Energy distribution and scheduling control method and system for liquid-cooled energy storage system

Abstract

The application provides a liquid-cooled energy storage system energy distribution and scheduling control method and system, relates to the technical field of energy management, and comprises the following steps: collecting battery module working parameters, calculating a health state index, performing performance matching and cluster analysis to obtain an optimal grouping scheme; combining power grid load demand and power generation power to calculate target charging and discharging power; using a graph neural network to construct a topological relationship, analyze thermal coupling effects and electrical characteristics, predict temperature changes and performance degradation of different power distribution schemes, and determine an optimal power distribution scheme. The application improves energy utilization efficiency of the energy storage system, prolongs the service life of the battery, and guarantees system operation stability.

Description

Technical Field

[0001] This invention relates to the field of energy management technology, and in particular to a method and system for energy distribution and scheduling control of a liquid-cooled energy storage system. Background Technology

[0002] With the development of new energy sources and the transformation of the energy structure, the load fluctuations and peak-shaving demands faced by the power grid are increasing. Liquid-cooled energy storage systems play an important role in grid stability and the integration of renewable energy due to their advantages such as high heat dissipation efficiency and good safety.

[0003] Liquid-cooled energy storage systems typically consist of multiple battery modules, each with different performance characteristics and health status. During system operation, energy distribution and scheduling control issues need to be addressed to ensure safe and efficient system operation.

[0004] Existing liquid-cooled energy storage systems still suffer from limitations in energy distribution and scheduling control technologies. These limitations include the inability to accurately allocate energy based on the actual state of battery modules, resulting in poor-performing battery modules becoming bottlenecks that restrict the overall system's charging and discharging capabilities and lifespan. Furthermore, the lack of effective consideration of the thermal coupling effect between battery modules and the failure to organically combine thermal management with power management lead to uneven system temperature distribution, accelerating the performance degradation of some battery modules. Additionally, the inability to perform predictive control based on changes in grid load demand and fluctuations in power generation makes it difficult to achieve an optimal balance between system performance and lifespan, thus failing to meet the grid's requirements for rapid response and stable operation of energy storage systems.

[0005] Therefore, a solution is urgently needed to address the problems existing in the current technology. Summary of the Invention

[0006] This invention provides a method and system for energy distribution and scheduling control of a liquid-cooled energy storage system, which can at least solve some of the problems existing in the prior art.

[0007] A first aspect of this invention provides a method for energy distribution and scheduling control of a liquid-cooled energy storage system, comprising:

[0008] Collect the operating parameters of the battery modules in the liquid-cooled energy storage system, and calculate the health status index of the battery modules based on the operating parameters;

[0009] The performance matching index between different battery modules is calculated based on the health status index. The feature vectors corresponding to the battery modules are constructed based on the performance matching index and cluster analysis is performed to obtain the initial grouping scheme. The battery modules are grouped based on the initial grouping scheme and the feature vectors of the battery modules in each group are calculated. It is detected whether the feature vectors deviate from the cluster center of the group by more than a preset threshold. If so, the cluster analysis and grouping are repeated until the grouping target is met and the optimal grouping scheme is obtained.

[0010] Acquire grid load demand data and power generation data, and calculate the target charging and discharging power;

[0011] Based on the optimal grouping scheme, the battery module performance parameters are collected, the topological relationship between battery modules is constructed through graph neural network, the thermal coupling effect and electrical characteristic influence between different groups are analyzed in combination with the target charge and discharge power, and the temperature change trend and performance degradation rate of battery modules under different preset power allocation schemes are predicted in combination with historical performance parameters. The optimal power allocation scheme is then determined and allocated to the battery modules.

[0012] The charging and discharging operation is performed based on the optimal power allocation scheme, and the working status of the battery module is monitored in real time. If there are abnormal fluctuations, the system switches to standby mode.

[0013] In one alternative implementation,

[0014] The operating parameters of the battery modules in the liquid-cooled energy storage system are collected, and the health status index of the battery modules is calculated based on the operating parameters, including:

[0015] The operating parameters corresponding to the battery module are collected, including the individual cell voltage and the total module voltage collected by a voltage sensor set on the surface of the battery module, the charging and discharging current collected by a current sensor, and the number of charging and discharging cycles recorded.

[0016] Based on the collected working parameters, the internal resistance change rate is obtained by calculating the ratio of voltage change to current change per unit time, the voltage change rate is obtained by calculating the voltage change per unit time, and the capacity decay rate is obtained by calculating the ratio of current capacity to rated capacity after charge and discharge cycles.

[0017] The health status index is obtained by weighting the internal resistance change rate, voltage change rate, and capacity decay rate according to preset weighting coefficients.

[0018] In one alternative implementation,

[0019] Based on the health status index, a performance matching index is calculated between different battery modules. Based on the performance matching index, feature vectors corresponding to the battery modules are constructed and cluster analysis is performed to obtain an initial grouping scheme. Based on the initial grouping scheme, the battery modules are grouped, and the feature vector of each battery module within each group is calculated. It is then checked whether the feature vector deviates from the cluster center of the group by more than a preset threshold. If so, the cluster analysis and grouping are repeated until the grouping objective is met, resulting in the optimal grouping scheme, which includes:

[0020] Based on the health status index of the battery module, the performance matching index is obtained by calculating the ratio of the difference between the health status indices of the two battery modules to the maximum value of the health status indices of the two battery modules. The performance matching index and the health status index of the battery module are combined to construct the feature vector corresponding to the battery module.

[0021] The K-means clustering algorithm is used to perform cluster analysis on the feature vectors to obtain an initial grouping scheme, and the cluster center corresponding to each initial grouping scheme is calculated;

[0022] Calculate the Euclidean distance between the feature vector of the battery module in each initial grouping scheme and the cluster center, assign weight coefficients based on the Euclidean distance and re-perform cluster analysis until the Euclidean distance is less than or equal to the preset threshold;

[0023] Calculate the intra-group consistency index within the same group and the inter-group difference index between different groups. Based on the intra-group consistency index and the inter-group difference index, dynamically adjust the number of clusters and introduce a group number balance constraint to obtain the optimal grouping scheme.

[0024] In one alternative implementation,

[0025] The initial grouping scheme obtained by performing cluster analysis on the feature vectors using the K-means clustering algorithm includes:

[0026] Obtain the feature vector corresponding to the battery module, calculate the ratio of the difference between the feature vector minus the minimum value of the feature vector and the difference between the maximum and minimum values ​​of the feature vector, generate a standardized feature vector, calculate the performance matching index difference and health status index difference between any two battery modules based on the standardized feature vector, substitute the difference into the exponential function and assign weight coefficients, and then sum them up by weight to generate an electrochemical characteristic similarity matrix.

[0027] Randomly select the first initial cluster center from all battery modules, calculate the minimum distance from the remaining battery modules to the initial cluster center, construct a probability distribution based on the minimum distance, select an optimized cluster center, and repeat the process until a preset number of times.

[0028] Construct a network flow graph and set source and sink points. Set battery modules as vertices and connect adjacent battery modules. Calculate the weight of the edges. The weight of the edges is calculated by weighting the elements of the electrochemical characteristic similarity matrix with an exponential function of the distance from the sample to the optimized cluster center.

[0029] Set an initial search radius and substitute it into the exponential decay function. Summing the radius with the minimum radius yields the dynamic neighborhood radius. Allocate battery modules based on the dynamic neighborhood radius. Calculate the average value of the feature vectors of battery modules within a class to obtain the cluster center. Determine whether the change in the cluster center and the maximum difference in the performance matching index within the class meet the preset threshold and balance requirements. If they meet, output the clustering result.

