Energy storage power station battery heat flow digital twin simulation and cold adaptive control method

Abstract

The application discloses a method for energy storage power station battery heat flow digital twin simulation and cold adaptive control, and relates to the technical field of battery heat flow data processing.The method comprises the following steps: S1, collecting and preprocessing battery characteristic data and cooling characteristic data, and constructing a standardized thermal state data set; S2, evaluating the overall heat flow density of the battery, and dynamically adjusting the cooling pump speed and the refrigerant flow distribution; S3, analyzing the simulation accuracy, and real-time correcting the cooling boundary conditions; S4, analyzing the cooling demand intensity, and generating a hierarchical adjustment instruction for the refrigerant flow and the pump speed based on the demand analysis results; and S5, generating and checking the cold distribution instruction, issuing the execution and performing the preventive adjustment.The method solves the problem that most current thermal management systems lack heat flow multi-dimensional space perception and model coupling capability, and three-dimensional heat distribution cannot be mapped in real time, thereby limiting fine regulation and control and early identification of heat distribution abnormalities.

Description

Technical Field

[0001] This invention relates to the field of battery thermal flow data processing technology, specifically to a digital twin simulation method for battery thermal flow and adaptive cooling control method for energy storage power stations. Background Technology

[0002] During the charging and discharging process of energy storage batteries, a large amount of electrical energy is converted into heat and accumulates rapidly within the battery cluster. Therefore, thermal management technology is widely used for battery operating status sensing, cooling circuit regulation, and system energy consumption optimization. Meanwhile, with the continuous improvement of battery cluster capacity and power levels, digital twin technology has gradually been introduced into the field of thermal management. Through synchronous modeling and interactive simulation of battery geometry, operating signals, and sensor data, a virtual model highly coupled with the actual battery pack is constructed, providing predictive support for cooling distribution and operating strategies.

[0003] For example, the invention patent with announcement number CN112818602B provides a battery digital twin control method and device based on big data analysis. The method includes: performing clustering processing based on vehicle information of multiple vehicles and battery information of the batteries in each vehicle to obtain at least one vehicle category and at least one battery category; selecting a target vehicle from among the multiple vehicles based on the at least one vehicle category; selecting a target battery from among the target vehicles based on the at least one battery category; obtaining the battery parameters of the target battery corresponding to the target vehicle through a third-party device, or obtaining the battery parameters of the target battery corresponding to the target vehicle when the target vehicle is under special operating conditions; and updating the parameters of the refined multiphysics model corresponding to the target battery based on the battery parameters of the target vehicle. By establishing refined multiphysics models only for a small number of batteries in a small number of vehicles, the problem of extremely high cloud computing resource consumption is solved.

[0004] For example, invention patent CN114004168B discloses a fuel cell integrated management system and method based on digital twins. The system includes a physical fuel cell stack for generating dynamic performance parameters during operation; a digital twin model for simulating the dynamic performance parameters of the physical fuel cell stack in real time; a data acquisition platform for acquiring the dynamic performance parameters of the physical fuel cell stack in real time, preprocessing and extracting features from the data, transmitting the extracted feature data to the digital twin model, and driving the model's operation; and a fuel cell and lithium battery terminal management module as a human-computer interface for remotely and in real-time visually presenting the operating status of the digital twin model and controlling the adjustment parameters of the physical fuel cell stack and the digital twin model. This invention enables remote, real-time, and visual monitoring of complex fuel cell systems; and allows for timely and accurate adjustment of various relevant physical quantities of the physical equipment based on the operating results of the digital twin model, achieving optimal performance.

[0005] However, existing thermal management methods largely rely on point temperature monitoring and single flow rate regulation, lacking real-time perception and coupled modeling of the three-dimensional heat flow distribution within the battery cluster. This makes it impossible to identify local hotspots and temperature gradients on a spatial scale. In cooling control, fixed thresholds or single flow rate increases / decreases are often used, ignoring the dynamic response characteristics of the refrigerant to thermal disturbances under different operating conditions. This can easily lead to overcooling or insufficient cooling, thus affecting the stability and energy efficiency of thermal management. Furthermore, while digital twins have been applied in manufacturing and structural simulation, the field of thermal control in energy storage power stations still lacks deep integration of battery geometry, operating signals, and sensor data. This results in discrepancies between the predicted model and the measured thermal field, making it difficult to achieve adaptive allocation of cooling supply.

[0006] To address the above issues, there is an urgent need for digital twin simulation of battery thermal flux and adaptive control methods for cooling capacity in energy storage power stations. Summary of the Invention

[0007] Technical problems to be solved

[0008] To address the shortcomings of existing technologies, this invention provides a digital twin simulation of battery thermal flow and adaptive control method for cooling capacity in energy storage power stations. This method solves the problem that most current thermal management systems lack multi-dimensional spatial perception and model coupling capabilities for thermal flow, and cannot map three-dimensional thermal distribution in real time, thus limiting refined control and early identification of thermal distribution anomalies.

[0009] Technical solution

[0010] To achieve the above objectives, this invention employs the following technical solution: a digital twin simulation of battery thermal flux and adaptive cooling control method for energy storage power stations, comprising: S1, collecting battery characteristic data and cooling characteristic data during battery operation, and preprocessing the collected battery characteristic data and cooling characteristic data to construct a standardized thermal state dataset; S2, based on the standardized thermal state dataset, evaluating the overall heat flux density of the battery, and dynamically adjusting the cooling pump speed and refrigerant flow distribution based on the heat flux evaluation results; S3, based on the standardized thermal state dataset, performing accuracy analysis on the simulation accuracy from the difference between the measured temperature and the predicted temperature from infrared thermal imaging, and correcting the cooling boundary conditions in real time based on the accuracy analysis results; S4, using the heat flux evaluation results and accuracy analysis results as input, performing demand analysis on the cooling demand intensity, and generating graded adjustment commands for refrigerant flow and pump speed based on the demand analysis results; S5, generating and verifying cooling capacity distribution commands in each cooling control cycle, issuing and executing them, and combining real-time deviation feedback and short-term prediction for preventative adjustment.

[0011] Furthermore, the specific steps for collecting battery characteristic data and cooling characteristic data during battery operation are as follows: collecting battery characteristic data during battery operation, including: single cell volume, battery heating power, highest battery surface temperature, lowest battery surface temperature, and infrared thermal imaging measured temperature, while simultaneously calculating and recording the standard deviation of battery temperature fluctuation; collecting cooling characteristic data, including: refrigerant flow rate, refrigerant inlet temperature, and refrigerant outlet temperature, while simultaneously calculating and recording the average inlet and outlet temperature difference of the refrigerant.

[0012] Furthermore, the specific steps for preprocessing the collected battery feature data and cooling feature data to construct a standardized thermal state dataset are as follows: For the battery feature data, the volume of a single battery cell is registered in the database as a geometric constant; the battery heating power is cross-validated and subjected to moving average filtering; the highest and lowest temperatures on the battery surface are extracted based on the measured temperatures from infrared thermal imaging to extract edge values ​​and construct the battery temperature field; the measured temperatures from infrared thermal imaging are processed by Gaussian denoising, and the standard deviation of the battery temperature fluctuation is calculated using a sliding window mechanism; For the cooling feature data, the refrigerant flow rate is smoothed over time and subjected to variability analysis to remove pulse interference; the instantaneous temperature difference is calculated after aligning the refrigerant inlet and outlet temperatures, and the average inlet and outlet temperature difference of the refrigerant is statistically analyzed; the standardized battery feature data and cooling feature data are then normalized to construct a standardized thermal state dataset.

