Intelligent phase change immersion cooling method based on double-cavity dynamic regulation

Abstract

The application relates to the technical field of battery cooling and discloses an intelligent phase change immersion cooling method based on double-cavity dynamic regulation. Firstly, the method performs sensor data collection to obtain temperature and internal resistance monitoring data of high internal resistance battery cavities and low internal resistance battery cavities; battery state analysis is carried out based on the data to determine high internal resistance state factors and low internal resistance state factors; the state factors are used as indexes to search for target nozzle working parameters in a nozzle parameter configuration space; the nozzle working state is adjusted according to the target parameters, and a phase change immersion cooling operation is performed; battery capacity calibration is regularly carried out, and a migration determination reminder is triggered when the calibration result meets migration conditions. Through real-time monitoring, dynamic analysis and intelligent regulation, the method realizes differentiated cooling of different internal resistance battery cavities, can adapt to dynamic changes of battery performance, improves cooling efficiency and precision, and guarantees battery operation stability.

Description

Technical Field

[0001] This invention relates to the field of battery cooling technology, specifically to an intelligent phase change immersion cooling method based on dual-cavity dynamic control. Background Technology

[0002] With the rapid development of new energy technologies, high-energy-density batteries are increasingly widely used in electric vehicles, energy storage systems, and other fields. Batteries generate a large amount of heat during charging and discharging. If this heat cannot be dissipated in time, the battery temperature will rise, affecting its performance and lifespan. Especially in large-scale battery pack applications, the internal resistance of different battery cells varies. Batteries with higher internal resistance generate more heat during operation, making them prone to localized overheating and creating a risk of thermal runaway.

[0003] Currently, common battery cooling methods include air cooling, liquid cooling, and phase change cooling. Air cooling has a simple structure but low cooling efficiency, making it difficult to meet the heat dissipation requirements of high-power batteries. Liquid cooling removes heat through liquid circulation, resulting in high cooling efficiency, but the system structure is complex, requiring additional pumps and piping, increasing equipment size and cost, and posing a risk of leakage. Phase change cooling utilizes phase change materials to absorb heat, achieving cooling with advantages such as high heat dissipation efficiency and stable temperature control. However, traditional phase change cooling methods are mostly passive and cannot dynamically adjust the cooling intensity according to the real-time state of the battery. When the battery's internal resistance and temperature distribution are uneven, precise heat dissipation is difficult to achieve, potentially leading to performance degradation in some batteries due to insufficient heat dissipation.

[0004] During long-term use, the performance of individual battery cells gradually degrades, and their internal resistance changes. Batteries that initially had low internal resistance may gradually transform into those with high internal resistance. Existing cooling systems lack a dynamic response mechanism to these changes in battery state, making it impossible to adjust cooling strategies in a timely manner and adapt to dynamic changes in battery performance. This negatively impacts the overall lifespan and safety of the battery pack. Therefore, a smart cooling method is needed that can dynamically adjust cooling intensity based on the battery's real-time internal resistance and temperature state, and adapt to changes in battery performance, to address the shortcomings of existing cooling technologies. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent phase change immersion cooling method based on dual-cavity dynamic control, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides an intelligent phase change immersion cooling method based on dual-cavity dynamic control, the method comprising:

[0007] Perform sensor data acquisition to obtain battery temperature monitoring data sets and internal resistance monitoring data sets for high internal resistance battery chambers and low internal resistance battery chambers;

[0008] Based on the battery temperature monitoring data set and the internal resistance monitoring data set, battery state analysis is performed to determine the high internal resistance state factor and the low internal resistance state factor.

[0009] Using the high internal resistance state factor and the low internal resistance state factor as indexes, a centralized search is performed in the nozzle parameter configuration space to determine the target nozzle operating parameters.

[0010] Based on the target nozzle operating parameters, adjust the nozzle operating states of the high internal resistance battery cavity and the low internal resistance battery cavity to perform phase change immersion cooling operation;

[0011] Periodically perform battery capacity calibration to obtain a calibration result set. When the calibration result set meets the migration conditions, trigger a migration judgment reminder.

[0012] Preferably, the sensor data acquisition includes:

[0013] Collect battery temperature monitoring data and internal resistance monitoring data of the high internal resistance battery cavity, and battery temperature monitoring data and internal resistance monitoring data of the low internal resistance battery cavity.

[0014] The battery temperature monitoring data of the high internal resistance battery cavity and the battery temperature monitoring data of the low internal resistance battery cavity are integrated to generate the battery temperature monitoring data set.

[0015] The internal resistance monitoring data of the high internal resistance battery cavity and the internal resistance monitoring data of the low internal resistance battery cavity are integrated to generate the internal resistance monitoring data set.

[0016] Preferably, the battery state analysis includes:

[0017] Data fluctuation analysis was performed on the battery temperature monitoring data set to obtain temperature fluctuation indicators;

[0018] Data stability analysis is performed on the internal resistance monitoring data set to obtain internal resistance stability index;

[0019] Based on the temperature fluctuation index and the internal resistance stability index, the high internal resistance state factor and the low internal resistance state factor are calculated.

[0020] Preferably, the data fluctuation analysis includes:

[0021] Extract time-series data points from the battery temperature monitoring dataset;

[0022] Calculate the rate of change between adjacent data points;

[0023] The average of all rates of change is used as the temperature fluctuation index.

[0024] Preferably, the centralized search in the nozzle parameter configuration space includes:

[0025] Obtain the high internal resistance state factors and low internal resistance state factors of multiple samples, as well as the corresponding nozzle operating parameters of multiple samples.

