outflow heat. Read the convective heat transfer coefficient as well as the surface area and volume of the unit, multiply the difference between the node temperature and the ambient air temperature by the convective heat transfer coefficient and the surface area, and then divide by the volume to obtain the convective outflow heat.

[0063] Finally, the respiration heat and phase change heat are added together, and then the conduction outflow heat and convection outflow heat are subtracted to obtain the net heat source intensity. If the net heat source intensity > 0, it indicates that the node is releasing net heat; if the net heat source intensity < 0, it indicates that the node is absorbing or dissipating net heat. Simultaneously, the net heat source intensity is used as a component in the energy equation (heat conduction-biological respiration coupling equation). ;

[0064] 1-6, After convergence is determined, the grid node of the current time layer... , , This refers to the temperature field, humidity field, and heat field at that moment. If convergence is not achieved, the process of "decoupling iteration → linear solution → convergence judgment" is repeated until the accuracy requirements are met. After completing one time step, the time layer n→n+1 is moved to the next iteration until the total simulation duration ends or the predetermined termination condition is met.

[0065] By numerically solving the partial differential equation model built on the server, the system outputs three-dimensional temperature, humidity, and heat field distribution data inside the grain pile. Utilizing various computational methods such as spatial discretization, temporal discretization, decoupled iteration, and solving sparse linear equations, it achieves the ability to efficiently and stably simulate the evolution of the grain pile. An iterative convergence mechanism ensures computational accuracy, and by performing heat conservation analysis (conduction, convection, phase change, and respiration heat) on each node, it comprehensively understands the energy balance of each micro-unit, ensuring the model reflects the real physical processes. Special consideration is given to the boundary condition changes brought about by the grain silo structure (such as silo walls and ventilation openings), as well as the computational challenges posed by the structural complexity of the grain pile itself (such as vertical stratification and local clumping). Through layered iterative and refined solutions, the accuracy of local anomaly detection is greatly improved. Its core contribution lies in successfully transforming the theoretical results of partial differential modeling into a high-precision dynamic computing engine that is feasible and executable within the digital twin system, thereby supporting the refined modeling and high-reliability early warning of the entire system.

[0066] The early warning management module uses a three-dimensional distribution model of temperature, humidity, and heat to analyze and accurately identify risks in grain warehouses and issue warnings. Specifically:

[0067] Extract three-dimensional models of temperature distribution, humidity distribution, and heat distribution of grain piles within the grain silo at various times. Set temperature ranges; if the temperature of any node in the three-dimensional temperature distribution model exceeds the upper limit of the range, it indicates a high temperature and is marked in red; if the temperature of any node is within the range, it is marked in yellow; if the temperature of any node is below the lower limit, it is marked in blue. On the discrete three-dimensional mesh, using 6-neighborhood or 26-neighborhood connectivity rules, nodes of the same risk level are clustered into risk clusters, thus dividing the three-dimensional temperature distribution model into several clusters. Extract temperature cluster feature parameters from the risk clusters: calculate the volume of each risk cluster (V), calculate the nearest distance between the risk cluster and the silo wall or outlet (L), calculate the average temperature of each node within the risk cluster (F), and select the maximum temperature from the nodes of the risk cluster (Fmax). Then, use the formula... and The temperature cluster feature parameters are normalized, and the normalized temperature cluster feature parameters are then weighted and fused to obtain the temperature warning value, which is denoted as follows. ,in The temperature representing the temperature state of a risk cluster is obtained by weighting the maximum temperature Fmax and the average temperature F within the risk cluster. The weighting formula is as follows: Vmin and Vmin are the minimum and maximum cluster volumes in the 3D temperature distribution model, respectively; Lmin is the minimum distance between the nearest cluster and the warehouse wall or vent in the 3D temperature distribution model; and Lmax is the maximum minimum distance. warm and T alarm These represent the lower and upper limits of the temperature range. It should be noted that larger risk clusters cover more grains, have a wider impact range, and are more likely to cause mold, spontaneous combustion, or even insect infestation, resulting in greater losses. At the same temperature, large-area overheating will cause more grains to suffer stress, naturally increasing the overall risk level. Risk clusters with smaller minimum distances are located near condensation / heat exchange dead zones, making them more likely to form high-humidity and high-heat "blind spots," which are more difficult to detect and handle, and are also more dangerous.

