Temperature control method and system for industrial personal computer mainboard

Abstract

The invention relates to the field of temperature control, in particular to a temperature control method and system for an industrial personal computer mainboard. The method comprises the following steps: acquiring a CT imaging structure diagram of a mainboard; performing inter-component physical connection analysis and component spatial topology mining on the mainboard CT imaging structure diagram, and constructing a mainboard three-dimensional topological structure model; calculating a real-time temperature value of each mainboard assembly; carrying out time sequence temperature fluctuation analysis and dynamic temperature distribution rendering on the mainboard three-dimensional topological structure model based on the real-time temperature value of each mainboard assembly so as to construct a dynamic heat distribution twinborn model; obtaining a mainboard state monitoring log; and carrying out load fluctuation trend evolution analysis on the mainboard state monitoring log, and carrying out multi-period periodic association mining so as to generate a multi-time-sequence load fluctuation evolution rule. Through efficient and dynamic temperature cooling control, long-term stability and reliability of mainboard performance are achieved.

Description

Technical Field

[0001] The present invention relates to the field of temperature control, and particularly to a temperature control method and system for an industrial computer motherboard. Background Art

[0002] With the continuous development of industrial automation and information technology, the industrial computer motherboard, as the core control unit, undertakes important computing, data processing, and control tasks. With the wide application of industrial computers in various fields such as production lines, robot systems, intelligent manufacturing, and data processing centers, the stability and reliability of the industrial computer motherboard have become the key to system performance. However, during long-term continuous operation, due to its characteristics of processing complex tasks, carrying high loads, and high-speed computing, various components on the motherboard will generate a large amount of heat. If these heats cannot be effectively controlled, the motherboard temperature will continue to rise, resulting in overheating of components, and then causing system failures, performance degradation, and even hardware damage.

[0003] Currently, motherboard temperature control mainly relies on traditional physical heat dissipation devices such as radiators and fans. Although these methods can reduce the motherboard temperature to a certain extent, it is still difficult to meet the requirements of precise temperature control under high loads and complex environments. In addition, traditional temperature monitoring methods often rely on a single temperature sensor and manual maintenance, unable to conduct detailed temperature analysis on each area of the motherboard in real time and accurately, and lacking dynamic temperature control strategy adjustment under different loads and operating states.

[0004] With the progress of technology, especially the development of sensor technology, data analysis technology, and artificial intelligence technology, the traditional temperature monitoring and control methods have gradually been unable to meet the temperature management requirements of modern industrial computers. To ensure the efficient operation of the system and extend the hardware life, an intelligent temperature control method for industrial computer motherboards is urgently needed. This method can not only monitor the temperature fluctuations of each component in real time, but also dynamically adjust the temperature control strategy according to factors such as load changes, environmental temperature, and working states, so as to accurately control the motherboard temperature, avoid various risks brought by overheating, and improve the stability and reliability of the system. Therefore, developing an intelligent and dynamically adjustable temperature control method for industrial computer motherboards has become an important topic to meet the needs of modern industrial automation and ensure the efficient and stable operation of industrial computers. Summary of the Invention

[0005] To solve the above technical problems, the present invention proposes a temperature control method and system for an industrial computer motherboard to solve at least one of the above technical problems.

[0006] To achieve the above object, the present invention provides a temperature control method for an industrial computer motherboard, including the following steps: Step S1: Collect the CT imaging structure diagram of the main board; perform physical connection analysis between components and component space topology mining on the CT imaging structure diagram of the main board to construct a three-dimensional topology structure model of the main board; Step S2: Calculate the real-time temperature value of each main board component; perform time-series temperature fluctuation analysis and dynamic temperature distribution rendering on the three-dimensional topology structure model of the main board based on the real-time temperature value of each main board component to construct a dynamic thermal distribution twin model; Step S3: Obtain the main board status monitoring log; perform load fluctuation trend evolution analysis on the main board status monitoring log and conduct multi-period cycle correlation mining to generate multi-time series load fluctuation evolution rules; Step S4: Predict the multi-time point load demand of the industrial control computer according to the multi-time series load fluctuation evolution rules and predict the power demand of each component one by one to construct a power demand prediction curve for each component; Step S5: Perform temperature rise simulation of the operating state of the industrial control computer on the dynamic thermal distribution twin model based on the power demand prediction curve of each component and conduct temperature rise trend mining for each area one by one to construct a temperature rise trend map for each area; Step S6: Make an intelligent cooling medium switching decision based on the temperature rise trend map of each area and perform dynamic cooling power adjustment to construct a dynamic cooling control model.

[0007] The present invention collects the CT imaging structure diagram of the main board and analyzes the physical connections between components to accurately understand the spatial distribution and interconnection relationships of each component on the main board. This provides reliable data support for subsequent temperature analysis, power prediction, etc. The constructed three-dimensional topological structure model can help design a more scientific cooling system layout and improve the thermal management efficiency. The accurate calculation of the real-time temperature of each component and the analysis of the timing fluctuations help to monitor the temperature changes of each part of the main board in real time and identify potential overheating risks in advance. Through the rendering of the dynamic temperature distribution, not only can the temperature changes be intuitively displayed, but also the temperature fluctuation trend prediction based on time can be realized, providing data support for the optimization and adjustment of the cooling system. Through the evolution analysis of the load fluctuation trend, the change trend of the main board load can be predicted in advance, providing support for subsequent power demand prediction and temperature management. Through the mining of periodic data, the regularity of the load fluctuation is identified, so as to customize and optimize the temperature control strategy according to the working characteristics of the main board in different time periods. Through the trend analysis of the load fluctuation, the power demand of each component at different time points can be accurately predicted. This helps to provide a more accurate power consumption model for the cooling system, so that the system can adjust the power supply in real time. Based on the power demand prediction, the power consumption of the industrial control computer can be more effectively managed, energy waste can be reduced, and the operation efficiency of the system can be improved. Through the temperature rise simulation analysis of the power demand prediction curve, the temperature rise situation of each component and area can be understood in real time, early warnings can be provided for the areas with excessive temperature, and the occurrence of overheating faults can be reduced. Constructing a regional temperature rise situation map can provide engineers with the heat distribution of each area, helping to optimize the cooling design and avoid performance degradation or damage caused by overheating in some areas. By intelligently switching the cooling medium and dynamically adjusting the cooling power, the efficient cooling of the main board is realized, and the performance degradation or hardware damage caused by overheating of the system is avoided. This step realizes the real-time response to the temperature change of the system, and the cooling strategy can be dynamically adjusted according to different working environments and temperature states, improving the adaptability and efficiency of the overall temperature control system.

[0008] In this specification, a temperature control system for an industrial control computer main board is provided, which is used to execute the temperature control method for an industrial control computer main board as described above, and includes: A three-dimensional topology module, configured to collect the CT imaging structure diagram of the main board; perform physical connection analysis between components and component space topology mining on the CT imaging structure diagram of the main board, and construct a three-dimensional topology structure model of the main board; A twin model module, configured to calculate the real-time temperature value of each main board component; perform timing temperature fluctuation analysis and dynamic temperature distribution rendering on the three-dimensional topology structure model of the main board based on the real-time temperature value of each main board component, so as to construct a dynamic thermal distribution twin model; A load fluctuation module, configured to obtain the main board status monitoring log; perform load fluctuation trend evolution analysis on the main board status monitoring log, and conduct multi-period cycle correlation mining, so as to generate multi-temporal load fluctuation evolution rules; A demand prediction module, configured to perform multi-point load demand prediction for the industrial control computer according to the multi-temporal load fluctuation evolution rules, and conduct power demand prediction for each component one by one, so as to construct a power demand prediction curve for each component; A temperature rise trend module, configured to perform temperature rise simulation of the running state of the industrial control computer on the dynamic thermal distribution twin model based on the power demand prediction curve of each component, and conduct temperature rise trend mining for each area one by one, so as to construct a temperature rise trend map for each area; An intelligent cooling control module, configured to make a decision on intelligent cooling medium switching based on the temperature rise trend map of each area, and conduct dynamic cooling power adjustment, so as to construct a dynamic cooling control model.

[0009] By obtaining the CT imaging structure diagram of the main board, the present invention can accurately restore the spatial layout of all components on the main board and their mutual connection relationships, provide accurate hardware structure data. The constructed three-dimensional topological structure model provides a necessary physical basis for subsequent analyses such as temperature fluctuations and load changes, enabling the system temperature control to be optimized according to the actual hardware layout. The clear component spatial layout helps in the design of the cooling system, can determine the heat source intensive areas, foresee the temperature control difficulties in advance, optimize the cooling scheme, obtain the temperature of each component in real time, and provide accurate data support for the thermal management of the main board. The real-time change trend of temperature can help detect overheating problems in a timely manner. Through the analysis of sequential temperature fluctuations, clearly understand the temperature changes of each component at different time points, and help engineers make dynamic adjustments. In addition, the temperature distribution rendering technology visually presents these data, making the temperature changes easier to understand. Through dynamic rendering and analysis, construct a thermal distribution model that truly reflects the temperature state of the main board, providing a scientific basis for future optimization and adjustment, helping to capture the fluctuation trend of the load during the operation of the industrial control computer, revealing the change law of the load. Through the correlation analysis of multiple time periods, the periodic law of load fluctuations can be found, which helps to optimize the temperature control strategy under different load conditions. Through the evolution analysis of the load fluctuation trend, the load peak and trough can be predicted in advance, providing reliable data support for the subsequent adjustment of temperature management and cooling strategy. Through the analysis of load fluctuations, the power demand of each component at different time points can be accurately predicted, which can effectively guide the working state of the cooling system, avoid over-cooling or insufficient cooling. The power demand prediction curve provides data support for the intelligent scheduling of the cooling system, helping the system to adjust the cooling strategy according to the change of power demand during actual operation, and optimizing energy consumption. By combining the power demand prediction with the thermal distribution model, the temperature rise trend of each area of the main board can be accurately simulated. By analyzing the simulation results, the overheated areas can be found, and measures can be taken in advance. Through intelligent decision-making, the cooling medium (such as air, liquid or other cooling methods) can be switched in real time, and the cooling power can be dynamically adjusted according to the area temperature rise situation map. This enables the cooling system to be optimized and adjusted according to the real-time state of the main board. Dynamically adjusting the cooling power can adjust the allocation of cooling resources according to demand, avoid unnecessary energy waste, and ensure that the temperature control effect is not affected. Effective cooling control can prevent the system from performance degradation or hardware damage caused by overheating, thus greatly improving the long-term stability and reliability of the industrial control computer main board. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 It is a schematic flow chart of the steps of a temperature control method for an industrial control computer main board according to the present invention; Figure 2 It is a schematic detailed implementation step flow chart of step S1; Figure 3 It is a schematic detailed implementation step flow chart of step S2; Figure 4 It is a schematic diagram of the detailed implementation steps of step S3. Specific implementation manners