[0026] A two-dimensional configuration space is pre-constructed, where the first dimension represents the range of high internal resistance state factors and the second dimension represents the range of low internal resistance state factors.

[0027] The multiple high internal resistance state factors and the multiple low internal resistance state factors of the samples are mapped to the two-dimensional configuration space to form multiple sample space points;

[0028] The multiple sample nozzle operating parameters are used to identify the multiple sample space points, thereby generating the nozzle parameter configuration space.

[0029] Preferably, the centralized search in the nozzle parameter configuration space further includes:

[0030] The high internal resistance state factor and the low internal resistance state factor are input into the two-dimensional configuration space as target points;

[0031] Calculate the distance metric between the target point and each point in the sample space;

[0032] Select the sample space point with the smallest distance metric as the matching point;

[0033] The working parameters of the sample nozzle corresponding to the matching point are used as the working parameters of the target nozzle.

[0034] Preferably, the periodic battery capacity calibration includes:

[0035] The battery capacity, internal resistance, and temperature rise sensitivity of the high internal resistance battery cavity and the low internal resistance battery cavity are measured quarterly.

[0036] The calibration result set is generated by integrating the battery capacity value, the internal resistance value, and the temperature rise sensitivity value.

[0037] Determine whether the calibration result set meets the migration conditions, wherein the migration conditions include a first preset percentage of battery capacity decay to the initial value of battery capacity, a second preset percentage of internal resistance increase to the initial value of internal resistance, or a temperature rise sensitivity greater than a preset threshold.

[0038] Preferably, the trigger migration determination reminder includes:

[0039] When the calibration result set satisfies the migration condition, a migration signal is generated;

[0040] Based on the migration signal, a migration command is output to prompt the staff to perform the battery migration and replacement operation.

[0041] Preferably, the method further includes:

[0042] Based on the high internal resistance state factor and the low internal resistance state factor, predict the trend of battery thermal behavior.

[0043] The operating parameters of the target nozzle are adjusted based on the battery thermal behavior trend.

[0044] Preferably, after performing the phase change immersion cooling operation, the process includes:

[0045] Monitor the cooling effect data of the high internal resistance battery cavity and the low internal resistance battery cavity;

[0046] Based on the cooling effect data, update the battery temperature monitoring data set and the internal resistance monitoring data set;

[0047] Repeat the steps of performing battery state analysis, performing centralized search in the nozzle parameter configuration space, adjusting the nozzle working state of the high internal resistance battery cavity and the low internal resistance battery cavity based on the target nozzle working parameters, and performing phase change immersion cooling operation.

[0048] Compared with the prior art, the beneficial effects of the present invention are:

[0049] By acquiring real-time temperature and internal resistance information for high-internal-resistance and low-internal-resistance battery cavities using sensor data, precise data is obtained for subsequent state analysis and cooling regulation. Based on this monitoring data, battery state analysis determines the high-internal-resistance state factors and low-internal-resistance state factors, which accurately reflect the differences in the operating state of the battery within different cavities, making cooling regulation more targeted.

[0050] By using state factors as an index to centrally search for target nozzle operating parameters in the nozzle parameter configuration space, precise matching of cooling parameters can be achieved, avoiding the blind setting of parameters in traditional cooling methods. Adjusting the nozzle operating state according to the target parameters can dynamically change the intensity and range of phase change cooling, ensuring that high internal resistance battery cavities obtain stronger cooling effects, while avoiding performance degradation of low internal resistance battery cavities due to overcooling, thus realizing differentiated cooling for different cavities.

[0051] Regularly performing battery capacity calibration and triggering alerts when migration conditions are met allows for timely detection of changes in battery performance, facilitating proper battery migration and ensuring the battery remains in a suitable cooling environment. This dynamic adjustment mechanism adapts to performance degradation and internal resistance changes during long-term use, maintaining a good fit between the cooling system and the battery's condition. Attached Figure Description

[0052] Figure 1 This is a schematic diagram illustrating the working principle of the intelligent phase change immersion cooling method based on dual-cavity dynamic control described in this invention.

[0053] Figure 2 A flowchart for sensor data acquisition;

[0054] Figure 3 A flowchart for constructing the nozzle parameter configuration space;

[0055] Figure 4 Flowchart for battery capacity calibration;

[0056] Figure 5 This is a flowchart for monitoring and updating after cooling. Detailed Implementation

[0057] 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.

[0058] Please see Figure 1 This invention provides an intelligent phase change immersion cooling method based on dual-cavity dynamic control, the method comprising:

[0059] The system collects battery temperature and internal resistance data from both high-resistance and low-resistance battery cavities using sensors, creating battery temperature and internal resistance monitoring datasets. Based on this data, the battery state is analyzed, and high-resistance and low-resistance state factors are calculated. Subsequently, using the state factors as an index, the system searches for matching target nozzle operating parameters in a pre-built nozzle parameter configuration space, and adjusts the operating state of the dual-cavity nozzle accordingly to perform phase change immersion cooling. The system also periodically performs battery capacity calibration; when the calibration results meet the migration conditions, a migration judgment reminder is triggered, prompting personnel to migrate or replace the battery.

