Energy storage system management method and system based on AI model analysis

Abstract

The invention discloses an energy storage system management method and system based on AI model analysis. The method comprises the steps of obtaining test data of a battery cell and inputting the test data into an AI prediction model to obtain a first health state score and a battery cell aging trend index; rejecting the battery cells with the first health state scores smaller than a certain value; executing a clustering algorithm according to the first health state scores of the healthy battery cells and the battery cell aging trend indexes, so that the healthy battery cells falling into the same cluster are assembled into the same battery pack during assembly; a second SOH (state of health) score of the battery pack is predicted, and the aging trend indexes of the battery pack are clustered, so that the packs falling into the same interval form a battery cluster, and the battery cluster is used for a single energy storage device; and acquiring operation data of the battery cluster in an operation stage, inputting the operation data into the AI model, outputting second health state scores of the battery packs and the battery cells in the battery cluster, and triggering an energy distribution adjustment instruction or human intervention information. Through the method, the service life of the battery cluster can be prolonged.

Description

Technical Field

[0001] This application relates to the field of battery management technology, and in particular to a method and system for managing energy storage systems based on AI model analysis. Background Technology

[0002] Lithium-ion batteries, with their advantages of high energy density, long cycle life, and no memory effect, have become the preferred technology for electrochemical energy storage systems. Battery state of health (SOH), as a core indicator for measuring its degradation degree, directly affects the safety, reliability, and economy of energy storage systems. High-precision assessment and management of battery SOH can improve battery performance.

[0003] Chinese patent CN222915686U, "A Battery Management System with Remote Monitoring of Battery Operating Status," proposes a battery management system integrating monitoring, protection, communication, and balancing circuits. This system assesses battery status and predicts lifespan in real time during operation, and improves battery pack performance by transferring charge through balancing circuits. Chinese patent CN119382298A, "A Cloud-Based BMS Battery Remote Monitoring and Data Analysis System," analyzes the performance and losses of various battery components to assess and manage the overall battery health.

[0004] However, both of the above patents rely on energy regulation during operation to maintain battery performance, which is a post-intervention mode. For some cells that already have performance degradation or potential failure risks at the manufacturing stage, this mode is difficult to identify in time, and the effect of later equalization maintenance is limited, and the overall life of the battery pack is still constrained. Summary of the Invention

[0005] Based on this, in response to the aforementioned technical problems, an energy storage system management method and system based on AI model analysis are provided to solve the problem of low overall battery pack life in existing battery management technologies.

[0006] Firstly, an energy storage system management method based on AI model analysis, the method comprising:

[0007] During the testing phase before the battery cells leave the factory, test data of the battery cells are acquired to form a battery cell factory test dataset; the battery cell factory test dataset is input into a pre-built AI prediction model to output the current first health status score (SOH) and the first battery cell aging trend index.

[0008] Cells with a current first health status score (SOH) less than a certain threshold are removed, and the remaining cells are recorded as healthy cells. Using the first health status score (SOH) of the healthy cells and the cell aging trend index as clustering features, a clustering algorithm is executed so that healthy cells falling into the same cluster are assembled into the same battery pack during assembly.

[0009] Acquire test data from multiple battery packs to form a battery pack factory test dataset. Input the battery pack factory test dataset into a pre-built first AI prediction model and output the current second health status score (SOH) and the second battery pack aging trend index.

[0010] The second health status score (SOH) and the aging trend index of the battery pack are used as clustering features to execute a clustering algorithm so that healthy battery packs falling into the same cluster are grouped into the same battery cluster during assembly; any one of the battery clusters is used for a single energy storage device.

[0011] During the operation of the energy storage device, real-time operating data of the battery cluster is collected and input into the AI ​​model. The system outputs the second state of health (SOH) score for all battery packs and individual cells in the battery cluster, which triggers energy distribution adjustment commands or human intervention information and sends them to the battery management system.

[0012] Optionally, in the above scheme, the test data of the battery cell includes: battery cell voltage, battery cell current, ambient temperature, ambient humidity, charging and discharging power, and capacity;

[0013] The test data for the battery pack include: battery pack voltage, battery pack current, ambient temperature, ambient humidity, and charging / discharging power.

[0014] The battery cluster operating data includes: cell voltage, cell current, battery pack voltage, battery pack current, ambient temperature, ambient humidity, charging and discharging power, and battery SOC.

