A method for optimizing defrosting efficiency of a commercial refrigerator

Abstract

The application discloses a commercial refrigerator defrosting efficiency optimization method, and relates to the technical field of refrigerator defrosting.The method comprises the following steps: step one, key parameter acquisition and sensor grid layout;step two, discrete thermal resistance network inversion;step three, multi-factor edge level feature extraction and frost index calculation;step four, frost area identification and space boundary generation;step five, defrosting priority calculation and partition scheduling;step six, iteration verification and end determination.Through the innovative link of thermal resistance network inversion, frost edge marking, space mapping and attribute defrosting, the method completely abandons the extensive defrosting of fixed time and one-size-fits-all, truly realizes on-demand, accurate and energy-saving defrosting based on the frost degree and the importance of the position, fundamentally solves the two extreme problems of excessive defrosting and insufficient defrosting, and optimizes the defrosting efficiency.

Description

Technical Field

[0001] This invention relates to the field of freezer defrosting technology, specifically a method for optimizing defrosting efficiency in commercial freezers. Background Technology

[0002] Once frost forms on the evaporator surface, the frost layer becomes a thermal resistance, severely weakening the heat exchange efficiency between the refrigerant and the air inside the cabinet. Decreased heat exchange efficiency leads to the compressor working more frequently or for longer periods, significantly increasing energy consumption. The frost layer hinders the circulation of cold air, easily forming hot spots inside the cabinet, preventing food from being fully cooled. To compensate for the temperature rise, the control system repeatedly starts the refrigeration cycle, causing frequent temperature fluctuations and affecting product quality. Regular defrosting of commercial freezers is not only a key step in maintaining normal equipment operation and saving energy, but also a necessary maintenance method to ensure food quality and safety and extend the service life of the equipment. Most existing commercial refrigerators use timed electric heating for defrosting. Regardless of the degree of frost or the intensity of use, the defrosting is started at a set interval, which can easily lead to either over-defrosting or under-defrosting. Over-defrosting wastes energy, while under-defrosting causes frost to accumulate and affect performance. Summary of the Invention

[0003] The purpose of this invention is to solve the problems mentioned in the background section by proposing a method for optimizing the defrosting efficiency of commercial freezers.

[0004] The objective of this invention can be achieved through the following technical solution: a method for optimizing defrosting efficiency in commercial freezers, comprising: Step 1: Deploy temperature sensors, heat flux meters, micro-differential pressure sensors, and line laser scanning modules in the evaporator area, shelf space layers, and back panel aisles to construct a node-edge 3D sensing mesh, and acquire a reference surface model under frost-free conditions; specifically: Several high-precision temperature sensors, miniature heat flux meters, and micro-differential pressure sensors are installed in key parts of the upright freezer. A wired laser scanning module (or structured light depth camera) is also installed to cover these key parts, including the evaporator area, shelf space layers, and back panel aisles. Each temperature-heat flux sensor location is designated as a node, and the heat conduction path between adjacent nodes is considered an edge, representing the equivalent thermal resistance. Based on the sensor mesh of each key part, nodes are numbered i=1, 2…N, and adjacent point pairs (i, j) are defined as forming thermal resistance edges. A batch of 3D point clouds is obtained by scanning the physical region containing the scanning path (i, j). Simultaneously, in a frost-free state, the same recognition method is used to scan the same physical region to fit a reference plane or curved surface of the bare metal or frost-free surface, collectively referred to as the reference surface model, denoted as: This model can be obtained by least squares plane fitting or higher-order surface fitting; Step 2: Based on Fourier's law and the nodal energy conservation equation, the discretized thermal resistance network is inverted, and the thermal resistance difference network under operating conditions and frost-free calibration is calculated. Step 3: Extract thermal resistance factor, heat transfer factor, flow resistance factor, and frost thickness factor using multi-factor edge-level features, and then... Heat flux factor Flow resistance factor Thickening factor of cream According to the formula The frosting index is calculated, and from this, the frosting index of each heat conduction path can be obtained. When the frost index is greater than the frost judgment threshold, the heat conduction path, i.e. the edge, is marked as a frost edge, thus the frost edges in the entire sensor network can be marked. Step 4: Identify the clustering subgraph of the frost edge set using the connected component algorithm, extract the 3D coordinates of the nodes to generate the spatial boundary, and map it to the CAD model to annotate the work area attributes; Step 5: Calculate the defrost value using a formula based on the cumulative frost index, maximum frost index, and area volume of the spatial boundary; then perform attribute-based defrost operations in descending order of the defrost values. Step 6: After defrosting, recheck the frost index. If frost is present, return to Step 1; otherwise, terminate the process.