[0060] Example 1: See Figure 2In the sensor data acquisition stage, temperature monitoring data is acquired through a distributed thermocouple network. Three thermocouple probes are deployed on the surface of each cell in both high-resistance and low-resistance battery cavities, located at the electrode connection, geometric center, and edge region, respectively. Thermocouples synchronously acquire temperature data with a sampling period of 100 milliseconds, which is then processed by an AD conversion module to generate a timestamped raw dataset. Internal resistance monitoring employs an AC injection method, applying a 1kHz test signal during the battery charging / discharging interval. The voltage and current phase difference is obtained through a four-terminal measurement circuit, and the internal resistance value is calculated and marked with the corresponding timestamp.

[0061] The raw data enters the preprocessing stage: temperature data undergoes median filtering to eliminate instantaneous interference, and internal resistance data is smoothed using a moving average method. The processed data are aligned according to time series; temperature and internal resistance data for high-internal-resistance battery cavities are formed into separate ordered arrays, while low-internal-resistance battery cavities simultaneously generate independent ordered arrays. Finally, these are integrated into a battery temperature monitoring dataset and an internal resistance monitoring dataset. The dataset structure uses a two-dimensional time-value matrix for storage, with row indices representing millisecond-level timestamps and column indices indicating the battery cavity number and parameter type.

[0062] The system extracts the time-series temperature curve for each battery cavity and uses a sliding window mechanism for dynamic segmentation. The window width is set to 30 seconds, and each sliding step is 5 seconds. Data fluctuation analysis is performed within each window: the rate of temperature change of adjacent data points within the window is calculated, and the arithmetic mean of all rates of change within the window is obtained after taking the absolute value. The average of all windows is then weighted, with the window weights positively correlated with their time proximity, and the final output is a temperature fluctuation index. This index reflects the thermal stability characteristics of the battery cavity over time.

[0063] For each battery cavity's internal resistance dataset, data blocks are divided with a 10-minute interval. The standard deviation of the internal resistance value is calculated within each data block, along with the difference between the maximum and minimum values. These two numerical values ​​are input into a stability assessment model, which outputs a stability coefficient in the 0-1 range. The stability coefficients of all data blocks are weighted according to time series, with more recent data blocks receiving higher weights, ultimately generating an internal resistance stability index. For high-internal-resistance battery cavities, two indices are input into a dedicated calculation unit containing dynamically adjusted weighting coefficients: the initial weight for temperature fluctuation is set to 0.65, and the weight for internal resistance stability is 0.35. These weighting coefficients are automatically adjusted based on ambient temperature; when the ambient temperature exceeds 35°C, the weight for temperature fluctuation increases to 0.7. The calculation unit performs linear weighted fusion, outputting a high internal resistance state factor in the 0-1 range. Low-internal-resistance battery cavities use an independent calculation unit, with a fixed weight for temperature fluctuation at 0.6 and a weight for internal resistance stability at 0.4, generating a low internal resistance state factor using the same algorithm.

[0064] The entire analysis process employs a pipelined architecture, with the temperature fluctuation analysis thread and the internal resistance stability analysis thread executing in parallel. When a new data packet arrives, the system prioritizes processing the temperature data thread, while the internal resistance data thread is allowed a maximum synchronization delay of 300 milliseconds. State factors are updated every 5 seconds, and the update results are written to a shared memory area for subsequent modules to access. Historical state factor data is stored in a circular buffer in chronological order, with the buffer capacity being the data from the most recent hour; older data is automatically overwritten.

[0065] An anomaly handling mechanism is implemented throughout the entire process: when thermocouple data samples exceed the reasonable temperature range (-20℃ to 80℃) for three consecutive times, the system automatically switches to the backup probe and marks the fault point. If the internal resistance measurement value undergoes a step change (the difference between adjacent sample values ​​exceeds 20%), the retest mechanism is triggered and the internal resistance calibration procedure is initiated. When an anomaly occurs during the state factor calculation, the system uses the moving average of the previous three valid values ​​to replace the current abnormal value. All abnormal events are recorded in the system log, including the timestamp, anomaly type, and handling measures.

[0066] The data acquisition module encapsulates the processed data into JSON format messages and pushes them to the status analysis message queue. After consuming the messages, the analysis module generates status factor data packets and broadcasts them to the nozzle control module via a publish-subscribe pattern. Message transmission uses the TCP protocol to ensure reliability, with a 500-millisecond timeout retransmission mechanism. The entire process runs in a real-time operating system environment, with critical threads given the highest priority to ensure the time determinism of the status factor update cycle.

[0067] The state factor data packet includes a timestamp, cell number, temperature fluctuation index, internal resistance stability index, and final state factor value. A CRC checksum is added to the data packet; the receiving end verifies the checksum before parsing and using it. Historical data storage employs a time-sharing archiving strategy, generating a data snapshot every minute and storing it in the time-series database, with a retention period of 30 days. The database establishes a joint index for the temperature fluctuation index and the internal resistance stability index, supporting rapid retrieval of historical state change trends by time range.

[0068] Example 2: See Figure 3 The construction of the nozzle parameter configuration space begins with the collection of historical operating data. The system extracts operating records from the past six months from the time-series database and filters data packages containing complete state factors and nozzle parameters. Each data package contains a timestamp, high internal resistance state factor values, low internal resistance state factor values, and corresponding nozzle operating parameter records. Specifically, the nozzle operating parameters include three sub-items: opening duration, injection frequency, and atomized particle diameter. Data filtering requires that the state factor values ​​be within a valid range and that the nozzle parameters be within the normal operating range.