[0015] In the above scheme, optionally, the battery aging trend indicators include: potential lifespan, degradation rate, and failure risk; the state of health (SOH) score is obtained by comprehensively considering the potential lifespan, degradation rate, and failure risk.

[0016] Optionally, in the above scheme, the construction of the AI ​​prediction model includes:

[0017] Collect historical full-lifecycle test data for battery cells, battery packs, and battery clusters from new products to failure, and establish a training sample set. The full-lifecycle test data includes cell test data, battery pack test data, and battery cluster test data. Cell test data includes: cell current, cell voltage, capacity, ambient temperature, ambient humidity, charge / discharge power, and the corresponding actual potential lifespan, degradation rate, and failure risk. Battery pack test data includes: battery pack current, battery pack voltage, capacity, ambient temperature, ambient humidity, charge / discharge power, and the corresponding actual potential lifespan, degradation rate, and failure risk. Battery cluster test data includes: battery cluster current, battery cluster voltage, capacity, ambient temperature, ambient humidity, charge / discharge power, and the corresponding actual potential lifespan, degradation rate, and failure risk.

[0018] Feature extraction is performed based on the training sample set to form a model input feature set. The model feature set is then input into the deep neural network-based model to obtain an AI prediction model.

[0019] In the above scheme, optionally, using the first health status score (SOH) of the healthy battery cell and the battery cell aging trend index as clustering features, the clustering algorithm further includes:

[0020] When the total number of healthy cells in the same cluster is less than a predetermined value, the healthy cells in that cluster are removed.

[0021] Optionally, in the above scheme, the energy distribution adjustment command includes: increasing the depth of charge and discharge for battery packs with a high second state of health score (SOH), and reducing power or bypassing battery packs with a low second state of health score (SOH).

[0022] Optionally, in the above scheme, the human intervention information includes uploading fault codes and issuing maintenance work orders.

[0023] Secondly, an energy storage system management system based on AI model analysis, the system comprising:

[0024] Cell testing module: used to acquire test data of the cells during the testing phase before the cells leave the factory, forming a cell factory test dataset; input the cell factory test dataset into a pre-built AI prediction model, and output the current first health status score (SOH) and the first cell aging trend index;

[0025] Cell assembly module: used to remove cells whose current first health status score (SOH) is less than a certain threshold, and record the remaining cells as healthy cells; using the first health status score (SOH) of the healthy cells and the cell aging trend index as clustering features, execute a clustering algorithm so that healthy cells falling into the same cluster are assembled into the same battery pack during assembly;

[0026] Battery pack testing module: used to acquire test data of multiple battery packs, form a battery pack factory test dataset, input the battery pack factory test dataset into a pre-built first AI prediction model, and output the current second health status score (SOH) and the second battery pack aging trend index.

[0027] Battery pack assembly module: used to perform a clustering algorithm with the second health status score (SOH) and the aging trend index of the battery pack as clustering features, so that healthy battery packs falling into the same cluster are assembled into the same battery cluster during assembly; any one of the battery clusters is used for a single energy storage device;

[0028] Battery operation monitoring module: used to collect real-time operating data of the battery cluster during the operation of the energy storage device, input the data into the AI ​​model, output the second state of health (SOH) score of all battery packs and individual cells in the battery cluster, and trigger energy distribution adjustment instructions or human intervention information based on this, and send them to the battery management system.

[0029] Thirdly, a computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the energy storage system management method based on AI model analysis described in the first aspect.

[0030] Fourthly, a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the energy storage system management method based on AI model analysis described in the first aspect.

[0031] This application has at least the following beneficial effects:

[0032] This application conducts testing at the cell manufacturing stage to evaluate the cell's health status and predict its aging trend. Cells with low health status scores are directly eliminated, avoiding the "weakest link" problem from the outset and significantly reducing the probability of safety accidents caused by cell inconsistencies or hidden defects in energy storage systems. Based on the first health status score (SOH), aging trend indicators, and internal resistance of healthy cells, a clustering algorithm is executed to group healthy cells belonging to the same cluster into the same battery pack during assembly. The entire battery pack is then tested to predict its second health status score (SOH) and second aging trend indicators, and clustered based on these indicators to group packs belonging to the same range into battery clusters. Each of these battery clusters can be used in a single energy storage device. This "optimal matching" intelligent grouping strategy ensures high consistency of the cells in the groups, effectively slowing down the rate at which premature degradation of individual cells drags down the overall system lifespan, thus improving the overall lifespan of the battery clusters. Meanwhile, during the operation of the battery cluster, the operating data of the battery cluster is also monitored, and the health scores of the cells and battery packs in the battery cluster are predicted. Based on this, energy allocation is carried out to improve the overall lifespan of the battery cluster. Attached Figure Description