[0011] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0012] The embodiments of the present application provide a temperature control method and system for an industrial control computer motherboard. The execution subjects of the temperature control method and system for the industrial control computer motherboard include, but are not limited to, the following general computing nodes that carry this system: mechanical equipment, data processing platforms, cloud server nodes, network uploading devices, etc. The data processing platform includes, but is not limited to, at least one of an audio and image management system, an information management system, and a cloud data management system.

[0013] Please refer to Figures 1 to 4 , the present invention provides a temperature control method for an industrial control computer motherboard. The temperature control method for the industrial control computer motherboard includes the following steps: Step S1: Collect the CT imaging structure diagram of the motherboard; perform physical connection analysis between components and component space topology mining on the CT imaging structure diagram of the motherboard to construct a three-dimensional topology structure model of the motherboard; Step S2: Calculate the real-time temperature value of each motherboard component; perform time-series temperature fluctuation analysis and dynamic temperature distribution rendering on the three-dimensional topology structure model of the motherboard based on the real-time temperature value of each motherboard component to construct a dynamic thermal distribution twin model; Step S3: Obtain the motherboard status monitoring log; perform load fluctuation trend evolution analysis on the motherboard status monitoring log and perform multi-period cycle correlation mining to generate multi-time-series load fluctuation evolution rules; Step S4: Perform industrial control computer multi-time-point load demand prediction according to the multi-time-series load fluctuation evolution rules and perform power demand prediction for each component one by one to construct a power demand prediction curve for each component; Step S5: Perform industrial control computer operating state temperature rise simulation on the dynamic thermal distribution twin model based on the power demand prediction curve of each component and perform temperature rise trend mining for each area one by one to construct a temperature rise trend map for each area; Step S6: Make an intelligent cooling medium switching decision based on the temperature rise trend map of each area and perform dynamic cooling power adjustment to construct a dynamic cooling control model.

[0014] The present invention collects the CT imaging structure diagram of the main board and conducts physical connection analysis between components to accurately understand the spatial distribution and interconnection relationship of each component on the main board. This provides reliable data support for subsequent temperature analysis, power prediction, etc. The constructed three-dimensional topological structure model can help design a more scientific cooling system layout and improve the thermal management efficiency. The accurate calculation of the real-time temperature of each component and the analysis of the timing fluctuations help monitor the temperature changes of each part of the main board in real time and identify potential overheating risks in advance. Through the rendering of the dynamic temperature distribution, not only can the temperature changes be intuitively displayed, but also the temperature fluctuation trend prediction based on time can be realized, providing data support for the optimization and adjustment of the cooling system. Through the evolution analysis of the load fluctuation trend, the change trend of the main board load can be predicted in advance, providing support for subsequent power demand prediction and temperature management. Through the mining of periodic data, the regularity of the load fluctuations can be identified, so as to customize and optimize the temperature control strategy according to the working characteristics of the main board at different time periods. Through the trend analysis of the load fluctuations, the power demand of each component at different time points can be accurately predicted. This helps to provide a more accurate power consumption model for the cooling system, so that the system can adjust the power supply in real time. Based on the power demand prediction, the power consumption of the industrial control computer can be managed more effectively, energy waste can be reduced, and the operating efficiency of the system can be improved. Through the temperature rise simulation analysis of the power demand prediction curve, the temperature rise trend of each component and area can be understood in real time, early warnings can be provided for areas with excessive temperature, and the occurrence of overheating failures can be reduced. Constructing a regional temperature rise trend map can provide engineers with the thermal distribution of each area, helping to optimize the cooling design and avoid performance degradation or damage caused by overheating in some areas. By intelligently switching the cooling medium and dynamically adjusting the cooling power, efficient cooling of the main board is achieved, avoiding performance degradation or hardware damage of the system caused by overheating. Real-time response to the system temperature changes is achieved, and the cooling strategy can be dynamically adjusted according to different working environments and temperature states, improving the adaptability and efficiency of the overall temperature control system.

[0015] In an embodiment of the present invention, refer to Figure 1 , which is a schematic diagram of the step flow of a temperature control method for an industrial control computer main board according to the present invention. In this example, the steps of the temperature control method for the industrial control computer main board include: Step S1: Collect the CT imaging structure diagram of the main board; conduct physical connection analysis between components and component space topology mining on the CT imaging structure diagram of the main board, and construct a three-dimensional topological structure model of the main board; In this embodiment, a motherboard sample is prepared and placed in a CT scanner. Appropriate CT scan parameters are selected, such as voltage (e.g., 80 - 120 kV) and current (e.g., 100 - 200 μA), to ensure high-resolution images are generated. The CT scan is performed to obtain multi-view slice image data of the motherboard. The scanning process generally takes several minutes and the scanning time is adjusted according to the device performance and the target resolution. After completion, the generated CT data is usually saved in the DICOM (Digital Imaging and Communications in Medicine) format. The acquired CT images are preprocessed, including noise removal and image enhancement. Image processing software (such as OsiriX or ITK-SNAP) is used to perform these operations, and filters (such as Gaussian filters) are used to reduce image noise and enhance image contrast. The preprocessed CT images are converted into three-dimensional volume data, usually through reconstruction algorithms (such as back-projection method). In the generated three-dimensional volume data, different components are identified through image segmentation techniques. Threshold segmentation, region growing, or machine learning algorithms (such as semantic segmentation in deep learning) are used to extract each component on the motherboard. Deep learning models such as U-Net are used to train the images to ensure accurate identification of components such as resistors, capacitors, and chips on the motherboard. After identifying each component, physical connection analysis is performed. By defining the connection relationships between components (such as pins, solder joints, etc.), a connection graph between components is constructed. Using graph theory methods, an adjacency matrix for each component is generated to represent the connection strength and type between components. If two components are connected by pins, the corresponding connection value is marked in the adjacency matrix. According to the identified components and their physical connections, a three-dimensional topological structure model is constructed. Computer-aided design (CAD) software (such as AutoCAD or SolidWorks) is used to visualize the components and connection relationships. During the construction process, ensure that the spatial positions, sizes, and relative positions of the components are accurately represented. Calibration parameters (such as the true size and position deviation of the components) are used to optimize the accuracy of the model. After completing the preliminary three-dimensional topological model, model verification is carried out. By comparing with the actual motherboard (such as size measurement and connection verification) to ensure the accuracy of the model. If errors or inconsistencies are found, return to the previous steps for adjustment until an accurate three-dimensional topological structure model is generated.

[0016] Step S2: Calculate the real-time temperature value of each motherboard component; perform temporal temperature fluctuation analysis and dynamic temperature distribution rendering on the motherboard three-dimensional topological structure model based on the real-time temperature value of each motherboard component to construct a dynamic thermal distribution twin model; In this embodiment, a suitable temperature sensor is selected, such as a thermocouple, an NTC thermistor, or an infrared thermometer. These sensors can monitor the temperature of each component on the motherboard in real time. Usually, devices with fast response time and high precision are selected. The sensors are arranged near key components, such as the processor, memory module, and power module. It is necessary to ensure that the position of the sensor can accurately reflect the operating temperature of the component, avoiding occlusion or interference. Design and build a data acquisition system to collect temperature data in real time. Use a microcontroller such as Arduino or Raspberry Pi, connect multiple temperature sensors, and send the data to the central processing unit through serial communication or a Wi-Fi module. Set the sampling frequency (such as once per second) to ensure that the instantaneous changes in temperature can be captured. The data acquisition system should have the function of real-time data recording and storage for subsequent analysis. Write a program in the data acquisition system to convert the raw data read by the sensor into the actual temperature value. According to the type and characteristics of the sensor, necessary calibration is carried out to ensure the accuracy of the temperature reading. Monitor the temperature changes of each component in real time and store the data in a database to form a temperature time series dataset, providing a basis for subsequent analysis. Extract the real-time temperature data of each component from the database and perform time series analysis. Use statistical analysis methods (such as moving average, standard deviation) to identify the trends and periodicities of temperature fluctuations. Perform FFT (Fast Fourier Transform) analysis on the temperature data to identify the frequency components, helping to understand the periodic characteristics of temperature fluctuations and the influencing factors. Use computer graphics software (such as MATLAB, Matplotlib of Python, or Plotly) to perform visual rendering of the dynamic temperature distribution. Map the real-time temperature data to the surface of a three-dimensional topological structure model to generate a heat map. Adopt color coding (such as heat mapping) so that different temperature ranges correspond to different colors to intuitively display the temperature distribution of the motherboard. After completing the rendering of the dynamic temperature distribution, combine the real-time temperature data with the three-dimensional topological structure model to construct a dynamic thermal distribution twin model. Use 3D modeling software such as Unity or Blender to integrate the temperature data with the model to achieve dynamic display. Design the time evolution process of the thermal distribution model to ensure that the temperature state of each component changing with time is reflected in the model.