[0069] The high internal resistance state factor originally ranged from 0 to 1.2, and was mapped to the standard range of 0 to 1.0 through a linear transformation. The low internal resistance state factor originally ranged from 0 to 0.9, and was simultaneously mapped to the range of 0 to 0.8. In the nozzle parameters, the opening duration was normalized to a value between 0 and 1 in seconds, the injection frequency was converted to a percentage in Hertz units, and the atomized particle diameter was scaled according to a range of 50-200 micrometers. All normalization operations preserved the conversion relationship between the original and standard values, and a bidirectional mapping table was established and stored in the configuration file.

[0070] The horizontal axis is defined as the high internal resistance state factor dimension, with scales in 0.01 increments. The vertical axis is the low internal resistance state factor dimension, with scales in 0.008 increments. The coordinate system covers the possible distribution area of ​​all sample data, with a 5% margin for expansion in the boundary areas. Each historical data packet generates a sample space point, whose coordinates are determined by the normalized state factor value. The sample point carries a nozzle parameter label, and the label data structure contains three fields: activation duration code, frequency code, and atomized particle code. Each field stores the normalized parameter value.

[0071] The system detects overlapping sample points, considering any two points with a distance less than 0.005 as overlapping points. The nozzle parameters of the overlapping point set are calculated using an arithmetic mean and merged into a single representative point. Sparse point sets in the spatial edge regions are supplemented with virtual points using an interpolation algorithm, generated based on the gradient variation trend of parameters in adjacent regions. The final nozzle parameter configuration space contains approximately 1200 valid sample points, each with unique coordinates and parameter labels. The spatial data is stored in a KD-tree structure to improve retrieval efficiency.

[0072] The system acquires the raw values ​​of the current high internal resistance state factor and low internal resistance state factor, converts them into standard coordinate values ​​through the mapping relationship in the preprocessing stage, and forms the target point coordinates. A range search is performed in the KD-tree structure: with the target point as the center, the initial search radius is set to 0.05. If a sample point exists within this radius, the Euclidean distance between the target point and each sample point is calculated; if no sample point exists, the search radius increases in steps of 0.01 until a valid point is found.

[0073] In addition to the difference in state factor coordinates, a time dimension weight is introduced: recent sample points are given an advantage in distance calculation, and the time decay coefficient is set according to an exponential function. The final distance value is the product of the spatial distance and the time adjustment factor. The system selects the three sample points with the smallest distance as the candidate point set. When the distance difference between the candidate points is less than 0.02, the nozzle parameters of the three points are taken as a weighted average, with the weight inversely proportional to the distance value. If the distance difference between the candidate points is large, the parameters of the nearest neighbor single point are selected.

[0074] The system checks the rationality of candidate parameters: whether the on-time is within the allowable range of 1-15 seconds, whether the spray frequency is within the safe range of 20-100Hz, and whether the atomized particle diameter meets the process requirements of 60-180 micrometers. If any parameter exceeds the threshold, the system switches to a backup strategy: expanding the search along the state factor gradient direction in the configuration space until compliant parameters are obtained. The final output of the target nozzle operating parameters includes three sub-items in the original value format, along with a parameter confidence score. The system compares the differences between the new parameters and the current operating parameters. When the change in on-time exceeds 2 seconds, the change in frequency exceeds 15Hz, or the change in particle diameter exceeds 30 micrometers, a step-by-step adjustment procedure is initiated. Each adjustment does not exceed 20% of the original value, and the adjustment interval is set to 10 seconds. During the adjustment process, the battery temperature change rate is continuously monitored. If the change rate exceeds 3℃ / second, the adjustment is paused. After the parameters stabilize, the actual operating effect is recorded, forming new sample points to supplement the configuration space.

[0075] A weekly integrity check is performed on the configuration space: This includes checking the sample point distribution density and adding markers to areas with a density below 0.5 points per unit area; scanning for expired sample points and moving points older than three months to the archive; and analyzing parameter label consistency by clustering and reorganizing point sets with parameter differences exceeding 20% ​​within the same coordinate region. Maintenance operations are performed during system idle periods, with a time limit of 5 minutes, during which cached space is used to maintain normal operation of the search function.

[0076] When a target point falls into a blank area with no sample points, the system generates virtual parameters based on the parameter gradients of surrounding points and triggers a sample collection command. When candidate parameters conflict, conflict details are recorded and a manual review process is initiated. All search operations generate logs containing target point coordinates, candidate point information, output parameters, and adjustment process data; logs are retained for three months. The main storage uses memory mirroring technology, synchronizing to the solid-state drive every two hours. Incremental backups are performed daily, and full backups are generated and uploaded to cloud storage weekly. Data version management records every spatial structure adjustment, supporting rollback to any version from the last seven days. Access control mechanisms restrict modification permissions in the configuration space, allowing only authorized accounts to operate after two-factor authentication.

[0077] The search interface receives JSON requests containing two state factors and returns a JSON response containing three nozzle parameters. The interface is configured with flow control, limiting the maximum number of requests processed per second to no more than 50. The service health monitoring module tracks response latency in real time and automatically scales up service instances when the average latency exceeds 200 milliseconds. The service status dashboard displays real-time operational metrics such as search hit rate and parameter adjustment frequency to assist in operational decision-making.

[0078] After each successful application of new parameters, the system records actual operational data, including temperature drop and internal resistance changes within 5 minutes of cooling startup. When the same set of parameters accumulates ten valid operational records, the system evaluates the parameter performance stability. Data packets that meet the performance evaluation criteria will be used as high-quality sample points and given priority in subsequent search and matching processes. Low-quality sample points will have their search weight reduced until they are moved to the isolation observation area.