[0030] In one alternative implementation,

[0031] Based on the optimal grouping scheme, battery module performance parameters are collected. A graph neural network is used to construct the topological relationships between battery modules. Combined with the target charge / discharge power analysis, the thermal coupling effect and electrical characteristic impact between different groups are examined. Historical performance parameters are used to predict the temperature change trend and performance degradation rate of battery modules under different preset power allocation schemes. The optimal power allocation scheme is then determined and assigned to the battery modules, including:

[0032] The performance parameters of the battery module are obtained and the battery module is set as a node in a graph neural network. The topological relationship between the battery modules is constructed to obtain an adjacency matrix. The performance parameters are written into a feature matrix and the topological features of the battery module are extracted.

[0033] A heat conduction equation is established based on the target charge and discharge power. The thermal coupling effect and electrical characteristic influence between different groups are analyzed, and the temperature field distribution of the battery module is calculated.

[0034] The historical performance parameters of the battery module are obtained, and the historical performance parameters are fused with the current state parameters. Based on the fused data, the temperature change trend and performance degradation rate of the battery module are calculated.

[0035] Based on the temperature field distribution, the temperature gradient and hot spot regions of the battery module are determined. Combining the topological features, and under the constraints of voltage, temperature and rate limits, the optimal power allocation scheme is iteratively solved by combining the temperature change trend and performance degradation rate, and the optimal power allocation scheme is allocated to the battery module.

[0036] In one alternative implementation,

[0037] A heat conduction equation is established based on the target charge and discharge power. The pre-acquired thermal characteristic parameters of the battery module are then substituted into the heat conduction equation to analyze the thermal coupling effect and electrical characteristic influence between different groups. The temperature field distribution of the battery module is calculated, including:

[0038] A phase change material is filled into the gaps between the battery modules. A heat conduction equation is established based on the target charge and discharge power of the battery modules and the thermal management characteristics of the phase change material. The phase change temperature of the phase change material is used as the reference temperature for the heat conduction equation.

[0039] The composite thermal characteristic parameters of the battery module and the phase change material are obtained, the battery module is set as a heat conduction node, the equivalent thermal conductivity between adjacent nodes is calculated, and a thermal characteristic parameter matrix is ​​constructed.

[0040] The temperature data and phase state data of the phase change material are collected in real time. The temperature data is compared with the phase change temperature to calculate the heat storage state of the phase change material. The heat storage efficiency is calculated based on the sensible heat storage and latent heat storage of the phase change material.

[0041] Substitute the thermal characteristic parameter matrix into the heat conduction equation, set boundary conditions based on the thermal storage efficiency, and calculate the transient temperature field distribution of the battery module.

[0042] In one alternative implementation,

[0043] Charging and discharging operations are performed based on the optimal power allocation scheme, and the battery module's operating status is monitored in real time. If abnormal fluctuations are detected, the system switches to standby mode, including:

[0044] The system obtains the optimal power allocation scheme to control the battery module to perform charging and discharging operations, and collects the temperature, current and voltage data of the battery module in real time.

[0045] The fluctuation values ​​of the temperature data, current data, and voltage data are compared with preset thresholds. When the fluctuation value of any parameter exceeds the corresponding preset threshold, a status abnormality signal is generated.

[0046] In response to the abnormal status signal, the battery module is switched to a pre-set standby operating state, where the power of the standby operating state is less than the power corresponding to the optimal power allocation scheme.

[0047] A second aspect of the present invention provides a system comprising:

[0048] The first unit is used to collect the operating parameters of the battery modules in the liquid-cooled energy storage system and calculate the health status index of the battery modules based on the operating parameters.

[0049] The second unit is used to calculate the performance matching index between different battery modules based on the health status index, construct the feature vector corresponding to the battery module based on the performance matching index and perform cluster analysis to obtain the initial grouping scheme, group the battery modules based on the initial grouping scheme and calculate the feature vector of the battery module in each group, detect whether the feature vector deviates from the cluster center of the group by more than a preset threshold, if so, repeat the cluster analysis and grouping until the grouping target is met and the optimal grouping scheme is obtained.

[0050] The third unit is used to acquire grid load demand data and power generation data and calculate the target charging and discharging power.

[0051] The fourth unit is used to collect battery module performance parameters based on the optimal grouping scheme, construct the topological relationship between battery modules through graph neural network, analyze the thermal coupling effect and electrical characteristics between different groups in combination with the target charge and discharge power, predict the temperature change trend and performance degradation rate of battery modules under different preset power allocation schemes in combination with historical performance parameters, determine the optimal power allocation scheme and allocate it to the battery modules.

[0052] The fifth unit is used to perform charging and discharging operations based on the optimal power allocation scheme, and to detect the working status of the battery module in real time. If there are abnormal fluctuations, it will switch to standby mode.

[0053] A third aspect of the present invention provides an electronic device, comprising:

[0054] A processor and a memory for storing processor-executable instructions, wherein the processor is configured to invoke instructions stored in the memory to perform the aforementioned method.

[0055] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0056] In this invention, the optimal grouping scheme is obtained by calculating the performance matching index between battery modules based on the health status index and performing cluster analysis. This achieves scientific grouping of battery modules with different performance states, avoiding the problem of overall efficiency reduction due to the performance degradation of a single module, and improving the overall service life of the system. By combining graph neural networks to construct the topological relationship between battery modules, the effects of thermal coupling and electrical characteristics are analyzed, making power allocation more accurate and reasonable, reducing hot spot problems and uneven aging, and ensuring long-term stable operation of the system. By real-time detection of the working status of battery modules and prediction of temperature change trends and performance degradation rates under different power allocation schemes, intelligent dynamic scheduling of the system is realized, improving the energy utilization efficiency and response speed of the energy storage system, and enhancing the system's adaptability to grid fluctuations. Attached Figure Description

[0057] Figure 1 This is a flowchart illustrating the energy distribution and scheduling control method of the liquid-cooled energy storage system according to an embodiment of the present invention;

[0058] Figure 2 This is a comparison chart showing the similarity between battery module clustering results and electrochemical characteristics in the energy distribution and scheduling control method of the liquid-cooled energy storage system according to an embodiment of the present invention.

[0059] Figure 3 The figures show experimental data and simulation results of the battery module phase change material thermal management technology in the liquid-cooled energy storage system energy distribution and scheduling control method according to an embodiment of the present invention. Detailed Implementation

[0060] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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.

[0061] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0062] Figure 1 This is a flowchart illustrating the energy distribution and scheduling control method for a liquid-cooled energy storage system according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0063] Collect the operating parameters of the battery modules in the liquid-cooled energy storage system, and calculate the health status index of the battery modules based on the operating parameters;

[0064] The performance matching index between different battery modules is calculated based on the health status index. The feature vectors corresponding to the battery modules are constructed based on the performance matching index and cluster analysis is performed to obtain the initial grouping scheme. The battery modules are grouped based on the initial grouping scheme and the feature vectors of the battery modules in each group are calculated. It is detected whether the feature vectors deviate from the cluster center of the group by more than a preset threshold. If so, the cluster analysis and grouping are repeated until the grouping target is met and the optimal grouping scheme is obtained.

[0065] Acquire grid load demand data and power generation data, and calculate the target charging and discharging power;

[0066] Based on the optimal grouping scheme, the battery module performance parameters are collected, the topological relationship between battery modules is constructed through graph neural network, the thermal coupling effect and electrical characteristic influence between different groups are analyzed in combination with the target charge and discharge power, and the temperature change trend and performance degradation rate of battery modules under different preset power allocation schemes are predicted in combination with historical performance parameters. The optimal power allocation scheme is then determined and allocated to the battery modules.

[0067] The charging and discharging operation is performed based on the optimal power allocation scheme, and the working status of the battery module is monitored in real time. If there are abnormal fluctuations, the system switches to standby mode.

[0068] In one alternative implementation,

[0069] The operating parameters of the battery modules in the liquid-cooled energy storage system are collected, and the health status index of the battery modules is calculated based on the operating parameters, including:

[0070] The operating parameters corresponding to the battery module are collected, including the individual cell voltage and the total module voltage collected by a voltage sensor set on the surface of the battery module, the charging and discharging current collected by a current sensor, and the number of charging and discharging cycles recorded.