[0068] Similarly, extract cluster feature parameters from risk clusters: calculate the volume of each risk cluster, calculate the nearest distance between the risk cluster and the warehouse wall or outlet, calculate the average humidity of each node within the risk cluster, select the node with the highest humidity from the nodes of the risk cluster, normalize the humidity cluster feature parameters, and then perform a weighted fusion calculation on the normalized humidity cluster feature parameters to obtain the humidity warning value, denoted as... ;

[0069] Similarly, heat cluster characteristic parameters are extracted from risk clusters: the volume of each risk cluster is calculated, the nearest distance between the risk cluster and the warehouse wall or outlet is calculated, the average heat of each node within the risk cluster is calculated, the maximum heat value is selected from each node of the risk cluster to normalize the heat cluster characteristic parameters, and the normalized heat cluster characteristic parameters are weighted and fused to obtain the heat warning value, which is denoted as... ;

[0070] Temperature warning value Humidity warning value Heat warning value According to the formula The comprehensive early warning index (ECR) is calculated, and an early warning range is set. If the comprehensive early warning index is greater than the maximum value of the early warning range, it means that at least one (or more) of temperature, humidity, or heat has reached or exceeded the corresponding alarm threshold, and their weighted combination has pushed the entire environment into a very critical state. In this case, the risk cluster is marked as a high-level alarm cluster, and high-level response measures are implemented. If the comprehensive early warning index is within the early warning range, the risk cluster is marked as a medium-level alarm cluster, and medium-level response measures are implemented. If the comprehensive early warning index is less than the lower limit of the early warning range, it means that temperature, humidity, and heat are all within the normal range. In this case, continuous monitoring and fixed level one ventilation are sufficient, and no other response measures are required.

[0071] Further, the advanced response measures are as follows: extract the location of the advanced alarm cluster, enhance the ventilation of this location to level three, and call the robotic arm to quickly turn over the pile at its location to break up the high temperature / high humidity clumps. After turning over the pile, enhance the ventilation in the grain warehouse to level three, maintain it for a fixed time, and then restore it to the normal level one ventilation intensity.

[0072] Further, the intermediate response measures are as follows: extract the location of the intermediate alarm cluster, upgrade the ventilation of this location to level two, and call the robotic arm to perform local small-scale turning of the pile at this location (turning the nearest 1 to 2 layers of grain pile is sufficient) to break up the high temperature / high humidity clumps. After turning the pile, maintain level two ventilation for a fixed period of time, and then restore normal level one ventilation.

[0073] Based on a three-dimensional temperature, humidity, and heat distribution model, and employing techniques such as regional clustering, threshold determination, cluster feature extraction, and comprehensive early warning index calculation, a precise, hierarchical, and responsive intelligent early warning system has been achieved. This system can not only extract high temperature, high humidity, and high heat risk points from a large number of nodes, but also form risk clusters using 6-neighbor or 26-neighbor connectivity clusters. For each risk cluster, characteristic parameters such as volume, average value, maximum value, and shortest distance are calculated to quantify the risk intensity and urgency of response. The ECR comprehensive early warning index, generated by weighted fusion of temperature, humidity, and heat risk warning values, can further subdivide risk clusters into low, medium, and high levels, precisely triggering response mechanisms. This achieves risk control through hazard mitigation and stability maintenance through proactive measures. The advanced response scheme emphasizes system self-healing, rapidly intervening in high-risk areas through automated data turnover and strong ventilation, while the medium-level response reflects energy conservation and strategic approach, addressing only localized areas. This improves the accuracy, efficiency, and timeliness of early warning, truly achieving the goals of early detection, rapid response, and minimal losses.

[0074] The above formulas are all derived from software simulation using a large amount of data, and are selected to be close to the actual values. The coefficients in the formulas are set by those skilled in the art based on the actual situation.

[0075] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0076] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A grain warehouse monitoring and early warning system based on digital twin technology, characterized in that, include: The server communicates with various sensors in the grain warehouse to obtain temperature and humidity information at various locations within the grain warehouse. At the same time, it builds a dynamic evolution coupled partial differential equation model and provides initial conditions and boundaries to dynamically simulate the physical state of the grain pile. The conduction analysis module solves the coupled partial differential equation model based on the server to obtain the temperature, humidity and heat at any point inside the grain pile, and then constructs a three-dimensional model of temperature, humidity and heat distribution based on this. The early warning management module divides the entire grain pile into several risk clusters based on a three-dimensional model of temperature, humidity and heat distribution. It performs temperature, humidity and heat risk analysis on each risk cluster to obtain a comprehensive early warning index, and compares and analyzes it with the early warning interval. If the comprehensive early warning index is greater than the upper limit of the early warning interval, the risk cluster is marked as a high-level alarm cluster and advanced response measures are implemented for it. If the comprehensive early warning index is within the warning range, the risk cluster is marked as a medium-level alarm cluster and medium-level response measures are implemented for it; if the comprehensive early warning index is less than the lower limit of the warning range, continuous monitoring and fixed level one ventilation are sufficient, and no other response measures are required.