[0017] Step S3: Obtain the motherboard status monitoring log; perform an analysis on the evolution trend of the load fluctuation in the motherboard status monitoring log, and perform multi-period correlation mining, so as to generate the multi-time series load fluctuation evolution law; In this embodiment, a motherboard status monitoring system is designed and deployed to ensure that key performance indicators (KPIs) such as CPU load, memory usage, power consumption, temperature, etc. can be recorded in real time. This is achieved using dedicated monitoring software (such as Prometheus, Zabbix, or Nagios). Set the time interval for logging in the system (e.g., log once every 5 seconds) to ensure sufficient monitoring data is obtained for subsequent analysis. The monitoring data should be saved in a structured format (such as CSV or JSON) to ensure easy processing. Store the obtained monitoring logs in a database (such as MySQL, PostgreSQL, or NoSQL database), and set up a data backup mechanism to prevent data loss. Ensure that the database can support fast querying and analysis of large-scale data. Implement data cleaning to handle missing values and outliers to ensure data accuracy and consistency. Use the Pandas library in Python for data cleaning and preprocessing for subsequent analysis. Extract the monitoring logs from the database and calculate the load fluctuations of key components of the motherboard. Use the sliding window method to perform periodic calculations on the data to analyze the load fluctuations within a specific time period. Set the window size (e.g., 10 minutes) to capture short-term load fluctuations. Apply time series analysis methods (such as the autoregressive integrated moving average model ARIMA) to model the load fluctuation data and identify long-term trends and seasonal components. By plotting time series graphs, ACF (autocorrelation function), and PACF (partial autocorrelation function) graphs, help identify the periodicity and trend changes in the data. Use frequency domain analysis methods such as the Fourier transform (FFT) to extract the frequency characteristics of load fluctuations and identify potential periodic patterns. Through FFT analysis, understand the impact degree of load fluctuations at different frequencies. Set the parameters for periodic analysis (such as sampling frequency and time window) to ensure effective capture of the periodic characteristics of load fluctuations. Based on the extracted load fluctuation data, apply association rule learning algorithms (such as Apriori or FP-Growth) for multi-period periodic association mining to identify the relationships between load fluctuations and other factors (such as temperature, power consumption). Set the support and confidence thresholds to filter out insignificant association rules and ensure the extraction of meaningful load fluctuation patterns.

[0018] Step S4: Predict the multi-point load requirements of the industrial control computer according to the multi-temporal load fluctuation evolution law, and predict the power requirements of each component one by one to construct the power requirement prediction curve for each component; In this embodiment, a suitable time series prediction model is selected. Methods such as ARIMA, SARIMA, or LSTM (Long Short-Term Memory Network) are used to capture the time dependence and periodicity of the load. The load demand data is preprocessed, including detrending, deseasonalizing, and normalizing, to improve the prediction accuracy of the model. The data is divided into a training set and a test set, usually 80% for training and 20% for validation. The prediction model is trained using the training set data, and the model parameters are adjusted to optimize the model performance. When using LSTM, hyperparameters such as the number of units in the hidden layer, learning rate, and batch size are adjusted. The accuracy of the model is verified by using the test set data for prediction and calculating performance metrics such as Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and R² value to evaluate the prediction ability of the model. The best model is selected according to the verification results, and the trained model is saved for subsequent multi-timepoint load demand prediction. The power demand data of each component of the industrial control computer is collected, including CPU, memory, hard disk, and network interface, etc. The power demand is recorded by referring to the device manual, the specification provided by the manufacturer, or using real-time monitoring tools (such as current sensors). The power demand data of each component is integrated into a database to form structured component power demand information for subsequent analysis. Using the information obtained from the overall load demand prediction, a power demand prediction model is constructed for each component one by one. Linear regression models, support vector machines, or deep learning models (such as neural networks) are used and trained in combination with the historical power demand data of each component. During the model construction process, characteristic variables (such as current load, temperature, usage period, etc.) are set, and appropriate target variables (such as power demand) are selected to ensure that the model can reflect the change trend of the component power demand. The trained component power demand prediction model is used to predict the future power demand. The prediction time period (such as the next 24 hours or 48 hours) is set, and a power demand prediction curve for each component is generated. The prediction results are compared with the actual monitoring data to analyze the accuracy and stability of the model. If obvious deviations are found, the previous steps need to be returned for model adjustment and optimization. The power demand prediction curves of each component are visually displayed using data visualization tools (such as Matplotlib, Plotly, or Tableau). Through line charts, bar charts, or heat maps, etc., the change trend of the power demand of each component is intuitively displayed. A detailed prediction report is generated, including the power demand prediction curves of each component, relevant parameters, and model performance evaluation results, and provided to relevant decision-makers for reference.

[0019] Step S5: Based on the power demand prediction curve of each component, perform temperature rise simulation of the operating state of the industrial control computer on the dynamic thermal distribution twin model, and conduct temperature rise trend mining for each area one by one to construct a temperature rise trend map for each area; In this embodiment, obtain the power demand prediction curves of each component from step S4, ensuring the consistency of the data timestamps for subsequent temperature rise simulation. The data should include the power demands of each component within the prediction time period, usually in watts (W). Integrate this data into a structured dataset (such as CSV or a database) and add necessary metadata, such as component type, operating conditions, and ambient temperature. Configure the thermal model parameters of each component according to its thermal characteristics (such as thermal conductivity, specific heat capacity, etc.), obtaining these parameters by referring to the technical documents provided by the manufacturer or through experimental measurements. Determine the overall thermal characteristics of the motherboard, considering the influence of additional cooling devices such as heat sinks and fans for accurate modeling in the simulation. Select a suitable thermal simulation software (such as ANSYS, COMSOL Multiphysics, or Flotherm) for temperature rise simulation. These software support the solution of complex thermal models and can handle multi-physics field coupling problems. Import the dynamic thermal distribution twin model into the simulation software, ensuring that the geometric structure, material properties, and boundary conditions of the model are correctly set. In the simulation software, set the simulation parameters, including the time step, total simulation duration, and solution accuracy. Usually, select a smaller time step (such as 1 second) to improve the accuracy of the simulation results. Input the power demand prediction curves of each component as heat sources for time-series thermal simulation, ensuring that the changes in the heat sources can reflect the actual working conditions. Start the simulation, monitor the computational performance and stability during the process, and adjust the computing resources (such as the number of CPU cores and memory) according to the simulation requirements to ensure the rapid completion of the simulation. Collect the temperature data generated during the simulation, usually output in the form of node temperature or average temperature. Extract the temperature rise data of each region from the simulation results, dividing the regions according to the model, for example, dividing the motherboard into CPU area, memory area, power supply area, etc. Calculate the temperature rise value of each region, usually the temperature at the end of the simulation minus the ambient temperature, to obtain the relative temperature rise value. Use data visualization tools (such as Matplotlib, Plotly, or ParaView) to plot the temperature rise trend diagrams of each region, selecting forms such as heat maps, contour maps, or 3D surface maps to display the temperature rise distribution. Set the color mapping so that different temperature intervals correspond to different colors for intuitive display of the temperature rise in each region. Analyze the generated temperature rise trend diagrams, identify the regions with higher temperature rise, and compare them with the power demand data to find potential thermal management problems. Record the temperature rise data of each region and generate a temperature rise report, including the temperature rise values of each region, influencing factors, and optimization suggestions.

[0020] Step S6: Make an intelligent decision on the switching of the cooling medium based on the temperature rise trend diagram of each region and perform dynamic cooling power adjustment to construct a dynamic cooling control model.

[0021] In this embodiment, the temperature rise trend diagrams of each region are obtained from step S5. These diagrams show the temperature distribution and temperature rise values of different regions, using forms such as heat maps and contour maps to ensure the intuitiveness of temperature information, and to ensure the accuracy and real-time nature of the temperature rise data. Record the maximum temperature rise value, average temperature rise value and their corresponding time points for each region. According to industry standards or equipment specifications, set the temperature rise threshold for each region. Set the temperature rise threshold for the CPU region to 75°C and the temperature rise threshold for the memory region to 70°C to ensure that the equipment operates within a safe temperature range. Record the safe temperature range for each region and establish a temperature rise monitoring database for reference in subsequent decision-making. Determine the available cooling media, including air cooling, liquid cooling or phase change materials, etc. According to the temperature conditions of each region in the temperature rise trend diagram, select the appropriate cooling media. When the temperature of the CPU region exceeds the set threshold, decide to switch to a liquid cooling system to achieve more efficient cooling. Use decision tree algorithms or fuzzy logic control to implement the switching decision of intelligent cooling media. Set the input variables, including the current temperature rise value, the performance of the cooling media (such as cooling rate, energy consumption, etc.) and the historical cooling effect. Set the decision rules. If the CPU temperature rise value > 75°C, then switch to liquid cooling. If the memory temperature rise value is between 65°C and 70°C, then maintain the current air cooling state. Encode the above rules into the control system to ensure that cooling decisions are automatically made during real-time monitoring. According to the selected cooling media, implement dynamic cooling power adjustment. For the air cooling system, it is achieved by adjusting the fan speed. For the liquid cooling system, adjust the flow rate of the pump and the temperature of the coolant. Set the parameters for power adjustment. For example: Fan speed: Set between 1000 - 3000 RPM and adjust dynamically according to the temperature rise value. Liquid cooling pump flow rate: Set between 0.5 - 2 L / min and adjust according to the CPU load condition. Implement a closed-loop control system to monitor the temperature changes of each region in real-time and dynamically adjust the cooling power according to the temperature feedback. When the CPU temperature drops to within the safe range (such as below 72°C), automatically reduce the fan speed or pump flow rate. Use the PID control algorithm (Proportional-Integral-Derivative control) to optimize the response time and steady-state error of the cooling power adjustment. Set the PID parameters (such as Kp, Ki, Kd) to improve the stability and accuracy of the control system. Integrate the above cooling decisions and dynamic adjustment strategies to construct a dynamic cooling control model. Use tools such as Matlab Simulink for modeling to achieve simulation and optimization. Test the cooling effect under different working conditions, analyze the relationship between the cooling response time, energy consumption and temperature rise situation, and optimize the cooling strategy. Record each cooling decision and its effect in the database for subsequent analysis and optimization. Generate a cooling control log to record the input, output and execution effect of each decision. Regularly review and update the cooling strategy. Based on new temperature rise data and system performance, perform model improvement and parameter adjustment to ensure the efficient operation of the cooling system.

