Control cabinet heat dissipation structure simulation and optimization design system combined with digital twinning technology

Abstract

This invention relates to the field of automated nondestructive testing technology, specifically a control cabinet heat dissipation structure simulation and optimization design system combining digital twin technology. It includes modules for data mapping, baseline reconstruction, defect injection, differential extraction, and coupling decision. The system is based on a thermal fluid network model, combining real-time load and environmental data to solve for the ideal baseline temperature field. Its core is to use dual-track differential extraction to extract the actual residual and theoretical residual, and calculate their topological similarity. When the similarity exceeds a threshold, structural performance degradation is determined, and optimization suggestions are generated. This invention effectively eliminates common-mode interference from the environment and load using differential operations, achieving accurate identification of internal structural heat dissipation degradation.

Description

Technical Field

[0001] This invention relates to the field of automated non-destructive testing technology, specifically a control cabinet heat dissipation structure simulation and optimization design system that combines digital twin technology. Background Technology

[0002] In complex application scenarios such as industrial control sites and large data centers, the control cabinet, as the physical carrier of core equipment, is affected by both external ambient temperature fluctuations and real-time changes in equipment load. To ensure stable equipment operation, existing technologies typically rely on sensor arrays to collect temperature data and execute over-temperature alarms by comparing with preset thresholds, or perform coarse monitoring based on basic physical models. However, this monitoring method based on absolute values ​​is difficult to effectively decouple data under varying operating conditions. It cannot accurately distinguish whether the temperature rise of the equipment is caused by common-mode interference due to high external environmental temperature or sudden load increase, or by a decrease in heat dissipation efficiency caused by internal structural defects such as filter dust accumulation, air duct baffle deformation, or fan characteristic attenuation. This leads to the failure to detect potential problems under low temperature or low load conditions, even if the heat dissipation structure has deteriorated, because the threshold has not been triggered. Under high temperature and high load conditions, false alarms are often caused by environmental noise. In addition, the existing solution lacks the ability to perform reverse analysis based on the fusion of real-time sensor data and geometric models, making it difficult to quantify the impact of structural defects on the flow field and failing to provide accurate digital guidance for the improvement of the heat dissipation structure. Therefore, how to use digital twin and data processing technologies to eliminate environmental and load interference, accurately extract structural heat dissipation defect characteristics, and thus achieve early fault identification and automated structural optimization design is a technical problem that urgently needs to be solved. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention provides a simulation and optimization design system for the heat dissipation structure of a control cabinet, incorporating digital twin technology. Specifically, the technical solution of this invention includes: The multidimensional data mapping module is configured to acquire the static geometric model of the control cabinet and real-time operating sensor data, and to construct a digital thermal fluid network model containing fluid nodes and thermal resistance paths based on the static geometric model. The ideal reference reconstruction module is configured to, based on the digital thermal fluid network model, set a preset theoretical minimum flow resistance coefficient and rated full-efficiency fan parameters, and combine the load power and ambient temperature in the real-time operation sensor data to calculate the ideal reference temperature field through thermal network solution. The defect feature injection module is configured to call a preset structural defect feature library and inject structural flow resistance gain parameters or thermal resistance parameters into the digital thermal fluid network model to generate a theoretical anomaly simulation field containing specific structural failure features. The dual-track differential extraction module is configured to calculate the real residual vector between the real-time running sensor data and the ideal reference temperature field, and the theoretical residual vector between the theoretical anomaly simulation field and the ideal reference temperature field, respectively. The topology coupling decision module is configured to calculate the numerical similarity between the actual residual vector and the theoretical residual vector. If the numerical similarity is greater than a preset threshold, it is determined that there is structural heat dissipation performance degradation, and structural optimization design suggestions are generated based on the structural defect features that cause the numerical similarity to be greater than the preset threshold.

[0004] Preferably, the method for obtaining the static geometric model and real-time operational sensor data of the control cabinet includes: The 3D design file of the control cabinet is analyzed to extract the cabinet dimensions, air duct topology and material thermophysical parameters as a static geometric model; By deploying a sensor array inside the control cabinet, the temperature of key heating components, the air temperature inside the cabinet, the ambient temperature at the air inlet, the fan speed, and the real-time power of the equipment are collected synchronously as real-time operation sensing data.

[0005] Preferred methods for calculating the ideal reference temperature field include: The lumped parameter method is used to discretize the nodes of the digital thermal fluid network model, and a set of energy conservation equations including heat capacity nodes and thermal resistance paths is established. Using the ambient temperature at the air inlet as the boundary condition and the real-time power of the equipment as the heat source input, under the ideal condition of no dust accumulation and no backflow in the air duct, the theoretical minimum temperature rise of each node under the current load is obtained by iteratively solving the energy conservation equations, which serves as the ideal reference temperature field.

[0006] Preferably, methods for generating theoretical anomaly simulation fields containing specific structural failure characteristics include: Select the flow resistance gain coefficient, thermal bypass factor, or fan PQ curve attenuation rate from the structural defect feature library; If the flow resistance gain coefficient is selected, it is added as a multiplication factor to the momentum resistance term of the digital thermal fluid network model; If a thermal bypass factor is selected, a parallel thermal resistance path is established between the air inlet node and the air outlet node, and the thermal bypass factor is converted into the thermal resistance value of the path to construct a simulation model with defects. Based on the aforementioned flawed simulation model, the temperature distribution of each node under the current load is recalculated as a theoretical anomaly simulation field.