[0005] In a preferred embodiment of the present invention, the process of inverting the discretized thermal resistance network is as follows: 2-1, Derivation of Network Equations: First, in steady state and assuming that heat flow is only conducted along the i→j direction, we can obtain the following from Fourier's law: T i and T j The temperatures of nodes i and j are respectively, obtained by temperature sensors. R represents the heat flux along the i→j direction, which is directly measured by a heat flux meter. ij This represents the equivalent thermal resistance between nodes i and j, including all contributions from the thermal conductivity of the metal fins, the air gap, and the additional impedance of the frost layer. 2-2. In actual network deployment, not all edges have heat flux meters. In this case, an equilibrium equation is constructed for the entire sensor network, and the equivalent thermal resistance R of all edges is collectively derived. ij ; 2-3, the thermal resistance network {R} under operating conditions ij} and the reference network thermal resistance under frost-free calibration { According to the formula The thermal resistance difference network {ΔR} is obtained by performing difference calculation. ij}

[0006] In a preferred embodiment of the present invention, the equilibrium equation and the collective inversion process are as follows: In steady state, the sum of the heat flow entering and leaving each node i must equal the heat flow exchanged between that node and the external environment. Energy conservation is applied to node i. ,in Let i represent the set of nodes adjacent to node i. Let be the external heat flow, representing the net heat flow absorbed by node i from external areas such as evaporative cooling flow or shelf surfaces. This is equal to the algebraic sum of the heat flux meter readings at that location. Then, a balance equation is constructed for each node: The left side of the equilibrium equation represents the convergence of each heat flux, while the right side represents the external heat flux input. Therefore, the known condition is: nodal temperature T. i External heat flow Unknown condition: Thermal resistance R of each side ij , Solving for the thermal resistance of the unknown conditions based on known conditions: Under frost-free calibration conditions, measure the temperature and heat flux at all network points and solve for the thermal resistance of the reference network. }, using the least squares method or constrained fitting, optimize {R} ij To minimize the error between the model heat flux and the observed heat flux, when The iteration stops when the value is less than μ, and the thermal resistance R of each side is obtained. ij， μ represents the convergence error threshold.

[0007] In a preferred embodiment of the present invention, the extraction process of thermal resistance factor and heat transfer factor is as follows: Thermal resistance factor extraction: For each heat conduction path in the sensor network, i.e., edge (i, j), read the thermal resistance difference of each edge, and extract the maximum thermal resistance difference, denoted as . The thermal resistance factor is obtained by normalizing the thermal resistance difference by dividing the thermal resistance difference of each heat conduction path by the maximum thermal resistance difference. The larger the value, the more significant the increase in thermal resistance; Heat flux factor extraction: Extracting the heat flux of each heat conduction path. and heat flux under frost-free calibration According to the formula Complete the heat flux The heat flux factor was obtained by normalization. When the heat flux drops significantly, the closer the normalized heat flux factor value is to 1, the more severe the blockage.

[0008] In a preferred embodiment of the present invention, the extraction process of the flow resistance factor and the frost thickness factor is as follows: Flow resistance factor extraction: Extract the wind speeds at nodes i and j, and calculate the average wind speed along the heat conduction path from i to j, denoted as . Simultaneously, the pressures at nodes i and j are extracted, and the difference between them is calculated and the absolute value is taken to obtain the pressure difference along the heat conduction path from i to j, denoted as . Extract the wind speed along the heat conduction path from i to j under frost-free calibration. and pressure difference The flow resistance factor is obtained by measuring the degree of blockage using relative change. The quantification formula is ; Frost thickness factor extraction: Extract the real-time point cloud of each heat conduction path (i, j). For each point, calculate its height difference from the frost-free reference surface. This yields the height differences of each point in the heat conduction path (i, j). The frost thickness of the heat conduction path (i, j) is then calculated by performing an arithmetic mean. For each heat conduction path in the mesh, i.e., each edge, there exists a frost layer thickness. This represents the average thickness of the frost layer along this heat conduction path; the maximum frost layer thickness selected from the sensor network is denoted as . According to the formula Frost thickness for all heat conduction paths The frost thickness factor was obtained by normalization and denoted as . Its value quantitatively reflects the physical thickness of the frost layer; the larger the value, the thicker the adhesion.

[0009] In a preferred embodiment of the present invention, the spatial boundary generation process is as follows: The entire sensor network is abstracted as a graph G = (V, E), where V represents all sensor nodes and E represents all adjacent node pairs, using frost edge sets. Constructing subgraphs In the subgraph The connected component algorithm is run on the refrigerator, starting from an unvisited frost edge and traversing all connected frost edges along the edge until no further extension is possible. Each traversal yields a connected component, which is a set of interconnected frost edges. Each connected component reflects a continuous frosting network formed inside the refrigerator, i.e., a local or cross-area frosting zone. Let the k-th connected component contain the set of edges. The corresponding set of nodes ,but{ , } represents the initial frost region; k is a positive integer representing the index of the connected component; After completing the frost edge clustering analysis and identifying each connected component, the spatial boundary is calculated for each connected component of the frost region—that is, the subgraph formed by all interconnected frost edges—to determine the approximate location of the preliminary frost region in the physical device and obtain the spatial boundary. Specifically, the three-dimensional coordinates of all endpoint nodes are extracted from each connected component. These coordinates are provided by the corresponding sensor module and have been mapped to a unified cabinet coordinate system through the global coordinate alignment module. Then, the minimum and maximum values ​​of the point set on the three orthogonal axes x, y, and z are taken, and these extreme coordinates are spliced ​​together to construct an axis-aligned minimum cuboid. The cuboid precisely encloses all the sensor nodes belonging to frost region k. Finally, each preliminary frost region k will be marked with a spatial boundary. By projecting the spatial boundary onto the 3D CAD model of the freezer, identifying the key parts where the frost zone is located, and labeling the spatial boundary according to the key parts, the working area and the attributes of each spatial boundary on the physical equipment are generated. The working area attribute is the name of the key part.