[0022] In this embodiment, referring to Figure 2 , which is a schematic diagram of the detailed implementation steps of step S1. In this embodiment, the detailed implementation steps of step S1 include: Step S11: Based on a CT scanner, perform high-precision scanning on the industrial control computer motherboard to collect the CT imaging structure diagram of the motherboard; Step S12: Perform in-depth visual recognition on the CT imaging structure diagram of the motherboard and mark each motherboard component node; Step S13: Perform three-dimensional space position registration calculation on each motherboard component node, thereby generating the three-dimensional space coordinates of each component; Step S14: Perform physical connection analysis between components on each motherboard component node to generate the physical connection relationship between components; Step S15: Based on the physical connection relationship between components and the three-dimensional space coordinates of each component, perform component space topology mining to obtain component space topology connection data; Step S16: According to the component space topology connection data, perform three-dimensional topological point cloud modeling on the CT imaging structure diagram of the motherboard to construct a three-dimensional topological structure model of the motherboard.

[0023] In this embodiment, a high-resolution CT scanner is selected to ensure that it has sufficient spatial resolution (e.g., 100 μm) to clearly capture the details and internal structure of the main board. Configure the scanning parameters, such as voltage (usually 80 - 120 kV) and current (e.g., 100 - 200 μA), and optimize them according to the density and thickness of the main board material. Fix the industrial computer main board on the scanning platform of the CT scanner to ensure that it does not move during the scanning process. Set an appropriate scanning range according to the size and structure of the main board to ensure that the entire main board is within the scanning area. Start the scanning and record the scanning time and parameter configuration. Generally, the CT scanning time is several minutes to more than ten minutes, depending on the scanning resolution and the complexity of the main board. After the scanning is completed, the system will generate a series of slice images, and use CT image reconstruction algorithms (such as filtered back projection or algebraic reconstruction technique) to reconstruct the slice images into a three-dimensional CT imaging structure diagram. Verify the imaging quality to ensure that there are no obvious artifacts or noises. If problems are found, adjust the scanning parameters and rescan. Use image processing software (such as MATLAB or OpenCV) to preprocess the CT imaging structure diagram, including denoising, enhancing contrast, and edge detection, to improve the accuracy of subsequent recognition. Apply a deep learning model (such as a convolutional neural network, CNN) for visual recognition of the main board components. First, construct a training data set, including labeled main board component images, to train the model to recognize different components (such as resistors, capacitors, chips, etc.). Apply the trained model to the CT imaging structure diagram to automatically label the nodes of each main board component. Record information such as the category, location, and size of each component. Manually verify the recognition results to ensure that each component is correctly labeled. If misrecognition or missed recognition occurs, adjust the model parameters or increase the training data set, and retrain and recognize again. Extract the two-dimensional coordinates of each component node from the deep vision recognition results and record its position in the CT image. Usually, it is represented in pixel units and converted to millimeter units for subsequent spatial position registration. Adopt a registration algorithm (such as rigid transformation or non-rigid transformation) to convert the two-dimensional coordinates of each component into three-dimensional space coordinates. Use the least squares method to determine the registration parameters to improve the registration accuracy. Combine the geometric model of the main board to determine the specific position (X, Y, Z coordinates) of the components in three-dimensional space to ensure that the spatial position of each component is consistent with the actual situation. Record the three-dimensional space coordinates of each component in the database to form a structured data table for subsequent analysis and application. Analyze the physical connections between components to identify connection points (such as solder joints, plugs, etc.). Determine which components are physically connected to each other through distance thresholds and geometric shape matching. Apply a graph theory model, regard each component as a node in the graph, and the physical connection as an edge, to construct a connection graph between components. Use graph traversal algorithms (such as depth-first search or breadth-first search) to analyze the connection relationships between components.Record the connection information of each component, including the connected component ID, connection type (such as welding, plugging), and the spatial coordinates of the connection points. According to the physical connection relationships between components obtained in the previous step, construct the spatial topology model of the components. Use topological analysis methods to identify the interdependencies between components and their connection methods. Adopt topological data analysis (TDA) techniques to mine the connection relationships between components and identify key connection nodes and edges. Use persistent homology methods to analyze topological features and identify the key components of the motherboard and their connection patterns. Record the results of topological mining and generate component spatial topology connection data for subsequent modeling. Based on the three-dimensional spatial coordinates of the components and the topological connection data, generate the three-dimensional point cloud model of the motherboard. Use point cloud processing software (such as PCL or MeshLab) to convert the vertices and connection relationships into the three-dimensional point cloud format. Reconstruct the surface of the generated point cloud data and use a triangular mesh generation algorithm (such as Delaunay triangulation) to construct the three-dimensional topological structure model of the motherboard. Optimize the details of the model to ensure the integrity and visualization effect of the model, and record the attribute information (such as name, type, etc.) of each component. Verify the constructed three-dimensional topological structure model to ensure its accuracy and integrity. Compare it with the actual motherboard to check whether the structure and component positions of the model are consistent. Apply the final three-dimensional topological structure model to subsequent maintenance, fault diagnosis, and optimization design to provide a reference basis.

[0024] In this embodiment, refer to Figure 3 , which is a schematic diagram of the detailed implementation steps of step S2. In this embodiment, the detailed implementation steps of step S2 include: Step S21: Calculate the real-time temperature value of each motherboard component based on the temperature monitoring sensor; Step S22: Conduct a time-series temperature fluctuation analysis on the real-time temperature value of each motherboard component to generate the temperature fluctuation curve of each component; Step S23: Conduct a motherboard heat distribution analysis on the temperature fluctuation curve of each component according to the three-dimensional spatial coordinates of each component, and construct a dynamic heat distribution change diagram; Step S24: Render the dynamic temperature distribution of the motherboard three-dimensional topological structure model according to the dynamic heat distribution change diagram to construct a dynamic heat distribution twin model.

[0025] In this embodiment, temperature monitoring sensors, such as thermocouples or digital temperature sensors, are installed on each motherboard component. Select suitable sensors to ensure that their measurement ranges cover the expected temperature range of the motherboard (-20°C to 100°C) with an accuracy within ±0.5°C. Before installing the sensors, calibrate them to ensure accurate readings for each sensor. Use standard temperature equipment for comparative calibration. Configure a data acquisition system to ensure that the temperature data of all sensors can be read in real time. Use a data acquisition module (such as Arduino, Raspberry Pi, or a professional DAQ system) to convert the sensor signals into digital signals and transmit them to a computer through a communication interface (such as USB, RS-232, or Wi-Fi). Set the data acquisition frequency, for example, collect temperature data once per second to obtain the real-time temperature changes of each component under dynamic operation. Store the real-time collected temperature data in a database or data table, recording the temperature readings of each component at each time point. Ensure that the data format is unified for subsequent analysis. Use data processing tools (such as Python, MATLAB) to perform preliminary cleaning on the real-time temperature data to remove outliers and noise and ensure data quality. Conduct time series analysis on the real-time temperature values of each component, calculate statistical metrics such as the mean, standard deviation, maximum, and minimum of the temperature to evaluate the temperature stability of each component. Generate a temperature fluctuation curve and use linear interpolation or smoothing algorithms (such as moving average) to smooth the temperature data and eliminate short-term fluctuation interference. Use data visualization tools (such as Matplotlib, Tableau) to plot the temperature fluctuation curves of each component. Set appropriate coordinate axes, mark the time and temperature values, and clearly display the temperature change trend. Identify the time periods with significant temperature fluctuations and temperature anomaly points, and record this data for subsequent analysis. Set a threshold for temperature fluctuations (such as ±5°C), monitor the temperature fluctuation curves of each component, and identify the occurrence time and duration of abnormal temperature fluctuations. Record the abnormal fluctuation data in the database and provide a basis for subsequent fault diagnosis. Combine the three-dimensional spatial coordinates (X, Y, Z) of each component with its corresponding temperature fluctuation data to construct a heat distribution model of the motherboard. Use three-dimensional interpolation methods (such as Kriging interpolation or spline interpolation) to estimate the temperature distribution on and inside the motherboard surface. Set the calculation area and resolution of the temperature distribution to ensure that the model can accurately reflect the heat distribution of the motherboard. Use heat distribution analysis tools (such as MATLAB or the SciPy library in Python) to calculate the heat distribution of the motherboard and generate a heat distribution map. Represent different temperature regions through color gradients to ensure that the graph is clear and easy to understand. Record the positions and temperature values of the high-temperature regions and analyze the causes of these regions for subsequent optimization and maintenance. Combine the dynamic heat distribution change map and the three-dimensional topological structure model, and use three-dimensional modeling software (such as Blender, 3ds Max) for dynamic temperature distribution rendering.Map the heat distribution data onto the surface of the topological structure model to generate a dynamically displayed temperature distribution effect. Set the rendering parameters (such as lighting, material properties) to ensure that the visual effect of the temperature distribution is realistic and readable. Demonstrate the change of temperature distribution over time through an animation effect to generate a dynamic thermal distribution twin model. This model can reflect the temperature changes of each component on the motherboard in real time, providing visual support for subsequent monitoring and maintenance. Record the temperature distribution data in each state for comparative analysis and fault diagnosis. Apply the dynamic thermal distribution twin model to the actual monitoring system to update the temperature status of the motherboard in real time, providing timely temperature warnings and maintenance suggestions to the operator. Regularly collect feedback data to evaluate the accuracy and practicality of the dynamic temperature distribution model, and optimize and adjust it according to the actual situation to improve the response speed and accuracy of the model.