[0007] Preferred methods for calculating the actual residual vector and the theoretical residual vector include: The temperature of the key heating element in the real-time running sensor data is compared with the temperature of the corresponding node in the ideal reference temperature field. The common-mode temperature rise caused by ambient temperature fluctuations and load changes is eliminated to generate a real residual vector. By performing a difference operation between the node temperature in the theoretical anomaly simulation field and the temperature of the corresponding node in the ideal reference temperature field, the temperature rise deviation distribution caused purely by the injected structural impedance factor is extracted, and a theoretical residual vector is generated.

[0008] Preferably, the method for calculating the topological similarity between the actual residual vector and the theoretical residual vector includes: Construct feature sequences of the actual residual vector and the theoretical residual vector in the time domain or spatial distribution; Calculate the cosine similarity or cross-correlation coefficient between two feature sequences; The calculated cosine similarity or cross-correlation coefficient is used as the topological similarity to quantify the degree of matching between the actual temperature rise pattern and the theoretical defect pattern.

[0009] Preferred methods for generating structural optimization design recommendations include: When it is determined that there is structural heat dissipation performance degradation, identify the structural defect feature type that causes the numerical similarity to be greater than a preset threshold; If the identified structural defect feature type is flow resistance gain coefficient, then geometric optimization parameters are generated to increase the air inlet area or adjust the guide vane angle. If the identified structural defect characteristic type is a thermal bypass factor, then a structural improvement scheme is generated by adding a sealing partition or optimizing the fan layout, as a structural optimization design suggestion.

[0010] Preferably, the topology coupling decision module is further used for: If the topological similarity does not meet the preset threshold condition, it is determined that the current abnormal temperature rise is caused by a non-structural external heat source or sensor drift. It suppresses the generation of structural optimization design suggestions and outputs environmental adaptability adjustment instructions or sensor calibration prompts.

[0011] Compared with the prior art, the present invention has the following beneficial effects: 1. This system effectively eliminates common-mode interference caused by ambient temperature fluctuations and equipment load changes through an ideal reference reconstruction module and a dual-track differential extraction module. The system uses differential operations to subtract the ideal reference temperature field from the real-time sensing data, thereby extracting the pure temperature rise characteristics caused only by structural factors. This mechanism significantly improves the signal-to-noise ratio of the detection and solves the technical problem of existing technologies in complex application scenarios where it is difficult to distinguish whether the equipment temperature rise is caused by high external environmental temperature and sudden load increase or by the deterioration of the internal heat dissipation structure. 2. This system employs a topology-coupled decision module, which determines faults by calculating the numerical similarity between the actual residual vector and the theoretical residual vector in spatial or temporal distributions, rather than relying on traditional absolute temperature thresholds. This allows the system to detect potential deterioration of the heat dissipation structure in advance, even if the overall temperature of the equipment does not trigger an over-temperature alarm under low-temperature or low-load conditions, as long as the temperature rise pattern matches the structural defect characteristics, effectively avoiding missed detections. At the same time, under high-temperature and high-load conditions, it can also eliminate false alarms caused by environmental noise through pattern matching, achieving a technological leap from extensive monitoring to accurate trend prediction. 3. This system possesses structural optimization capabilities based on defect feature inversion. When structural heat dissipation performance degradation is determined, it can accurately identify the defect type causing the anomaly. Based on fluid mechanics principles and equivalent impedance formulas, the system can transform abstract dimensionless defect parameters into specific geometric optimization suggestions, such as accurately calculating the required increase in air inlet area, adjusting the angle of the guide vane, or increasing the filling width of the sealing partition. This fills the gap in existing solutions lacking reverse analysis capabilities, providing precise digital guidance for improving heat dissipation structures and realizing a closed loop from fault diagnosis to engineering rectification. 4. This system introduces negative decision logic and a high-fidelity multi-dimensional data mapping mechanism. On the one hand, if the topological similarity does not meet the preset conditions, the system will automatically determine that the abnormal temperature rise is caused by non-structural external heat sources or sensor drift, thereby suppressing the generation of structural optimization suggestions and prompting environmental adjustments or sensor calibration, thus avoiding ineffective structural modifications. On the other hand, by parsing the 3D design file to directly extract the geometric model and integrate real-time sensor data, the physical consistency of the digital twin model is ensured, effectively reducing the risk of operational errors caused by model errors or asynchronous sensor data. Attached Figure Description

[0012] The present invention will be further explained below with reference to the accompanying drawings and embodiments: Figure 1 This is a structural diagram of the system of the present invention. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0014] Example 1: Please see Figure 1 A control cabinet heat dissipation structure simulation and optimization design system combining digital twin technology, including: The multidimensional data mapping module is configured to acquire the static geometric model of the control cabinet and real-time operating sensor data, and to construct a digital thermal fluid network model containing fluid nodes and thermal resistance paths based on the static geometric model. The ideal reference reconstruction module is configured to be based on a digital thermal fluid network model, setting the preset theoretical minimum flow resistance coefficient and rated full-efficiency fan parameters, and combining the load power and ambient temperature in the real-time operation sensor data to solve the ideal reference temperature field through thermal network solution. The defect feature injection module is configured to call a pre-set structural defect feature library and inject structural flow resistance gain parameters or thermal resistance parameters into the digital thermal fluid network model to generate a theoretical anomaly simulation field containing specific structural failure features. The dual-track differential extraction module is configured to calculate the real residual vectors of the real-time running sensor data and the ideal reference temperature field, as well as the theoretical residual vectors of the theoretical anomaly simulation field and the ideal reference temperature field, respectively. The topology coupling decision module is configured to calculate the numerical similarity between the actual residual vector and the theoretical residual vector. If the numerical similarity is greater than a preset threshold, it is determined that there is a structural heat dissipation performance degradation, and structural optimization design suggestions are generated based on the structural defect features that cause the numerical similarity to exceed the preset threshold.