[0013] Furthermore, the specific steps for evaluating the overall heat flux density of the battery based on the standardized thermal state dataset are as follows: divide the volume of a single battery cell by the battery's heating power to obtain a volumetric heating intensity value; subtract the lowest temperature on the battery surface from the highest temperature on the battery surface and multiply it by the corresponding temperature weighting factor to obtain a surface temperature difference correction value; add one to the refrigerant flow rate and divide it by the refrigerant flow rate and multiply it by the corresponding cooling weighting factor to obtain a cooling flow rate weighted value; add the volumetric heating intensity value, the surface temperature difference correction value, and the cooling flow rate weighted value to obtain a three-dimensional heat flux evaluation value.

[0014] Furthermore, the specific steps for dynamically adjusting the cooling pump speed and refrigerant flow distribution based on the heat flow assessment results are as follows: Real-time comparison of the three-dimensional heat flow assessment value with the heat flow threshold: When the three-dimensional heat flow assessment value is less than the heat flow threshold, the cooling pump operating speed remains unchanged, the refrigerant maintains the default flow rate, and only the normal heat flow distribution is output in the battery twin model; When the heat flow assessment value is equal to the heat flow threshold, the cooling pump is adjusted to 25% to 40% of the rated value to increase the refrigerant flow rate, highlighting potential hot spots in the battery twin model in real time, and generating a cooling capacity allocation plan in advance; When the three-dimensional heat flow assessment value exceeds the heat flow threshold, the cooling pump is adjusted to the highest speed, the refrigerant flow rate is increased to the maximum, and a high-load operation command is immediately sent to the cooling capacity adjustment device, while the three-dimensional heat distribution in the battery twin model is updated in real time and high-risk areas are marked, triggering an operation and maintenance alarm.

[0015] Furthermore, the specific steps for analyzing the simulation accuracy based on the difference between the measured and predicted temperatures from infrared thermal imaging, using a standardized thermal state dataset, are as follows: Input data is used as training data, including: the battery's three-dimensional geometry and single-cell volume established based on CAD refinement, refrigerant flow rate, refrigerant inlet and outlet temperatures, and the battery temperature field extracted from infrared thermal imaging; on the modeling side, transient conjugate heat transfer is solved on the input data using CFD to obtain high-fidelity temperature and heat flow distribution; then, a fast prediction model is constructed using principal component dimensionality reduction and thermal RC network equivalence methods, and an extended Kalman filter is used... Online calibration is performed; on the control side, the fast prediction model is coupled with Simulink, and short-time rolling optimization is used to solve the inverse problem under constraints to meet the cooling requirements of the seat belt. On the output side, the predicted temperature and predicted cooling flow rate are output in real time; the absolute value of the difference between the measured temperature and the predicted temperature of the battery by infrared thermal imaging is divided by one and the sum of the standard deviation of temperature fluctuation, and then one is added and the natural logarithm is taken to obtain the temperature difference deviation correction value; the value of the predicted cooling flow rate plus one is divided by the refrigerant flow rate minus the predicted cooling flow rate, and then one is added to obtain the cooling deviation correction value; the temperature difference deviation correction value and the cooling deviation correction value are multiplied to obtain the model deviation correction value.

[0016] Furthermore, the specific steps for real-time correction of cooling boundary conditions based on accuracy analysis results are as follows: When the calculated results of three consecutive model deviation correction values ​​continue to increase, it is determined that the deviation between model prediction and actual measurement has widened. The measured temperature and flow rate data from infrared thermal imaging are fed back to the battery twin model to dynamically correct the thermal conductivity and boundary conditions, and trigger the cooling capacity adjustment device to enter an early warning state. When the fluctuations of the calculated results of three consecutive model deviation correction values ​​are all less than the tolerance threshold, the battery twin model access data is continuously updated, and the cooling pump speed is adjusted to 25% to 40% of the rated value to maintain stable cooling circuit flow. At the same time, the deviation area is marked in the battery twin model and monitoring is prompted. When the calculated results of three consecutive model deviation correction values ​​continue to decrease, the battery twin model parameters are kept unchanged, the cooling pump speed is reduced, and the cooling circuit enters energy-saving mode.

[0017] Furthermore, the specific steps for conducting a demand analysis on cooling demand intensity using heat flow assessment results and accuracy analysis results as input are as follows: multiply the three-dimensional heat flow assessment value by the model deviation correction value plus one, and then take the square root to obtain the heat flow deviation coupling value; add one to the average inlet and outlet temperature difference of the refrigerant and divide it by the current refrigerant inlet temperature minus the refrigerant outlet temperature to obtain the cooling temperature difference ratio; add the heat flow deviation coupling value to the cooling temperature difference ratio to obtain the adaptive allocation value of cooling capacity.

[0018] Furthermore, the specific steps for generating graded adjustment commands for refrigerant flow and pump speed based on the demand analysis results are as follows: Real-time comparison of the adaptive cooling capacity allocation value and the cooling capacity allocation threshold, where the cooling capacity allocation threshold includes a first cooling capacity threshold and a second cooling capacity threshold, with the first cooling capacity threshold being higher than the second cooling capacity threshold: When the adaptive cooling capacity allocation value is less than or equal to the second cooling capacity threshold, the cooling pump is kept at its lowest speed, the output of the cooling capacity adjustment device remains unchanged, thermal imaging image enhancement is turned off, only the original values ​​of battery characteristic data are recorded, the simulation incremental calculation process is turned off, the three-dimensional thermal distribution map is not refreshed, and only the temperature sequence of the central node is appended; when the adaptive cooling capacity allocation value is greater than the second cooling capacity threshold... When the threshold value is less than or equal to the first cooling threshold, the cooling pump speed is increased, the image analysis algorithm is started to reconstruct the boundaries of the hotspot area, a simplified heat map panel is generated and synchronized to the control interface, and the current cooling distribution range information is updated; when the cooling adaptive allocation value is greater than the first cooling threshold, the cooling pump is immediately switched to the highest speed, the dynamic rendering of the three-dimensional thermal distribution map and the hotspot migration trajectory tracking are continuously started, the full cycle charge and discharge task data are loaded and a third-order time step prediction sequence is generated; the simulation engine starts the complete reconstruction path and synchronizes the real-time battery feature data and cooling feature data, while recording the current battery feature data and cooling feature data as a high-frequency data snapshot.

[0019] Furthermore, the specific steps for generating and verifying the cooling capacity allocation command within each cooling control cycle, issuing and executing it, and combining real-time deviation feedback and short-term prediction for preventative adjustment are as follows: Within each cooling control cycle, a cooling capacity allocation command is generated based on the three-dimensional heat flow evaluation value and the model deviation correction value. The command covers the cooling pump speed and refrigerant flow rate. A shadow simulation is performed in the battery twin model. When the predicted temperature is equal to or greater than the safety threshold, the cooling capacity allocation command is rolled back and re-verified until the predicted temperature is less than the safety threshold before being issued to the execution component. During the execution phase of the cooling capacity allocation command, the average inlet and outlet temperature difference of the refrigerant is continuously collected and compared with the predicted temperature. At the same time, the deviation information between the average inlet and outlet temperature difference of the refrigerant and the twin predicted temperature is fed back to the battery twin model. Before the end of the cooling control cycle, a short-term prediction is performed based on the adaptive cooling capacity allocation value. When a local hot spot trend appears, the cooling pump speed is increased in advance, and the execution and readback results are recorded as baseline data.