[0079] The algorithm analyzes the sample point density and parameter dispersion in each region of the configuration space, increasing the sample point collection frequency in regions with high dispersion. It identifies regions sensitive to parameter changes and increases the sample point density by 50% in regions where the state factor gradient is greater than 0.3. The optimized spatial structure generates an evaluation report, showcasing the trends in key indicators such as spatial coverage and parameter consistency.

[0080] Example 3: See Figure 4 The battery capacity calibration operation starts automatically according to a preset cycle. Before execution, the system pauses the phase change immersion cooling process to ensure the battery is in a static state. The calibration trigger signal is generated by the time management module, which synchronously coordinates the calibration sequence of multiple battery chambers to avoid power load fluctuations caused by simultaneous measurements. During the calibration preparation stage, the system disconnects the load connection of the battery under test and connects to a dedicated measurement circuit, which includes a constant current discharge unit, a voltage sampling circuit, and a temperature compensation module.

[0081] When measuring battery capacity using the constant current discharge method, the discharge current is set to 0.2 times the battery's rated capacity, and the cutoff voltage is automatically matched according to the battery's chemistry. The discharge time-voltage curve is recorded during the measurement process. Discharge is stopped when the voltage drops to the cutoff point, and the actual capacity value is obtained through integration. The integration formula is as follows:

[0082]

[0083] in: To calculate the battery capacity value, It is a constant current discharge current. and These represent the start and end times of the discharge, respectively. It is the temperature compensation coefficient, which is obtained from the compensation curve based on the real-time measured battery surface temperature.

[0084] The test signal is a 1kHz sine wave, and the current amplitude is controlled within 5% of the battery's rated capacity. The measurement system synchronously acquires the voltage response signal, extracts the same-frequency component through a digital lock-in amplifier, and calculates the magnitude of the complex impedance as the internal resistance value. Each battery compartment is measured three times repeatedly, and the median is taken as the final internal resistance value. The measurement interval is no less than 30 seconds to avoid the influence of battery polarization. The system controls the ambient temperature to rise linearly at a rate of 5℃ / min, while continuously measuring the battery internal resistance at 10-second intervals. The least-squares fit slope of the temperature-internal resistance curve is the temperature rise sensitivity value. A safety protection mechanism is set in the measurement process to immediately terminate the test when the internal resistance suddenly increases by more than 15% or the temperature reaches 60℃.

[0085] A retest is triggered when the capacity value differs from the previous calibration result by more than 20%; the calibration procedure is initiated when the internal resistance value exceeds the battery's technical specifications; and the temperature sensor status is checked when the temperature rise sensitivity value becomes negative. Data that passes calibration is stored categorized by battery chamber number, generating a calibration result set containing timestamps, measured values, and calibration flags.

[0086] The battery capacity degradation judgment uses a dual standard: a warning is recorded when the current capacity is below 85% of the initial value, and migration conditions are met when it is below 80%. The internal resistance growth judgment introduces a time-weighted algorithm, with recent growth data having higher weight; a judgment is triggered if the internal resistance growth value within three months reaches 50% of the initial value. The temperature rise sensitivity threshold is dynamically adjusted according to the battery type: 8Ω / ℃ for lithium iron phosphate batteries and 12Ω / ℃ for ternary lithium batteries. The capacity degradation condition has the highest priority, and a migration signal is generated immediately once it is met; the internal resistance growth condition requires two consecutive calibrations exceeding the threshold for confirmation; the temperature rise sensitivity condition requires comprehensive judgment based on ambient temperature, with the threshold appropriately relaxed in high-temperature environments. A composite alarm is generated when multiple conditions are met simultaneously, indicating the specific values ​​of all parameters exceeding the standard. Raw measurement data is retained in a high-speed cache for 30 days for real-time analysis; calibration results are stored in a time-series database, establishing a joint index by battery number and calibration time; migration judgment records are written to a dedicated event log, linked with the maintenance work order system. The database is backed up daily incrementally, retaining historical data from the past five years for trend analysis.

[0087] Automatic retry is performed when power fluctuations occur during measurement; switching to a backup measurement scheme is initiated when data verification fails; and manual review is submitted when there are disputes regarding boundary conditions in threshold determination. Detailed contextual information is recorded for all abnormal events, including the system status, environmental parameters, and preceding operation records at the time of the anomaly.

[0088] The migration early warning generation module is activated when the judgment conditions are met. The early warning information includes the battery cavity location number, a list of parameters exceeding the standard, the percentage of exceeding the standard, and a suggested handling time limit. Early warning levels are divided into three levels based on the extent of exceeding the standard: Level 1 requires handling within 72 hours, Level 2 requires handling within 168 hours, and Level 3 is only for observation and recording. Early warning information is pushed synchronously through multiple channels, including pop-ups on the system monitoring interface, email notifications, and mobile APP reminders. The system analyzes the changing trends of historical calibration data and automatically shortens the calibration interval for battery cavities with rapid performance degradation. The initial default cycle is 90 days, which can be adjusted to a minimum of 15 days. The cycle adjustment algorithm considers factors such as battery usage frequency, ambient temperature fluctuation range, and historical migration records, with each adjustment not exceeding 30% of the original cycle. The constant current source automatically performs current accuracy verification monthly, triggering a calibration procedure when the deviation exceeds 0.5%; the voltage measurement channel uses a reference voltage source for periodic self-testing; the temperature sensor is calibrated by comparison with a platinum resistance standard. Calibration data is recorded in the equipment file, and the calibration status serves as a prerequisite for the validity of the measurement data. Calibration tasks for multiple battery chambers are queued and executed according to power load capacity; measurements of large-capacity batteries are scheduled during off-peak hours; and the system automatically switches to low-power mode between measurements. Energy consumption monitoring displays the total power consumption of the current calibration task in real time, and automatically pauses subsequent tasks when the preset limit is exceeded.

