A fully immersed liquid-cooled energy storage system and control method thereof

Abstract

The present invention discloses a fully immersed liquid-cooled energy storage system and a control method thereof, which relates to the field of energy storage control technology, including installing multiple sensors in a liquid immersion tank and calibrating them, using the calibrated multiple sensors to collect status data of battery modules and cooling liquid in real time, preprocessing the status data of the battery modules and cooling liquid, uploading it to a control center, constructing a Transformer model in the control center, and using historical battery module and cooling liquid status data for training to obtain a trained Transformer model, inputting the preprocessed battery module and cooling liquid status data into the trained Transformer model, and obtaining prediction results of the location of high-risk areas and expected temperature change values. The present invention proposes an intelligent control strategy and multi-sensor network based on the Transformer model, which realizes local cooling by real-time monitoring of temperature changes of liquid-cooled energy storage equipment and real-time adjustment of the flow rate and path of the coolant.

Description

Technical Field

[0001] The present invention relates to the field of energy storage control technology, and in particular to a fully immersed liquid-cooled energy storage system and a control method thereof. Background Art

[0002] As an emerging and efficient cooling solution, fully immersed liquid-cooled energy storage has been widely used in recent years. Its working principle is to completely immerse the battery module in coolant and use the high thermal conductivity of the liquid to effectively remove heat, thereby keeping the battery's operating temperature within a safe range. However, its existing technology still has some shortcomings.

[0003] On the one hand, traditional fully submerged liquid-cooled energy storage control strategies are relatively simple and lack adaptive capabilities, making it difficult to precisely adjust the coolant flow path and flow rate based on real-time changing load conditions. On the other hand, existing data collection techniques are incomplete. Traditional data collection methods can only obtain limited status information and lack comprehensive monitoring of the temperature distribution within the entire battery module. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a fully immersed liquid-cooled energy storage system and a control method thereof to solve the problems of insufficient intelligence of cooling control strategies and incomplete data collection.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a fully immersed liquid-cooled energy storage control method, which comprises:

[0008] Install multiple sensors in a liquid immersion tank and perform calibration;

[0009] Utilize calibrated multi-sensors to collect real-time status data of battery modules and cooling liquid;

[0010] Pre-process the status data of the battery module and cooling liquid and upload it to the control center;

[0011] Build a Transformer model in the control center and use historical battery module and cooling liquid status data for training to obtain the trained Transformer model;

[0012] The pre-processed battery module and cooling liquid status data are fed into the trained Transformer model to obtain predictions of high-risk area locations and expected temperature changes.

[0013] Based on the predicted results of the location of high-risk areas and expected temperature change values, the control center automatically adjusts the working status of the micropump and microvalve, changes the flow path and flow rate of the cooling liquid, and achieves local cooling.

[0014] As a preferred solution of the fully immersed liquid-cooled energy storage control method of the present invention, the steps of installing multiple sensors in the liquid immersion tank and calibrating the sensors specifically include the following steps:

[0015] Place the temperature sensor in a constant temperature bath, measure the reading, and compare it with the temperature in the constant temperature bath. Use linear regression to calibrate the temperature sensor.

[0016] Place the current sensor under a standard current source with known current, measure the reading, and compare it with the known current value. Use the least squares method to calibrate the current sensor.

[0017] Place the flow sensor in a standard fluid with a known flow rate, measure the reading, and compare it with the known flow rate value. Use linear interpolation to calibrate the flow sensor.

[0018] The standard current source is defined based on a known current value, and the standard fluid is defined based on a known flow rate value and constant temperature and pressure conditions.

[0019] As a preferred solution of the fully immersed liquid-cooled energy storage control method of the present invention, wherein: the use of calibrated multi-sensors to collect status data of the battery module and the cooling liquid in real time specifically includes the following steps:

[0020] Use the calibrated temperature sensors to collect temperature data of the battery module and cooling liquid by arranging embedded temperature monitoring points and performing timed sampling.

[0021] The calibrated current sensor is used to collect the charging status data of the battery module using the coulomb counting method combined with temperature compensation.

[0022] The calibrated flow rate sensor is installed on the coolant flow path to collect the flow rate data of the cooling liquid.

[0023] As a preferred solution of the fully immersed liquid-cooled energy storage control method of the present invention, wherein: the preprocessing includes outlier processing, standardization processing and difference processing;

[0024] The pre-processed status data is formatted and encrypted using TLS. A secure connection is established with the control center via the HTTP protocol to upload the status data.