[0071] Based on the collected working parameters, the internal resistance change rate is obtained by calculating the ratio of voltage change to current change per unit time, the voltage change rate is obtained by calculating the voltage change per unit time, and the capacity decay rate is obtained by calculating the ratio of current capacity to rated capacity after charge and discharge cycles.

[0072] The health status index is obtained by weighting the internal resistance change rate, voltage change rate, and capacity decay rate according to preset weighting coefficients.

[0073] The system collects the operating parameters of the battery module by installing multiple voltage sensors on its surface to collect individual cell voltages and the total module voltage. For example, for a battery module consisting of 12 cells connected in series, a voltage sensor is installed between the positive and negative terminals of each cell to collect voltage data. Simultaneously, a current sensor is installed at the output of the battery module to collect charge and discharge current data. The battery management system records the number of charge and discharge cycles of the battery module. The data acquisition frequency can be set to once per second to ensure real-time data accuracy. The collected data is transmitted to a data processing unit for processing and analysis via a dedicated communication interface.

[0074] Based on the collected operating parameters, the ratio of voltage change to current change per unit time is calculated to obtain the internal resistance change rate. Two adjacent sampling times, t1 and t2, are selected, and the voltage values ​​V1 and V2, and the current values ​​I1 and I2 of the battery module are recorded at these two times, respectively. The voltage change ΔV = V2 - V1 and the current change ΔI = I2 - I1 are calculated, and the internal resistance value R in this sampling interval is R = ΔV / ΔI. By comparing with the initial internal resistance value R0, the internal resistance change rate ΔR = (R - R0) / R0 is calculated. For example, if the initial internal resistance value is 20 milliohms and the currently measured internal resistance value is 24 milliohms, then the internal resistance change rate is (24 - 20) / 20 = 0.2, which means an increase of 20%.

[0075] The voltage change rate is obtained by calculating the voltage change per unit time. Select two moments t1 and t2 during a charging or discharging process, and record the battery module's voltage values ​​V1 and V2 at these two moments. Calculate the voltage change ΔV = V2 - V1, and the time interval Δt = t2 - t1, then the voltage change rate is ΔV / Δt. The voltage stability of the battery module is evaluated by comparing it with a standard voltage change rate. For example, under standard discharge conditions, if the battery module's voltage drops from 4.2V to 3.8V in one hour, the voltage change is 0.4V, and the standard voltage change rate is 0.4V / hour. If the actual measured voltage change rate is 0.5V / hour, then the voltage change rate deviation is (0.5 - 0.4) / 0.4 = 0.25, which is 25% deviation from the standard value.

[0076] In addition, the capacity decay rate is obtained by calculating the ratio of the current capacity to the rated capacity after a certain number of charge-discharge cycles. After the battery module completes one full charge-discharge cycle, the actual discharge capacity Cn during that cycle is recorded and compared with the rated capacity C0 of the battery module to calculate the capacity decay rate ΔC = (C0 - Cn) / C0. For example, if the rated capacity of the battery module is 100 amp-hours, and after 200 cycles the actual measured discharge capacity is 90 amp-hours, then the capacity decay rate is (100 - 90) / 100 = 0.1, meaning the capacity has decreased by 10%.

[0077] Based on the calculated internal resistance change rate, voltage change rate, and capacity decay rate, a health status index is obtained by weighting these factors according to preset weighting coefficients. In this embodiment, the weighting coefficient for the internal resistance change rate is set to 0.3, the weighting coefficient for the voltage change rate is 0.2, and the weighting coefficient for the capacity decay rate is 0.5. When calculating the health status index, the internal resistance change rate, the voltage change rate, and the capacity decay rate are multiplied by their respective weighting coefficients. The products of these three factors are then added together to obtain a comprehensive decay rate index. Subtracting this index from 1 yields the health status index.

[0078] For example, if the internal resistance change rate of a battery module is 0.2, the voltage change rate deviation is 0.25, and the capacity decay rate is 0.1, then the comprehensive decay rate index is 0.2×0.3+0.25×0.2+0.1×0.5=0.06+0.05+0.05=0.16, and the health status index is 1-0.16=0.84, that is, the health status of the battery module is maintained at 84%.

[0079] In this embodiment, voltage and current sensors are arranged on the surface of the battery module to collect operating parameters such as individual cell voltage, total module voltage, and charge / discharge current in real time. Combined with the cumulative record of charge / discharge cycles, a complete parameter acquisition network is established. By calculating the dynamic change characteristics of voltage and current per unit time, two key indicators, internal resistance change rate and voltage change rate, are obtained. These indicators can sensitively reflect the dynamic degradation process of battery performance. The capacity decay rate is calculated by the ratio of the current capacity to the rated capacity, which effectively characterizes the degree of degradation of the battery's energy storage capacity. This significantly improves the accuracy and timeliness of battery health status diagnosis and provides strong support for the optimization decision-making of the battery management system.

[0080] In one alternative implementation,

[0081] Based on the health status index, a performance matching index is calculated between different battery modules. Based on the performance matching index, feature vectors corresponding to the battery modules are constructed and cluster analysis is performed to obtain an initial grouping scheme. Based on the initial grouping scheme, the battery modules are grouped, and the feature vector of each battery module within each group is calculated. It is then checked whether the feature vector deviates from the cluster center of the group by more than a preset threshold. If so, the cluster analysis and grouping are repeated until the grouping objective is met, resulting in the optimal grouping scheme, which includes:

[0082] Based on the health status index of the battery module, the performance matching index is obtained by calculating the ratio of the difference between the health status indices of the two battery modules to the maximum value of the health status indices of the two battery modules. The performance matching index and the health status index of the battery module are combined to construct the feature vector corresponding to the battery module.

[0083] The K-means clustering algorithm is used to perform cluster analysis on the feature vectors to obtain an initial grouping scheme, and the cluster center corresponding to each initial grouping scheme is calculated;

[0084] Calculate the Euclidean distance between the feature vector of the battery module in each initial grouping scheme and the cluster center, assign weight coefficients based on the Euclidean distance and re-perform cluster analysis until the Euclidean distance is less than or equal to the preset threshold;

[0085] Calculate the intra-group consistency index within the same group and the inter-group difference index between different groups. Based on the intra-group consistency index and the inter-group difference index, dynamically adjust the number of clusters and introduce a group number balance constraint to obtain the optimal grouping scheme.

[0086] Obtain the health status index of multiple battery modules to be grouped, which represents the current health level of the battery modules. For example, a batch has 20 battery modules, and their health status indices are: 0.92, 0.89, 0.85, 0.95, 0.88, 0.79, 0.93, 0.81, 0.87, 0.90, 0.84, 0.78, 0.92, 0.86, 0.91, 0.83, 0.89, 0.82, 0.94, and 0.80.

[0087] Based on the acquired health status index, a performance matching index is calculated between different battery modules. The performance matching index is calculated as the ratio of the difference between the health status indices of two battery modules to the maximum value of their respective health status indices. For example, for two battery modules with health status indices of 0.92 and 0.89, their performance matching index is (0.92-0.89) / 0.92 = 0.033. Through similar calculations, the performance matching index between any two battery modules can be obtained.

[0088] The performance matching index and health status index of the battery module are combined to construct a feature vector. For each battery module, an n-dimensional feature vector is constructed, where n is the total number of battery modules. The first element in the feature vector is the health status index of the battery module, and the remaining n-1 elements are the performance matching indices of the battery module with the other n-1 battery modules. For example, for a battery module with a health status index of 0.92, its feature vector is [0.92, 0.033, 0.076, 0.032, 0.043, 0.141, 0.011, 0.120, 0.054, 0.022, 0.087, 0.152, 0, 0.065, 0.011, 0.098, 0.033, 0.109, 0.022, 0.130].