[0020] Beneficial effects

[0021] The present invention has the following beneficial effects:

[0022] (1) The battery thermal flow digital twin simulation and cooling capacity adaptive control method of the energy storage power station achieves accurate characterization of the internal thermal distribution state of the battery by introducing a three-dimensional thermal flow evaluation mechanism based on battery geometric modeling and infrared measured data fusion.

[0023] (2) The battery thermal flow digital twin simulation and cooling capacity adaptive control method of the energy storage power station effectively eliminates the error accumulation between prediction and measurement by establishing model deviation correction value and introducing feedback closed loop, and enhances the stability of control response.

[0024] (3) The battery heat flow digital twin simulation and cooling capacity adaptive control method of the energy storage power station realizes the refined allocation and energy-saving management of cooling resources by dynamically calculating the cooling capacity adaptive allocation value and triggering the control strategy in stages.

[0025] (4) The battery thermal flow digital twin simulation and cooling capacity adaptive control method of the energy storage power station ensures that the cooling capacity command has a temperature safety zone prediction guarantee before execution by embedding shadow simulation and parameter verification mechanism in each control cycle.

[0026] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0027] Figure 1 This is a flowchart of the digital twin simulation of battery thermal flux and adaptive control of cooling capacity in the energy storage power station according to the present invention.

[0028] Figure 2 This is a line graph showing the adaptive allocation of cooling capacity involved in this invention.

[0029] Figure 3 This is a schematic diagram of the signal principle involved in this invention;

[0030] Figure 4 This is a flowchart illustrating the signal generation process involved in this invention. Detailed Implementation

[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] Please see Figures 1-4 This invention provides a technical solution: a digital twin simulation of battery thermal flux and adaptive cooling control method for energy storage power stations, comprising: S1, collecting battery characteristic data and cooling characteristic data during battery operation, and preprocessing the collected battery characteristic data and cooling characteristic data to construct a standardized thermal state dataset; S2, based on the standardized thermal state dataset, evaluating the overall heat flux density of the battery, and dynamically adjusting the cooling pump speed and refrigerant flow distribution based on the heat flux evaluation results; S3, based on the standardized thermal state dataset, performing accuracy analysis on the simulation accuracy from the difference between the measured temperature and the predicted temperature from infrared thermal imaging, and correcting the cooling boundary conditions in real time based on the accuracy analysis results; S4, using the heat flux evaluation results and accuracy analysis results as input, performing demand analysis on the cooling demand intensity, and generating graded adjustment commands for refrigerant flow and pump speed based on the demand analysis results; S5, generating and verifying cooling capacity distribution commands in each cooling control cycle, issuing and executing them, and combining real-time deviation feedback and short-term prediction for preventive adjustment.

[0033] Specifically, the steps for collecting battery characteristic data and cooling characteristic data during battery operation are as follows: Collect battery characteristic data during battery operation, including: single cell volume, used to establish a normalized benchmark for heat flux density; battery heating power, verified by combining the product of bus voltage and current with controller power consumption, reflecting the intensity of the internal heat source per unit time; the highest and lowest battery surface temperatures, extracted from the infrared thermal imaging matrix to construct thermal distribution gradient features; and the measured infrared thermal imaging temperature, acquired using a high-frequency frame rate area array infrared sensor, serving as direct input for dynamic thermal field reconstruction and simulation feedback. Simultaneously, calculate and record the standard deviation of battery temperature fluctuations, using a sliding time window to statistically analyze temperature dispersion, reflecting the thermal stability within the current cycle.

[0034] Cooling characteristic data is collected, including: refrigerant flow rate, which is continuously monitored by a flow sensor to provide a basis for cooling capacity adjustment intensity; refrigerant inlet temperature and refrigerant outlet temperature, which are synchronously collected at the inlet and outlet positions and time-stamped by a temperature sensing unit; and the average inlet and outlet temperature difference of the refrigerant is calculated and recorded as a key indicator to measure the current heat exchange intensity and cooling efficiency, which is used to drive the adaptive cooling capacity allocation logic.

[0035] This implementation plan constructs a fundamental data system to support heat flow assessment and cooling capacity regulation. By collecting and statistically analyzing battery characteristic data and cooling characteristic data generated during battery operation with high precision and in multiple dimensions, it provides complete and dynamically updated data support for subsequent three-dimensional thermal field modeling, heat intensity estimation, cooling capacity evaluation, and adaptive control strategies. This step not only achieves simultaneous perception of the battery's internal heating behavior and external cooling response state, but also enhances the quantitative expression of thermal stability and heat transfer efficiency through the calculation of key indicators such as the standard deviation of temperature fluctuation and the average inlet and outlet temperature difference. It is a prerequisite for achieving accurate twin modeling and closed-loop cooling capacity regulation.

[0036] Specifically, the collected battery feature data and cooling feature data are preprocessed to construct a standardized thermal state dataset. The specific steps are as follows: For battery feature data, the volume of a single cell is registered as a geometric constant in the database for subsequent heat flux density normalization calculation and parameter binding for 3D geometric modeling; the battery heating power is obtained through cross-validation of DC bus voltage and current with controller power consumption, and instantaneous spike noise is removed by moving average filtering to obtain a stable heat source intensity curve; the highest and lowest temperatures on the battery surface are extracted based on edge values ​​from infrared thermal imaging data, and further interpolated and fitted with the matrix temperature distribution to construct a dynamically updated battery surface temperature field; the measured temperature from infrared thermal imaging is Gaussian denoising processed before entering the data platform to reduce the influence of sensor noise and environmental interference, and the standard deviation of battery temperature fluctuation is calculated using a sliding time window mechanism as a key indicator characterizing the current operational stability.

[0037] For cooling characteristic data, refrigerant flow rate is processed by time-series smoothing and variability analysis to eliminate instability caused by mechanical pulses and instantaneous disturbances, and to generate a continuous flow sequence. After aligning the refrigerant inlet and outlet temperatures on the timestamp, the instantaneous inlet and outlet temperature difference is calculated point by point, and the average inlet and outlet temperature difference of the refrigerant is statistically obtained over multiple periods to reflect the continuous heat exchange capacity and overall efficiency level of the cooling process.

[0038] Finally, the standardized battery feature data and cooling feature data are unified and normalized, aligned and encapsulated at the same time resolution to form a standardized thermal state dataset, providing a complete, reliable and continuously updated data foundation for subsequent three-dimensional heat flow assessment, deviation correction and adaptive allocation of cooling capacity.

[0039] In this implementation scheme, battery characteristic data and cooling characteristic data are transformed from their original acquisition state into a unified, standardized thermal state dataset. Through geometric constant registration, power curve smoothing, temperature field reconstruction, fluctuation feature extraction, flow rate time-series correction, and temperature difference statistical processing, the uncertainties caused by sampling noise and environmental disturbances are eliminated, while ensuring the alignment and consistency of various data in both time and physical dimensions. The resulting dataset provides an accurate, stable, and directly usable input foundation for three-dimensional heat flow assessment, model bias correction, and adaptive cooling allocation, thereby supporting high-precision simulation and dynamic control of the digital twin model.

[0040] Specifically, the steps for evaluating the overall heat flux density of the battery based on a standardized thermal state dataset are as follows:

[0041] Divide the volume of a single cell by the cell's heating power to obtain the volumetric heating intensity value; subtract the lowest temperature from the highest surface temperature of the cell and multiply it by the corresponding temperature weighting factor to obtain the surface temperature difference correction value; add one to the refrigerant flow rate, divide it by the refrigerant flow rate, and multiply it by the corresponding cooling weighting factor to obtain the cooling flow rate weighted value; add the volumetric heating intensity value, the surface temperature difference correction value, and the cooling flow rate weighted value to obtain the three-dimensional heat flow evaluation value.