[0025] As a preferred solution of the fully immersed liquid-cooled energy storage control method of the present invention, wherein: the Transformer model is constructed in the control center and trained using the historical battery module and cooling liquid status data to obtain the trained Transformer model, specifically including the following steps:

[0026] Select the TensorFlow deep learning framework in the control center and build a Transformer model using a multi-head self-attention mechanism, a feedforward neural network, residual connections, and layer normalization;

[0027] Input historical battery module and cooling liquid status data into the Transformer model for training;

[0028] During the training process, batch gradient descent is used for iterative training, and L2 regularization and early stopping are used to prevent overfitting, forward propagation is used to calculate prediction values ​​and losses, back propagation is used to calculate gradients, and the Adam optimizer is used to update model parameters.

[0029] As a preferred solution of the fully immersed liquid-cooled energy storage control method of the present invention, the method of inputting the pre-processed battery module and cooling liquid state data into a trained Transformer model to obtain the prediction results of the high-risk area location and expected temperature change value specifically includes the following steps:

[0030] Converting the pre-processed state data of the pool module and the cooling liquid into a three-dimensional tensor structure;

[0031] The three-dimensional tensor structure includes the number of samples, the number of time steps and the number of feature dimensions;

[0032] The three-dimensional tensor is used with a multi-head self-attention mechanism to calculate the attention weights and perform weighted summation to obtain the locations of high-risk areas and expected temperature changes.

[0033] As a preferred solution of the fully immersed liquid-cooled energy storage control method of the present invention, the control center automatically adjusts the working state of the micropump and microvalve based on the prediction results of the high-risk area location and the expected temperature change value, changes the flow path and flow rate of the cooling liquid, and realizes local cooling, which specifically includes the following steps:

[0034] Based on the acquired high-risk area location information, the battery module locations that need priority cooling are automatically identified, and the working status of the micropump and microvalve are adjusted based on the expected temperature change value;

[0035] By adjusting the output power of the micropump, pulse width modulation and PID control algorithm are used to control the flow of coolant to high-risk areas;

[0036] By controlling the opening of the microvalve, using stepper motor control and closed-loop feedback, the flow path of the coolant is changed, and the coolant is covered to the position of the battery module that needs cooling, thereby achieving local cooling.

[0037] In a second aspect, the present invention provides a fully immersed liquid-cooled energy storage control system, including a calibration module, a data acquisition module, a preprocessing module, a model training module, a result prediction module, and an adjustment module;

[0038] Calibration module for mounting and calibrating multiple sensors in a liquid immersion tank;

[0039] A data acquisition module is used to collect status data of the battery module and cooling liquid in real time using calibrated multi-sensors;

[0040] A pre-processing module is used to pre-process the status data of the battery module and the cooling liquid and upload it to the control center;

[0041] The model training module is used to build a Transformer model in the control center and use historical battery module and cooling liquid status data for training to obtain a trained Transformer model;

[0042] The result prediction module is used to input the pre-processed battery module and cooling liquid status data into the trained Transformer model to obtain the prediction results of the high-risk area location and expected temperature change value;

[0043] The adjustment module is used to input the pre-processed battery module and cooling liquid status data into the trained Transformer model to obtain the prediction results of the high-risk area location and expected temperature change value.

[0044] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the fully immersed liquid-cooled energy storage control method as described in the first aspect of the present invention is implemented.

[0045] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the fully immersed liquid-cooled energy storage control method as described in the first aspect of the present invention is implemented.

[0046] The beneficial effects of this invention include: proposing an intelligent control strategy based on the Transformer model, predicting future temperature changes based on real-time monitoring data and automatically adjusting the coolant flow path and flow rate, significantly improving dynamic response capabilities and adaptability, ensuring precise temperature control even under complex operating conditions, and improving battery safety and service life. Utilizing a multi-sensor network, comprehensive monitoring of the temperature distribution within the entire battery module is achieved, and data preprocessing technology is used to improve data quality. This not only provides richer status information, but also enhances the ability to identify potential high-temperature risk areas, thereby effectively preventing failures. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0048] Figure 1 This is a flow chart of the fully immersed liquid-cooled energy storage control method in Example 1.