[0089] The K-means clustering algorithm is used to cluster the feature vectors to obtain an initial grouping scheme. Assuming the battery modules need to be divided into 4 groups, the K-means algorithm is used to cluster the feature vectors. The initial cluster centers can be randomly selected or selected based on a certain strategy. After iterative calculation, the final clustering result, i.e., the initial grouping scheme, is obtained. For example, the initial grouping results are: Group 1 contains modules 1, 7, 13, 15, and 19; Group 2 contains modules 4, 10, and 16; Group 3 contains modules 2, 5, 9, and 17; and Group 4 contains modules 3, 6, 8, 11, 12, 14, 18, and 20.

[0090] Calculate the cluster center for each group, which is the mean of the feature vectors within the same group. For example, the cluster centers for group 1 are [0.924, 0.034, 0.075, 0.032, 0.044, 0.140, 0.012, 0.121, 0.054, 0.023, 0.088, 0.153, 0.002, 0.066, 0.012, 0.097, 0.034, 0.110, 0.003, 0.131].

[0091] The battery modules are grouped based on the initial grouping scheme, and the Euclidean distance between the feature vector of each battery module in each group and the cluster center is calculated. The Euclidean distance is obtained by taking the square root of the sum of the squares of the differences in each dimension of the two vectors. A preset threshold of 0.05 is set; when the Euclidean distance is greater than 0.05, it indicates that the matching degree between the battery module and its group is not ideal.

[0092] The detection revealed that the Euclidean distance between the feature vector of module 9 in group 3 and the cluster center was 0.058, exceeding the preset threshold of 0.05. Therefore, the health status index in the feature vector was assigned a higher weight coefficient than the performance matching index. For example, the weight of the health status index was set to 2, while the weight of the performance matching index remained at 1. The weighted feature vector was then used for re-clustering analysis.

[0093] After re-clustering, module 9 was moved to group 2. The Euclidean distance was calculated again, and it was found that the Euclidean distance between the feature vector of all battery modules and the cluster center of their respective groups was less than the preset threshold of 0.05, which met the grouping requirements.

[0094] Calculate the mean performance matching index among battery modules within the same group to determine the intra-group consistency index for each group. For example, the intra-group consistency index for group 1 is 0.018, for group 2 it is 0.022, for group 3 it is 0.015, and for group 4 it is 0.027. The smaller the intra-group consistency index, the more consistent the performance of the battery modules within that group.

[0095] Calculate the minimum difference between cluster centers of different groups to determine the inter-group difference index. For example, if the Euclidean distances between group 1 and the cluster centers of groups 2, 3, and 4 are 0.15, 0.12, and 0.21 respectively, then the inter-group difference index for group 1 is 0.12; similarly, the inter-group difference indices for other groups are calculated. The larger the inter-group difference index, the more significant the difference between that group and other groups.

[0096] Based on the intra-group consistency index and the inter-group difference index, the number of clusters is dynamically adjusted and a constraint on the balance of the number of groups is introduced. If the intra-group consistency index of a group is too high or the inter-group difference index is too low, it may be necessary to increase the number of clusters; if the number of battery modules in some groups varies too much, the grouping scheme needs to be adjusted to achieve a more balanced distribution.

[0097] Through multiple iterative adjustments, the optimal grouping scheme was determined as follows: Group 1 includes modules 1, 7, 13, 15, and 19; Group 2 includes modules 4, 9, 10, and 16; Group 3 includes modules 2, 5, and 17; and Group 4 includes modules 3, 6, 8, 11, 12, 14, 18, and 20.

[0098] In this embodiment, by standardizing the differences in health status index into a performance matching index, a quantitative evaluation standard for performance differences between battery modules is established, providing a more accurate performance matching metric. An adaptive weight adjustment mechanism is adopted to dynamically adjust the weight of the health status index according to Euclidean distance during the clustering process, ensuring that the grouping results focus more on the consistency of battery performance. Through the comprehensive evaluation of intra-group consistency and inter-group difference indicators, adaptive optimization of the number of clusters is achieved, balancing the number distribution of each group while ensuring the quality of grouping.

[0099] In one alternative implementation,

[0100] The initial grouping scheme obtained by performing cluster analysis on the feature vectors using the K-means clustering algorithm includes:

[0101] Obtain the feature vector corresponding to the battery module, calculate the ratio of the difference between the feature vector minus the minimum value of the feature vector and the difference between the maximum and minimum values ​​of the feature vector, generate a standardized feature vector, calculate the performance matching index difference and health status index difference between any two battery modules based on the standardized feature vector, substitute the difference into the exponential function and assign weight coefficients, and then sum them up by weight to generate an electrochemical characteristic similarity matrix.

[0102] Randomly select the first initial cluster center from all battery modules, calculate the minimum distance from the remaining battery modules to the initial cluster center, construct a probability distribution based on the minimum distance, select an optimized cluster center, and repeat the process until a preset number of times.

[0103] Construct a network flow graph and set source and sink points. Set battery modules as vertices and connect adjacent battery modules. Calculate the weight of the edges. The weight of the edges is calculated by weighting the elements of the electrochemical characteristic similarity matrix with an exponential function of the distance from the sample to the optimized cluster center.

[0104] Set an initial search radius and substitute it into the exponential decay function. Summing the radius with the minimum radius yields the dynamic neighborhood radius. Allocate battery modules based on the dynamic neighborhood radius. Calculate the average value of the feature vectors of battery modules within a class to obtain the cluster center. Determine whether the change in the cluster center and the maximum difference in the performance matching index within the class meet the preset threshold and balance requirements. If they meet, output the clustering result.

[0105] The feature vector corresponding to the battery module is obtained. The feature vector contains two indicators: performance matching index and health status index. By standardizing the feature vector, the data is scaled to the interval [0, 1] to eliminate the influence of differences in the dimensions of different features. The standardization calculation method is to subtract the minimum value of the feature vector and then divide it by the difference between the maximum and minimum values. For example, for the feature vector of a certain battery module [85.2, 0.92], assuming that the performance matching index range is [80, 100] and the health status index range is [0.8, 1.0], then the standardized feature vector is [(85.2-80) / (100-80), (0.92-0.8) / (1.0-0.8)] = [0.26, 0.6].

[0106] Based on standardized feature vectors, the performance matching index difference and health status index difference between any two battery modules are calculated. For standardized battery module A [0.26, 0.6] and battery module B [0.35, 0.55], the calculated difference is [|0.26-0.35|, |0.6-0.55|] = [0.09, 0.05]. Substituting these differences into the exponential function exp(-difference) and assigning weight coefficients, the results are weighted and summed to generate an electrochemical characteristic similarity matrix. Assuming the weight coefficients are 0.6 and 0.4, the similarity between battery modules A and B is 0.6×exp(-0.09) + 0.4×exp(-0.05) = 0.6×0.914 + 0.4×0.951 = 0.929. In this way, the similarity between all battery module pairs is calculated, forming an electrochemical characteristic similarity matrix.

[0107] For initial cluster center selection, the K-means++ algorithm is used. The first initial cluster center is randomly selected from all battery modules. The minimum distance from the remaining battery modules to the selected cluster center is calculated. A probability distribution is constructed based on these minimum distances; the larger the distance, the higher the probability that the battery module will be selected as the next cluster center, thus selecting optimized cluster centers. This process is repeated until a preset number of cluster centers are selected. For example, for a system containing 200 battery modules, if the preset number of clusters is 4, then 4 well-distributed initial cluster centers are selected using the above method.

[0108] A network flow graph is constructed, with battery modules as vertices, and adjacent battery modules are connected to form edges. The edge weight is calculated as a weighted sum of the corresponding elements of the electrochemical characteristic similarity matrix and the exponential function of the distance from the sample to the optimized cluster center. For example, for the edge weight between battery modules A and B, if their similarity is 0.929, and the distance from A to its nearest cluster center is 0.15, and the distance from B to its nearest cluster center is 0.25, and the distance weight coefficient is 0.5, then the edge weight can be calculated as 0.929 + 0.5 × (exp(-0.15) + exp(-0.25)) / 2 = 0.929 + 0.5 × (0.861 + 0.779) / 2 = 1.149.