[0089] Example 4: The migration judgment reminder triggering process manifests as a multi-level linked alarm response mechanism in actual operation. When any parameter in the calibration result set meets the migration conditions, the system first generates a structured record in the event log. Taking the high internal resistance battery cavity numbered B7-12 as an example, its calibration data shows that the current capacity is 78% of the initial value, the internal resistance has increased to 158% of the initial value, and the temperature rise sensitivity reaches 9.3Ω / ℃, simultaneously triggering both capacity decay and internal resistance increase migration conditions. The system automatically marks this battery cavity as a state to be migrated and records the detailed values ​​of the parameters exceeding the standard and the trigger time in the event log.

[0090] The migration signal generation module generates differentiated signal content based on the combination of exceeding parameters. The signal data structure includes four fields: battery location identifier, migration urgency rating, primary exceeding parameter, and secondary exceeding parameter. The urgency rating is divided into five levels based on the magnitude of the exceeding and the parameter type. The rating criteria corresponding to different exceeding situations are shown in Table 1.

[0091] Table 1: Rating Criteria for the Urgency of Migration Signals.

[0092]

[0093] For the B7-12 battery compartment case, the system generates a Level 1 emergency signal, including the complete location path (Area B - Row 7 - Slot 12), key parameters exceeding limits (capacity 78% / internal resistance 158%), and associated temperature rise sensitivity data. Immediately after signal generation, the multi-channel notification system is activated: a red-bordered warning box pops up on the large screen in the central control room, a high-priority alarm message is pushed to the mobile terminals of maintenance personnel, and a spare parts query request is sent to the warehouse management system.

[0094] The command output process adapts to different formats depending on the device type. Commands sent to the intelligent inspection robot are encoded in binary, containing the target location coordinates and the sequence of operation steps. Commands pushed to the engineer's tablet display a visual navigation path, marking the specific location of the battery compartment within the rack. Request messages transmitted to the warehouse management system contain the battery model code and serial number verification information. All commands are accompanied by a timestamp and issuer identifier, forming a complete operation traceability chain.

[0095] In practical applications, battery thermal behavior trend prediction manifests as a dynamically updated thermogram. The system continuously records the temporal changes of high and low internal resistance state factors, calculating a moving average over a 30-minute window. When the state factor of the B7-12 battery cavity shows a monotonically increasing trend for three consecutive windows, and the slope of the increase exceeds a threshold, the prediction model determines that the battery has a risk of thermal runaway. The model output includes a temperature rise prediction curve for the next 15 minutes, which is overlaid with real-time monitoring data and displayed on the warning interface.

[0096] Based on the predicted results for the B7-12 battery cavity, the system first gradually reduced the injection frequency from 80Hz to 65Hz, with each adjustment not exceeding 5Hz; then, it increased the atomized particle diameter from 120 micrometers to 150 micrometers, completing the adjustment in three steps; finally, it extended the on-time by 2 seconds. After each adjustment, the rate of change of the state factor was monitored. If the rate of change did not slow down, a backup strategy was activated, implementing coordinated cooling in adjacent nozzle groups.

[0097] When the system detects that a battery compartment simultaneously meets all three migration conditions, it automatically generates a task awaiting review. The operations supervisor's terminal device receives a review request containing complete calibration data, and the interface displays historical calibration data comparison curves and current parameter deviation analysis. The supervisor can zoom in on key data segments using gestures to view detailed measurement logs, and ultimately choose to confirm the migration or mark it as a false alarm. The entire review process is recorded, including the operation time and decision basis, and incorporated into the quality audit file.

[0098] When the migration signal reaches an urgency level of 3 or higher, the system automatically searches the inventory database and locates a matching spare battery cell. Taking the B7-12 battery compartment as an example, the system locates the nearest replaceable cell in the warehouse's 3D map based on its model identifier LFP-120Ah and generates a work order containing the picking route and quality inspection points. After receiving the work order, the warehouse robot performs a rapid battery performance test to ensure that the spare cell capacity is not less than 95% of the nominal value and the internal resistance deviation is within 5%.

[0099] After the new battery is installed, the system performs three verification tests: the first test measures the open-circuit voltage and internal resistance reference values; the second test detects voltage drop under a 5A load current; and the third test records the self-discharge rate after 8 hours of static operation. All test data are compared with the factory standard values, and a reinstallation process is triggered if the deviation exceeds the allowable range. After successful verification, the new battery information is written to the system database, and the old battery is marked as awaiting recycling and enters the life analysis queue. The trend prediction model's self-optimization is achieved through online learning. Each deviation between the prediction result and the actual monitoring data is recorded as a learning sample, and the model parameters are updated when the accumulated samples reach 100. The update process uses a sliding window mechanism, retaining valid samples from the most recent three months and gradually eliminating earlier data. Model version management records the performance index changes for each update and supports rollback to the previous two stable versions.