[0015] This embodiment discloses a control cabinet heat dissipation structure simulation and optimization design system that combines digital twin technology. The system aims to solve the technical problem in the prior art that it is difficult to distinguish whether the temperature rise of the control cabinet is caused by external environment or load fluctuations or by internal structural defects. The multidimensional data mapping module constructs a digital base for the physical entity, acquiring the static geometric model and real-time operational sensor data of the control cabinet. The system constructs a digital thermal fluid network model through a spatial discretization algorithm. Specifically, the spatial discretization algorithm refers to a meshing strategy that divides the internal space of the control cabinet into several independent air control volume nodes based on the modeling requirements of the lumped parameter method, ensuring that the model nodes match the subsequent thermal network equations. This model is based on graph theory topology, mapping fluid nodes to air control volumes in different areas within the control cabinet, and mapping thermal resistance paths to convective heat transfer or thermal conductivity impedance between nodes. The ideal reference reconstruction module establishes a perfect reference system that excludes structural aging, sets the preset theoretical minimum flow resistance coefficient and rated full-efficiency fan parameters, and combines the load power and ambient temperature in the real-time operating sensor data to calculate the ideal reference temperature field through the thermal network solver. The defect feature injection module synthesizes fault states using digital twin technology, calls a pre-set structural defect feature library, and injects structural flow resistance gain parameters or thermal resistance parameters as disturbance factors into the digital thermofluid network model to generate a theoretical anomaly simulation field containing specific structural failure features. On this basis, the dual-track differential extraction module eliminates common-mode interference caused by environmental and load fluctuations, and calculates the actual residual vector and the theoretical residual vector respectively. The topology coupling decision module qualitatively identifies faults through morphological matching, calculates the numerical similarity between the two residual vectors, and determines that there is structural heat dissipation performance degradation when the numerical similarity is greater than a preset threshold, and generates corresponding structural optimization design suggestions. This embodiment, by constructing an ideal benchmark and performing dual-track differential analysis, can eliminate ambient temperature fluctuations and equipment load changes as common-mode signals. This allows for the accurate extraction of temperature rise characteristics caused solely by structural factors, such as dust accumulation, blockage, or baffle deformation, in complex industrial control scenarios. This enables the system to detect potential problems in advance through topological matching of residual morphology, even when the equipment has not triggered an overheating alarm, such as under low-load conditions in winter but the heat dissipation structure has deteriorated. This represents a technological leap from simple threshold alarms to trend prediction.

[0016] Example 2: Methods for obtaining the static geometric model and real-time operational sensor data of the control cabinet include: The 3D design file of the control cabinet is analyzed to extract the cabinet dimensions, air duct topology and material thermophysical parameters as a static geometric model; By deploying a sensor array inside the control cabinet, the temperature of key heating components, the air temperature inside the cabinet, the ambient temperature at the air inlet, the fan speed, and the real-time power of the equipment are collected synchronously as real-time operation sensing data.

[0017] This embodiment further specifies the multidimensional data mapping module in Embodiment 1, detailing the data acquisition path. The system executes the static geometric model analysis step, directly reading the standard three-dimensional design file of the control cabinet, extracting the cabinet dimensions to determine the computational domain boundary, extracting the air duct topology to identify the physical path of fluid flow, and extracting material thermal property parameters, specifically including material density. Specific heat capacity and thermal conductivity Simultaneously retrieve fluid physical property parameters, including aerodynamic viscosity. Prandtl numbers thermal conductivity of air ; To eliminate model errors caused by using fixed parameters under non-standard operating conditions, such as high temperature or high altitude environments, the specific method involves the system accessing a pre-set fluid property database. The FluidPropertyDB.json file reads the characteristic coefficients of the physical property calculation model based on the media identifier Air index, including the Sutherland formula constants. Reference viscosity And thermal conductivity fitting factor, combined with the real-time air temperature inside the cabinet collected by the sensor. The built-in physical property calculation engine dynamically calculates the physical parameters at the current moment: dynamic viscosity. According to the Sutherland formula:

[0018] Real-time calculation of thermal conductivity With Prandtl number Based on temperature polynomial function

[0019] Dynamic corrections are performed to obtain floating parameters that match the actual thermal state, and these parameters are loaded into global variables in memory as the basic constants for thermal resistance calculation. In order to achieve automated mapping from three-dimensional geometric entities to one-dimensional thermal fluid network topology, the extraction of duct topology is specifically carried out using a voxelized connected domain analysis algorithm: Boolean subtraction is performed on the three-dimensional model, that is, all entity component models are subtracted from the inner bounding box of the cabinet to obtain the fluid domain, and the fluid domain is discretized into a voxel mesh using an octree algorithm. To construct a low-order network that meets the requirements of the lumped parameter method, the system performs a region aggregation step based on geometric features: Fans, vents, or locations with abrupt changes in cross-sectional area (i.e., a rate of change greater than 30%) are used as dividing boundaries. This 30% threshold is an engineering parameter selected based on fluid dynamics simulation experience to balance model calculation accuracy and node size. It is generally believed that when the rate of change in cross-section exceeds this value, a non-negligible local drag loss will occur. Connected voxel sets are aggregated into independent air control volume nodes. The centroid coordinates of the aggregation region are calculated as the node positions, and the hydraulic diameter of the dividing boundary is calculated according to the formula:

[0020] in, Let the area be the area of ​​the connected surface. The wetted perimeter is used as the basis for calculating the Euclidean distance between the centroids of two adjacent air control volume nodes, which is also used as the length of the duct branch. The total area of ​​all voxel surfaces on the dividing boundary is calculated by integration as the cross-sectional area of ​​the branch. These three key geometric parameters The geometric properties of the connecting edges are stored together, and a graph-theoretic topology consisting of fluid nodes and resistance edges is constructed by traversing the adjacency connectivity between aggregate voxels. This ensures the physical authenticity of the airflow path identification and lays the foundation for subsequent flow resistance formulas. The calculation provides explicit geometric input parameters; The system performs dynamic data acquisition steps, synchronously collecting real-time operating data through a sensor array arranged inside the control cabinet. The sensor array collects the temperature of key heat-generating components, such as the surface temperature of the inverter IGBT module, at a specific frequency, while also collecting the air temperature distributed in the air inlet, core, and outlet areas of the cabinet. In addition, the system also collects the ambient temperature at the air inlet to determine thermodynamic boundary conditions, collects the fan speed to correct the convective heat transfer coefficient, and collects the real-time power of the equipment through a current transformer as the basis for heat source input. This embodiment ensures the physical consistency of the digital twin model by integrating static design parameters and dynamic sensing data. Especially in large data center or factory workshop scenarios, the topology of the air duct is directly extracted by parsing the 3D design file, avoiding the geometric errors introduced by manual modeling. At the same time, the synchronous acquisition mechanism of the multi-point sensor array provides high-fidelity data input for subsequent residual calculation, effectively reducing the time deviation noise caused by asynchronous input data and ensuring the real-time mapping accuracy of the thermal fluid network model.

[0021] Example 3: Methods for calculating the ideal reference temperature field include: The lumped parameter method is used to discretize the nodes of the digital thermal fluid network model, and a set of energy conservation equations including heat capacity nodes and thermal resistance paths is established. Using the ambient temperature at the air inlet as the boundary condition and the real-time power of the equipment as the heat source input, under the ideal condition of no dust accumulation and no backflow in the air duct, the theoretical minimum temperature rise of each node under the current load is obtained by iteratively solving the energy conservation equations, which serves as the ideal reference temperature field.

[0022] This embodiment is a further specification of the ideal benchmark reconstruction module in Embodiment 2; this step uses the lumped parameter method to discretize the nodes of the digital thermofluid network model and construct a fluid-thermal coupled network; to ensure the physical authenticity of the thermal network parameters, the system performs fluid network solving based on Kirchhoff's flow law. Branch pressure drop equation In this step, in order to accurately map the static geometric model to the fluid network parameters, the flow resistance is... It is not arbitrarily set, but calculated based on the following physical correlation formula:

[0023] in, The theoretical minimum flow resistance coefficient is a preset value. To distinguish it from the subsequent local resistance coefficient and gain coefficient, it specifically refers to the dimensionless frictional resistance coefficient, characterizing the frictional properties of an ideal smooth airflow channel. The theoretical minimum flow resistance coefficient is set according to the Blasius formula in fluid mechanics. Dynamic calculations, or direct adoption of engineering empirical values ​​for smooth metal ducts, such as 0.02, are used to define the physical lower limit under dust-free conditions. , , These are the branch lengths, cross-sectional areas, and hydraulic diameters of the air duct, extracted analytically from the static geometric model, respectively. Given the air density; after constructing a complete fluid impedance network based on the preset theoretical minimum flow resistance coefficient, the rated full-efficiency fan parameters, specifically the standard PQ characteristic curve entered from the equipment specification, are used as the power source to iteratively solve for the mass flow rate of each duct branch under ideal conditions. To address the nonlinear relationship between pressure drop and flow rate in fluid networks, the Newton-Raphson method is employed for iterative solution: nodal pressure residual equations are constructed.

[0024] And using the Jacobian matrix Update and iterate To ensure the stability of numerical calculations and prevent singularities under zero-flow conditions, the Jacobian matrix... The element structure is as follows: diagonal elements

[0025] off-diagonal elements When node When connected, among them, The preset numerical stability constant is set to a value of [value missing]. And assign it a physical unit With the square root The terms maintain dimensional consistency, and the elements of the Jacobian matrix... The physical meaning is the rate of change of mass flow rate caused by nodal pressure perturbations, i.e., the admittance slope, in units of... until the pressure residual is less than Pa; Based on the obtained Compute the convective heat transfer admittance between nodes This leads to the establishment of a set of energy conservation equations that include heat capacity nodes and thermal resistance paths; for any node, the energy conservation equations are defined as follows:

[0026] in, :node The heat capacity originates from the material properties and volume in the static geometric model. :node The temperature to be determined, here referring to the fluid node temperature. :node Internal heat source power Adjacent node set index Internode thermal conductivity :node With ambient temperature The equivalent thermal conductivity between them is specifically configured as follows:

[0027] in, Let be the outer surface area of ​​the node. The natural convection heat transfer coefficient is... and The cabinet wall thickness and material thermal conductivity are extracted from the static geometric model, respectively. A composite thermal resistance model is used to correct the error caused by neglecting the thermal resistance of the wall surface. Calculations are performed based on the classical natural convection criterion to ensure the correctness of the physical principles: According to the formula:

[0028] Calculate the Grashof number, where, It is the acceleration due to gravity. The coefficient of thermal expansion of air. For kinematic viscosity; combined with Prandtl number According to the formula Calculate the Nusel number by... To obtain the heat transfer coefficient; in this formula, the coefficient is... These are dimensionless geometric constants, automatically determined by the system based on the normal vectors of the surfaces in the geometric model: for perpendicular surfaces, take... For horizontal surfaces ; Feature length The definition is based on adaptive adjustment of surface orientation: for vertical surfaces, it is defined as the projected height of the physical component corresponding to the node in the direction of gravity; For a horizontal surface, it is defined as the ratio of surface area to perimeter. To avoid computational singularities caused by zero projection height. In the process of solving for the ambient temperature at the air inlet, in order to obtain the ideal reference temperature field, that is, the temperature rise distribution of the system under thermal equilibrium, the system will use the time differential term in the above equation. Set to zero; given In Node temperature The function causes the system of algebraic equations to exhibit nonlinear characteristics. The system is solved using the Picard iteration method: initializing the temperature of each node, calculating the initial... Construct a system of linear equations The updated temperature field was obtained by using the LU decomposition method, and the calculation was then performed again. Repeat the above steps until the maximum temperature residual between two adjacent iterations is less than [a certain value]. At this point, the air temperature of the fluid node is obtained. ; To maintain physical consistency with the temperatures of key heat-generating components, such as the surface temperature of IGBTs, in real-time sensing data, the system further performs solid-fluid coupling mapping calculations: for nodes containing heat sources, the formula is used... Calculate the surface temperature of the component; where, The forced convection heat transfer coefficient is calculated based on the mass flow rate of that branch obtained from the fluid network solution. Real-time computation, using a variant of the Dittus-Boelter relational model:

[0029] in, These are fluid physical property parameters; thus, a complete dataset containing the component surface temperature and fluid temperature is constructed as an ideal reference temperature field.

[0030] Example 4: Methods for generating theoretical anomaly simulation fields that include specific structural failure characteristics include: Select flow resistance gain coefficient, thermal bypass factor, or fan PQ curve attenuation rate from the structural defect feature library; If the flow resistance gain coefficient is selected, it is added as a multiplication factor to the momentum drag term of the digital thermal fluid network model; If a thermal bypass factor is selected, a parallel thermal resistance path is established between the air inlet node and the air outlet node, and the thermal bypass factor is converted into the thermal resistance value of the path to construct a simulation model with defects. Based on the flawed simulation model, the temperature distribution of each node under the current load is recalculated as a theoretical anomaly simulation field.

[0031] This embodiment is a further specification of the defect feature injection module in Embodiment 3. It generates fault samples based on the synthetic analysis approach. The system selects specific fault parameters from the structural defect feature library, including flow resistance gain coefficient, thermal bypass factor, or fan PQ curve attenuation rate. In response to the selected flow resistance gain coefficient, the system uses it as a multiplication factor and adds it to the momentum drag term of the digital thermal fluid network model. The corrected inter-node fluid resistance formula is as follows:

[0032] in, The ideal fluid resistance originates from the static geometric model. The injected flow resistance gain coefficient, here referring to a dimensionless multiplication factor, is used to quantify the degree of resistance multiplication relative to the ideal state. It differs from the friction coefficient or local resistance coefficient and originates from a pre-set structural defect feature library. In response to the selection of a thermal bypass factor, the system establishes a virtual parallel thermal resistance path between the inlet and outlet nodes. The thermal resistance value of this path is calculated as follows:

[0033] in, Specific heat capacity of air; Total system mass flow rate; The thermal bypass factor characterizes the proportional response of the return airflow. Based on the selected fan PQ curve attenuation rate, the system derating the characteristic equation of the fan node. To ensure the computability of the simulation model, the fan characteristic equation... Specifically defined as a quadratic polynomial fitting function:

[0034] in, The volumetric flow rate passing through the fan node. The maximum static pressure of the corresponding fan , and The fitting coefficients are obtained by least squares regression based on standard PQ curve data points provided by the fan manufacturer. The standard PQ curve data points come from the system's built-in component specification database, which stores the performance tables corresponding to fan models in JSON format, containing a series of discrete flow-static pressure coordinate pairs. The system obtains data points by parsing the JSON file and calls the polynomial fitting function in a numerical analysis library, such as NumPy, to calculate the above coefficients, which satisfy the following conditions: The domain constraint; where, The selected attenuation rate is used to simulate the decrease in static voltage output capability caused by fan aging or voltage fluctuations. Based on the aforementioned simulation model with defect parameters, and employing a solution strategy identical to that in the ideal baseline reconstruction module—that is, setting the same iterative convergence criteria, such as the temperature residual between two adjacent iterations being less than 100°C—the simulation model incorporates defect parameters. ℃, and relaxation factor, such as 0.5, recalculate the temperature distribution of each node under the current load and ambient temperature, and the output result is the theoretical anomaly simulation field; maintaining the consistency of the solution parameters is to ensure that the difference between the theoretical anomaly field and the ideal reference field comes purely from the injected structural defect parameters, rather than the iteration error or truncation error in the numerical calculation process. This embodiment uses a parameterized injection mechanism to actively simulate various potential structural faults. This forward modeling method can not only simulate known fault modes, but also quantitatively evaluate the temperature rise performance under different fault severity by adjusting the gain coefficient and bypass factor, providing a solid theoretical basis for subsequent fault classification and enabling the system to have the ability to anticipate faults that have never occurred.