[0049] Figure 2 Schematic diagram of the fully immersed liquid-cooled energy storage system in Example 1. DETAILED DESCRIPTION

[0050] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0051] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0052] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0053] Example 1, with reference to Figure 1 and Figure 2 , which is the first embodiment of the present invention, provides a fully immersed liquid-cooled energy storage system and a control method thereof, comprising the following steps:

[0054] S1. Install multiple sensors in a liquid immersion tank and calibrate them.

[0055] The specific operations include the following:

[0056] When calibrating a temperature sensor, select a high-precision constant temperature bath as a reference. Place the temperature sensor to be calibrated in the bath and set different temperature points. At each temperature point, record the sensor reading and compare it with the actual temperature value of the bath. Use linear regression analysis to determine the relationship between the sensor output and the actual temperature. Then, find the best-fitting line by minimizing the sum of squared errors. Based on the slope and intercept of the line, adjust the sensor output for calibration.

[0057] When calibrating a current sensor, use a standard current source with a known current as a reference device. Connect the current sensor to the standard current source, record the sensor reading at each set current value, compare it with the known current value, and use the least squares method to analyze it to obtain the best fit relationship between the sensor output and the actual current value. Based on this fit, adjust the sensor's gain and offset for calibration.

[0058] When calibrating a flow sensor, use a standard fluid device, place the flow sensor in it, and test it under different preset flow conditions. Select multiple flow points, record the sensor readings, compare them with the known flow values, and use linear interpolation to adjust the sensor output for calibration.

[0059] The above standard current sources are defined based on known current values. They are devices used to provide stable current output and can accurately generate the required current intensity according to the set value. The standard fluids are defined based on known flow rate values ​​and constant temperature and pressure conditions. They are designed for calibrating flow rate sensors and are intended to eliminate interference from external factors in the calibration process.

[0060] S2. Use calibrated multi-sensors to collect status data of the battery module and cooling liquid in real time.

[0061] The specific operations include the following:

[0062] When collecting temperature data for battery modules and cooling liquids, an embedded design is used, with calibrated temperature sensors placed inside the battery modules and along the cooling liquid flow path, to directly monitor local temperature changes. A timed sampling method is also used to ensure data continuity and accuracy. For example, a temperature reading can be set every 5 seconds or 1 minute. This method not only captures instantaneous temperature changes but also analyzes their trends over time, providing accurate data support for subsequent thermal management.

[0063] When collecting battery module state of charge data, the coulomb counting method is used to estimate the remaining capacity of the battery. The coulomb counting method calculates the state of charge by measuring the total charge flowing through the battery. In specific operation, the current sensor continuously monitors the current flowing into and out of the battery and accumulates this data to calculate the battery module state of charge. At the same time, the charge state estimate is adjusted by combining temperature compensation technology, and the current measurement value is adjusted according to the operating temperature of the battery, thereby improving the accuracy of the collected state of charge data.

[0064] When collecting the flow rate data of the cooling liquid, the flow rate sensor is installed on the coolant flow path, usually at the pump outlet and the main branch point of the cooling circuit. Commonly used flow rate sensors include ultrasonic flow meters and turbine flow meters, which can provide high-precision flow rate measurement. At the same time, in order to ensure the accuracy of the data, a pulse output mode can be used, that is, a pulse signal is generated each time the fluid passes through the sensor, and the flow rate is calculated by counting the number of pulses. In addition, the data from the temperature sensor can be combined to comprehensively analyze the operating status of the liquid-cooled energy storage to ensure the accuracy of the collected flow rate data.

[0065] S3. Pre-process the status data of the battery module and cooling liquid and upload them to the control center.

[0066] The specific steps include:

[0067] First, data denoising and outlier removal are performed. In the specific operation, a low-pass filter is used to remove high-frequency noise. The low-pass filter allows frequencies below the set filter threshold to pass through, while attenuating frequency components above the filter threshold. This can be achieved using first-order and second-order infinite impulse response filters. Then, the Z-Score method is used to detect and remove outliers. The Z-Score calculates the standard score of each data point, that is, the difference between the point and the mean divided by the standard deviation. If the absolute value of the Z-Score of a data point exceeds the outlier removal threshold, it is considered an outlier and is removed, thereby ensuring that the data quality of subsequent analysis is not affected by outliers;

[0068] Data normalization is then performed, using the Min-Max normalization algorithm to map the collected data features to the same scale. Min-Max normalization uses a linear transformation to scale the data to a specific interval, usually [0, 1]. This not only makes data with different features comparable, but also facilitates the training and optimization of subsequent machine learning models.