[0100] This implementation method ensures the accuracy of migration decisions through a structured signal generation mechanism, and achieves rational resource allocation through a multi-level urgency system. The closed-loop operation of the predictive model and real-time control effectively prevents the risk of thermal runaway, while the human-machine collaborative verification mechanism balances automation efficiency and decision reliability. The entire migration process is seamlessly integrated with the spare parts management system, forming a complete operation and maintenance chain from detection to replacement.

[0101] Example 5: See Figure 5 Cooling effect monitoring is initiated immediately after the phase change immersion cooling operation. An infrared thermal imager array performs a full-area scan of both high-resistance and low-resistance battery cavities, with a scan frequency set to twice per second. The monitoring area for each battery cavity is divided into grid cells, each cell being one-hundredth the surface area of ​​a single battery cell. The thermal imager acquires temperature data from each cell, generating a temperature distribution matrix. The cooling effect data includes three core indicators: the maximum temperature drop is obtained by comparing the values ​​of the highest temperature points before and after cooling; the temperature uniformity coefficient is calculated as the reciprocal of the standard deviation of the surface temperature after cooling; and the cooling rate is the average temperature change slope from the start to the end of cooling.

[0102] The newly acquired cooling effect data enters the preprocessing channel: First, sensor deviation correction is performed, with each thermal imager probe compensating for measurement errors according to the calibration parameter table; then, data alignment is performed, matching the temperature matrix with the battery location mapping table; finally, timestamp synchronization is performed to ensure consistency with the timeline of the internal resistance monitoring data. The processed temperature data replaces the oldest 20% of the data in the battery temperature monitoring dataset, while the internal resistance monitoring dataset is supplemented with measurements at three time points after the cooling process ends, set at 30 seconds, 2 minutes, and 5 minutes after the cooling process ends.

[0103] When the temperature data update volume in the new dataset exceeds 15% of the total capacity, or when three new internal resistance data records are added, the system activates the battery state analysis module. The analysis process uses the same algorithm flow as the initial calculation, but introduces incremental calculation optimization: only the time window involved in the updated data is reprocessed, and the results of historical window data are reused and cached. The regenerated high internal resistance state factor and low internal resistance state factor are compared with the previous version, and the amount and direction of change are recorded.

[0104] The system determines the search region based on the direction of change in the state factor: if the state factor increases by more than 5%, the search region shifts towards the positive gradient direction of the configuration space; if it decreases by more than 5%, it expands towards the negative gradient direction. The search radius is dynamically adjusted according to the magnitude of change; for every 2% increase in the magnitude of change, the radius expands by 0.01 units. Within the locked region, the system only retrieves sample points newly added in the last three months, prioritizing tag data that matches the current battery aging level.

[0105] The initial search results are input into the parameter change evaluator, which calculates the difference between the new parameters and the current operating parameters. Parameter sets with a difference of less than 10% are ignored; parameter sets with a difference between 10% and 20% proceed to the safety verification stage; parameter sets with a difference exceeding 20% ​​require manual review. The safety verification stage checks the process feasibility of the parameter combinations, including the matching degree between atomized particle diameter and injection frequency, and the coordination between operating duration and coolant flow rate.

[0106] The parameter adjustment scheme, which has passed safety verification, is broken down into multiple sub-steps, each modifying only a single parameter dimension. The duration adjustment is performed step-by-step in seconds, with each increase or decrease not exceeding 10% of the current value; the injection frequency adjustment is performed step-by-step in Hertz, with each change controlled within 5 Hertz; and the atomized particle diameter adjustment is achieved in three levels using a piezoelectric controller. After each sub-step is completed, the system pauses for two minutes to collect temperature response data; if the temperature change rate exceeds three degrees Celsius per second, it reverts to the previous step.

[0107] The system continuously compares the deviation of the current battery temperature from the target temperature range. A stable state is determined when the temperature at all monitoring points remains within the target range for five consecutive minutes. If stability is not achieved after six rounds of parameter adjustments, the system initiates an exception handling procedure: freezing the current parameter configuration, generating a diagnostic report, and switching to the basic cooling mode. The basic cooling mode uses a fixed parameter combination selected from a preset scheme library based on the ambient temperature.

[0108] Each closed-loop control cycle's complete record includes fields such as initial state factors, retrieval parameters, adjustment steps, final parameters, and stabilization time. These records are indexed by battery chamber number and date. Raw monitoring data is retained for thirty days, analysis results for ninety days, and operation logs are permanently stored. Archived data is used to periodically generate performance reports, showing the correlation trend between nozzle parameter adjustment frequency and cooling effect. Self-inspection includes infrared thermal imager lens cleanliness detection, determining whether cleaning is needed by analyzing background thermal images; nozzle blockage detection, comparing the difference between commanded flow rate and actual flow rate; and communication link integrity testing, sending test messages to verify the response of each node. Issues discovered during self-inspection are prioritized and processed accordingly; minor issues that do not affect the normal control process are delayed until the maintenance window for repair.

[0109] This implementation achieves quantitative evaluation of cooling effect through high-resolution temperature monitoring, and improves parameter adjustment efficiency through an incremental search mechanism. A step-by-step loading strategy avoids the risk of parameter abrupt changes, and dynamic termination conditions balance the control effect with energy consumption costs. The entire closed-loop system forms a continuously optimizing feedback mechanism, adapting to dynamic changes in battery operating status.

[0110] 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.