[0042] The formula for calculating the three-dimensional heat flux assessment value is:

[0043]

[0044] In the formula: This represents the volume of a single cell, used to normalize the heat flux density, and is derived from battery geometric modeling. This represents the battery's heat generation power, used to characterize the instantaneous heat generated during charging and discharging, and is derived from calculations based on the DC bus voltage and current and the controller's power consumption. This indicates the highest temperature on the battery surface, used to characterize surface hotspots and gradients, and is derived from an infrared thermal imager. This indicates the lowest surface temperature of the battery, used to characterize surface hotspots and gradients, and is derived from an infrared thermal imager. This indicates the refrigerant flow rate, used to reflect the intensity of bottom convection, and is derived from the cooling circuit flow sensor. This represents a temperature weighting factor, ranging from 0.5 to 2, used to dynamically adjust the weight of the battery surface temperature difference in the three-dimensional heat flow assessment. It originates from the deviation between the temperature gradient characteristics of the current operating segment of the battery and the historical steady-state temperature distribution. Specifically, the calculation first extracts the difference between the highest and lowest temperatures within the current sliding time window from the infrared thermal imaging matrix, and calculates its rate of change and duration to characterize the dynamic evolution trend of the temperature difference. Then, it retrieves the temperature difference benchmark sequence of the battery during historical steady-state operation, including the long-term average temperature difference, median fluctuation amplitude, and high-temperature zone... Domain persistence indexes are used to form a temperature difference template. Then, the current temperature difference characteristics are compared with the benchmark template to obtain the amplitude difference rate, duration offset value, and matching degree of abnormal temperature rise segment. Combined with the temperature distribution stability rating under different load conditions, weighted aggregation is performed to finally generate a temperature weighting factor. When the battery surface temperature difference expands rapidly and deviates from the steady-state template, the temperature weighting factor increases, thereby enhancing the weight of the temperature difference term in the heat flow assessment. When the temperature difference changes slowly and is highly consistent with historical patterns, the temperature weighting factor decreases to avoid over-amplification of stable temperature distributions. The cooling weighting factor, ranging from 1 to 3, is used to adjust the contribution of refrigerant flow rate in the three-dimensional heat flux assessment. It originates from a comparative analysis between the current flow fluctuation and the stability of heat transfer capacity in the cooling loop. Specifically, the instantaneous fluctuation sequence of refrigerant flow rate within the current sliding time window is first extracted, and its average value, slope, and fluctuation frequency are calculated to characterize flow stability. Then, the steady-state flow characteristics of historical cooling loops are extracted, including rated flow rate, typical flow deviation rate, and cooling capacity retention, to establish a flow baseline model. Next, the current flow characteristics are aligned with the baseline model, including mean difference rate, fluctuation frequency matching degree, and heat transfer efficiency retention rate. Combined with the stability level of the cooling unit under different operating conditions, multi-dimensional weighted aggregation is performed to obtain the cooling weighting factor. When the refrigerant flow rate fluctuates drastically and differs significantly from the baseline, the cooling weighting factor increases, thereby enhancing the inhibitory regulation of the cooling flow rate term in the heat flux assessment. Conversely, when the flow rate is stable and the heat transfer efficiency is consistent with the historical pattern, the cooling weighting factor decreases to maintain the balanced contribution of the cooling intensity term in the overall assessment.

[0045] In this implementation scheme, the heat generation intensity, surface temperature difference characteristics, and cooling flow state during battery operation are comprehensively quantified to generate a three-dimensional heat flow evaluation value H, thereby providing a unified metric for battery heat flow distribution modeling and cooling capacity adjustment. By reflecting the normalized results of single-cell volume and instantaneous heat generation power as heat flow density per unit volume, and combining the surface temperature difference term corrected by the temperature weighting factor and the refrigerant flow correction term adjusted by the cooling weighting factor, the formula can simultaneously reflect the coupling effect of internal heat source intensity, surface temperature gradient, and external cooling capacity, forming a dynamic evaluation benchmark for the overall thermal state of the battery, providing core computational support for subsequent twin simulation correction and adaptive cooling capacity allocation.

[0046] Specifically, the steps for dynamically adjusting the cooling pump speed and refrigerant flow distribution based on the heat flow assessment results are as follows: Real-time comparison of the three-dimensional heat flow assessment value with the heat flow threshold, and implementation of graded control measures according to different ranges:

[0047] When the three-dimensional heat flux assessment value is less than the heat flux threshold, the cooling pump is maintained at the current low-speed operation state, the refrigerant maintains the default flow rate, and the cooling circuit is kept in the basic maintenance mode. At the same time, only the conventional heat flux distribution image is output in the battery twin model to record the basic evolution of the temperature field and archive it as control data, without triggering additional adjustment commands.

[0048] When the three-dimensional heat flux assessment value equals the heat flux threshold, the cooling pump operating speed is automatically adjusted to 25% to 40% of the rated speed, and the refrigerant flow rate is moderately increased to enhance the cooling coverage of potential hot spots. The hot spot identification function is activated in the battery twin model, highlighting the surface and local temperature rise trends in real time to form a visual warning area. At the same time, a cooling capacity allocation plan is generated in the background, including possible subsequent flow rate classification, pump speed adjustment and cooling capacity compensation strategies, so as to quickly switch when entering the next stage.

[0049] When the three-dimensional heat flux assessment value exceeds the heat flux threshold, the cooling pump is immediately switched to the highest speed, the refrigerant flow is rapidly increased to the equipment's allowable upper limit, and a high-load operation command is sent to the cooling capacity regulation device to drive the cooling unit into full-power operation. In the battery twin model, the three-dimensional heat distribution map is updated and rendered in real time, and hot spot migration trajectories and high-risk areas are automatically marked. The operation and maintenance alarm mechanism is triggered simultaneously, pushing the high heat load status to the operation and maintenance monitoring terminal, and recording key parameters for subsequent safety analysis and operation decisions.

[0050] In this implementation plan, a tiered cooling control mechanism is established by comparing the three-dimensional heat flux assessment value with the heat flux threshold in real time, enabling a step-by-step response from routine maintenance and advance planning to full-load emergency response. Specifically, this step can maintain a basic cooling mode when the battery heat load is low, avoiding unnecessary energy consumption; when the heat flux approaches the critical point, the cooling capacity is increased in advance and a cooling capacity allocation plan is generated, leaving a buffer time for subsequent regulation; when the heat flux exceeds the threshold, the highest intensity cooling and operation and maintenance alarm are quickly triggered, ensuring that risk hotspots are identified and dealt with in real time. This provides a key execution link for digital twin simulation and adaptive cooling capacity control, ensuring the safety and stability of energy storage battery operation.