[0069] Next, we perform time synchronization and interpolation. We synchronize the collected data based on the timestamps and use interpolation to fill in missing time point data. Since sensors may operate at different sampling frequencies, we need to ensure that the time axis of all data is consistent. By checking the timestamp of each data point, we can identify discontinuous time periods. For linear interpolation, we can predict the value of the missing point by calculating the equation of the line between two adjacent points. For spline interpolation, we use a piecewise polynomial function to fit the data curve. The specific operation can be performed using the Pandas library in Python to perform interpolation processing to ensure the consistency and integrity of the time series data.

[0070] Finally, the pre-processed state data is formatted. This format conversion typically involves converting the state data into a JSON format that is easily parsed. For example, Python's json module is used to serialize the data. Once the data format conversion is complete, TLS encryption is used to prevent the data from being eavesdropped or tampered with during network transmission. TLS encryption can be automatically implemented through the HTTPS protocol to ensure end-to-end security. For example, in Python, the requests library is used to send HTTP requests with TLS encryption to upload the data to the control center.

[0071] Through the above operations, the preprocessing process from denoising and standardization to time synchronization and encrypted upload is realized. These high-quality preprocessed data provide a solid foundation for subsequent Transformer model training, improving the overall performance and reliability of energy storage equipment.

[0072] S4. Build a Transformer model in the control center and use historical battery module and cooling liquid status data for training to obtain a trained Transformer model.

[0073] The specific steps include:

[0074] TensorFlow was selected as the deep learning framework in the control center, and the Transformer model was built based on TensorFlow. The core components of this Transformer model include a multi-head self-attention mechanism, a feedforward neural network, residual connections, and layer normalization. The multi-head self-attention mechanism uses multiple independent attention heads to capture the dependencies between different positions in the input sequence. Each attention head calculates the attention weight by the dot product of the query, key, and value matrices, thereby highlighting important time-step features. The feedforward neural network is used to further extract deep features. It consists of two layers of linear transformations and activation functions and can capture complex patterns in the data. Residual connections and layer normalization are used to alleviate the gradient vanishing problem in deep networks and ensure that the output of each layer has a stable distribution, which helps accelerate convergence and improve model performance.

[0075] The preprocessed battery module and cooling liquid status data is fed into the Transformer model for training. This status data is organized as a time series. Each input sample contains data from multiple time steps, and each time step contains multiple feature dimensions, such as temperature and flow rate. This provides rich information for the Transformer model.

[0076] During the training process, batch gradient descent is used for iterative training. In each iteration, a small batch of data is randomly sampled from the dataset to calculate the gradient. This method improves computational efficiency and helps the model escape from the local optimal solution. To prevent overfitting, L2 regularization and early stopping are used. L2 regularization limits model complexity by adding a weight decay term to the loss function, thereby avoiding overfitting of the training data. Early stopping terminates training early when performance on the validation set no longer improves, preventing model overfitting.

[0077] The specific steps of training include forward propagation to calculate the predicted value and loss, and back propagation to calculate the gradient. The specific mathematical formula is:

[0078] ;

[0079] in, represents the loss value, represents the number of samples, represents the sample index, Indicates the The true value of the sample, Indicates the The predicted value of samples;

[0080] The mathematical formula for calculating weight gradient is:

[0081] ;

[0082] in, Indicates the loss value Weight The gradient, A real-valued vector, represents the predicted value vector, Represents input data;

[0083] The mathematical formula for calculating the bias gradient is:

[0084] ;

[0085] Among them, represents the loss value Bias The gradient, A real-valued vector, represents the predicted value vector;

[0086] At the same time, during the iteration, the Adam optimizer updates the first-order moment estimate and the second-order moment estimate based on the calculated gradient. The first-order moment estimate helps the model accelerate the learning rate along the steep direction, while the second-order moment estimate is adaptively adjusted according to the historical gradient size of the parameter, so that the learning rate can change dynamically for each parameter. The Adam optimizer uses the updated first-order moment estimate, second-order moment estimate, and bias-corrected values ​​to calculate a new adaptive learning rate for each parameter, and adjusts the parameter value accordingly to gradually optimize the model performance.

[0087] Through the above operations, a Transformer model was successfully built and trained in the control center, ensuring the accuracy and generalization ability of the model, and providing a solid foundation for further analysis and decision-making.