[0109] The initial search radius is set to 0.3, the decay coefficient to 0.05, and the minimum radius to 0.1. As iterations proceed, the initial search radius is updated using an exponential decay function: Dynamic neighborhood radius = Minimum radius + Initial search radius × exp(-decay coefficient × current iteration number). For example, in the 5th iteration, the dynamic neighborhood radius = 0.1 + 0.3 × exp(-0.05 × 5) = 0.1 + 0.3 × 0.779 = 0.334.

[0110] Battery modules are assigned based on their dynamic neighborhood radius. Each battery module is assigned to the nearest cluster center that is within the dynamic neighborhood radius. If the distance from a battery module to all cluster centers is greater than the dynamic neighborhood radius, it is assigned to the cluster with the smallest distance. For each cluster, the average of the standardized feature vectors of the battery modules within the cluster is calculated as the new cluster center. For example, if a cluster contains battery modules [0.26, 0.6], [0.35, 0.55], and [0.3, 0.58], then the new cluster center is [(0.26+0.35+0.3) / 3, (0.6+0.55+0.58) / 3] = [0.303, 0.577].

[0111] The clustering process checks whether the variation in cluster centers is less than a preset threshold (e.g., 0.01) and whether the maximum difference in performance matching indices within each cluster meets the balance requirements (e.g., not exceeding 0.15). When these conditions are met, the clustering results are output as the initial grouping scheme. For example, if four clusters are ultimately obtained, each containing approximately 50 battery modules, and the maximum difference in performance matching indices within each group does not exceed 0.15 after standardization, it indicates that the grouping results meet the battery balance requirements.

[0112] In this embodiment, the influence of dimensions is eliminated by standardizing the maximum and minimum values ​​of the feature vectors. An electrochemical characteristic similarity matrix is ​​constructed by using exponential function mapping and weighted summation. A cluster center optimization mechanism based on probability distribution is introduced to guide the selection of cluster centers through the minimum distance probability distribution.

[0113] In existing technologies, simple feature comparison or fixed threshold grouping is usually used, which ignores the complex electrochemical characteristics between battery modules. Furthermore, the random selection of cluster centers can easily lead to unstable grouping results. The lack of a dynamic adjustment mechanism for the grouping process makes it difficult to ensure the balance and reliability of the grouping results.

[0114] This embodiment uses battery modules as network nodes, reflects the electrochemical similarity between modules through edge weights, introduces a dynamic neighborhood mechanism, and dynamically adjusts the search radius through an exponential decay function to achieve adaptive optimization of the grouping process. It adopts dual constraints of cluster center change and intra-cluster performance matching exponential difference to ensure that the grouping results meet both performance balance requirements and good stability, which helps to extend the battery pack's lifespan and improve energy utilization efficiency. It can flexibly adjust parameters according to actual application needs to meet the grouping optimization requirements in different scenarios.

[0115] Figure 2 This is a comparison chart of the relationship between battery module clustering results and electrochemical characteristic similarity in the energy distribution and scheduling control method of the liquid-cooled energy storage system according to an embodiment of the present invention. The chart shows the distribution of the relationship between battery module clustering results and electrochemical characteristic similarity. The horizontal axis represents the electrochemical characteristic similarity value, and the vertical axis represents the frequency of battery pairs.

[0116] Figure 2 The study reveals significant differences in the similarity distribution between intra-group and inter-group battery pairs. The similarity of intra-group battery pairs primarily ranges from 0.85 to 0.97, with a mean similarity of 0.913 and a standard deviation of 0.042. In contrast, the similarity of inter-group battery pairs mainly ranges from 0.58 to 0.83, with a mean similarity of 0.742 and a standard deviation of 0.087. The data distribution shows that the frequency of intra-group battery pairs is only 0.008 at a similarity of 0.58, gradually increasing with increasing similarity, reaching a peak frequency of 0.283 at 0.93, then decreasing to 0.249 at 0.98, and finally dropping to 0.081 at 1.00. In contrast, the inter-group battery pairs were more concentrated in the low similarity region, with frequencies of 0.087, 0.146, 0.192, 0.247, 0.185 and 0.108 at similarity levels of 0.58, 0.63, 0.68, 0.73, 0.78 and 0.83, respectively, showing a trend of first increasing and then decreasing, reaching a peak frequency of 0.247 at a similarity level of 0.73.

[0117] The distribution is sparse in the high similarity region (0.88-1.00), with frequencies of 0.068, 0.042, 0.018, and 0.007, respectively. Within the similarity range greater than 0.90, intra-group battery pairs account for as high as 72.6%, while inter-group battery pairs account for only 3.5%. The distribution characteristics shown in the figure verify the clustering effect of this technical solution based on the electrochemical characteristic similarity matrix, grouping battery modules with similar electrochemical characteristics into the same group. Particularly in the similarity range of 0.88-0.92, the frequency of intra-group battery pairs reaches a peak of 0.283, far exceeding the frequency of inter-group battery pairs (0.042) in this range.

[0118] The obvious separation effect proves the effectiveness of this technical solution in battery module grouping. It can accurately identify the similarity between batteries based on two key characteristics: battery performance matching index and health status index, thereby forming battery packs with highly consistent electrochemical characteristics, laying the foundation for improving the overall performance and lifespan of the battery pack.

[0119] In one alternative implementation,

[0120] Based on the optimal grouping scheme, battery module performance parameters are collected. A graph neural network is used to construct the topological relationships between battery modules. Combined with the target charge / discharge power analysis, the thermal coupling effect and electrical characteristic impact between different groups are examined. Historical performance parameters are used to predict the temperature change trend and performance degradation rate of battery modules under different preset power allocation schemes. The optimal power allocation scheme is then determined and assigned to the battery modules, including:

[0121] The performance parameters of the battery module are obtained and the battery module is set as a node in a graph neural network. The topological relationship between the battery modules is constructed to obtain an adjacency matrix. The performance parameters are written into a feature matrix and the topological features of the battery module are extracted.

[0122] A heat conduction equation is established based on the target charge and discharge power. The thermal coupling effect and electrical characteristic influence between different groups are analyzed, and the temperature field distribution of the battery module is calculated.

[0123] The historical performance parameters of the battery module are obtained, and the historical performance parameters are fused with the current state parameters. Based on the fused data, the temperature change trend and performance degradation rate of the battery module are calculated.

[0124] Based on the temperature field distribution, the temperature gradient and hot spot regions of the battery module are determined. Combining the topological features, and under the constraints of voltage, temperature and rate limits, the optimal power allocation scheme is iteratively solved by combining the temperature change trend and performance degradation rate, and the optimal power allocation scheme is allocated to the battery module.

[0125] Battery module performance parameters, including key parameters such as voltage, current, temperature, internal resistance, and state of charge, are collected based on the optimal grouping scheme. After collection, normalization processing is performed to unify parameters with different dimensions into the [0, 1] interval. For example, if the battery module temperature range is 0℃ to 60℃ and the measured temperature is 25℃, the normalized temperature value is 0.417.

[0126] Each battery module is set as a node in a graph neural network. Topological relationships are constructed based on the physical connections and heat conduction paths between modules. If there are 10 battery modules, a 10×10 adjacency matrix is ​​generated, with a value of 1 for physically adjacent modules and 0 for non-adjacent modules. Simultaneously, normalized performance parameters are written into a feature matrix, with each row representing a battery module and each column representing a performance parameter. A three-layer graph convolutional network is used to convolve the feature matrix, with 16 convolutional kernels per layer and the ReLU activation function, to extract the topological features of the battery modules, resulting in a feature representation that includes the interrelationships between battery modules.