[0010] In a preferred embodiment of the present invention, the defrosting priority calculation and partition scheduling process for spatial boundaries is as follows: Extract all spatial boundaries The spatial boundary is calculated by summing the frost indices of all frost edges within it, along with their corresponding frost indices. The cumulative frost index is denoted as Selecting spatial boundaries The highest frosting index Calculate the spatial boundary The volume of space is denoted as From all spatial boundaries The largest spatial volume is denoted as . According to the formula Calculations are performed to obtain the defrost value. It should be noted that the spatial boundary The larger the area volume, the higher the defrost priority and the higher the defrost value; the higher the maximum frost index, the higher the defrost priority and the higher the defrost value; the higher the cumulative frost index, the higher the priority and the higher the defrost value. All spatial boundaries are sorted sequentially according to their defrost values ​​from largest to smallest, and defrosting is performed sequentially in this order.

[0011] In a preferred embodiment of the present invention, the specific defrosting operation of the spatial boundary is as follows: S101: Identify the work area and its attributes at the spatial boundary. If there is a work area that has not been defrosted, then execute S102; otherwise, it indicates that the defrosting operation at the spatial boundary is complete. S102, determine whether the attribute of the work area is an evaporator. If so, perform defrosting operation on this work area using an electric heating wire; otherwise, proceed to S103. S103, determine whether the attribute of the work area is a shelf gap. If so, use directional hot air pulse to perform defrosting operation; otherwise, jump to S104. S104, determine whether the attribute of the work area is a back panel channel. If so, use hot air circulation to perform defrosting operation; otherwise, jump to S101.

[0012] Compared with the prior art, the beneficial effects of the present invention are: 1. By treating each pair of adjacent sensing nodes as equivalent thermal resistance edges and applying Fourier's law and the node energy conservation equation, the temperature difference and heat flux measurement data are mapped into a network thermal resistance distribution with clear physical meaning. By comparing and solving the reference thermal resistance and the actual thermal resistance under frost-free calibration and operating conditions, the thermal resistance increment of each edge can be quantified. This allows the dispersed sensing signals to be fused into a globally consistent thermal resistance network, so that the complex three-dimensional heat flow field can be compressed into discretized network parameters. The physical information contained in the small temperature difference and heat flow error can be inverted through the least squares method, thereby achieving accurate positioning of the heat transfer capacity attenuation under the influence of frost, and providing quantifiable and comparable core indicators for intelligent control decision-making. 2. By extracting multi-factor side-level features and calculating the frost index, the physical quantities of thermal resistance increment, heat flux decrease, airflow blockage, and frost thickness are normalized and weighted and fused to generate a unified frost index. This overcomes the uncertainty of a single indicator and uses cross-validation of physical quantities from multiple angles and levels to significantly improve the robustness and accuracy of frost condition judgment. The final frost index has both a comprehensive judgment capability under statistical significance and takes into account physical intuition, thus providing a high-confidence decision basis for accurately selecting defrosting areas. 3. By identifying frost zones and generating spatial boundaries, all frost paths are automatically clustered into several independent frost zones. The minimum axis-aligned bounding box is calculated using sensor node coordinates to accurately obtain the boundary of each frost zone in three-dimensional space. This eliminates the subjective errors present in traditional manual inspections and achieves automated and reproducible area extraction based on the entire network of sensor data. At the same time, the geometric overlap determination between the spatial boundary and the CAD model can quickly label the area, clearly identifying whether it belongs to functional components such as evaporator blades, shelf gaps, or back panel channels. This provides clear operational units for subsequent zone scheduling and differentiated defrosting process selection. Maintenance personnel can intuitively understand the distribution of frost zones, improving operational efficiency and minimizing errors. 4. By comprehensively considering the boundary volume, maximum frost index, and cumulative frost score of each spatial boundary, a quantifiable defrost value is generated, and these values ​​are sorted from largest to smallest. The optimal defrost strategy is then executed sequentially for each sub-area. This eliminates the limitations of a single indicator within a multi-dimensional decision-making system, ensuring that critical areas with large volume, high thermal resistance, and high frost volume are prioritized for processing, maximizing defrost efficiency and system operational safety. Simultaneously, through the streamlined design of attribute judgment processes, intelligent and modular operation scheduling is achieved, significantly reducing energy consumption and shortening operation time. It can also be flexibly expanded to new parts or new technologies, enabling easy maintenance and upgrades in the future. 5. By iteratively verifying and determining the end of the defrosting process after completing defrosting operations at all spatial boundaries, the overall frost index is recalculated and the remaining frost edges are checked to determine whether it is necessary to re-enter the defrosting cycle. This not only verifies the actual effect of the defrosting strategy in real time, avoiding potential hazards left by surface defrosting, but also captures new frost conditions caused by heat flow redistribution or sudden environmental changes during the defrosting process, achieving dynamic adaptive management. When the thermal resistance of all heat conduction paths returns to below the safe threshold, the system intelligently determines that the current defrosting cycle is complete, thereby avoiding unnecessary repetitive operations and energy consumption. This iterative closed loop ensures the continuity and reliability of the defrosting work, truly combining efficient defrosting with energy-saving optimization, and realizing the best practice for intelligent defrosting of large upright commercial freezers. In summary, this solution, through an innovative process of thermal resistance network inversion, frost edge annotation, spatial mapping, and attribute-based defrosting, completely abandons the crude, time-based defrosting method. It truly achieves on-demand, precise, and energy-saving defrosting based on the degree of frost formation and the importance of the affected area, fundamentally solving the two extreme problems of over-defrosting and under-defrosting, and optimizing defrosting efficiency. Attached Figure Description