[0026] In this embodiment, the specific steps of step S24 are as follows: Perform component-by-component heat diffusion evolution based on the three-dimensional topological structure model of the motherboard to obtain the heat diffusion path of each component; Calculate the temperature gradient based on the real-time temperature values of each motherboard component to generate a motherboard temperature gradient map; Conduct heat conduction loss analysis on the motherboard temperature gradient map to generate inter-component heat conduction loss data; Perform heat diffusion dynamics mining based on the inter-component heat conduction loss data and the heat diffusion path of each component to generate the heat diffusion evolution characteristics of the motherboard; Render the dynamic temperature distribution of the three-dimensional topological structure model of the motherboard according to the heat diffusion evolution characteristics of the motherboard and the dynamic thermal distribution change map to construct a dynamic thermal distribution twin model.

[0027] In this embodiment, using the three-dimensional topological structure model of the main board, the finite element analysis (FEA) method is adopted to simulate the heat diffusion process. An appropriate heat diffusion equation (such as the heat conduction equation) is selected, and physical properties such as the thermal conductivity, specific heat capacity, and density of the materials are defined in the model. Considering the thermal conductivity of the PCB material is 0.3 W / (m·K) and that of the circuit components is 5 W / (m·K). The initial conditions and boundary conditions are set. The initial temperature value is the real-time temperature value of each component, and the boundary conditions simulate the ambient temperature (such as 25°C) and heat dissipation conditions (such as natural convection) of the main board. The numerical simulation software (such as ANSYS or COMSOL Multiphysics) is used to perform the heat diffusion evolution simulation and calculate the heat diffusion path of each component. During the simulation process, the time step of the heat diffusion is recorded to ensure the accuracy of the simulation results. A heat diffusion path diagram is generated to show how heat diffuses from the heat source component to the surrounding components, with a focus on the connection relationship between the high-temperature region and the low-temperature region. According to the three-dimensional spatial coordinates and real-time temperature values of each component, the temperature gradient is calculated. Numerical methods (such as the central difference method) are used to estimate the temperature gradient, and an appropriate mesh division is set to ensure the accuracy of the temperature calculation. The minimum distance (such as 1 cm) between each component is selected as the mesh spacing. Data visualization tools (such as MATLAB, Matplotlib in Python) are used to generate the temperature gradient diagram of the main board. Different temperature gradient regions are represented by color gradients, with the red region indicating a high temperature gradient and the blue region indicating a low temperature gradient. The gradient values of each region are recorded for subsequent heat conduction loss analysis. Based on the temperature gradient diagram, the heat conduction loss between each component is analyzed. Fourier's law is used to calculate the heat flux density, q = −k∇T, where q is the heat flux density, k is the thermal conductivity of the material, and ∇T is the temperature gradient. The thermal conductivity and contact area of the connecting components are set to facilitate the calculation of the actual heat conduction loss. By calculating the heat flux density between each component, the inter-component heat conduction loss data, including the lost heat and the loss location, are generated. The analysis results are recorded in the database to form a structured data table for subsequent analysis and application. Using the inter-component heat conduction loss data and the heat diffusion path, a heat diffusion kinetic model is established. An appropriate mathematical model (such as a kinetic model or a heat transfer model) is selected to quantitatively analyze the heat diffusion process. By simulating the heat transfer between components, the main heat diffusion channels and paths are identified, with a focus on the high-loss regions. Kinetic analysis methods, such as time series analysis or kinetic system modeling, are applied to extract the characteristics of heat diffusion, such as the diffusion rate, diffusion range, and heat retention time. The key parameters of heat diffusion are recorded for subsequent analysis of the heat diffusion evolution characteristics. Unity) is used for dynamic temperature distribution rendering. The heat diffusion characteristics are mapped onto the three-dimensional topological structure model of the main board to generate a dynamic visualization effect. Animation parameters are set to show the change of temperature distribution over time and provide an intuitive heat diffusion process.Verify the accuracy and visibility of the rendering effect to ensure that the dynamic temperature distribution can truly reflect the actual situation. Check the accuracy of the model by comparing it with the actual temperature data. Optimize the rendering effect based on the feedback, adjust the color mapping, lighting, and material properties to ensure that the visualization effect of the dynamic thermal distribution twin model is clear and easy to understand.

[0028] In this embodiment, refer to Figure 4 , which is a schematic diagram of the detailed implementation steps of step S3. In this embodiment, the detailed implementation steps of step S3 include: Step S31: Obtain the motherboard status monitoring log; calculate the CPU utilization rate for the motherboard status monitoring log to obtain the real-time CPU utilization rate; Step S32: Quantify the memory usage based on the motherboard status monitoring log to generate a quantified value of the memory occupancy rate; Step S33: Calculate the real-time load of the motherboard according to the real-time CPU utilization rate and the quantified value of the memory occupancy rate to generate the real-time load characteristics of the motherboard; Step S34: Analyze the evolution of the load fluctuation trend for the real-time load characteristics of the motherboard to extract the load fluctuation trend characteristics; Step S35: Conduct multi-period cycle correlation mining on the load fluctuation trend characteristics to generate the multi-temporal load fluctuation evolution law.

[0029] In this embodiment, a system monitoring tool (such as the top command in Linux or the Performance Monitor in Windows) is used to regularly obtain the motherboard status monitoring logs. Set the log recording frequency, for example, record once per minute to ensure the timeliness of the data. Format the obtained monitoring logs into structured data (such as CSV or JSON format) for subsequent analysis and processing. Each log record should contain key metrics such as timestamp, CPU utilization, and memory usage. Extract the CPU utilization data from the monitoring logs. CPU utilization is usually expressed as a percentage, and CPU utilization = total active time / total time × 100%. By analyzing the CPU activity status in each time period, calculate the real-time CPU utilization. If the CPU active time in a certain time period is 30 seconds and the total time is 60 seconds, then the CPU utilization is 50%. Extract the current memory usage status from the status monitoring logs, including information such as total memory, used memory, and available memory. Memory usage is usually in bytes, and record the memory status of each monitoring. Calculate the memory occupancy rate according to the extracted data, memory occupancy rate = used memory / total memory × 100%, and record the calculated quantitative value of the memory occupancy rate in the database to form comprehensive performance monitoring data combined with CPU utilization. Use a visualization tool to draw the change curve of the memory occupancy rate for convenient subsequent analysis. Combine the real-time CPU utilization and memory occupancy rate to calculate the real-time load characteristics of the motherboard. Adopt a weighted average method, set weights (such as CPU accounting for 75% and memory accounting for 25%), and the formula is: real-time load = 0.75 × CPU utilization + 0.25 × memory occupancy rate, to obtain a comprehensive load index reflecting the overall load situation of the motherboard. Record the calculated real-time load characteristics in the database, marked with timestamps, to form a continuous load analysis data set. Perform time series analysis on the real-time load characteristic data to identify the trend of load fluctuations. Use the sliding window technique to calculate the moving average of the load to smooth short-term fluctuations. Set an appropriate window size (such as 5 minutes) to smooth the load at each time point and avoid the interference of accidental fluctuations. Identify the rules and periodicity of load fluctuations. Analyze the change trends of the load in different time periods (such as weekdays and weekends, day and night), and extract the characteristics of high load and low load. Record parameters such as the amplitude, frequency, and period of load fluctuations for subsequent correlation analysis. Use time series analysis techniques (such as autocorrelation function and Fourier transform) to analyze the characteristics of load fluctuation trends and identify potential periodic patterns. By calculating the autocorrelation coefficient, judge the correlation of the load in different time periods. Set appropriate analysis parameters, such as the period length (such as hours, days) and the analysis window, to capture the periodicity of load changes. Based on the analysis results, generate the evolution law of multi-time series load fluctuations, and record the load change characteristics and their influencing factors in different time periods. Identify the load characteristics during peak periods (such as working hours) and trough periods (such as night).Record the discovered rules in the database to form historical data of load fluctuations, providing support for subsequent services.

[0030] In this embodiment, step S4 includes the following steps: Step S41: Predict the load requirements of the industrial computer at multiple time points according to the multi-temporal load fluctuation evolution rule, so as to generate load requirement prediction data for multiple time points of the industrial computer; Step S42: Calculate the power consumption of each motherboard component node one by one based on the motherboard status monitoring log to generate the power consumption parameters of each component; Step S43: Predict the component power requirements for the power consumption parameters of each component according to the load requirement prediction data of the industrial computer at multiple time points to obtain the multi-temporal power requirement prediction data of each component; Step S44: Perform time series discrete fitting on the multi-temporal power requirement prediction data of each component to construct the power requirement prediction curve of each component.