[0035] Example 5: Methods for calculating the actual residual vector and the theoretical residual vector include: The temperature of the key heating element in the real-time running sensor data is compared with the temperature of the corresponding node in the ideal reference temperature field. The common-mode temperature rise caused by ambient temperature fluctuations and load changes is eliminated to generate a real residual vector. By performing a difference operation between the node temperature in the theoretical anomaly simulation field and the temperature of the corresponding node in the ideal reference temperature field, the temperature rise deviation distribution caused purely by the injected structural impedance factor is extracted, and a theoretical residual vector is generated.

[0036] This embodiment is a further specification of the dual-track differential extraction module in Embodiment 4. The core of this embodiment lies in removing background noise through differential operations. The system executes a real residual vector generation step, extracting the temperature of key heating elements from the real-time operating sensor data and performing a differential operation with the temperature of the corresponding node in the ideal reference temperature field to generate a real residual vector. The operation process is as follows:

[0037] in, The actual residual vector represents the degree to which the actual observed value deviates from the ideal value. The sensor's measured temperature vector is derived from real-time operating sensor data. The ideal reference temperature vector originates from the theoretical residual vector generation step performed by the ideal reference reconstruction module system. This involves extracting the nodal temperatures from the theoretical anomaly simulation field and performing a difference operation between these temperatures and the corresponding node temperatures in the ideal reference temperature field to generate the theoretical residual vector. The operation process is as follows:

[0038] in, Theoretical residual vector, characterizing the distribution of pure temperature rise deviation caused by specific structural defects. The theoretical anomaly simulation temperature vector, derived from the defect feature injection module, effectively eliminates common-mode temperature rise caused by environmental temperature fluctuations, such as high summer temperatures, and load changes, such as full-load operation, by introducing an ideal reference temperature as a subtrahend. The generated residual vector retains only the deviation relative to the perfect state, which makes the tiny structural heat dissipation efficiency attenuation signal, which is usually submerged in the high ambient temperature, stand out, greatly improving the signal-to-noise ratio of the detection and ensuring the clarity of fault features in complex thermal environments.

[0039] Example 6: Methods for calculating the topological similarity between the actual residual vector and the theoretical residual vector include: Construct feature sequences of the actual residual vector and the theoretical residual vector in the time domain or spatial distribution; Calculate the cosine similarity or cross-correlation coefficient between two feature sequences; The calculated cosine similarity or cross-correlation coefficient is used as the topological similarity to quantify the degree of matching between the actual temperature rise pattern and the theoretical defect pattern.

[0040] This embodiment is a further specification of the topological coupling decision module in Embodiment 5; the system constructs feature sequences of the actual residual vector and the theoretical residual vector in spatial distribution, assuming a total of If there are 1 key temperature measurement point, then the characteristic sequence is: dimensional vectors; the system calculates topological similarity. If cosine similarity is used to quantify the degree of matching between two residual vectors in terms of spatial distribution, the formula is as follows:

[0041] in, Topological similarity, with values ​​ranging from -1 to 1. :No. The actual residual value at each measuring point :No. Theoretical residual values ​​at each measuring point Numerical stability constant, for example This is to prevent computational overflow caused by the denominator being zero when the residual vector approaches zero, i.e., when the device is in an ideal state. Furthermore, if a feature sequence on a time-domain distribution is selected, corresponding to the time-domain features of the embodiment, and the cross-correlation coefficient is calculated, the system continuously samples the residual vector at a preset frequency, such as 1Hz, constructing a sequence of length... time series vector and And based on the Pearson correlation coefficient formula, the cross-correlation value with zero lag or moving lag is calculated: This allows the system to capture waveform similarities over time caused by dynamic faults such as fan surge; the calculated similarity is then used to quantify the degree of matching between the actual temperature rise pattern and the theoretical defect pattern. Regarding the logic for setting the threshold, this embodiment uses statistical methods to determine it: During the initial system calibration phase, historical residual data of the equipment in a fault-free state is collected, and the similarity distribution of its background noise is calculated. The preset threshold is set as the mean of this distribution plus three times the standard deviation, i.e., 3... In principle, the threshold is usually set between 0.85 and 0.9 to ensure that false alarms are effectively suppressed while maintaining high sensitivity. This embodiment uses cosine similarity instead of Euclidean distance as the criterion because in actual working conditions, the severity of the fault may change over time, but the temperature rise distribution pattern caused by the fault, i.e., the direction of the vector, is relatively fixed. For example, air inlet blockage always leads to the maximum temperature rise in the air inlet area. By calculating topological similarity and introducing a small amount of The algorithm enhances robustness in low-noise environments by focusing on the shape of the temperature rise rather than its absolute value, thereby enabling qualitative identification of fault types. It has strong anti-interference capabilities and can accurately identify fault types even when the temperature rise is small in the early stages of a fault.