[0013] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0014] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

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

[0016] Please see Figure 1 As shown, a method for optimizing defrosting efficiency in commercial freezers is presented. This method is applied to large upright commercial freezers and includes: Step 1: Key parameter acquisition and sensor grid deployment; Several high-precision temperature sensors, miniature heat flux meters, and micro-differential pressure sensors are installed in key parts of the upright freezer. A wired laser scanning module (or structured light depth camera) is also installed to cover these key parts, including the evaporator area, shelf space layers, and back panel aisles. Each temperature-heat flux sensor location is designated as a node, and the heat conduction path between adjacent nodes is considered an edge, representing the equivalent thermal resistance. Based on the sensor mesh of each key part, nodes are numbered i=1, 2…N, and adjacent point pairs (i, j) are defined as forming thermal resistance edges. A batch of 3D point clouds is obtained by scanning the physical region containing the scanning path (i, j). Simultaneously, in a frost-free state, the same recognition method is used to scan the same physical region to fit a reference plane or curved surface of the bare metal or frost-free surface, collectively referred to as the reference surface model, denoted as: This model can be obtained by least squares plane fitting or higher-order surface fitting; By deploying high-precision temperature sensors, miniature heat flux meters, micro differential pressure sensors, and line laser scanning modules in key areas such as the evaporator area, shelf space layers, and back panel aisles of the vertical freezer, a node-edge three-dimensional sensing mesh is constructed. Simultaneously, a reference surface model is acquired under frost-free conditions, significantly improving the system's observability of the internal thermal and flow states of the freezer, laying a complete data foundation for subsequent refined diagnostics. The meshed layout not only ensures coverage of key heat exchange paths such as evaporator fin gaps, shelf gaps, and back panel air ducts, but also provides a directly mountable topology for subsequent networked thermal resistance inversion and frost zone location. This enables the entire system to possess high-resolution, traceable data acquisition capabilities, laying a solid foundation for intelligent defrosting.