[0051] Specifically, based on a standardized thermal state dataset, the accuracy analysis of the simulation is conducted by examining the difference between the measured and predicted temperatures from infrared thermal imaging. The specific steps are as follows: Input data is used as training data, including: the battery's three-dimensional geometry and individual cell volume, refined based on CAD, used to define the battery's spatial boundary conditions and thermal normalization parameters; refrigerant flow rate, refrigerant inlet temperature, and refrigerant outlet temperature, collected in real-time by sensors and used as input for cooling capacity constraints; and the battery surface temperature field obtained and extracted by infrared thermal imaging, used to construct dynamic verification data for the external temperature distribution. On the modeling side, transient conjugate heat transfer solutions are performed on the above multi-source inputs using a CFD platform to obtain a high-fidelity temperature field and heat flow distribution, including battery body heat conduction, coolant convection, and boundary heat transfer. Principal component analysis is then introduced to reduce the dimensionality of the high-dimensional thermal field data, extracting key feature variables to reduce computational overhead. Finally, the equivalent thermal RC network method is combined, i.e., a rapid thermal prediction model is established through a resistance-capacitance equivalent network, achieving a simplified expression of complex thermal processes. Based on this, an extended Kalman filter is used for online model calibration, fusing measured sensor data with model predictions to dynamically correct key parameters such as thermal conductivity and convection coefficient, thereby maintaining high consistency between the model and actual operating conditions. On the control side, the rapid prediction model is co-simulated with the Simulink platform. Through a short-time-domain rolling optimization strategy, the cooling demand curve that meets the temperature safety zone requirements is obtained under constraints, generating real-time adjustment signals. On the output side, the predicted temperature and predicted cooling flow rate are provided in real time as the control basis for the cooling pump speed adjustment and valve opening adjustment, forming a closed-loop mechanism of prediction-calibration-control. The temperature difference deviation correction value is obtained by dividing the absolute value of the difference between the measured temperature and the predicted temperature of the battery by one and the sum of the standard deviation of temperature fluctuation, then adding one and taking the natural logarithm. The cooling deviation correction value is obtained by dividing the predicted cooling flow rate plus one by the refrigerant flow rate minus the predicted cooling flow rate, then adding one. The model deviation correction value is obtained by multiplying the temperature difference deviation correction value and the cooling deviation correction value.

[0052] The formula for calculating the model bias correction value is:

[0053] ;

[0054] In the formula: This indicates the actual temperature measured by infrared thermal imaging, used to reflect hot spots on the battery surface, and is derived from the infrared thermal imager. This represents the model's predicted temperature, used to compare the differences between simulation and actual measurements, and is derived from the output of the digital twin model. This represents the standard deviation of temperature fluctuations, used to smooth out abnormal fluctuations, and is derived from sliding window calculations. This indicates the refrigerant flow rate, used to reflect the intensity of bottom convection, and is derived from the cooling circuit flow sensor. This represents the predicted cooling flow rate, used as a benchmark for deviation comparison, and is derived from the twin model output.

[0055] In this implementation plan, the degree of difference between the digital twin prediction results and the actual monitoring data is quantified, and the model parameters are dynamically adjusted accordingly. The formula normalizes the difference between the measured temperature from infrared thermal imaging and the model-predicted temperature, and then smooths out abnormal fluctuations by incorporating the standard deviation of temperature fluctuations, resulting in a stable temperature deviation factor. Simultaneously, the ratio of the measured refrigerant flow rate to the model-predicted cooling flow rate is introduced to reflect the deviation of cooling capacity between simulation and reality. Multiplying these two values ​​and taking a logarithmic compression yields the final correction value R. This correction value suppresses the amplification of deviations caused by sudden anomalies while ensuring that differences in cooling capacity are effectively mapped, thus providing a reliable basis for online updates, parameter calibration, and adaptive allocation of cooling capacity in the digital twin model.

[0056] Specifically, the steps for real-time correction of cooling boundary conditions based on accuracy analysis results are as follows:

[0057] When the calculated results of the model deviation correction values ​​show a continuous increasing trend for three consecutive times, it is determined that the difference between the digital twin prediction and the measured results is widening. At this time, the temperature matrix data acquired by infrared thermal imaging and the measured flow sequence obtained by the cooling circuit sensor are immediately synchronously transmitted back to the battery twin model, triggering the dynamic parameter calibration process to correct the thermal conductivity, heat transfer boundary conditions and local cooling efficiency factor online. At the same time, a new set of boundary conditions is generated and written into the simulation kernel, and the cooling capacity adjustment device automatically enters the early warning state, presets the acceleration commands of the cooling pump, fan and valve group to ensure a rapid response capability to deal with sudden thermal risks.

[0058] When the fluctuation range of the calculated model deviation correction value is less than the tolerance threshold for three consecutive times, it is determined that the twin prediction and the actual measurement are within the allowable error range. The model parameters and the collected data are kept updated in real time and the cooling pump speed is adjusted to 25% to 40% of the rated value to ensure that the flow rate is in a stable operating range. At the same time, the area with slight deviation is automatically marked in the three-dimensional thermal field of the battery twin model, and monitoring prompt information is generated and pushed to the operation and maintenance end to continuously track the potential evolution trend of local hot spots.

[0059] When the calculated results of the model deviation correction values ​​continue to decrease for three consecutive times, it is determined that the fit between the twin prediction and the actual measurement is gradually improving and the deviation is gradually converging. Therefore, the current parameters of the battery twin model are kept unchanged, and no additional boundary condition correction is performed. The cooling pump speed is automatically reduced to low speed, the refrigerant flow is reduced, the cooling circuit enters energy-saving mode, and only low-frequency sampling and parameter recording are retained in the background to reduce energy consumption and extend the service life of the equipment.

[0060] In this implementation plan, the consistency between the digital twin model and the measured data is dynamically verified by continuously monitoring the changing trend of the model deviation correction value, and differentiated cooling control strategies are triggered accordingly. When the deviation continues to expand, the measured data can be fed back in a timely manner and the model parameters can be corrected; when the deviation is within the tolerance range, the twin model is kept updated stably and the cooling pump operates at a medium speed, and the deviation area is marked in the model for operation and maintenance monitoring; when the deviation gradually converges, the model parameters are kept unchanged and the cooling pump speed is reduced, so that the cooling loop enters the energy-saving mode. Through this logic, closed-loop correction between prediction and measurement, graded execution of cooling capacity adjustment, and dynamic balance between energy consumption and safety are achieved.

[0061] Specifically, using the heat flow assessment results and accuracy analysis results as input, the specific steps for demand analysis of cooling demand intensity are as follows: multiply the three-dimensional heat flow assessment value by the model deviation correction value plus one, and then take the square root to obtain the heat flow deviation coupling value; add one to the average inlet and outlet temperature difference of the refrigerant and divide it by the current refrigerant inlet temperature minus the refrigerant outlet temperature to obtain the cooling temperature difference ratio; add the heat flow deviation coupling value to the cooling temperature difference ratio to obtain the adaptive allocation value of cooling capacity.

[0062] The formula for calculating the adaptive allocation value of cooling capacity is:

[0063] ;

[0064] In the formula: This represents a three-dimensional heat flow assessment value, used to characterize the intensity of heat distribution inside the battery; This represents the model bias correction value, used to quantify the degree of difference between simulation and actual measurement. This represents the average inlet and outlet temperature difference, used to construct an adaptive reference, and is derived from sliding window statistics. It indicates the refrigerant inlet temperature, used to characterize the instantaneous heat absorption capacity of the cooler, and is derived from the temperature sensor in the cooling circuit. It indicates the refrigerant outlet temperature, used to characterize the instantaneous heat absorption capacity of the cooler, and is derived from the cooling circuit temperature sensor.

[0065] In this implementation example, the three-dimensional heat flux evaluation value of Example 1 is set to 0.80, the model deviation correction value is set to 0.20, the average inlet and outlet temperature difference is set to 2.0, the refrigerant inlet temperature is set to 22.0, and the refrigerant outlet temperature is set to 27.0.

[0066] In Example 2, the three-dimensional heat flux evaluation value was set to 1.10, the model bias correction value was set to 0.35, the average inlet and outlet temperature difference was set to 3.5, the refrigerant inlet temperature was set to 24.0, and the refrigerant outlet temperature was set to 29.5.