[0111] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A smart phase change immersion cooling method based on dual-cavity dynamic control, characterized in that, The method includes: Perform sensor data acquisition to obtain battery temperature monitoring data sets and internal resistance monitoring data sets for high internal resistance battery chambers and low internal resistance battery chambers; Based on the battery temperature monitoring data set and the internal resistance monitoring data set, battery state analysis is performed to determine the high internal resistance state factor and the low internal resistance state factor. Using the high internal resistance state factor and the low internal resistance state factor as indexes, a centralized search is performed in the nozzle parameter configuration space to determine the target nozzle operating parameters. Based on the target nozzle operating parameters, adjust the nozzle operating states of the high internal resistance battery cavity and the low internal resistance battery cavity to perform phase change immersion cooling operation; Periodically perform battery capacity calibration to obtain a calibration result set. When the calibration result set meets the migration conditions, trigger a battery migration determination reminder. The battery state analysis includes: Data fluctuation analysis was performed on the battery temperature monitoring data set to obtain temperature fluctuation indicators; Data stability analysis is performed on the internal resistance monitoring data set to obtain internal resistance stability index; Based on the temperature fluctuation index and the internal resistance stability index, calculate the high internal resistance state factor and the low internal resistance state factor. The temperature fluctuation index and internal resistance stability index of the high internal resistance battery cavity are input into a dedicated calculation unit. The dedicated calculation unit contains dynamically adjusted weighting coefficients. The dedicated calculation unit performs linear weighted fusion and outputs a high internal resistance state factor in the 0-1 range. The low internal resistance battery cavity uses an independent calculation unit and generates a low internal resistance state factor using the same algorithm.

2. The intelligent phase change immersion cooling method based on dual-cavity dynamic control as described in claim 1, characterized in that, The sensor data acquisition process includes: Collect battery temperature monitoring data and internal resistance monitoring data of the high internal resistance battery cavity, and battery temperature monitoring data and internal resistance monitoring data of the low internal resistance battery cavity. The battery temperature monitoring data of the high internal resistance battery cavity and the battery temperature monitoring data of the low internal resistance battery cavity are integrated to generate the battery temperature monitoring data set. The internal resistance monitoring data of the high internal resistance battery cavity and the internal resistance monitoring data of the low internal resistance battery cavity are integrated to generate the internal resistance monitoring data set.

3. The intelligent phase change immersion cooling method based on dual-cavity dynamic control as described in claim 1, characterized in that, The data fluctuation analysis includes: Extract time-series data points from the battery temperature monitoring dataset; Calculate the rate of change between adjacent data points; The average of all rates of change is used as the temperature fluctuation index.

4. The intelligent phase change immersion cooling method based on dual-cavity dynamic control as described in claim 1, characterized in that, The centralized search in the nozzle parameter configuration space includes: Obtain the high internal resistance state factors and low internal resistance state factors of multiple samples, as well as the corresponding nozzle operating parameters of multiple samples. A two-dimensional configuration space is pre-constructed, where the first dimension represents the range of high internal resistance state factors and the second dimension represents the range of low internal resistance state factors. The multiple high internal resistance state factors and the multiple low internal resistance state factors of the samples are mapped to the two-dimensional configuration space to form multiple sample space points; The multiple sample nozzle operating parameters are used to identify the multiple sample space points, thereby generating the nozzle parameter configuration space.

5. The intelligent phase change immersion cooling method based on dual-cavity dynamic control as described in claim 4, characterized in that, The centralized search in the nozzle parameter configuration space also includes: The high internal resistance state factor and the low internal resistance state factor are input into the two-dimensional configuration space as target points; Calculate the distance metric between the target point and each point in the sample space; Select the sample space point with the smallest distance metric as the matching point; The working parameters of the sample nozzle corresponding to the matching point are used as the working parameters of the target nozzle.

6. The intelligent phase change immersion cooling method based on dual-cavity dynamic control as described in claim 1, characterized in that, The periodic battery capacity calibration includes: The battery capacity, internal resistance, and temperature rise sensitivity of the high internal resistance battery cavity and the low internal resistance battery cavity are measured quarterly. The calibration result set is generated by integrating the battery capacity value, the internal resistance value, and the temperature rise sensitivity value. Determine whether the calibration result set meets the migration conditions, wherein the migration conditions include a first preset percentage of battery capacity decay to the initial value of battery capacity, a second preset percentage of internal resistance increase to the initial value of internal resistance, or a temperature rise sensitivity greater than a preset threshold.

7. The intelligent phase change immersion cooling method based on dual-cavity dynamic control as described in claim 6, characterized in that, The migration determination reminder for the trigger battery includes: When the calibration result set satisfies the migration condition, a migration signal is generated; Based on the migration signal, a migration command is output to prompt the staff to perform the battery migration and replacement operation.

8. The intelligent phase change immersion cooling method based on dual-cavity dynamic control as described in claim 1, characterized in that, The method further includes: Based on the high internal resistance state factor and the low internal resistance state factor, predict the trend of battery thermal behavior. The operating parameters of the target nozzle are adjusted based on the battery thermal behavior trend.

9. The intelligent phase change immersion cooling method based on dual-cavity dynamic control as described in claim 1, characterized in that, After performing the phase change immersion cooling operation, the following is included: Monitor the cooling effect data of the high internal resistance battery cavity and the low internal resistance battery cavity; Based on the cooling effect data, update the battery temperature monitoring data set and the internal resistance monitoring data set; Repeat the steps of performing battery state analysis, performing centralized search in the nozzle parameter configuration space, adjusting the nozzle working state of the high internal resistance battery cavity and the low internal resistance battery cavity based on the target nozzle working parameters, and performing phase change immersion cooling operation.

Images