[0017] Step 2: Discretized thermal resistance network inversion; Network equation derivation: First, in steady state and assuming heat flow is only conducted along the i→j direction, we can obtain the following from Fourier's law: T i and T j The temperatures of nodes i and j are respectively, obtained by temperature sensors. R represents the heat flux along the i→j direction, which is directly measured by a heat flux meter. ij This represents the equivalent thermal resistance between nodes i and j, including contributions from metal fin thermal conduction, air gaps, and additional impedance from frost. The local resistance R during frost formation is... ij The thermal resistance is significantly increased; if temperature difference and heat flux sensors are simultaneously installed at nodes i and j, the equivalent thermal resistance R between nodes can be directly calculated. ij This includes the thermal resistance of the base metal and the additional thermal resistance of any frost layer that may be present. In actual network deployment, not all edges have heat flux meters. In this case, it is necessary to construct an equilibrium equation for the entire sensor network and collectively deduce the equivalent thermal resistance R of all edges. ijUnder steady state, the sum of the heat flow entering and leaving each node i must be equal to the heat flow exchanged between that node and the external environment (which can be measured by the heat flux meter at that point or assumed to be zero). Energy conservation is applied to node i (inflow = outflow): ,in Let i represent the set of nodes adjacent to node i. Let be the external heat flow, representing the net heat flow absorbed by node i from external areas such as evaporative cooling flow or shelf surfaces. This is equal to the algebraic sum of the heat flux meter readings at that point (which can be positive or negative). Then, a balance equation is constructed for each node: The left side of the equilibrium equation represents the convergence of each heat flux, while the right side represents the external heat flux input. Therefore, the known condition is: nodal temperature T. i External heat flow Unknown condition: Thermal resistance R of each side ij The thermal resistance is calculated based on known and unknown conditions: Temperature and heat flux at all network points are measured under frost-free calibration conditions, and the thermal resistance of the reference network is determined. }, using the least squares method or constrained fitting, optimize {R} ij To minimize the error between the model heat flux and the observed heat flux, when The iteration stops when the value is less than μ (convergence error threshold), and the thermal resistance R of each side is obtained. ij ; The thermal resistance network {R} under operating conditions ij} and the reference network thermal resistance under frost-free calibration { According to the formula The thermal resistance difference network {ΔR} is obtained by performing difference calculation. ij}; By treating each pair of adjacent sensing nodes as equivalent thermal resistance edges and applying Fourier's law and the nodal energy conservation equation, temperature difference and heat flux measurement data are mapped into a network thermal resistance distribution with clear physical meaning. By comparing and solving the benchmark thermal resistance and the actual thermal resistance under frost-free calibration and operating conditions, not only can the thermal resistance increment of each edge be quantified, but also the local heat transfer bottlenecks caused by frost adhesion or flow channel blockage can be revealed. The greatest value of this step is that it integrates the dispersed sensing signals into a globally consistent thermal resistance network, so that the complex three-dimensional heat flow field can be compressed into discretized network parameters. The physical information contained in the small temperature difference and heat flow error can be inverted by the least squares method, thereby realizing the accurate location of the heat transfer capacity attenuation under the influence of frost, and providing quantifiable and comparable core indicators for intelligent control decision-making.

[0018] Step 3: Multi-factor side-level feature extraction and frost index calculation; For each heat conduction path in the sensor network, i.e., edge (i, j), read the thermal resistance difference of each edge, and extract the maximum thermal resistance difference as . The thermal resistance factor is obtained by normalizing the thermal resistance difference by dividing the thermal resistance difference of each heat conduction path by the maximum thermal resistance difference. The larger the value, the more significant the increase in thermal resistance; Extracting the heat flux of each heat conduction path and heat flux under frost-free calibration According to the formula Complete the heat flux The heat flux factor was obtained by normalization. When the heat flux drops significantly, the closer the normalized heat flux factor value is to 1, the more severe the blockage. Extract the wind speeds at nodes i and j, and calculate their average to obtain the wind speed along the heat conduction path from i to j, denoted as . Simultaneously, the pressures at nodes i and j are extracted, and the difference between them is calculated and the absolute value is taken to obtain the pressure difference along the heat conduction path from i to j, denoted as . Extract the wind speed along the heat conduction path from i to j under frost-free calibration. and pressure difference The flow resistance factor is obtained by measuring the degree of blockage using relative change. The quantification formula is ; Extract the real-time point cloud for each heat conduction path (i, j). For each point, calculate its height difference from the frost-free reference surface. This yields the height differences between points along the heat conduction path (i, j). The frost thickness along the heat conduction path (i, j) is then calculated by performing an arithmetic mean. For each heat conduction path in the mesh, i.e., each edge, there exists a frost layer thickness. This represents the average thickness of the frost layer along this heat conduction path; the maximum frost layer thickness selected from the sensor network is denoted as . According to the formula Frost thickness for all heat conduction paths The frost thickness factor was obtained by normalization and denoted as . Its value quantitatively reflects the physical thickness of the frost layer; the larger the value, the thicker the adhesion. thermal resistance factor Heat flux factor Flow resistance factor Thickening factor of cream According to the formula The frosting index is calculated, and from this, the frosting index of each heat conduction path can be obtained. Set the frosting detection threshold. ,when > If significant frost adhesion and airflow blockage occur on the heat conduction path (i, j), then the paths of nodes i and j are marked in red and labeled as frost edges. In this way, frost edges in the entire sensor network can be labeled. By extracting multi-factor side-level features and calculating the frost index, the physical quantities of thermal resistance increment, heat flux decrease, airflow obstruction, and frost thickness are normalized and weighted to generate a unified frost index. This overcomes the uncertainty of a single indicator and uses cross-validation of physical quantities from multiple angles and levels to significantly improve the robustness and accuracy of frost condition judgment. The final frost index has both a comprehensive judgment capability under statistical significance and takes into account physical intuition, thus providing a high-confidence decision-making basis for accurately selecting defrosting areas.