[0067] In Example 3, the three-dimensional heat flux evaluation value was set to 0.95, the model deviation correction value was set to 0.15, the average inlet and outlet temperature difference was set to 1.8, the refrigerant inlet temperature was set to 21.5°C, and the refrigerant outlet temperature was set to 26.0°C.

[0068] In Example 4, the three-dimensional heat flux evaluation value was set to 1.60, the model bias correction value was set to 0.40, the average inlet and outlet temperature difference was set to 4.2, the refrigerant inlet temperature was set to 25.0, and the refrigerant outlet temperature was set to 30.0.

[0069] In Example 5, the three-dimensional heat flux evaluation value was set to 1.30, the model bias correction value was set to 0.10, the average inlet and outlet temperature difference was set to 2.7, the refrigerant inlet temperature was set to 23.5, and the refrigerant outlet temperature was set to 28.0.

[0070] In Example 6, the three-dimensional heat flux evaluation value was set to 2.10, the model bias correction value was set to 0.50, the average inlet and outlet temperature difference was set to 5.0, the refrigerant inlet temperature was set to 26.0, and the refrigerant outlet temperature was set to 31.0.

[0071] In Example 7, the three-dimensional heat flux evaluation value was set to 1.75, the model bias correction value was set to 0.25, the average inlet and outlet temperature difference was set to 3.2, the refrigerant inlet temperature was set to 23.0°C, and the refrigerant outlet temperature was set to 29.0°C. The adaptive cooling capacity allocation values ​​for each example were calculated, as shown in Table 1.

[0072] Table 1. Cooling Capacity Adaptive Allocation Values ​​Data Table

[0073]

[0074] like Figure 2As shown, this is a line graph of the adaptive cooling capacity allocation value provided in this application example. (See Table 1 and...) Figure 2 As can be seen, Instance 6 has the highest adaptive cooling capacity allocation value, indicating that the three-dimensional heat flux assessment value and model deviation correction value are both high in this scenario, and the average inlet and outlet temperature difference is also the largest. This reflects the high heat load intensity, obvious model fitting error, and urgent cooling demand, requiring more cooling capacity to be allocated to cope with transient thermal risks and achieve rapid recovery of temperature control stability. Instance 3 has the lowest adaptive cooling capacity allocation value. Although its model deviation correction value is small and the actual cooling temperature difference is also low, it indicates that the thermal state changes are stable and the energy consumption pressure is small under this condition. It is suitable to adopt a conservative cooling capacity allocation strategy to maintain energy efficiency balance and avoid resource waste. The line graph of the adaptive cooling capacity allocation value clearly shows the cooling resource response intensity of each instance under typical operating conditions. The higher the value, the more actively cooling is needed, which is suitable for the formulation and execution priority ranking of the twin-driven dynamic thermal management strategy.

[0075] Specifically, the steps for generating graded adjustment instructions for refrigerant flow and pump speed based on the demand analysis results are as follows: real-time comparison of the adaptive allocation value of cooling capacity and the cooling capacity allocation threshold. The cooling capacity allocation threshold includes a first cooling capacity threshold and a second cooling capacity threshold. The first cooling capacity threshold is higher than the second cooling capacity threshold, which is used to realize graded response and resource allocation optimization in dynamic cooling control.

[0076] When the adaptive cooling capacity allocation value is less than or equal to the second cooling capacity threshold, the current operating condition is identified as being in a low heat load range. The cooling pump is kept running at the lowest speed, and the cooling capacity adjustment device is controlled to maintain the current output power unchanged. At this time, in order to reduce the computational load, the infrared thermal imaging image enhancement function is automatically turned off, and only the original measured values ​​of battery characteristic data are collected and stored. The simulation module suspends the execution of the incremental calculation process, does not refresh the three-dimensional thermal distribution map in real time, and only records the temperature sequence of the central node by appending to it for subsequent trend tracking and baseline comparison, so as to avoid the waste of redundant computing resources.

[0077] When the adaptive cooling capacity allocation value is greater than the second cooling capacity threshold and less than or equal to the first cooling capacity threshold, it is determined that the current heat load is in a medium fluctuation range. The cooling pump speed is automatically increased to medium speed operation to ensure that the cooling flow rate responds within the safety zone. At the same time, the image analysis algorithm is activated to reconstruct the boundaries and extract the area of ​​hot spots in the infrared image, and to extract the features of the high-temperature core area. The simulation module generates a simple heat map panel, marks the key areas of concern, and displays the visualization results synchronously on the control interface for easy manual inspection and strategy fine-tuning. The spatial coverage information of the current cooling capacity distribution range is updated synchronously and written to the cooling capacity control log for retrieval.

[0078] When the adaptive cooling capacity allocation value exceeds the first cooling capacity threshold, the current operating condition is determined to be in a high-heat-load, high-risk state. The cooling pump is immediately switched to its highest speed to achieve extreme cooling capacity response. Simultaneously, the simulation module is triggered to enter an enhanced operation mode, continuously launching the dynamic rendering engine of the three-dimensional thermal distribution map to globally visualize and track changes in the temperature field of the battery surface and internal environment, and constructing hotspot migration trajectories in real time. Historical data and strategy models of the full-cycle charge-discharge task are loaded, and a short-term prediction sequence is constructed based on a third-order time step to simulate the evolution trend of the heat load in advance. The simulation path executes a complete reconstruction strategy, synchronously introducing real-time collected battery characteristic data and cooling characteristic data for joint calculation, and forming a high-frequency data snapshot of all key parameters at that moment to facilitate post-fault cause investigation and source analysis.

[0079] In this implementation plan, a hierarchical response mechanism is constructed based on the real-time comparison between the adaptive cooling capacity allocation value and the cooling capacity allocation threshold. This enables refined control of cooling resource scheduling and dynamic linkage adjustment of the simulation model. The core objective is to flexibly adjust the cooling strategy and simulation calculation process under different heat load conditions, thereby improving overall operating efficiency and computing resource utilization. When the cooling demand is low, this step effectively suppresses redundant resource calls and records only key node data to maintain basic monitoring. Under medium heat load conditions, the cooling capacity and image analysis intensity are appropriately enhanced, while ensuring that the visualization of thermal data and the control interface are updated synchronously. Under high heat load scenarios, the maximum cooling capacity is rapidly released, and the simulation engine, control platform, and data recording module are linked to comprehensively enhance the response accuracy and predictive control capabilities for high-risk operating conditions, thereby constructing an intelligent cooling control closed loop that combines energy saving, real-time performance, and safety.

[0080] Specifically, the steps for generating and verifying cooling capacity allocation commands within each cooling control cycle, issuing and executing them, and combining real-time deviation feedback and short-term prediction for preventative adjustments are as follows:

[0081] Within each cooling control cycle, based on the current battery thermal state data, the three-dimensional heat flow assessment value and model deviation correction value are extracted in real time, and a set of cooling capacity allocation instructions are generated accordingly. These instructions explicitly define the target speed of the cooling pump and the dynamic flow rate allocation of the refrigerant to achieve precise delivery of cooling resources. Before the instruction is issued, a shadow simulation process is simultaneously initiated within the battery twin model to perform feedforward prediction of the heat diffusion process under the current instruction conditions. If the temperature at any node during the simulation is equal to or higher than the safety threshold, the current cooling capacity allocation instruction is automatically rolled back, and relevant parameters are readjusted for verification simulation until the predicted temperature is completely lower than the safety threshold. Then, the cooling capacity allocation instruction is officially issued to the execution component to ensure that the control strategy has sufficient safety margin.