[0042] Example 7: Methods for generating structural optimization design recommendations include: When it is determined that there is structural heat dissipation performance degradation, identify the structural defect feature type that causes the numerical similarity to be greater than a preset threshold; If the identified structural defect feature type is flow resistance gain coefficient, then geometric optimization parameters are generated to increase the air inlet area or adjust the guide vane angle. If the identified structural defect characteristic type is a thermal bypass factor, then a structural improvement scheme is generated by adding a sealing partition or optimizing the fan layout, as a structural optimization design suggestion.

[0043] This embodiment is a further specification of the step of generating structural optimization design suggestions in embodiment 6; in response to the topology coupling decision module determining that there is structural heat dissipation performance degradation, the system executes the defect type identification step to identify the structural defect feature types that lead to high similarity; The system generates specific quantitative optimization suggestions based on the identification results: if the identified type is flow resistance gain coefficient That is, the current resistance is the ideal value. The system generates compensatory geometric optimization parameters based on the equivalent impedance principle: calculating the required inlet area expansion coefficient. This coefficient is calculated based on the physical law that fluid resistance is inversely proportional to the square of the flow area. In order to offset To achieve a drag gain of several times, the area needs to be increased. This multiplier generates a specific value for increasing the inlet mesh area; or, for adjusting the guide vane angle, the system calls a preset flow resistance coefficient-angle characteristic curve, i.e. , here Refers to the local drag coefficient; Where A and B are wind tunnel test calibration constants. In the system's preset parameters, for commonly used guide vane structures, typical example values ​​are taken: A = 1.2 and B = 0.035. and set The unit is degrees, to ensure the exponential term. It is a dimensionless number, thus satisfying the requirement of physical dimension consistency, and is used to fit the local resistance characteristics of the guide vane at different opening and closing angles, through the inverse function.

[0044] Calculate the amount of guide vane angle adjustment required to offset the flow resistance gain and generate an accurate recommendation; If the identified type is a thermal bypass factor Characterizing the proportion of return airflow, the system inversely calculates the required sealing geometry parameters based on the parallel airflow splitting formula: assuming leakage occurs at topologically adjacent gaps, based on the quadratic resistance law defined in fluid networks. Based on the principle that the voltage drop of parallel branches is equal, the following is derived: This allows for the calculation of the current equivalent leakage thermal resistance.

[0045] in, This refers to the equivalent flow resistance of the main duct branch to which the node belongs under ideal conditions, as determined by the ideal reference reconstruction module. Confirmed; further based on the orifice outflow formula , here The preset gap outflow coefficient is set to 0.65. This value originates from the standard empirical coefficient in engineering fluid mechanics for non-circular sharp-edged gaps, and is used to approximately estimate the leakage flow rate and inversely calculate the equivalent leakage area that needs to be sealed. Combined with the gap length in the geometric model Calculate the recommended sealing partition thickness or fill width. This generates an improved structural design by adding a sealing strip of width W at coordinates [X,Y,Z]; where the gap length... The edge length of the non-duct gap entity connecting the air inlet and air outlet areas is obtained by searching in the static geometric model. If it cannot be directly searched, the hydraulic diameter of the path is taken as the estimated value. This embodiment realizes a closed loop from fault diagnosis to design optimization. Through inversion derivation, the abstract dimensionless defect factor is transformed into specific physical dimension modification suggestions, which directly guides the engineering rectification.

[0046] Example 8: The topology coupling decision module is also used for: If the topological similarity does not meet the preset threshold condition, it is determined that the current abnormal temperature rise is caused by a non-structural external heat source or sensor drift. It suppresses the generation of structural optimization design suggestions and outputs environmental adaptability adjustment instructions or sensor calibration prompts.

[0047] This embodiment supplements the logic of the topology coupling decision module in Embodiment 1. In response to a calculated topology similarity less than a preset threshold, meaning the actual temperature rise distribution does not match any known structural defect, the system determines that the current abnormal temperature rise is caused by a non-structural external heat source, such as direct sunlight on the cabinet causing localized overheating, or sensor drift. The system automatically suppresses the generation of structural optimization design suggestions to prevent unnecessary structural modifications due to misjudgment. The system generates and outputs environmental adaptability adjustment instructions, such as suggesting a reduction in ambient temperature, or outputting sensor calibration prompts, marking sensors with abnormally large residuals as requiring calibration. To verify the technical effectiveness of this system, a field test was conducted in a certain MW-level frequency converter control cabinet. The experimental ambient temperature was set to... The equipment is operating at Rated load; under initial conditions, the system calculates the topological similarity. greater than the threshold The system determined that there was structural degradation in heat dissipation efficiency; the system identified the structural defect as having a flow resistance gain coefficient characteristic. And locate it to the air inlet filter area; according to the optimization suggestions generated in Example 7, the effective area of ​​the air inlet is increased. After rectification, under the same operating conditions, the surface temperature of the IGBT module increased from [a certain value] times; Descending to Furthermore, the calculated new topological similarity decreased to This verifies the effectiveness of the system in accurately identifying structural defects and guiding optimization under complex working conditions; The negative decision logic introduced in this embodiment greatly improves the robustness of the system; it effectively avoids false alarms of structural design defects in the event of non-structural failures, such as simple environmental high temperature or sensor failure, ensuring that the output optimization suggestions have a very high degree of confidence, and reducing the risk of ineffective work and misoperation by maintenance personnel in the actual operation and maintenance process.