[0019] Step 4: Frost area identification and spatial boundary generation; The entire sensor network is abstracted as a graph G = (V, E), where V represents all sensor nodes and E represents all adjacent node pairs, using frost edge sets. Constructing subgraphs In the subgraph The connected component algorithm is run on the refrigerator, starting from an unvisited frost edge and traversing all connected frost edges along the edge until no further extension is possible. Each traversal yields a connected component, which is a set of interconnected frost edges. Each connected component reflects a continuous frosting network formed inside the refrigerator, i.e., a local or cross-area frosting zone. Let the k-th connected component contain the set of edges. The corresponding set of nodes ,but{ , } represents the initial frost region; k is a positive integer representing the index of the connected component; After completing the frost edge clustering analysis and identifying each connected component, the spatial boundary is calculated for each connected component of the frost region—that is, the subgraph formed by all interconnected frost edges—to determine the approximate location of the preliminary frost region in the physical device, thus obtaining the spatial boundary. Specifically, the three-dimensional coordinates of all endpoint nodes are extracted from each connected component. These coordinates are provided by the corresponding sensor modules and mapped to a unified rack coordinate system through a global coordinate alignment module. Then, the minimum and maximum values ​​of the point set on the three orthogonal axes x, y, and z are taken, and these extreme coordinates are concatenated to construct an axis-aligned minimum cuboid. The cuboid precisely encloses all sensor nodes belonging to frost region k. Finally, each preliminary frost region k is marked with a spatial boundary. ; to define the spatial boundary Projecting the image onto the 3D CAD model of the freezer, the key areas where the frost zone is located are clearly identified, and the spatial boundaries are determined based on these key areas. By adding the corresponding labels, each spatial boundary is generated. The work area on the physical equipment and its attributes, where the work area attribute is the name of the key part (i.e., function); for example: if the spatial boundary... If the blades of the evaporator partially overlap, then the overlapping part is the working area, and the attribute of the working area is the evaporator. By identifying frost zones and generating spatial boundaries, all frost paths (frost edges) are automatically clustered into several independent frost zones. The minimum axis-aligned bounding box is calculated using sensor node coordinates to accurately obtain the boundary of each frost zone in three-dimensional space. This eliminates the subjective errors present in traditional manual inspections and achieves automated and reproducible area extraction based on full-network sensor data. At the same time, the geometric overlap determination between the spatial boundary and the CAD model can quickly label the area, clearly identifying whether it belongs to functional components such as evaporator blades, shelf gaps, or back panel channels. This provides clear operational units for subsequent zone scheduling and differentiated defrosting process selection. Maintenance personnel can intuitively understand the distribution of frost zones, improving operational efficiency and minimizing errors.

[0020] Step 5, Defrosting Priority Calculation and Zone Scheduling: Extract all spatial boundaries The spatial boundary is calculated by summing the frost indices of all frost edges within it, along with their corresponding frost indices. The cumulative frost index is denoted as Selecting spatial boundaries The highest frost index Calculate the spatial boundary The volume of space is denoted as From all spatial boundaries The largest spatial volume is denoted as . According to the formula Calculations are performed to obtain the defrost value. It should be noted that the spatial boundary The larger the area volume, the higher the defrost priority and the greater the defrost value; the higher the maximum frost index, the higher the defrost priority and the greater the defrost value; the higher the cumulative frost index, the higher the priority and the greater the defrost value. According to the frost removal value All spatial boundaries in descending order The order is determined, and defrosting is performed sequentially according to this order; specifically, the spatial boundaries are... The defrosting procedure is as follows: S101: Identify spatial boundaries The work area and its attributes are defined. If there are work areas that have not undergone defrosting, then S102 is executed; otherwise, the spatial boundary is indicated. The defrosting operation is complete; S102, determine whether the attribute of the work area is an evaporator. If so, perform defrosting operation on this work area using an electric heating wire; otherwise, proceed to S103. S103, determine whether the attribute of the work area is a shelf gap. If so, use directional hot air pulse to perform defrosting operation; otherwise, jump to S104. S104: Determine if the work area is a back panel passage. If so, use hot air circulation for defrosting; otherwise, proceed to S101. It should be explained that in this application scenario, the actual experiments conducted by technicians only focused on key areas such as the evaporator area, shelf space, air ducts, and back panel. This was because technicians found during their research that these locations are more prone to frost buildup, and frost has a significant impact on the freezer. Therefore, this judgment step could actually end at S104. However, for the sake of rigor and to reflect the logical thinking behind this defrosting process, S104 ends and immediately transitions to S101. If other key areas or new defrosting technologies are added in the future, simply add the corresponding judgment branch after S104; there is no need to refactor the overall logic. By comprehensively considering the boundary volume, maximum frost index, and cumulative frost score of each spatial boundary, a quantifiable defrosting value is generated and sorted from largest to smallest. The optimal defrosting strategy is then executed sequentially for each sub-area. The limitations of a single indicator are eliminated by integrating it into a multi-dimensional decision-making system, ensuring that critical areas with large volume, high thermal resistance, and high frost volume are treated first, maximizing defrosting efficiency and system operational safety. At the same time, through the process design of attribute judgment, intelligent and modular operation scheduling is achieved, significantly reducing energy consumption and shortening operation time. It can also be flexibly expanded to new parts or new technologies, making it easy to maintain and upgrade in the future.