2. The grain warehouse monitoring and early warning system based on digital twin technology according to claim 1, characterized in that, Server-coupled partial differential equation model construction: Establish the heat conduction-biological respiration coupling equation: ,in Let ρ be the temperature field of the grain pile, and ρ be the density of the grain pile. ρ is the specific heat capacity, k is the thermal conductivity. For respiratory exothermic terms, The convective heat transfer coefficient is... The surface area of ​​the grain pile; The outside air temperature, It is a spatial term, and ▽ is the gradient operator. For convective heat exchange; Establish the moisture diffusion-phase transition coupling equation: ,in The moisture content of the grain pile, The water diffusion coefficient is... For phase transition source terms, For ambient humidity, For diffusion term, This is an environmental exchange item.

3. A grain warehouse monitoring and early warning system based on digital twin technology according to claim 2, characterized in that, The solution process for the coupled partial differential equation model is as follows: 1-1, Spatial Discretization: The entire three-dimensional region of the grain pile is divided into several regular small units, i.e., grid nodes. The spatial step size Δx, Δy, Δz is selected according to the size of the silo, and the nodes are evenly distributed. The initial temperature, moisture content, grain pile density, specific heat capacity parameters corresponding to each node are matched with the sensor measurement points or initial assumed values. The outermost nodes are marked as boundary nodes, and the heat transfer flux and humidity exchange coefficient with the silo wall or air are recorded to provide boundary conditions for subsequent calculations. 1-2, Time Discreteness: The total simulation period is divided into several hourly steps Δt. Each step updates a corresponding time level. The current time is represented by an integer n, and the values ​​of all fields at n steps are denoted as... , , Initialize the temperature and humidity of all nodes with sensor measurements when n=0; The fields represent temperature, humidity, or heat. 1-3, Decoupling Iteration: The temperature obtained in the previous step Humidity (moisture content) Calculate the coupling parameters of the current grid, including thermal conductivity k, diffusion coefficient, breathing source term, and phase transition term. Fix the above coupling parameters and solve the heat conduction-breathing equation separately to obtain a new intermediate value of the temperature field. With the updated temperature field, solve the moisture diffusion-phase transition equation separately to obtain a new intermediate value of the humidity field. The coupling parameters are recalculated using the new intermediate values ​​of the temperature and humidity fields, and used as input for the next iteration. By repeatedly updating the parameters, solving for temperature and humidity, and updating the parameters again within a time step, the coupling solution of the two fields can be gradually approximated. 1-4, Linear Solution: In each decoupling iteration, the heat conduction equation and the moisture diffusion equation are discretized into a system of linear algebraic equations in space. The temperature and humidity updates correspond to a sparse linear relationship, which is solved using an efficient sparse matrix algorithm. At the boundary nodes, the contributions of convective heat transfer and humidity exchange are distributed to the coefficients and right-hand side terms of the linear equations. The intermediate values ​​of the temperature field and humidity field obtained from the solution are the intermediate field values ​​of this iteration. 1-5, Iterative Judgment: After completing one alternating update of temperature and humidity, it is necessary to check whether the coupled iteration has met the convergence requirements; calculate the maximum difference between the latest solution and the previous solution, which is the maximum temperature difference and the maximum humidity difference, and set a minimum threshold. If the maximum temperature difference and the maximum humidity difference are both less than the corresponding minimum threshold, then the coupled iteration in this time step is considered to have converged; if converged, the iteration in this time step is exited; if not converged, the current new solution is used as input, and the decoupled iteration step is returned to continue updating parameters and linear solution; the heat is calculated based on the temperature and humidity of the converged iteration. 1-6, After convergence is determined, the grid node of the current time layer... , , That is, the temperature field, humidity field and heat field at that moment. If it does not converge, repeat the decoupling iteration → linear solution → convergence judgment until the convergence requirement is met. After completing one time step, the time layer n → n+1, and then the next iteration is carried out until the total simulation time ends or the predetermined termination condition is reached.