[0088] S5. Input the pre-processed battery module and cooling liquid status data into the trained Transformer model to obtain the prediction results of the high-risk area location and the expected temperature change value.

[0089] The specific steps include:

[0090] Convert the preprocessed battery module and cooling liquid state data into a three-dimensional tensor structure. This process includes the number of samples, the number of time steps, and the number of feature dimensions. For example, if there are 100 samples, each containing 50 time steps, and each time step has 10 features, such as temperature and flow rate, this data can be organized into a tensor with a shape of (100, 50, 10). This step is implemented in the deep learning framework TensorFlow, using the view function to adjust the data shape to ensure that it meets the input requirements of the Transformer model.

[0091] Next, the three-dimensional tensor is processed using a multi-head self-attention mechanism. The core of the multi-head self-attention mechanism is to calculate the attention weight for each time step. For each time step, a query matrix, a key matrix, and a value matrix are generated through linear transformation. Then, an attention score is calculated. This process involves a dot product operation and a softmax function to ensure that the sum of the attention weights is 1. In this way, the model can automatically identify which time steps are most important for prediction. For example, when detecting battery overheating, the mechanism can automatically focus on time periods where the temperature rises rapidly, thereby improving prediction accuracy.

[0092] Based on the attention weights obtained from the above operations, the location of the high-risk area is further calculated. The specific formula is:

[0093] ;

[0094] in, The coordinates representing the location of the high-risk area, Indicates that high-risk areas are located in three-dimensional space Position on the axis, Indicates that high-risk areas are located in three-dimensional space Position on the axis, Indicates that high-risk areas are located in three-dimensional space Position on the axis, represents a 3×d-dimensional weight matrix, represents the number of samples, represents the sample index, Indicates the The attention weight of time steps, Indicates the The feature vector of each time step, Represents the bias term used to adjust the output position;

[0095] The output value coordinates provide a specific three-dimensional coordinate to identify the location of high-risk areas. Due to the use of the attention mechanism, the model can focus on the time periods and features that are most critical for prediction, thereby improving positioning accuracy.

[0096] The specific formula for calculating the expected temperature change value is:

[0097] ;

[0098] in Indicates the expected temperature change value, represents the number of samples, represents the sample index, Indicates the Samples at time point The temperature value, Indicates the Samples at time point The temperature value, Indicates the time point index;

[0099] The output value provides a standardized temperature change indicator to help evaluate the trend of temperature change. If the expected temperature change value is close to 1, it indicates that the temperature is generally rising. If it is close to -1, it indicates that the temperature is generally falling. If it is close to 0, it indicates that there is no significant temperature change trend.

[0100] Through the above operations, not only can the location of high-risk areas be accurately identified, but future temperature changes can also be predicted. This information provides critical data support for the control center, thereby effectively managing the temperature of energy storage equipment, extending equipment life and improving safety.

[0101] S6. Based on the predicted results of the location of the high-risk area and the expected temperature change value, the control center automatically adjusts the working status of the micropump and microvalve, changes the flow path and flow rate of the cooling liquid, and achieves local cooling.

[0102] The specific steps include:

[0103] Based on the predicted results of high-risk areas, the system automatically identifies battery module locations that require priority cooling. Combined with expected temperature trends, it determines which areas are most likely to overheat, allowing immediate cooling measures to be taken.

[0104] To regulate the coolant flow to high-risk areas, the control center adjusts the output power of the micropump and uses pulse-width modulation technology and PID control algorithms to precisely control the coolant flow rate. Pulse-width modulation adjusts the speed of the micropump by changing the duty cycle of the signal, thereby flexibly controlling the coolant flow rate. The PID control algorithm dynamically adjusts the operating parameters of the micropump based on real-time feedback data, such as the difference between the current temperature and the set temperature, to ensure that the coolant flow rate is always maintained at the optimal level. This method not only improves cooling efficiency but also effectively avoids energy waste caused by overcooling.

[0105] At the same time, by controlling the opening of the microvalve, the flow path of the coolant is changed so that it accurately covers the battery module position that needs cooling. The opening of the microvalve is controlled by a stepper motor, which has high movement precision and fast response speed, and can quickly adapt to changes. In addition, a closed-loop feedback mechanism is used to monitor the actual flow direction and temperature distribution of the coolant to ensure that the coolant can accurately reach the target area and achieve the expected cooling effect. This precise control method ensures the efficient operation and safety of the energy storage equipment even under complex and changing operating conditions.