[0127] A heat conduction equation is established based on the target charge / discharge power, taking into account the heat generation power, thermal conductivity, and heat capacity of the battery module. The heat conduction equation can be expressed as follows: Where ρ is density, Cp is specific heat capacity, λ is thermal conductivity, T is temperature, t is time, and q is the power generated per unit volume. For a battery module, the power generated q can be calculated using Joule heating: q = I 2 R, where I is the charging / discharging current and R is the internal resistance. For example, when the charging power is set to 50kW, Joule heat is calculated based on the internal resistance of each module. A typical 18650 lithium battery module has an internal resistance of 25mΩ and generates approximately 2.5W of heat at a current of 10A. Pre-obtained thermal characteristic parameters are substituted into the heat conduction equation, including thermal conductivity of 3.4W / (m·K), specific heat capacity of 1100J / (kg·K), and density of 2200kg / m³. Numerical calculation methods, such as the finite difference method, are used to analyze the thermal coupling effect between modules, calculate the temperature field distribution, and obtain real-time temperature data for each module. For example, at a charging power of 50kW, the temperature of the central module can reach 43℃, while the temperature of the edge modules is approximately 38℃.

[0128] Historical performance parameters of the battery module over the past 100 charge-discharge cycles are obtained, including the highest temperature, average temperature, rate of change of state of charge, and rate of increase of internal resistance for each cycle. The historical data is fused with the current state parameters using a Kalman filter algorithm, with a filter gain of 0.7 to assign higher weight to the current state. Based on the fused data, a Long Short-Term Memory (LSTM) network is used to predict the temperature change trend for the next 10 cycles. The network structure contains 64 hidden units with a time step of 5. Simultaneously, the performance degradation rate is calculated based on a capacity degradation model. Under standard conditions, a typical lithium-ion battery experiences approximately 0.5% capacity degradation per 100 cycles, while under high-temperature conditions (above 45°C), the degradation rate can increase to 0.8%.

[0129] Based on the temperature field distribution, the temperature gradient and hotspot regions of the battery modules are determined, and the maximum temperature difference between adjacent modules is calculated. Under normal operating conditions, this difference should be controlled within 5℃. Considering the topological characteristics of the battery modules, a target temperature of 35℃ is set, and the sum of squares of the temperature differences between each module and the target temperature is calculated. During the optimization process, voltage limits are set to 2.5V-4.2V, temperature limits to 0℃-45℃, charging rate limits to 0.5C-2C, and discharging rate limits to 0.5C-3C. The gradient descent algorithm is used to iteratively calculate the power allocation scheme, with a learning rate of 0.01 and a maximum of 1000 iterations. The iteration is terminated early when the rate of change of the sum of squares of the differences is less than 0.001.

[0130] The optimal power allocation scheme distributes the total power to each battery module according to a specific ratio. For example, in a system with a total charging power of 50kW, the central module with a higher temperature is allocated a lower power, such as 4.2kW, while the edge modules with a lower temperature are allocated a higher power, such as 5.8kW, thereby balancing the system temperature distribution. After allocation, the charging and discharging current of each module is adjusted in real time by the power control unit of the battery management system to ensure that each module operates according to the allocation scheme.

[0131] In this embodiment, graph convolution operations are used to extract topological features, which not only captures the performance characteristics of individual battery modules, but also deeply analyzes the mutual influence between modules, providing a global perspective for subsequent power allocation optimization. By analyzing the thermal coupling effect and electrical characteristic influence between different groups, the temperature distribution of battery modules under different power allocation schemes is accurately predicted, effectively avoiding the occurrence of local overheating. The data fusion method not only improves the accuracy of temperature change trend and performance degradation prediction, but also provides a more comprehensive decision basis for power allocation optimization.

[0132] In one alternative implementation,

[0133] A heat conduction equation is established based on the target charge and discharge power. The pre-acquired thermal characteristic parameters of the battery module are then substituted into the heat conduction equation to analyze the thermal coupling effect and electrical characteristic influence between different groups. The temperature field distribution of the battery module is calculated, including:

[0134] A phase change material is filled into the gaps between the battery modules. A heat conduction equation is established based on the target charge and discharge power of the battery modules and the thermal management characteristics of the phase change material. The phase change temperature of the phase change material is used as the reference temperature for the heat conduction equation.

[0135] The composite thermal characteristic parameters of the battery module and the phase change material are obtained, the battery module is set as a heat conduction node, the equivalent thermal conductivity between adjacent nodes is calculated, and a thermal characteristic parameter matrix is ​​constructed.

[0136] The temperature data and phase state data of the phase change material are collected in real time. The temperature data is compared with the phase change temperature to calculate the heat storage state of the phase change material. The heat storage efficiency is calculated based on the sensible heat storage and latent heat storage of the phase change material.

[0137] Substitute the thermal characteristic parameter matrix into the heat conduction equation, set boundary conditions based on the thermal storage efficiency, and calculate the transient temperature field distribution of the battery module.

[0138] The battery module is assembled from multiple individual cells connected in series and parallel. The gaps between the modules are filled with phase change material. The selected phase change material is a paraffin-based composite phase change material with a phase change temperature of 45℃, a latent heat value of 195J / g, and a thermal conductivity of 0.4W / (m·K). The phase change material filling process adopts a vacuum injection method to ensure uniform filling without air bubbles, achieving a filling rate of over 98%.

[0139] When establishing the heat conduction equation, the changes in charging and discharging power of the battery module under different operating conditions are considered. For example, when the battery module is operating at a 1C discharge rate with a target discharge power of 3.2kW, each individual cell generates approximately 8.5W of heat. The heat conduction equation couples the heat source term, the heat conduction term, and the latent heat of phase change term, setting the phase change temperature of the phase change material (45℃) as the reference temperature point for determining the thermal equilibrium state.

[0140] Thermophysical properties of the battery modules and phase change materials were tested separately. The battery module has a heat capacity of 880 J / (kg·K) and a thermal conductivity of 15 W / (m·K). The phase change material has a heat capacity of 1600 J / (kg·K) in the solid state and 2100 J / (kg·K) in the liquid state. Each battery module was designated as a heat conduction node, and the nodes were connected by the phase change material. For two adjacent battery module nodes, the equivalent thermal conductivity was calculated considering their geometric dimensions and positional relationships. For example, two battery modules 25 mm apart, with phase change material filling the gap, had an equivalent thermal conductivity of 0.32 W / (m·K). A thermal characteristic parameter matrix was constructed by combining the thermal characteristic parameters of all nodes and the equivalent thermal conductivity between nodes. The matrix dimensions corresponded to the number of battery modules, and each element represented the heat transfer characteristics between the corresponding nodes.

[0141] Temperature data of the phase change material (PCM) is collected in real time using a temperature sensor network at a sampling frequency of 1 Hz. Temperature sensors are positioned at key locations, including the surface of the battery module, the interior of the PCM, and its boundaries. Phase state data is extrapolated by combining pre-obtained phase change characteristic curves using differential scanning calorimetry with real-time temperature data. When the collected temperature is 44.8℃, compared to a phase change temperature of 45℃, it indicates that the PCM is in the initial stage of phase change.

[0142] The heat storage state of the phase change material (PCM) is calculated based on temperature data. When the PCM temperature is below the phase change temperature, heat is stored as sensible heat; when the temperature reaches the phase change temperature, heat is stored as latent heat. For example, for 200g of PCM, the sensible heat storage is 4800J when the temperature rises from 30℃ to 45℃; the latent heat storage is 39000J when the phase change is complete. The heat storage efficiency is calculated as the ratio of the actual heat storage to the theoretical maximum heat storage; in this example, the heat storage efficiency is approximately 92%.

[0143] The constructed thermal characteristic parameter matrix was substituted into the heat conduction equation, and boundary conditions were set based on the thermal storage efficiency. The boundary conditions included an ambient temperature of 25℃ and a heat dissipation coefficient of 20 W / (m²·K). The heat conduction equation was numerically solved using the finite difference method, with a time step of 0.1 s and a spatial step set according to the actual geometric dimensions of the battery module. The transient temperature field distribution of the battery module during charging and discharging was calculated. The results showed that after 30 minutes of 1C discharge, the highest surface temperature of the battery module was 52.3℃, the lowest was 48.1℃, and the temperature difference was 4.2℃. The temperature distribution in the phase change material region was uniform, with most areas maintaining a temperature around 45℃, indicating that the phase change material effectively absorbed the heat generated by the battery and maintained a relatively stable temperature environment through the phase change process.