[0048] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A control cabinet heat dissipation structure simulation and optimization design system combining digital twin technology, characterized in that, include: The multidimensional data mapping module is configured to acquire the static geometric model of the control cabinet and real-time operating sensor data, and to construct a digital thermal fluid network model containing fluid nodes and thermal resistance paths based on the static geometric model. The ideal reference reconstruction module is configured to, based on the digital thermal fluid network model, set a preset theoretical minimum flow resistance coefficient and rated full-efficiency fan parameters, and combine the load power and ambient temperature in the real-time operation sensor data to calculate the ideal reference temperature field through thermal network solution. The defect feature injection module is configured to call a preset structural defect feature library and inject structural flow resistance gain parameters or thermal resistance parameters into the digital thermal fluid network model to generate a theoretical anomaly simulation field containing specific structural failure features. The dual-track differential extraction module is configured to calculate the real residual vector between the real-time running sensor data and the ideal reference temperature field, and the theoretical residual vector between the theoretical anomaly simulation field and the ideal reference temperature field, respectively. The topology coupling decision module is configured to calculate the numerical similarity between the actual residual vector and the theoretical residual vector. If the numerical similarity is greater than a preset threshold, it is determined that there is structural heat dissipation performance degradation, and structural optimization design suggestions are generated based on the structural defect features that cause the numerical similarity to be greater than the preset threshold.

2. The control cabinet heat dissipation structure simulation and optimization design system combining digital twin technology according to claim 1, characterized in that, Methods for obtaining the static geometric model and real-time operational sensor data of the control cabinet include: The 3D design file of the control cabinet is analyzed to extract the cabinet dimensions, air duct topology and material thermophysical parameters as a static geometric model; By deploying a sensor array inside the control cabinet, the temperature of key heating components, the air temperature inside the cabinet, the ambient temperature at the air inlet, the fan speed, and the real-time power of the equipment are collected synchronously as real-time operation sensing data.

3. The control cabinet heat dissipation structure simulation and optimization design system combining digital twin technology according to claim 2, characterized in that, Methods for calculating the ideal reference temperature field include: The lumped parameter method is used to discretize the nodes of the digital thermal fluid network model, and a set of energy conservation equations including heat capacity nodes and thermal resistance paths is established. Using the ambient temperature at the air inlet as the boundary condition and the real-time power of the equipment as the heat source input, under the ideal condition of no dust accumulation and no backflow in the air duct, the theoretical minimum temperature rise of each node under the current load is obtained by iteratively solving the energy conservation equations, which serves as the ideal reference temperature field.

4. The control cabinet heat dissipation structure simulation and optimization design system combining digital twin technology according to claim 3, characterized in that, Methods for generating theoretical anomaly simulation fields that include specific structural failure characteristics include: Select the flow resistance gain coefficient, thermal bypass factor, or fan PQ curve attenuation rate from the structural defect feature library; If the flow resistance gain coefficient is selected, it is added as a multiplication factor to the momentum resistance term of the digital thermal fluid network model; If a thermal bypass factor is selected, a parallel thermal resistance path is established between the air inlet node and the air outlet node, and the thermal bypass factor is converted into the thermal resistance value of the path to construct a simulation model with defects. Based on the aforementioned flawed simulation model, the temperature distribution of each node under the current load is recalculated as a theoretical anomaly simulation field.

5. The control cabinet heat dissipation structure simulation and optimization design system combining digital twin technology according to claim 4, characterized in that, Methods for calculating the actual residual vector and the theoretical residual vector include: The temperature of the key heating element in the real-time running sensor data is compared with the temperature of the corresponding node in the ideal reference temperature field. The common-mode temperature rise caused by ambient temperature fluctuations and load changes is eliminated to generate a real residual vector. By performing a difference operation between the node temperature in the theoretical anomaly simulation field and the temperature of the corresponding node in the ideal reference temperature field, the temperature rise deviation distribution caused purely by the injected structural impedance factor is extracted, and a theoretical residual vector is generated.

6. The control cabinet heat dissipation structure simulation and optimization design system combining digital twin technology according to claim 5, characterized in that, Methods for calculating the topological similarity between the actual residual vector and the theoretical residual vector include: Construct feature sequences of the actual residual vector and the theoretical residual vector in the time domain or spatial distribution; Calculate the cosine similarity or cross-correlation coefficient between two feature sequences; The calculated cosine similarity or cross-correlation coefficient is used as the topological similarity to quantify the degree of matching between the actual temperature rise pattern and the theoretical defect pattern.

7. The control cabinet heat dissipation structure simulation and optimization design system combining digital twin technology according to claim 6, characterized in that, Methods for generating structural optimization design recommendations include: When it is determined that there is structural heat dissipation performance degradation, the structural defect feature type that causes the numerical similarity to be greater than a preset threshold is identified; If the identified structural defect feature type is flow resistance gain coefficient, then geometric optimization parameters are generated to increase the air inlet area or adjust the guide vane angle. If the identified structural defect characteristic type is a thermal bypass factor, then a structural improvement scheme is generated by adding a sealing partition or optimizing the fan layout, as a structural optimization design suggestion.

8. The control cabinet heat dissipation structure simulation and optimization design system combining digital twin technology according to claim 1, characterized in that, The topology coupling decision module is also used for: If the topological similarity does not meet the preset threshold condition, it is determined that the current abnormal temperature rise is caused by a non-structural external heat source or sensor drift. It suppresses the generation of structural optimization design suggestions and outputs environmental adaptability adjustment instructions or sensor calibration prompts.

Images