[0021] Step 6, Iterative Verification and Termination Determination: When all spatial boundaries are completed... During defrosting operations, the spatial boundaries are recalculated. The frost index of all edges is calculated. If there are frost edges, return to step one; if there are no frost edges, it means that all paths have returned to the safe thermal resistance level, and the current defrosting operation ends. After completing defrosting operations at all spatial boundaries, iterative verification and termination determination are performed. The overall frost index is recalculated and the remaining frost edges are checked to determine whether the defrosting cycle needs to be restarted. This not only verifies the actual effectiveness of the defrosting strategy in real time, avoiding potential hazards left by surface defrosting, but also captures new frost conditions caused by heat flow redistribution or sudden environmental changes during the defrosting process, achieving dynamic adaptive management. When the thermal resistance of all heat conduction paths returns to below the safe threshold, the system intelligently determines that the current defrosting cycle is complete, thereby avoiding unnecessary repetitive operations and energy consumption. This iterative closed loop ensures the continuity and reliability of the defrosting work, truly combining efficient defrosting with energy-saving optimization, and realizing the best practice for intelligent defrosting of large upright commercial freezers.

[0022] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for optimizing defrosting efficiency in commercial freezers, characterized in that, include: Step 1: Deploy temperature sensors, heat flux meters, micro-differential pressure sensors, and line laser scanning modules in the evaporator area, shelf space layers, and back panel aisles to construct a node-edge 3D sensing mesh. Acquire a reference surface model under frost-free conditions. Acquiring the reference surface model under frost-free conditions includes: using the same recognition method to scan the same physical area under frost-free conditions, fitting a reference plane or curved surface of the bare metal or frost-free surface; this is collectively referred to as the reference surface model. ; Step two: Based on Fourier's law and the nodal energy conservation equation, the discretized thermal resistance network is inverted, and the thermal resistance difference network under operating conditions and frost-free calibration is calculated; the process of inverting the discretized thermal resistance network is as follows: 2-1, Derivation of Network Equations: First, in steady state and assuming that heat flow is only conducted along the i→j direction, we can obtain the following from Fourier's law: T i and T j The temperatures of nodes i and j are respectively, obtained by temperature sensors. R represents the heat flux along the i→j direction, which is directly measured by a heat flux meter. ij This represents the equivalent thermal resistance between nodes i and j, including all contributions from the thermal conductivity of the metal fins, the air gap, and the additional impedance of the frost layer. 2-2. In actual network deployment, not all edges have heat flux meters. In this case, an equilibrium equation is constructed for the entire sensor network, and the equivalent thermal resistance R of all edges is collectively derived. ij ; 2-3, the thermal resistance network {R} under operating conditions ij } and the reference network thermal resistance under frost-free calibration { According to the formula The thermal resistance difference network {ΔR} is obtained by performing difference calculation. ij }; Step 3: Extract thermal resistance factor, heat transfer factor, flow resistance factor and frost thickness factor through multi-factor edge-level features, and fuse them to calculate and generate edge-level frost index. When the frost index is greater than the frost judgment threshold, the heat conduction path, i.e. the edge, is marked as a frost edge. In this way, frost edges in the entire sensor network can be marked. Step four: Identify the clustering subgraph of the frost edge set using the connected component algorithm, extract the 3D coordinates of the nodes to generate the spatial boundary, and map it to the CAD model to annotate the work area attributes; the spatial boundary generation process is as follows: The entire sensor network is abstracted as a graph G = (V, E), where V represents all sensor nodes and E represents all adjacent node pairs, using frost edge sets. Constructing subgraphs In the subgraph The connected component algorithm is run on the refrigerator, starting from an unvisited frost edge and traversing all connected frost edges along the edge until no further extension is possible. Each traversal yields a connected component, which is a set of interconnected frost edges. Each connected component reflects a continuous frosting network formed inside the refrigerator, i.e., a local or cross-area frosting zone. Let the k-th connected component contain the set of edges. The corresponding set of nodes ,but{ , } represents the initial frost region; k is a positive integer representing the index of the connected component; After completing the frost edge clustering analysis and identifying each connected component, spatial boundary calculation is performed on each connected component of the frost area—that is, the subgraph formed by all interconnected frost edges—to determine the approximate location of the preliminary frost area in the physical equipment and obtain the spatial boundary. Specifically, the three-dimensional coordinates of all endpoint nodes are extracted from each connected component. These coordinates are provided by the corresponding sensor module and have been mapped to a unified cabinet coordinate system through the global coordinate alignment module. Then, the minimum and maximum values ​​of the point set on the three orthogonal axes x, y, and z are taken, and these extreme coordinates are spliced ​​together to construct an axis-aligned minimum cuboid. The cuboid precisely encloses all sensor nodes belonging to frost area k. Finally, each preliminary frost area k is marked with a spatial boundary. The spatial boundary is projected onto the three-dimensional CAD model of the freezer to identify the key parts where the frost area is located. According to the key parts to which it belongs, the corresponding labels are marked on the spatial boundary, thus generating the working area and the attributes of each spatial boundary on the physical equipment. The working area attribute is the name of the key part. Step 5: Calculate the defrost value using a formula based on the cumulative frost index, maximum frost index, and area volume of the spatial boundaries. Perform attribute-based defrosting operations in descending order of defrost values. Specifically, this includes: extracting the frost edges of all spatial boundaries and their corresponding frost indices; summing the frost indices of all frost edges within each boundary to obtain the cumulative frost index; selecting the largest frost index among the spatial boundaries; calculating the spatial volume of the spatial boundaries; selecting the largest spatial volume from all spatial boundaries; and weighted and merging the cumulative frost index, maximum frost index, spatial volume, and largest spatial volume to obtain the defrost value; sorting all spatial boundaries in descending order of defrost values ​​and performing defrosting accordingly. Step 6: After defrosting, recheck the frost index. If frost is present, return to Step 1; otherwise, terminate the process.