4. A grain warehouse monitoring and early warning system based on digital twin technology according to claim 3, characterized in that, The heat is calculated based on the temperature and humidity that converge during the iterative process. The respiratory heat and phase change heat are calculated using the most recent temperature and humidity. Then, based on the heat conduction exchange between the node and its six neighboring nodes, the temperature gradient terms are calculated in the x, y, and z directions respectively. The difference is obtained by multiplying the difference by the average local thermal conductivity in the corresponding direction. The net thermal power density is obtained by dividing the difference in each direction by the corresponding grid spacing. The results in the three directions are summed to obtain the heat "lost" by the node due to internal conduction, i.e., the heat outflow. For nodes that are in direct contact with the warehouse wall or air, it is also necessary to consider the heat carried away by the convective heat exchange to accurately calculate the convective outflow heat. Read the convective heat transfer coefficient as well as the surface area and volume of the unit, multiply the difference between the node temperature and the ambient air temperature by the convective heat transfer coefficient and the surface area, and then divide by the volume to obtain the convective outflow heat. Finally, the heat of respiration and the heat of phase change are added together, and then the heat of conduction and the heat of convection are subtracted to obtain the net heat source intensity. This net heat source intensity is then used as a component in the energy equation. .

5. A grain warehouse monitoring and early warning system based on digital twin technology according to claim 4, characterized in that, Risk cluster classification: A three-dimensional model of the temperature distribution of grain piles in the grain warehouse at various times is extracted. Temperature intervals are set. If the temperature of any node in the three-dimensional model is greater than the upper limit of the temperature interval, the node is marked in red; if the temperature of any node in the three-dimensional model is within the temperature interval, the node is marked in yellow; if the temperature of any node in the three-dimensional model is less than the lower limit of the temperature interval, the node is marked in blue. On the discrete three-dimensional mesh, nodes of the same risk level are clustered into risk clusters using 6-neighborhood or 26-neighborhood connectivity rules. Thus, the three-dimensional model of temperature distribution can be divided into several clusters.

6. A grain warehouse monitoring and early warning system based on digital twin technology according to claim 5, characterized in that, Extract characteristic parameters of temperature risk cluster, humidity risk cluster, and heat risk cluster from the risk cluster: from Extracting temperature cluster characteristic parameters from risk clusters: Calculate the volume of each risk cluster, calculate the nearest distance between the risk cluster and the warehouse wall or outlet, calculate the average temperature of each node within the risk cluster, and select the maximum temperature from each node of the risk cluster. Thus, the temperature risk cluster characteristic parameters can be obtained. Similarly, humidity cluster characteristic parameters are extracted from risk clusters: calculate the volume of each risk cluster, calculate the nearest distance between the risk cluster and the warehouse wall or outlet, calculate the average humidity of each node in the risk cluster, and select the node with the highest humidity from the nodes of the risk cluster. Thus, the humidity risk cluster characteristic parameters can be obtained. Similarly, heat cluster characteristic parameters are extracted from risk clusters: calculate the volume of each risk cluster, calculate the nearest distance between the risk cluster and the warehouse wall or outlet, calculate the average heat of each node in the risk cluster, and select the node with the maximum heat from each node of the risk cluster. Thus, the heat risk cluster characteristic parameters can be obtained.

7. A grain warehouse monitoring and early warning system based on digital twin technology according to claim 6, characterized in that, Calculation method of comprehensive early warning index: The temperature risk cluster feature parameters, humidity risk cluster feature parameters, and heat risk cluster feature parameters are normalized and weighted fusion calculations are performed to obtain the temperature warning value, humidity warning value, and heat warning value. The temperature warning value, humidity warning value and heat warning value are then weighted and integrated to obtain the comprehensive warning index.

8. A grain warehouse monitoring and early warning system based on digital twin technology according to claim 7, characterized in that, Advanced response measures and intermediate response measures: The advanced response measures are as follows: extract the location of the advanced alarm cluster, enhance the ventilation of the area to level three, and call the robotic arm to fully turn over the pile in the area to break up the clumps. After turning over the pile, enhance the ventilation in the grain warehouse to level three, maintain it for a fixed time, and then restore the ventilation to the normal level one intensity. The intermediate response measures are as follows: extract the location of the intermediate alarm cluster, upgrade the ventilation of this location to level two, and call in a robotic arm to locally turn over the clumps in the area to break them up. After turning over the clumps, maintain level two ventilation for a fixed period of time, and then restore normal level one ventilation.

Images