[0106] Through the above steps, the fully immersed liquid-cooled energy storage control device can intelligently adjust the working status of the micropump and microvalve based on real-time prediction results, optimize the flow path and flow rate of the cooling liquid, thereby effectively managing the temperature of the energy storage device and providing a solid guarantee for long-term stable operation.

[0107] This embodiment also provides a fully immersed liquid-cooled energy storage control system, including: a calibration module, a data acquisition module, a preprocessing module, a model training module, a result prediction module, and an adjustment module;

[0108] Calibration module for mounting and calibrating multiple sensors in a liquid immersion tank;

[0109] A data acquisition module is used to collect status data of the battery module and cooling liquid in real time using calibrated multi-sensors;

[0110] A pre-processing module is used to pre-process the status data of the battery module and the cooling liquid and upload it to the control center;

[0111] The model training module is used to build a Transformer model in the control center and use historical battery module and cooling liquid status data for training to obtain a trained Transformer model;

[0112] The result prediction module is used to input the pre-processed battery module and cooling liquid status data into the trained Transformer model to obtain the prediction results of the high-risk area location and expected temperature change value;

[0113] The adjustment module is used to input the pre-processed battery module and cooling liquid status data into the trained Transformer model to obtain the prediction results of the high-risk area location and expected temperature change value.

[0114] This embodiment also provides a computer device suitable for the case of a fully immersed liquid-cooled energy storage control method, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the fully immersed liquid-cooled energy storage control method proposed in the above embodiment.

[0115] The computer device may be a terminal, comprising a processor, memory, a communication interface, a display, and an input device connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores an operating system and computer programs. The internal memory provides an environment for the operating system and computer programs stored in the non-volatile storage media. The communication interface of the computer device is used to communicate with external terminals via wired or wireless communication. Wireless communication may be achieved via Wi-Fi, a carrier network, NFC (near-field communication), or other technologies. The display of the computer device may be a liquid crystal display or an electronic ink display. The input device may be a touchscreen overlay on the display, buttons, a trackball, or a touchpad on the computer device housing, or an external keyboard, touchpad, or mouse.

[0116] This embodiment also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for controlling fully submerged liquid-cooled energy storage as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0117] In summary, the present invention proposes an intelligent control strategy based on the Transformer model, predicts future temperature changes based on real-time monitoring data, and automatically adjusts the coolant flow path and flow rate, significantly improving dynamic response capabilities and adaptability, ensuring precise temperature control even under complex operating conditions, and improving battery safety and service life. Utilizing a multi-sensor network, comprehensive monitoring of the temperature distribution within the entire battery module is achieved, and data preprocessing technology is used to improve data quality. This not only provides richer status information, but also enhances the device's ability to identify potential high-temperature risk areas, thereby effectively preventing failures.

[0118] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A fully immersed liquid-cooled energy storage control method, characterized by: include, Install multiple sensors in a liquid immersion tank and perform calibration; Utilize calibrated multi-sensors to collect real-time status data of battery modules and cooling liquid; Pre-process the status data of the battery module and cooling liquid and upload it to the control center; Build a Transformer model in the control center and use historical battery module and cooling liquid status data for training to obtain the trained Transformer model; Inputting the preprocessed battery module and cooling liquid status data into the trained Transformer model to obtain prediction results of the high-risk area location and expected temperature change value, including converting the preprocessed battery module and cooling liquid status data into a three-dimensional tensor structure, wherein the three-dimensional tensor structure includes the number of samples, the number of time steps, and the number of feature dimensions, using a multi-head self-attention mechanism to calculate the attention weights of the three-dimensional tensor, performing weighted summation, and obtaining the high-risk area location and expected temperature change value; Based on the predicted results of the high-risk area location and the expected temperature change value, the control center automatically adjusts the working status of the micropump and microvalve, changes the flow path and flow rate of the cooling liquid, and realizes local cooling, including automatically identifying the battery module position that needs priority cooling treatment based on the acquired high-risk area location information, and adjusting the working status of the micropump and microvalve in combination with the expected temperature change value. By adjusting the output power of the micropump, pulse width modulation and PID control algorithms are used to control the flow of coolant to the high-risk area position. By controlling the opening of the microvalve, stepper motor control and closed-loop feedback are used to change the flow path of the coolant, and the coolant is covered to the battery module position that needs cooling, thereby realizing local cooling.