[0031] In this embodiment, based on the multi-temporal load fluctuation evolution rule obtained from the previous analysis, a load requirement prediction model is constructed. Use time series prediction methods, such as ARIMA (Autoregressive Integrated Moving Average Model) or LSTM (Long Short-Term Memory Network), to capture the temporal characteristics of load changes. Collect historical load data, set model parameters (such as periodicity, trend component), and ensure that the model can accurately reflect the evolution trend of load requirements. Regularly train and validate the model to improve the prediction accuracy. Use the constructed model to predict the load requirements at multiple future time points. Set the prediction time range (such as the next 24 hours, 48 hours), and generate the corresponding load requirement prediction data. Record the load requirement prediction values at each time point and form a structured data table. Visualize the prediction results and display the future load trend through charts for easy analysis and decision-making. Extract the real-time load data and working status of each component from the motherboard status monitoring log. Power consumption calculation usually depends on the power characteristics of components, such as the power consumption characteristics of processors, memory, and other peripherals. Adopt a power consumption model, such as the Dynamic Power Model, and calculate the power consumption of each component through formulas: , where P is the power consumption, C is the capacitance under the load state, V is the voltage, and f is the frequency. Calculate the power consumption of each component one by one to generate a dataset containing the power consumption parameters of each component. The dataset should include information such as timestamps, component IDs, power consumption values, etc., for subsequent analysis. Use a data visualization tool to generate the change curve of the power consumption of each component, intuitively showing the power consumption dynamics. According to the predicted data of the industrial control computer load demand at multiple time points, combined with the power consumption parameters of each component, construct a power demand prediction model. Use machine learning methods such as linear regression, support vector machine, or neural network to predict the power demand of each component. Set the input features of the model, including the predicted load demand value, historical power consumption data, etc., to ensure that the model can take into account the impact of load changes on power demand. Use the constructed model to predict the power demand of each component, generating power demand prediction data at multiple time points. Predict the power demand for the next hour and day. Record the prediction results in the database to form a dataset for component power demand prediction and visualize it for subsequent analysis. Perform time series discrete fitting on the multi-time point power demand prediction data of each component. Select polynomial fitting, spline interpolation, or other interpolation methods to smooth the power demand data to ensure the continuity and readability of the data. Set appropriate fitting parameters (such as the polynomial order or the number of interpolation nodes) and adjust according to the actual data characteristics. Use a data fitting tool (such as the NumPy and SciPy libraries in Python) to fit the power demand data, generating the power demand prediction curve for each component. Record the fitting error metrics (such as the root mean square error RMSE) to evaluate the fitting effect. Visualize the fitting results and draw a comparison chart of the power demand prediction curve and the actual data to intuitively show the accuracy and trend of the fitting.

[0032] In this embodiment, step S5 includes the following steps: Step S51: Perform topological classification and division on the dynamic thermal distribution twin model to obtain multiple thermal distribution sub-region models; Step S52: Based on the power demand prediction curve of each component, perform temperature rise simulation for the industrial control computer in the operating state on multiple thermal distribution sub-region models, and collect temperature rise simulation data in the operating state; Step S53: Calculate the regional temperature rise rate for the temperature rise simulation data in the operating state to obtain the temperature rise rate parameter of each region; Step S54: Identify the temperature rise amplitude according to the temperature rise simulation data in the operating state, and extract the temperature rise amplitude value of each region; Step S55: Based on the temperature rise rate parameter of each region and the temperature rise amplitude value of each region, perform temperature rise trend mining for each region one by one to construct the temperature rise trend map of each region.

[0033] In this embodiment, the dynamic thermal distribution twin model is analyzed to identify the thermal distribution characteristics in the model. The clustering algorithm (such as K-means or DBSCAN) is used to classify the thermal distribution data, and an appropriate number of clusters (5 sub-regions) is set to capture the main thermal distribution patterns. The thermal distribution data is transformed into feature vectors, including parameters such as temperature, heat flux density, and thermal diffusivity, to ensure that the clustering can effectively reflect the changes in thermal distribution. The clustering results are verified to ensure that each sub-region model can effectively represent specific thermal distribution characteristics. Metrics such as the Silhouette Score are used to evaluate the clustering effect to ensure a high degree of differentiation between each sub-region. The characteristic parameters of each sub-region, such as the temperature range and heat source location, are recorded, and this information is organized into structured data for subsequent simulation and analysis. According to the power demand prediction curve of each component, a running-state temperature rise simulation model is constructed. Using thermal conduction simulation software (such as ANSYS Fluent or COMSOL Multiphysics), the power distribution of each sub-region is input. The simulation parameters are set, including environmental temperature, thermal conductivity, and specific heat capacity. Ensure that the environmental conditions can reflect the actual operating state of the industrial control computer. Start the simulation and record the temperature rise data at different time points. Set the time step (such as 10 times per second) during the simulation to ensure that the dynamic characteristics of the temperature rise change can be captured. Collect the simulation results, which involve temperature data and heat flux density information of each sub-region, to form a complete running-state temperature rise simulation dataset. Extract the historical data of the temperature change over time for each sub-region from the running-state temperature rise simulation data. The temperature rise rate calculation formula is: temperature rise rate = ΔT / Δt, where ΔT is the temperature change and Δt is the time change. For each sub-region, calculate its temperature rise rate and store the results in the database to form a temperature rise rate parameter dataset for each region. Use a visualization tool to display the temperature rise rate data as a chart to visually show the differences in temperature rise rates in different regions. Extract the temperature change range of each region from the running-state temperature rise simulation data. The temperature rise amplitude is defined as the difference between the highest temperature and the starting temperature at the end of the simulation, temperature rise amplitude = T(max) - T(initial). Calculate the temperature rise amplitude for each sub-region and record the results in the database. Form a structured temperature rise amplitude dataset, including region ID, starting temperature, highest temperature, and temperature rise amplitude value. Apply a data visualization tool to generate a comparison chart of the temperature rise amplitude to show the differences in temperature rise amplitudes in different regions for subsequent analysis. Combine the temperature rise rate and temperature rise amplitude values of each region to conduct a temperature rise trend analysis for each region one by one. Use the state space analysis method to establish a regional temperature rise model to identify the dynamic characteristics of the temperature rise. Set appropriate analysis parameters, considering the time changes in the temperature rise rate and amplitude, to capture the temperature rise trends in different regions. Visualize the analysis results to generate a temperature rise trend chart for each region, showing the evolution of the temperature rise rate and amplitude over time. Different colors and markers are used in the chart to visually display the changes in the temperature rise trend.Record the key parameters of the temperature rise trend graph to provide decision-making support for subsequent thermal management and optimization.

[0034] In this embodiment, step S6 includes the following steps: Step S61: Make an intelligent cooling medium switching decision for the temperature rise trend graph of each area and construct an intelligent cooling adjustment strategy; Step S62: Identify local temperature mutations in the temperature rise trend graph of each area to mark the areas with abnormal local temperature rises; Step S63: Calculate the temperature rise deviation trend for the areas with abnormal local temperature rises to extract the temperature rise deviation trend of the abnormal areas; Step S64: Calculate the cooling compensation power for the temperature rise deviation trend of the abnormal areas to obtain the cooling compensation parameters of the abnormal areas; Step S65: Dynamically adjust the cooling power of the intelligent cooling adjustment strategy according to the cooling compensation parameters of the abnormal areas to construct a dynamic cooling control model.

[0035] In this embodiment, the temperature rise trend graph of each area is analyzed, and a threshold value (temperature exceeding 70 °C) is set as the trigger condition for the cooling medium switch. A suitable cooling medium (such as a fan, liquid cooling, or phase change material) is selected for decision-making according to the actual temperature and load requirements of the area. A decision-making model is established using a decision tree or a fuzzy logic system, considering multiple factors (such as the current temperature, historical load data, environmental conditions, etc.) to make the cooling strategy more intelligent and adaptive. Cooling medium switching rules are formulated. When the temperature of a certain area continuously exceeds the set threshold, the cooling medium of this area is switched to a more efficient liquid cooling system. Conversely, when the temperature drops to the safe range, it is switched back to the standard air cooling system. Record the decision-making basis and implementation effect of each switch to evaluate the effectiveness of the strategy and optimize subsequent decisions. Use data analysis tools to perform local temperature mutation analysis on the temperature rise trend graph. Adopt a threshold-based detection method, and set the temperature change rate exceeding a certain critical value (such as 5 °C / minute) as the mutation flag. Implement moving average filtering (MAF) or other smoothing techniques to eliminate short-term fluctuations and ensure the accuracy of mutation identification. Once a local temperature mutation is identified, immediately mark these areas as abnormal areas, and record the time of mutation and the temperature change situation. Generate database records of abnormal areas, including information such as area ID, mutation time, and mutation amplitude. Mark these abnormal areas on the temperature rise trend graph through a visualization tool for subsequent analysis and decision-making. Extract the temperature values before and after the mutation from the temperature rise data of the local temperature rise abnormal area, and calculate the temperature rise deviation. The temperature rise deviation is defined as the difference between the current temperature and the expected temperature (such as the normal operating temperature): [\text{Temperature rise deviation} = T_{\text{current}} - T_{\text{expected}}] Conduct trend analysis, use linear regression or moving average methods to model the temperature rise deviation, and identify the change trend of the deviation over time. Record the temperature rise deviation trend data of the abnormal area in the database to form a structured data set, including time stamps, deviation values, and trend characteristics. Use a visualization tool to draw a temperature rise deviation trend graph to visually display the temperature rise changes in the abnormal area for subsequent analysis. According to the temperature rise deviation trend of the abnormal area, calculate the required cooling compensation power. Use a simple heat balance equation, set the temperature difference between the target temperature and the current temperature, and combine the heat conduction characteristics and cooling capacity of the area: P cool = K×(T current−T target) K is the cooling capacity parameter. Calculate the cooling compensation power of each abnormal area and record the results in the database to form a cooling compensation parameter data set. Include information such as area ID, required cooling power, current temperature, and target temperature. Visualize the cooling compensation power data to facilitate observing the differences in cooling requirements in different areas. According to the calculated cooling compensation parameters, establish a dynamic cooling control model to adjust the power output of the cooling system in real time. Adopt a PID control algorithm to perform dynamic adjustment according to the temperature rise state and cooling requirements.Set up a feedback mechanism to adjust the cooling power output by monitoring the temperature and load in real time to ensure that the temperature remains within a safe range. Start the cooling control system and perform dynamic power regulation based on real-time data and cooling compensation parameters. The monitoring system should be able to collect feedback data in real time and adjust the cooling power promptly. Record the parameters and effects of each adjustment for subsequent optimization of the cooling strategy and model.