[0082] During the execution of the cooling capacity allocation command, the inlet and outlet temperatures of the refrigerant in the cooling loop are continuously collected, and the average temperature difference is dynamically calculated. The actual heat exchange capacity is compared and analyzed with the temperature prediction results of the battery twin model, and the deviation information is extracted as a feedback parameter to update the thermal response state of the twin model in real time, maintaining the model's accurate mapping of the current cooling effect. Before the end of the cooling control cycle, a short-term trend prediction is performed based on the adaptive cooling capacity allocation value. If a significant heat accumulation trend is found in a local hot spot area, the cooling pump speed will be increased in advance to interrupt the hot spot development path. The control execution value and the actual reading are recorded together as baseline data for subsequent cycle modeling and control strategy iteration, thereby gradually building an adaptive and evolving cooling control mechanism.

[0083] like Figure 3 The diagram illustrates the signal principle involved in this invention. Starting from an IoT node, front-end sensing devices collect thermal state and environmental data in real time. This data is then aggregated by the data acquisition module and transmitted to the data center for unified processing and management. In the data center, infrared thermal imaging data flows into the modeling module to construct a three-dimensional thermal behavior simulation model of the battery and heat-sensitive components. After modeling, the data enters the simulation module to predict heat distribution and analyze trends in the current operating state. The simulation results are sent to the decision-making module to generate control commands for cooling intensity adjustment, fluid distribution optimization, and risk response. Simultaneously, the decision results are also fed back to the regulation module to dynamically correct operating parameters in the data center, such as cooling pump speed and valve opening, achieving adaptive optimization of cooling capacity regulation and thermal control. The entire process demonstrates the real-time coupling relationship between modeling, simulation, and regulation, supporting continuous monitoring, rapid evaluation, and precise intervention during operation. It is an important digital twin support structure for battery safety and heat dissipation efficiency optimization scenarios.

[0084] like Figure 4The diagram illustrates the signal generation flowchart of this invention, which begins with the acquisition of two types of data: physical structure data of the battery and dynamic state data of the battery during charging and discharging, including electrical signal information such as voltage, current, and power. Based on this data, a three-dimensional model of the battery is first constructed to realistically reproduce its spatial structural characteristics. Subsequently, the charging and discharging electrical signal data is input into the model for dynamic simulation, deduce the heat distribution and changing trends under different current conditions, and thus obtain the heat source evolution process during charging and discharging. On this basis, a three-dimensional thermodynamic diffusion model is established to further map the aforementioned heat source data into heat propagation paths and cooling demand distribution areas, providing quantitative support for subsequent cooling control strategies. When electrical signals indicating the start of discharging and charging are detected, the thermal simulation process is automatically initiated. Combined with the current operating condition evaluation results, the required refrigerant flow rate is predicted in advance, and a cooling capacity generation signal is output to achieve pre-adjustment of the cooling device. This simulation prediction method based on the heat source generation and diffusion mechanism helps to achieve intelligent, feedforward control of battery thermal management and is a key step in building digital twin thermal management.

[0085] This implementation scheme constructs a closed-loop cooling control mechanism with feedforward verification, real-time feedback, and adaptive adjustment capabilities. Its core objective is to ensure the battery remains below a safe temperature threshold throughout the thermal management process and to dynamically respond to changes in heat flow for precise cooling allocation. Cooling allocation commands are generated through three-dimensional heat flow assessment and model deviation correction. Shadow simulation is performed in the battery twin model to preemptively identify potential overheating risks, preventing thermal runaway caused by unreasonable commands from the outset. During actual execution, the temperature difference between the refrigerant inlet and outlet is collected and compared with the prediction results, enabling continuous correction of the model's prediction accuracy and improving the reliability and accuracy of cooling control. Furthermore, short-term prediction of local hotspots based on adaptive cooling allocation values ​​at the end of the cycle demonstrates a proactive response to sudden heat accumulation, helping to intervene in potential risk points in advance and ensuring the stability and energy efficiency of the entire thermal management process. This step, as a key execution unit of the cooling strategy, provides strong technical support for achieving battery thermal safety and energy efficiency optimization.

[0086] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0087] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A digital twin simulation method for battery thermal flux and adaptive cooling control in energy storage power stations, characterized by: include: S1. Collect battery characteristic data and cooling characteristic data during battery operation, and preprocess the collected battery characteristic data and cooling characteristic data to construct a standardized thermal state dataset. S2, based on a standardized thermal state dataset, performs a thermal flux assessment on the overall heat flux density of the battery, and dynamically adjusts the cooling pump speed and refrigerant flow distribution based on the thermal flux assessment results; S3, based on a standardized thermal state dataset, performs accuracy analysis on the difference between the measured temperature and the predicted temperature from infrared thermal imaging, and corrects the cooling boundary conditions in real time based on the accuracy analysis results. S4 takes the heat flow assessment results and accuracy analysis results as input, performs demand analysis on the cooling demand intensity, and generates graded adjustment commands for refrigerant flow and pump speed based on the demand analysis results. S5 generates and verifies cooling capacity allocation instructions within each cooling control cycle, issues and executes them, and performs preventative adjustments by combining real-time deviation feedback and short-term prediction.

2. The method for digital twin simulation of battery thermal flux and adaptive control of cooling capacity in energy storage power stations according to claim 1, characterized in that: The specific steps for collecting battery characteristic data and cooling characteristic data during battery operation are as follows: Collect battery characteristic data during battery operation, including: single cell volume, battery heat generation power, highest battery surface temperature, lowest battery surface temperature, and infrared thermal imaging measured temperature. At the same time, calculate and record the standard deviation of battery temperature fluctuation. Collect cooling characteristic data, including refrigerant flow rate, refrigerant inlet temperature and refrigerant outlet temperature, and calculate and record the average inlet and outlet temperature difference of the refrigerant.

3. The method for digital twin simulation of battery thermal flux and adaptive control of cooling capacity in energy storage power stations according to claim 1, characterized in that: The specific steps for preprocessing the collected battery feature data and cooling feature data to construct a standardized thermal state dataset are as follows: For battery characteristic data, the volume of a single cell is recorded as a geometric constant in the database, and the battery heat generation power is cross-validated and subjected to moving average filtering. The highest and lowest temperatures on the battery surface are extracted based on the measured temperature from infrared thermal imaging to extract edge values ​​and construct the battery temperature field. The measured temperature from infrared thermal imaging is processed by Gaussian noise reduction, and the standard deviation of the battery temperature fluctuation is calculated according to the sliding window mechanism. For cooling characteristic data, time-series smoothing and variability analysis are performed on refrigerant flow to remove pulse interference. After aligning the refrigerant inlet and outlet temperatures, the instantaneous temperature difference is calculated, and the average inlet and outlet temperature difference of the refrigerant is statistically analyzed. The standardized battery feature data and cooling feature data are normalized to construct a standardized thermal state dataset.

4. The method for digital twin simulation of battery thermal flux and adaptive control of cooling capacity in energy storage power stations according to claim 1, characterized in that: The specific steps for evaluating the overall heat flux density of the battery based on the standardized thermal state dataset are as follows: Divide the volume of a single battery cell by the battery's heating power to obtain the volumetric heating intensity value; The surface temperature difference correction value is obtained by subtracting the lowest surface temperature from the highest surface temperature of the battery and then multiplying it by the corresponding temperature weighting factor. Add one to the refrigerant flow rate, divide by the refrigerant flow rate, and multiply by the corresponding cooling weighting factor to obtain the cooling flow rate weighted value; The volumetric heat intensity value, the surface temperature difference correction value, and the cooling flow rate weighted value are added together to obtain the three-dimensional heat flow evaluation value.