2. The fully submerged liquid-cooled energy storage control method according to claim 1, wherein: The method of installing multiple sensors in a liquid immersion tank and calibrating the sensors specifically includes the following steps: Place the temperature sensor in a constant temperature bath, measure the reading, and compare it with the temperature in the constant temperature bath. Use linear regression to calibrate the temperature sensor. Place the current sensor under a standard current source with known current, measure the reading, and compare it with the known current value. Use the least squares method to calibrate the current sensor. Place the flow sensor in a standard fluid with a known flow rate, measure the reading, and compare it with the known flow rate value. Use linear interpolation to calibrate the flow sensor. The standard current source is defined based on a known current value, and the standard fluid is defined based on a known flow rate value and constant temperature and pressure conditions.

3. The fully submerged liquid-cooled energy storage control method according to claim 2, wherein: The method of using the calibrated multi-sensor to collect the status data of the battery module and the cooling liquid in real time specifically includes the following steps: Use the calibrated temperature sensors to collect temperature data of the battery module and cooling liquid by arranging embedded temperature monitoring points and performing timed sampling. The calibrated current sensor is used to collect the charging status data of the battery module using the coulomb counting method combined with temperature compensation. The calibrated flow rate sensor is installed on the coolant flow path to collect the flow rate data of the cooling liquid.

4. The fully submerged liquid-cooled energy storage control method according to claim 3, wherein: The preprocessing includes outlier processing, standardization processing and difference processing; The pre-processed status data is formatted and encrypted using TLS. A secure connection is established with the control center via the HTTP protocol to upload the status data.

5. The fully submerged liquid-cooled energy storage control method according to claim 4, wherein: The Transformer model is constructed in the control center and trained using the historical battery module and cooling liquid status data to obtain the trained Transformer model. Specifically, the following steps are included: Select the TensorFlow deep learning framework in the control center and build a Transformer model using a multi-head self-attention mechanism, a feedforward neural network, residual connections, and layer normalization; Input historical battery module and cooling liquid status data into the Transformer model for training; During the training process, batch gradient descent is used for iterative training, and L2 regularization and early stopping are used to prevent overfitting, forward propagation is used to calculate prediction values ​​and losses, back propagation is used to calculate gradients, and the Adam optimizer is used to update model parameters.

6. A fully submerged liquid-cooled energy storage system, based on the fully submerged liquid-cooled energy storage control method according to any one of claims 1 to 5, characterized in that: include, Calibration module, data acquisition module, preprocessing module, model training module, result prediction module, adjustment module; Calibration module for mounting and calibrating multiple sensors in a liquid immersion tank; A data acquisition module is used to collect status data of the battery module and cooling liquid in real time using calibrated multi-sensors; A pre-processing module is used to pre-process the status data of the battery module and the cooling liquid and upload it to the control center; The model training module is used to build a Transformer model in the control center and use historical battery module and cooling liquid status data for training to obtain a trained Transformer model; The result prediction module is used to input the preprocessed battery module and cooling liquid status data into the trained Transformer model to obtain the prediction results of the high-risk area location and expected temperature change value. The module converts the preprocessed battery module and cooling liquid status data into a three-dimensional tensor structure, which includes the number of samples, the number of time steps, and the number of feature dimensions. The three-dimensional tensor uses a multi-head self-attention mechanism to calculate the attention weights and perform weighted summation to obtain the high-risk area location and expected temperature change value. The regulation module is used to input the pre-processed battery module and cooling liquid status data into the trained Transformer model to obtain the prediction results of the high-risk area location and the expected temperature change value, including automatically identifying the battery module location that needs priority cooling treatment based on the obtained high-risk area location information, adjusting the working status of the micropump and microvalve in combination with the expected temperature change value, and controlling the flow of coolant to the high-risk area location by adjusting the output power of the micropump, using pulse width modulation and PID control algorithms, and changing the flow path of the coolant by controlling the opening of the microvalve, using stepper motor control and closed-loop feedback, so that the coolant covers the battery module location that needs cooling, thereby achieving local cooling.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the fully immersed liquid-cooled energy storage control method according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the fully immersed liquid-cooled energy storage control method according to any one of claims 1 to 5 are implemented.

Images