[0144] In this embodiment, a heat conduction model that better reflects actual working conditions is established by using the phase change temperature as the reference temperature for the heat conduction equation. This effectively utilizes the temperature regulation characteristics of the phase change material. By calculating the equivalent thermal conductivity between adjacent nodes, a complete thermal characteristic parameter matrix is ​​constructed, and a real-time monitoring mechanism for the phase change material is introduced. By collecting temperature and phase state data, the heat storage state of the phase change material is dynamically calculated, and a heat storage efficiency evaluation system based on sensible and latent heat is established.

[0145] In the existing technology, traditional thermal management systems often ignore the latent heat characteristics of phase change materials and fail to make full use of their thermal buffering capacity, resulting in uneven temperature field distribution, affecting the service life of the battery pack. At the same time, a single active heat dissipation method is difficult to cope with changes in transient heat load.

[0146] This embodiment not only considers the thermal characteristics of the battery module itself, but also incorporates the influence of phase change materials, making the description of thermal conduction characteristics more accurate. The real-time monitoring and evaluation mechanism provides reliable data support for temperature field optimization. It fully considers the dynamic characteristics of the phase change process, and can more accurately predict and control the temperature distribution. This not only improves the thermal safety of the battery pack, but also extends its service life, providing a reliable guarantee for the stable operation of the battery system.

[0147] Figure 3 The experimental data and simulation results of the phase change material thermal management technology of the battery module in the liquid-cooled energy storage system energy distribution and scheduling control method of the present invention are presented. The analysis of the temperature uniformity of the battery module under various discharge rate conditions is shown for four different thermal management schemes.

[0148] Under standard 1C discharge rate conditions, the maximum temperature difference of this technical solution is 4.2°C, which is significantly lower than 9.7°C (56.7% improvement) of the natural air cooling method, 7.3°C (42.5% improvement) of the graphene thermal pad method, and 6.2°C (32.3% improvement) of the liquid cooling system method.

[0149] As the discharge rate increases to 1.5C, the temperature difference of this technical solution is 7.3°C, still within the safe threshold; the temperature difference is 10.4°C at a discharge rate of 2C; 13.6°C at 2.5C; and reaches 13.3°C at 3C. Under the same conditions, the temperature differences for air natural cooling, graphene thermal pads, and liquid cooling systems rise to 23.5°C, 17.3°C, and 16.2°C, respectively. When the discharge rate continues to increase to 3.5C, the temperature difference of this technical solution is 15.1°C; 17.7°C at 4C; and 20.4°C at 4.5C. Under the high discharge rate of 5C, the maximum temperature difference of this technical solution is 18.2°C, significantly better than other methods (air natural cooling 24.3°C, graphene thermal pad 21.1°C, liquid cooling system 22.2°C).

[0150] The black dashed line in the figure represents the safe temperature difference threshold of 8°C. It can be seen that this technical solution can maintain the temperature difference within the safe threshold at a discharge rate of approximately 1.7C, while the natural air cooling method can only achieve this standard at approximately 0.7C, the graphene thermal pad method meets the standard at approximately 1.1C, and the liquid cooling system method meets the standard at approximately 1.3C.

[0151] The slope of the temperature difference growth curve of this technical solution is significantly smaller than that of the other three methods, indicating that it is less sensitive to changes in discharge rate and has stronger temperature stability. When the phase change material undergoes a phase change near 45°C, it can absorb a large amount of heat energy while the temperature remains almost unchanged, forming a heat buffer and effectively blocking the formation of hot spots. Traditional air natural cooling methods mainly rely on natural convection heat dissipation, which has high and uneven thermal resistance. Although graphene thermal pads have good thermal conductivity, they lack the thermal buffering effect of phase change materials. Through the thermal buffering characteristics of phase change materials, a safer and more stable thermal environment is provided for the battery.

[0152] In one alternative implementation,

[0153] Charging and discharging operations are performed based on the optimal power allocation scheme, and the battery module's operating status is monitored in real time. If abnormal fluctuations are detected, the system switches to standby mode, including:

[0154] The system obtains the optimal power allocation scheme to control the battery module to perform charging and discharging operations, and collects the temperature, current and voltage data of the battery module in real time.

[0155] The fluctuation values ​​of the temperature data, current data, and voltage data are compared with preset thresholds. When the fluctuation value of any parameter exceeds the corresponding preset threshold, a status abnormality signal is generated.

[0156] In response to the abnormal status signal, the battery module is switched to a pre-set standby operating state, where the power of the standby operating state is less than the power corresponding to the optimal power allocation scheme.

[0157] The system obtains the optimal power allocation scheme, selects the most suitable power allocation scheme from the model library based on the current operating conditions, and controls the battery module to perform charging and discharging operations through the power conversion unit (PCU).

[0158] During charging and discharging, temperature, current, and voltage data of the battery module are collected in real time via a sensor network. Temperature data is acquired using a thermistor array with multiple sampling points on the battery module surface, at a sampling frequency of 1Hz and an accuracy of ±0.5℃. Current data is acquired using a Hall effect current sensor, at a sampling frequency of 10Hz and an accuracy of ±0.1A. Voltage data is acquired using a high-precision voltage sampling circuit, at a sampling frequency of 10Hz and an accuracy of ±0.01V. All acquired data undergoes preliminary filtering before being transmitted to the central processing unit for analysis.

[0159] The data processing unit calculates the fluctuation values ​​of each parameter and compares them with preset thresholds. The fluctuation calculation uses a sliding window method. For temperature parameters, the rate of temperature change over 30 consecutive seconds is calculated; a temperature fluctuation exceeding 2°C / minute is considered abnormal. For current parameters, the rate of current change over 5 consecutive seconds is calculated; a current fluctuation exceeding 0.5C / second (where C is the battery's rated capacity) is considered abnormal. For voltage parameters, the rate of voltage change over 5 consecutive seconds is calculated; a voltage fluctuation exceeding 50mV / second is considered abnormal. These thresholds are set based on extensive historical data and battery safety characteristics, ensuring timely detection of anomalies while avoiding misjudgments.

[0160] When any fluctuation in any parameter exceeds the corresponding preset threshold, a status anomaly signal is generated. For example, during normal charging of an 18650 lithium battery module, the voltage of a single cell rises from 3.7V to 3.85V in 100 seconds, with a voltage change rate of 1.5mV / second, which is below the threshold of 50mV / second. However, suddenly in the next 3 seconds, the voltage rises rapidly to 4.0V, with a voltage change rate of 50mV / second, exceeding the preset threshold, at which point a status anomaly signal is immediately generated.

[0161] In response to an abnormal status signal, the battery module is switched to a pre-set standby operating state. The standby operating state is a low-power, high-safety mode where the power setting is less than the power corresponding to the optimal power allocation scheme. The power control unit immediately reduces the power output, lowering the charging power to 30%-50% of the normal operating power, or the discharging power to 40%-60% of the normal operating power. The current limit is adjusted, limiting the charging current to 0.3C of the rated current and the discharging current to 0.5C of the rated current. Third, the auxiliary cooling system is activated to enhance thermal management capabilities and keep the temperature within a safe range.

[0162] In standby mode, continue monitoring battery parameters. If the abnormal condition persists or worsens, further reduce the power to 10%-20% of the normal operating power and issue a warning signal. If the parameters return to normal and remain stable for more than a preset time (usually 5-10 minutes), execute a self-test procedure. After confirming that the battery is in normal condition, gradually restore it to normal operating status. The recovery process uses a stepped power increase strategy, increasing the power by 10%-20% each time, observing for 5 minutes, and if there are no abnormalities, increasing to the next step until the power level corresponding to the optimal power allocation scheme is restored.