5. The method for digital twin simulation of battery thermal flux and adaptive control of cooling capacity in an energy storage power station according to claim 1, characterized in that: The specific steps for dynamically adjusting the cooling pump speed and refrigerant flow distribution based on the heat flow assessment results are as follows: Real-time comparison of 3D heat flux assessment values ​​with heat flux thresholds: When the three-dimensional heat flux assessment value is less than the heat flux threshold, the cooling pump operating speed is kept constant, the refrigerant maintains the default flow rate, and only the normal heat flux distribution is output in the battery twin model; When the heat flow assessment value equals the heat flow threshold, adjust the cooling pump to 25% to 40% of the rated value to increase the refrigerant flow rate, highlight potential hot spots in the battery twin model in real time, and generate a cooling capacity allocation plan in advance. When the three-dimensional heat flow assessment value exceeds the heat flow threshold, the cooling pump is adjusted to the highest speed, the refrigerant flow is increased to the maximum, and a high-load operation command is immediately sent to the cooling capacity regulation device. At the same time, the three-dimensional heat distribution in the battery twin model is updated in real time and high-risk areas are marked, triggering an operation and maintenance alarm.

6. The method for digital twin simulation of battery thermal flux and adaptive control of cooling capacity in an energy storage power station according to claim 1, characterized in that: The specific steps for analyzing the simulation accuracy based on the difference between the measured and predicted temperatures from infrared thermal imaging, using a standardized thermal state dataset, are as follows: Input access data is used as training data. The access data includes: the three-dimensional geometry of the battery and the volume of a single cell based on CAD refinement, refrigerant flow rate, refrigerant inlet temperature and outlet temperature, and battery temperature field extracted by infrared thermal imaging. On the modeling side, transient conjugate heat transfer is solved on the access data in CFD to obtain high-fidelity temperature and heat flow distribution. Then, a fast prediction model is constructed using principal component dimensionality reduction and thermal RC network equivalence method, and online calibration is performed using extended Kalman filter. The control side combines the fast prediction model with Simulink and uses short time-domain rolling optimization to solve the problem under constraints to meet the cooling requirements of the seat belt. The predicted temperature and predicted cooling flow are output in real time on the output side. The temperature deviation correction value is obtained by dividing the absolute value of the difference between the measured temperature and the predicted temperature of the battery by one and the sum of the standard deviation of temperature fluctuation, and then adding one and taking the natural logarithm. Divide the value of predicted cooling flow plus one by the value of refrigerant flow minus predicted cooling flow, and then add one to obtain the cooling deviation correction value; Multiply the temperature difference deviation correction value by the cooling deviation correction value to obtain the model deviation correction value.

7. The method for digital twin simulation of battery thermal flux and adaptive control of cooling capacity in energy storage power stations according to claim 1, characterized in that: The specific steps for real-time correction of cooling boundary conditions based on accuracy analysis results are as follows: When the calculated results of the model deviation correction value continue to increase for three consecutive times, it is determined that the deviation between the model prediction and the actual measurement has widened. The measured temperature and flow rate data of infrared thermal imaging are fed back to the battery twin model to dynamically correct the thermal conductivity and boundary conditions, and trigger the cooling capacity adjustment device to enter the early warning stage. When the fluctuation of the calculated results of the model deviation correction value is less than the tolerance threshold for three consecutive times, the battery twin model access data is continuously updated, and the cooling pump speed is adjusted to 25% to 40% of the rated value to keep the cooling circuit flow stable. At the same time, the deviation area is marked in the battery twin model and monitoring is prompted. When the calculated results of the model deviation correction value continue to decrease for three consecutive times, the battery twin model parameters are kept unchanged, the cooling pump speed is reduced, and the cooling circuit enters the energy-saving mode.

8. The method for digital twin simulation of battery thermal flux and adaptive control of cooling capacity in energy storage power stations according to claim 1, characterized in that: The specific steps for conducting a demand analysis of cooling demand intensity, using heat flow assessment results and accuracy analysis results as input, are as follows: Multiply the 3D heat flow evaluation value by the model deviation correction value plus one, and then take the square root to obtain the heat flow deviation coupling value. Add one to the average inlet and outlet temperature difference of the refrigerant and divide by the value of the current refrigerant inlet temperature minus the refrigerant outlet temperature to obtain the cooling temperature difference ratio; The adaptive allocation value of cooling capacity is obtained by adding the heat flow deviation coupling value to the cooling temperature difference ratio.

9. The method for digital twin simulation of battery thermal flux and adaptive control of cooling capacity in an energy storage power station according to claim 1, characterized in that: The specific steps for generating graded adjustment commands for refrigerant flow and pump speed based on the demand analysis results are as follows: Real-time comparison of the adaptive cooling capacity allocation value with the cooling capacity allocation threshold. The cooling capacity allocation threshold includes a first cooling capacity threshold and a second cooling capacity threshold, with the first cooling capacity threshold being higher than the second cooling capacity threshold. When the adaptive allocation value of cooling capacity is less than or equal to the second cooling capacity threshold, the cooling pump is kept at the lowest speed, the output of the cooling capacity adjustment device is kept unchanged, the thermal imaging image enhancement is turned off, only the original value of the battery characteristic data is recorded, the simulation incremental calculation process is turned off, the three-dimensional thermal distribution map is not refreshed, and only the temperature sequence of the central node is appended and written. When the adaptive cooling capacity allocation value is greater than the second cooling capacity threshold and less than or equal to the first cooling capacity threshold, the cooling pump speed is increased, the image analysis algorithm is started to reconstruct the boundaries of the hot spot area, a simplified heat map panel is generated and synchronized to the control interface, and the current cooling capacity delivery range information is updated. When the adaptive cooling capacity allocation value is greater than the first cooling capacity threshold, the cooling pump is immediately switched to the highest speed, and the dynamic rendering of the three-dimensional thermal distribution map and hot spot migration trajectory tracking are continuously started. The full-cycle charge and discharge task data is loaded and a third-order time step prediction sequence is generated. The simulation engine starts the complete reconstruction path and synchronizes the real-time battery feature data and cooling feature data, while recording the current battery feature data and cooling feature data as a high-frequency data snapshot.

10. The method for digital twin simulation of battery thermal flux and adaptive control of cooling capacity in an energy storage power station according to claim 1, characterized in that: The specific steps for generating and verifying the cooling capacity allocation command in each cooling control cycle, issuing and executing it, and combining real-time deviation feedback and short-term prediction for preventative adjustment are as follows: Within each cooling control cycle, a cooling capacity distribution command is generated based on the three-dimensional heat flow assessment value and the model deviation correction value. The command covers the cooling pump speed and refrigerant flow rate. A shadow simulation is performed in the battery twin model. When the predicted temperature is equal to or greater than the safety threshold, the cooling capacity allocation command is rolled back and re-verified until the predicted temperature is less than the safety threshold before being sent to the execution component. During the execution phase of the cooling capacity distribution command, the average inlet and outlet temperature difference of the refrigerant is continuously collected and compared with the predicted temperature. At the same time, the deviation information between the average inlet and outlet temperature difference of the refrigerant and the twin predicted temperature is fed back to the battery twin model. Before the end of the cooling control cycle, short-term predictions are made based on the adaptive allocation value of cooling capacity. When a local hot spot trend appears, the cooling pump speed is increased in advance, and the execution and readback results are recorded as baseline data.

Images