[0036] In this embodiment, the specific steps of step S61 are as follows: Conduct a comprehensive evaluation of the temperature rise state characteristics for the temperature rise trend map of each region to obtain the temperature rise state evaluation value of each region; Define the temperature rise rate and amplitude in the normal state, and perform an analysis of the normal state temperature rise threshold limit based on the temperature rise rate and amplitude in the normal state to obtain the normal state temperature rise threshold; Compare the temperature rise state evaluation value of each region based on the normal state temperature rise threshold. When the normal state temperature rise threshold is greater than the temperature rise state evaluation value, make an air-cooling decision for the region to obtain an air-cooling strategy; When the normal state temperature rise threshold is less than or equal to the temperature rise state evaluation value, perform an immediate liquid-cooling switch for the region to construct an immediate liquid-cooling switch strategy; Conduct real-time monitoring of the temperature rise trend and intelligent cooling switch control according to the air-cooling strategy and the immediate liquid-cooling switch strategy, thereby constructing an intelligent cooling adjustment strategy.

[0037] In this embodiment, using the temperature data in the temperature rise trend map, define the comprehensive temperature rise state characteristics, including the temperature rise rate, amplitude, and temperature distribution. Adopt statistical methods, such as mean, standard deviation, and maximum value, etc., to quantify the temperature rise characteristics of each region. Calculate the temperature rise rate of each region, and the temperature rise rate = where and are the maximum and minimum temperatures of the calculated region, and For the corresponding time point, integrate the above features to form a comprehensive temperature rise state evaluation value, which is calculated by the method of weighted average. The weights can be set according to the importance of each feature. Evaluation value = W1 × Temperature rise rate + W2 × Temperature rise amplitude, where W1 and W2 are weight coefficients and W1 + W2 = 1. Record the evaluation value of each area and generate a structured data set for subsequent analysis and decision-making. Define the temperature rise rate and amplitude of the normal state through historical operation data. Calculate the average rate and amplitude within a certain time range as the reference value of the normal state. Set the normal state temperature rise rate to 2 °C / minute and the temperature rise amplitude to 10 °C. Record the normal state parameters for subsequent threshold limit analysis. Combine the rate and amplitude of the normal state to conduct threshold limit analysis of the temperature rise. Use statistical methods to set the temperature rise threshold to a certain multiple (such as 1.5 times) of the normal state temperature rise rate and amplitude to ensure safety. Temperature rise threshold = 1.5 × Normal state amplitude. The calculated temperature rise threshold should be recorded in the database for subsequent comparison. Compare the temperature rise state evaluation value of each area with the normal state temperature rise threshold. Use simple conditional judgment to determine whether the evaluation value exceeds the threshold. Record the comparison result. If the normal state temperature rise threshold is greater than the temperature rise state evaluation value, mark it as safe; otherwise, mark it as needing cooling. Based on the comparison result, make a decision on the air cooling strategy for the area that needs cooling. Set the specific parameters of the air cooling strategy, such as fan speed, cooling time, etc., and generate a cooling control instruction. Record the air cooling strategy in the database to form a decision-making document for subsequent analysis and implementation. When the normal state temperature rise threshold is less than or equal to the temperature rise state evaluation value, formulate an immediate liquid cooling switching strategy. Set the parameters of the liquid cooling system, such as flow rate, temperature set point, etc., to ensure timely and effective cooling. Combine the dynamic characteristics of the liquid cooling system to set the switching conditions and response time, such as switching to liquid cooling within 5 seconds after the temperature exceeds the threshold. Record the immediate liquid cooling switching strategy in the database, including switching conditions, implementation steps, expected effects and other information to ensure the traceability of operations. Through data visualization tools, display the implementation effect of the liquid cooling strategy for subsequent effect evaluation. Build a real-time temperature rise situation monitoring system, integrate temperature sensors and data acquisition modules to monitor the temperature rise of each area in real time. Set the monitoring frequency (such as once every 5 seconds) to ensure the timeliness of data. Input the monitoring data into the central control system for intelligent cooling strategy decision-making. Based on the air cooling and liquid cooling strategies, implement intelligent cooling switching control. Through control algorithms (such as fuzzy control or PID control), dynamically adjust the cooling method according to real-time temperature data to ensure that the temperature is kept within the safe range. Record the input data and output results of each cooling decision to form a complete cooling control log for subsequent analysis and optimization.

[0038] In this embodiment, a temperature control system for an industrial computer motherboard is provided, which is used to execute the temperature control method for the industrial computer motherboard as described above, including: 3D topological module, used to collect the CT imaging structure diagram of the main board; perform physical connection analysis between components and component space topology mining on the CT imaging structure diagram of the main board, and construct a 3D topological structure model of the main board; Twin model module, used to calculate the real-time temperature value of each main board component; perform time-series temperature fluctuation analysis and dynamic temperature distribution rendering on the 3D topological structure model of the main board based on the real-time temperature value of each main board component, so as to construct a dynamic thermal distribution twin model; Load fluctuation module, used to obtain the main board status monitoring log; perform load fluctuation trend evolution analysis on the main board status monitoring log, and perform multi-period cycle correlation mining, so as to generate multi-time series load fluctuation evolution rules; Demand prediction module, used to perform multi-time point load demand prediction for the industrial control computer according to the multi-time series load fluctuation evolution rules, and perform power demand prediction for each component one by one, so as to construct the power demand prediction curve of each component; Temperature rise trend module, used to perform temperature rise simulation of the industrial control computer operating state on the dynamic thermal distribution twin model based on the power demand prediction curve of each component, and perform temperature rise trend mining for each area one by one, so as to construct the temperature rise trend map of each area; Intelligent cooling control module, used to make intelligent cooling medium switching decisions based on the temperature rise trend map of each area, and perform dynamic cooling power adjustment, so as to construct a dynamic cooling control model.

[0039] By obtaining the CT imaging structure diagram of the main board, the present invention can accurately restore the spatial layout of all components on the main board and their interconnection relationships, providing accurate hardware structure data. The constructed three-dimensional topological structure model provides a necessary physical basis for subsequent analyses such as temperature fluctuations and load changes, enabling the system temperature control to be optimized according to the actual hardware layout. The clear spatial layout of components helps in the design of the cooling system, can determine the heat source intensive areas, anticipate the temperature control difficulties in advance, optimize the cooling scheme, and obtain the temperature of each component in real time, providing accurate data support for the thermal management of the main board. The real-time change trend of temperature can help detect overheating problems in a timely manner. Through the sequential temperature fluctuation analysis, clearly understand the temperature changes of each component at different time points, and help engineers make dynamic adjustments. In addition, the temperature distribution rendering technology visually presents these data, making the temperature changes easier to understand. Through dynamic rendering and analysis, construct a thermal distribution model that truly reflects the temperature state of the main board, providing a scientific basis for future optimization and adjustment, helping to capture the fluctuation trend of the load during the operation of the industrial control computer, revealing the change law of the load. Through the correlation analysis of multiple time periods, the periodic law of load fluctuations can be discovered, which helps to optimize the temperature control strategy under different load conditions. Through the evolution analysis of the load fluctuation trend, the load peaks and troughs can be predicted in advance, providing reliable data support for the subsequent adjustment of temperature management and cooling strategies. Through the analysis of load fluctuations, the power demand of each component at different time points can be accurately predicted, which can effectively guide the working state of the cooling system, avoid over-cooling or under-cooling. The power demand prediction curve provides data support for the intelligent scheduling of the cooling system, helping the system to adjust the cooling strategy according to the change of power demand during actual operation, and optimizing energy consumption. By combining the power demand prediction with the thermal distribution model, the temperature rise trend of each area of the main board can be accurately simulated. By analyzing the simulation results, the areas with overheating can be found, and measures can be taken in advance. Through intelligent decision-making, the cooling medium (such as air, liquid or other cooling methods) can be switched in real time, and the cooling power can be dynamically adjusted according to the area temperature rise situation map. This enables the cooling system to be optimized and adjusted according to the real-time state of the main board. Dynamically adjusting the cooling power can adjust the allocation of cooling resources according to the demand, avoid unnecessary energy waste, and at the same time ensure that the temperature control effect is not affected. Effective cooling control can avoid the performance reduction or hardware damage caused by overheating of the system, thus greatly improving the long-term stability and reliability of the industrial control computer main board.