[0163] In this embodiment, a comprehensive status monitoring system is established by collecting temperature, current, and voltage data in real time. This system can promptly capture various abnormal states during the operation of the battery module, providing a reliable data foundation for the system's safety protection. The threshold-based judgment method can not only respond quickly to sudden anomalies but also detect gradual faults, improving the system's early warning capability. By using a preset backup working state switching mechanism, an active protection system is constructed, which not only ensures the continuous operation of the system but also effectively reduces safety risks, demonstrating the system's flexible adjustment capability.

[0164] A second aspect of the present invention provides an energy distribution and scheduling control system for a liquid-cooled energy storage system, comprising:

[0165] The first unit is used to collect the operating parameters of the battery modules in the liquid-cooled energy storage system and calculate the health status index of the battery modules based on the operating parameters.

[0166] The second unit is used to calculate the performance matching index between different battery modules based on the health status index, construct the feature vector corresponding to the battery module based on the performance matching index and perform cluster analysis to obtain the initial grouping scheme, group the battery modules based on the initial grouping scheme and calculate the feature vector of the battery module in each group, detect whether the feature vector deviates from the cluster center of the group by more than a preset threshold, if so, repeat the cluster analysis and grouping until the grouping target is met and the optimal grouping scheme is obtained.

[0167] The third unit is used to acquire grid load demand data and power generation data and calculate the target charging and discharging power.

[0168] The fourth unit is used to collect battery module performance parameters based on the optimal grouping scheme, construct the topological relationship between battery modules through graph neural network, analyze the thermal coupling effect and electrical characteristics between different groups in combination with the target charge and discharge power, predict the temperature change trend and performance degradation rate of battery modules under different preset power allocation schemes in combination with historical performance parameters, determine the optimal power allocation scheme and allocate it to the battery modules.

[0169] The fifth unit is used to perform charging and discharging operations based on the optimal power allocation scheme, and to detect the working status of the battery module in real time. If there are abnormal fluctuations, it will switch to standby mode.

[0170] A third aspect of the present invention provides an electronic device, comprising:

[0171] A processor and a memory for storing processor-executable instructions, wherein the processor is configured to invoke instructions stored in the memory to perform the aforementioned method.

[0172] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0173] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0174] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for energy distribution and dispatch control of a liquid-cooled energy storage system, characterized in that, The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device.

2. The method of claim 1, wherein, The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping method and device. The application relates to a battery module grouping According to the collected working parameters, the ratio of the voltage change amount to the current change amount per unit time is calculated to obtain the internal resistance change rate, the voltage change amount per unit time is calculated to obtain the voltage change rate, and the ratio of the current capacity to the rated capacity under the number of charge and discharge cycles is calculated to obtain the capacity attenuation rate; According to the internal resistance change rate, the voltage change rate and the capacity attenuation rate, the health state index is obtained by weighted calculation according to the preset weight coefficient.

3. The method of claim 1, wherein, Based on the health state index, the performance matching index between different battery modules is calculated, the feature vector corresponding to the battery module is constructed based on the performance matching index, and clustering analysis is performed to obtain an initial grouping scheme. The battery modules are grouped based on the initial grouping scheme, and the feature vector of the battery modules in each group is calculated. It is detected whether the feature vector deviates from the cluster center of the group by more than a preset threshold. If so, repeat the clustering analysis and grouping until the grouping target is met to obtain an optimal grouping scheme including: Based on the health state index of the battery module, the performance matching index is obtained by calculating the ratio of the difference between the health state indexes of two battery modules to the maximum value of the health state indexes of the two battery modules. The performance matching index and the health state index of the battery module are combined to construct the feature vector corresponding to the battery module. The K-means clustering algorithm is used to perform clustering analysis on the feature vector to obtain an initial grouping scheme and calculate the cluster center corresponding to each initial grouping scheme. The Euclidean distance between the feature vector of the battery module in each initial grouping scheme and the cluster center is calculated, and the weight coefficient is assigned based on the Euclidean distance and the clustering analysis is performed again until the Euclidean distance is less than or equal to the preset threshold. The intra-group consistency index in the same group and the inter-group difference index between different groups are calculated. Based on the intra-group consistency index and the inter-group difference index, the number of clusters is dynamically adjusted and the number of groups is balanced to obtain an optimal grouping scheme.

4. The method of claim 3, wherein, The K-means clustering algorithm is used to perform clustering analysis on the feature vector to obtain an initial grouping scheme including: The feature vector corresponding to the battery module is obtained, the ratio of the feature vector minus the minimum value of the feature vector to the difference between the maximum value and the minimum value of the feature vector is calculated, and a standardized feature vector is generated. The performance matching index difference and the health state index difference between any two battery modules are calculated based on the standardized feature vector. The difference values are respectively substituted into the exponential function and weighted summed after assigning a weight coefficient to generate an electrochemical property similarity matrix. A first initial cluster center is randomly selected from all battery modules, the minimum distance from the remaining battery modules to the initial cluster center is calculated, a probability distribution is constructed based on the minimum distance, and an optimized cluster center is selected. Repeat the execution until a preset number is reached. A network flow graph is constructed and a source point and a sink point are set. The battery modules are set as vertices and connected adjacent battery modules. The weight of the edge is calculated. The weight of the edge is calculated by the exponential function of the element of the electrochemical property similarity matrix and the distance from the sample to the optimized cluster center. The initial search radius is set and substituted into an exponential decay function, and the dynamic neighborhood radius is obtained by summing the minimum radius. The battery modules are distributed according to the dynamic neighborhood radius, the average value of the feature vectors of the battery modules in the class is calculated to obtain the cluster center, and it is judged whether the change amount of the cluster center and the maximum difference of the performance matching index meet the preset threshold value and the balance requirement. When the preset threshold value and the balance requirement are met, the clustering result is output.

5. The method of claim 1, wherein, Based on the optimal power distribution scheme, the charging and discharging operation is performed, and the working state of the battery module is detected in real time. If there is abnormal fluctuation, it is switched to a standby state, including: An optimal power distribution scheme is obtained to control the battery module to perform charging and discharging operation, and the temperature data, current data and voltage data of the battery module are collected in real time; The fluctuation value of the temperature data, current data and voltage data is compared with the preset threshold value, and a state abnormal signal is generated when the fluctuation value of any parameter exceeds the corresponding preset threshold value; In response to the state abnormal signal, the battery module is switched to a pre-set standby working state, and the power of the standby working state is less than the power corresponding to the optimal power distribution scheme.

6. Energy distribution and dispatch control system for a liquid-cooled energy storage system, for implementing the method according to any one of the preceding claims 1-5, characterized in that, It includes: The first unit is used for collecting the working parameters of the battery module in the liquid-cooled energy storage system, and calculating the health state index of the battery module based on the working parameters; The second unit is used for calculating the performance matching index between different battery modules based on the health state index, constructing the feature vector corresponding to the battery module based on the performance matching index, and performing clustering analysis to obtain an initial grouping scheme. The battery modules are grouped based on the initial grouping scheme, and the feature vectors of the battery modules in each group are calculated. It is detected whether the feature vector deviates from the cluster center of the group by more than a preset threshold value. If so, repeat the clustering analysis and grouping until the grouping target is met to obtain an optimal grouping scheme; The third unit is used for obtaining grid load demand data and power generation data and calculating target charging and discharging power; The fourth unit is used for collecting battery module performance parameters based on the optimal grouping scheme, constructing the topological relationship between battery modules through a graph neural network, analyzing the thermal coupling effect and electrical characteristic influence between different groups in combination with the target charging and discharging power, and predicting the temperature change trend and performance decay rate of the battery module under different preset power distribution schemes in combination with historical performance parameters to determine the optimal power distribution scheme and distribute it to the battery module; The fifth unit is used for performing charging and discharging operation based on the optimal power distribution scheme, and detecting the working state of the battery module in real time. If there is abnormal fluctuation, it is switched to a standby state.

7. An electronic device, comprising: It includes: A processor; A memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method of any one of claims 1 to 5.

8. A computer-readable storage medium having stored thereon computer program instructions, wherein, The computer program instructions are executed by the processor to implement the method of any one of claims 1 to 5.

Images