2. The method for optimizing defrosting efficiency of a commercial freezer according to claim 1, characterized in that, The equilibrium equations and collective inversion process are as follows: In steady state, the sum of the heat flow entering and leaving each node i must equal the heat flow exchanged between that node and the external environment. Energy conservation is applied to node i. ,in Let i represent the set of nodes adjacent to node i. Let be the external heat flow, representing the net heat flow absorbed by node i from external areas such as evaporative cooling flow or shelf surfaces. This is equal to the algebraic sum of the heat flux meter readings at that location. Then, a balance equation is constructed for each node: The left side of the equilibrium equation represents the convergence of each heat flux, while the right side represents the external heat flux input. Therefore, the known condition is: nodal temperature T. i External heat flow Unknown condition: Thermal resistance R of each side ij , Solving for the thermal resistance of the unknown conditions based on known conditions: Under frost-free calibration conditions, measure the temperature and heat flux at all network points and solve for the thermal resistance of the reference network. }, using the least squares method or constrained fitting, optimize {R} ij To minimize the error between the model heat flux and the observed heat flux, when The iteration stops when the value is less than μ, and the thermal resistance R of each side is obtained. ij， μ represents the convergence error threshold.

3. The method for optimizing defrosting efficiency of a commercial freezer according to claim 2, characterized in that, The extraction process for thermal resistance factor and heat flux factor is as follows: Thermal resistance factor extraction: For each heat conduction path in the sensor network, i.e. edge (i, j), read the thermal resistance difference of each edge, extract the maximum thermal resistance difference, and divide the thermal resistance difference of each heat conduction path by the maximum thermal resistance difference to normalize the thermal resistance difference and obtain the thermal resistance factor. Heat flux factor extraction: The heat flux of each heat conduction path and the heat flux under frost-free calibration are extracted, and the heat flux is normalized to obtain the heat flux factor.

4. The method for optimizing defrosting efficiency of a commercial freezer according to claim 3, characterized in that, The extraction process for flow resistance factor and frost thickness factor is as follows: Flow resistance factor extraction: Extract the wind speeds at nodes i and j, and calculate the average value to obtain the wind speed of the heat conduction path from i to j. Simultaneously, extract the pressures at nodes i and j, and calculate the difference and take the absolute value to obtain the pressure difference of the heat conduction path from i to j. Extract the wind speed and pressure difference of the heat conduction path from i to j under frost-free calibration, and use the relative change to measure the degree of blockage to obtain the flow resistance factor. Frost thickness factor extraction: Extract the real-time point cloud of each heat conduction path (i, j). For each point, calculate its height difference with the frost-free reference surface. This yields the height difference between each point in the heat conduction path (i, j). Calculate the frost thickness of the heat conduction path (i, j) by performing an arithmetic average. For each heat conduction path in the mesh, i.e., each edge, there exists a frost thickness, representing the average thickness of the frost on this heat conduction path. Select the largest frost thickness from the sensor mesh and normalize the frost thickness of each heat conduction path based on the largest frost thickness to obtain the frost thickness factor.

5. The method for optimizing defrosting efficiency of a commercial freezer according to claim 4, characterized in that, The specific defrosting operation at the spatial boundary is as follows: S101: Identify the work area and its attributes at the spatial boundary. If there is a work area that has not been defrosted, then execute S102; otherwise, it indicates that the defrosting operation at the spatial boundary is complete. S102, determine whether the attribute of the work area is an evaporator. If so, perform defrosting operation on this work area using an electric heating wire; otherwise, proceed to S103. S103, determine whether the attribute of the work area is a shelf gap. If so, use directional hot air pulse to perform defrosting operation; otherwise, jump to S104. S104, determine whether the attribute of the work area is a back panel channel. If so, use hot air circulation to perform defrosting operation; otherwise, jump to S101.

Images