[0040] Therefore, from any perspective, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, it is intended to cover all changes falling within the meaning and scope of the equivalent elements of the application documents within the present invention.

[0041] The above are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the broadest scope consistent with the principles and novel features invented herein.

Claims

1. A temperature control method for an industrial computer motherboard, characterized in that, Including the following steps: Step S1: Collect the CT imaging structure diagram of the main board; perform physical connection analysis between components and component space topology mining on the CT imaging structure diagram of the main board to construct a 3D topological structure model of the main board; Step S2: Calculate the real-time temperature value of each main board component; Based on the real-time temperature values of each main board component, perform time-series temperature fluctuation analysis and dynamic temperature distribution rendering on the 3D topological structure model of the main board to construct a dynamic thermal distribution twin model; Step S3: Obtain the main board status monitoring log; perform load fluctuation trend evolution analysis on the main board status monitoring log and conduct multi-period cycle correlation mining to generate multi-time-series load fluctuation evolution rules; Step S4: Predict the multi-time-series load demand of the industrial control computer according to the multi-time-series load fluctuation evolution rules, and perform power demand prediction for each component one by one to construct a power demand prediction curve for each component; Step S5: Based on the power demand prediction curve of each component, perform temperature rise simulation of the industrial control computer in the running state on the dynamic thermal distribution twin model, and conduct temperature rise trend mining for each area one by one to construct a temperature rise trend map for each area; Step S6: Based on the temperature rise trend map of each area, make an intelligent cooling medium switching decision and perform dynamic cooling power adjustment to construct a dynamic cooling control model.

2. The temperature control method for an industrial computer motherboard according to claim 1, characterized in that, The specific steps of Step S1 are as follows: Step S11: Based on a CT scanner, perform high-precision scanning on the main board of the industrial control computer to collect the CT imaging structure diagram of the main board; Step S12: Perform deep visual recognition on the CT imaging structure diagram of the main board and mark each main board component node; Step S13: Perform 3D spatial position registration calculation on each main board component node to generate the 3D spatial coordinates of each component; Step S14: Perform physical connection analysis between each main board component node to generate the physical connection relationship between components; Step S15: Based on the physical connection relationship between components and the 3D spatial coordinates of each component, perform component space topology mining to obtain component space topology connection data; Step S16: According to the component space topology connection data, perform 3D topological point cloud modeling on the CT imaging structure diagram of the main board to construct a 3D topological structure model of the main board.

3. The temperature control method for an industrial computer mainboard according to claim 1, wherein The specific steps of Step S2 are as follows: Step S21: Based on temperature monitoring sensors, calculate the real-time temperature value of each main board component; Step S22: Perform time-series temperature fluctuation analysis on the real-time temperature values of each main board component to generate a temperature fluctuation curve for each component; Step S23: According to the 3D spatial coordinates of each component, perform main board heat distribution analysis on the temperature fluctuation curve of each component to construct a dynamic thermal distribution change diagram; Step S24: According to the dynamic thermal distribution change diagram, perform dynamic temperature distribution rendering on the 3D topological structure model of the main board to construct a dynamic thermal distribution twin model.

4. The temperature control method for an industrial control computer motherboard according to claim 3, characterized in that, The specific steps of Step S24 are as follows: Based on the 3D topological structure model of the main board, perform thermal diffusion evolution for each component one by one to obtain the thermal diffusion path of each component; Based on the real-time temperature values of each main board component, perform temperature gradient calculation to generate a main board temperature gradient map; Perform thermal conduction loss analysis on the main board temperature gradient diagram to generate thermal conduction loss data between components; Perform heat diffusion dynamics mining based on the thermal conduction loss data between components and the heat diffusion path of each component to generate the main board heat diffusion evolution characteristics; Perform dynamic temperature distribution rendering on the main board three-dimensional topological structure model according to the main board heat diffusion evolution characteristics and the dynamic heat distribution change diagram to construct a dynamic heat distribution twin model.

5. The temperature control method for an industrial control computer mainboard according to claim 1, characterized in that, The specific steps of step S3 are as follows: Step S31: Obtain the main board status monitoring log; calculate the CPU utilization rate for the main board status monitoring log to obtain the real-time CPU utilization rate; Step S32: Quantify the memory usage based on the main board status monitoring log to generate a memory occupancy rate quantification value; Step S33: Calculate the real-time load of the main board according to the real-time CPU utilization rate and the memory occupancy rate quantification value to generate the real-time load characteristics of the main board; Step S34: Perform load fluctuation trend evolution analysis on the real-time load characteristics of the main board to extract the load fluctuation trend characteristics; Step S35: Perform multi-period cycle correlation mining on the load fluctuation trend characteristics to generate the multi-time series load fluctuation evolution law.

6. The temperature control method for an industrial control computer motherboard according to claim 1, characterized in that, The specific steps of step S4 are as follows: Step S41: Perform multi-time point load demand prediction for the industrial control computer according to the multi-time series load fluctuation evolution law to generate industrial control computer load demand prediction data at multiple time points; Step S42: Calculate the power consumption of each main board component node one by one based on the main board status monitoring log to generate the power consumption parameters of each component; Step S43: Perform component power demand prediction on the power consumption parameters of each component according to the industrial control computer load demand prediction data at multiple time points to obtain the multi-time point power demand prediction data of each component; Step S44: Perform time series discrete fitting on the multi-time point power demand prediction data of each component to construct the power demand prediction curve of each component.

7. The temperature control method for the industrial control computer mainboard according to claim 1, characterized in that, The specific steps of step S5 are as follows: Step S51: Perform topological classification and division on the dynamic heat distribution twin model to obtain multiple heat distribution sub-region models; Step S52: Perform industrial control computer operating state temperature rise simulation on multiple heat distribution sub-region models based on the power demand prediction curve of each component, and collect the operating state temperature rise simulation data; Step S53: Calculate the regional temperature rise rate for the operating state temperature rise simulation data to obtain the temperature rise rate parameters of each region; Step S54: Identify the temperature rise amplitude according to the operating state temperature rise simulation data, and extract the temperature rise amplitude value of each region; Step S55: Perform temperature rise trend mining for each region based on the temperature rise rate parameters of each region and the temperature rise amplitude value of each region to construct the temperature rise trend diagram of each region.

8. The temperature control method for the industrial computer mainboard according to claim 1, characterized in that The specific steps of step S6 are as follows: Step S61: Make an intelligent cooling medium switching decision on the temperature rise trend diagram of each region to construct an intelligent cooling adjustment strategy; Step S62: Identify local temperature mutations on the temperature rise trend diagram of each region to mark the local temperature rise abnormal regions; Step S63: Calculate the temperature rise deviation trend for the local temperature rise abnormal regions to extract the temperature rise deviation trend of the abnormal regions; Step S64: Calculate the cooling compensation power for the temperature rise deviation trend in the abnormal area to obtain the cooling compensation parameters for the abnormal area; Step S65: Dynamically adjust the cooling power of the intelligent cooling adjustment strategy according to the cooling compensation parameters of the abnormal area to construct a dynamic cooling control model.

9. The temperature control method for an industrial computer mainboard according to claim 8, characterized in that, The specific steps of Step S61 are as follows: Comprehensively evaluate the temperature rise state characteristics of the temperature rise trend map of each area to obtain the temperature rise state evaluation value of each area; Define the temperature rise rate and amplitude of the normal state, and perform the analysis of the normal state temperature rise threshold limit according to the temperature rise rate and amplitude of the normal state, so as to obtain the normal temperature rise threshold; Based on the normal temperature rise threshold, compare the temperature rise state evaluation value of each area. When the normal temperature rise threshold is greater than the temperature rise state evaluation value, make an air-cooling decision for the area to obtain an air-cooling strategy; When the normal temperature rise threshold is less than or equal to the temperature rise state evaluation value, perform an immediate liquid cooling switch for the area to construct an immediate liquid cooling switch strategy; Perform real-time temperature rise trend monitoring and intelligent cooling switch control according to the air-cooling strategy and the immediate liquid cooling switch strategy, so as to construct an intelligent cooling adjustment strategy.

10. A temperature control system for an industrial computer motherboard, characterized in that, Used to execute the temperature control method for the industrial computer motherboard as described in claim 1, including: A three-dimensional topology module for collecting the main board CT imaging structure diagram; performing physical connection analysis between components and component space topology mining on the main board CT imaging structure diagram to construct a three-dimensional topology structure model of the main board; A twin model module for calculating the real-time temperature value of each main board component; performing time-series temperature fluctuation analysis and dynamic temperature distribution rendering on the three-dimensional topology structure model of the main board based on the real-time temperature value of each main board component to construct a dynamic thermal distribution twin model; A load fluctuation module for obtaining the main board status monitoring log; performing load fluctuation trend evolution analysis on the main board status monitoring log and performing multi-period cycle correlation mining to generate multi-time series load fluctuation evolution rules; A demand prediction module for predicting the multi-point load demand of the industrial computer according to the multi-time series load fluctuation evolution rules and predicting the power demand of each component one by one to construct the power demand prediction curve of each component; A temperature rise trend module for performing temperature rise simulation of the industrial computer operating state on the dynamic thermal distribution twin model based on the power demand prediction curve of each component and performing temperature rise trend mining for each area one by one to construct the temperature rise trend map of each area; An intelligent cooling control module for making an intelligent cooling medium switching decision based on the temperature rise trend map of each area and performing dynamic cooling power adjustment to construct a dynamic cooling control model.

Images