[0033] Figure 1 A flowchart illustrating an energy storage system management method based on AI model analysis, provided as an embodiment of this application;

[0034] Figure 2 A flowchart illustrating the operation of an energy storage battery according to one embodiment of this application;

[0035] Figure 3 This is a flowchart of the data acquisition process for an energy storage system in one embodiment of this application;

[0036] Figure 4 A flowchart illustrating the health scoring process using an AI model for an energy storage system, as provided in one embodiment of this application;

[0037] Figure 5 This is a flowchart illustrating an embodiment of the health score analysis provided in this application. Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0039] In the description of this application: unless otherwise stated, "a plurality of" means two or more. The terms "first," "second," "third," etc., in this application are intended to distinguish the objects referred to and do not have any special meaning in terms of technical connotation (e.g., they should not be construed as an emphasis on importance or order). Expressions such as "including," "comprising," and "having" also mean "not limited to" (certain units, components, materials, steps, etc.).

[0040] In one embodiment, such as Figure 1 As shown, an energy storage system management method based on AI model analysis is provided, the method comprising:

[0041] Step S1: During the testing phase before the battery cells leave the factory, acquire test data of the battery cells to form a battery cell factory test dataset; input the battery cell factory test dataset into a pre-built AI prediction model to output the current first health status score (SOH) and the first battery cell aging trend index.

[0042] Step S2: Eliminate cells whose current first health status score SOH is less than a certain threshold, and record the remaining cells as healthy cells; use the first health status score SOH of the healthy cells and the cell aging trend index as clustering features to execute a clustering algorithm so that healthy cells that fall into the same cluster are assembled into the same battery pack during assembly;

[0043] Step S3: Obtain test data from multiple battery packs to form a battery pack factory test dataset. Input the battery pack factory test dataset into a pre-built first AI prediction model and output the current second health status score (SOH) and the second battery pack aging trend index.

[0044] Step S4: Using the second health status score (SOH) and the aging trend index of the battery pack as clustering features, execute a clustering algorithm to ensure that healthy battery packs falling into the same cluster are grouped into the same battery cluster during assembly; any one of the battery clusters is used for a single energy storage device.

[0045] Step S5: During the operation phase of the energy storage device, real-time operating data of the battery cluster is collected and input into the AI ​​model. The second health status score (SOH) of all battery packs and individual cells in the battery cluster is output, and energy distribution adjustment instructions or human intervention information are triggered and sent to the battery management system.

[0046] In step S5, as Figure 2 The operation of energy storage equipment includes the following steps:

[0047] (1) Power on the entire energy storage system.

[0048] (2) After the energy storage system is powered on, the initialization work of the equipment is carried out, including the battery management system (BMS) in the battery compartment, liquid cooler, dehumidifier, and calibration of the startup parameters of the energy storage system management platform.

[0049] (3) After the energy storage system is successfully initialized, it enters the energy storage system shutdown state.

[0050] (4) After the energy storage system is shut down, a preliminary test of the whole system is carried out, including communication between the battery management system (BMS), liquid cooler, dehumidifier and energy storage system management platform, and whether the temperature and humidity parameters have been collected.

[0051] (5) The energy storage system is not faulty, and the energy storage system management platform enters standby mode after reading the temperature and humidity parameters.

[0052] (6) The energy storage system is in standby mode and waiting for the start command.

[0053] (7) The energy storage system enters the start-up state after receiving the start-up command.

[0054] (8) When the energy storage system is in the startup state, all the contacts in the high-voltage boxes close and enter the operating state.

[0055] (9) During operation, charging and discharging can be performed, and battery data can be collected during this stage. The collected data includes: battery cluster voltage, battery cluster temperature, battery pack voltage, battery pack temperature, battery cell voltage, battery cell temperature, battery SOC, charging and discharging power, battery cluster compartment temperature, and battery cluster compartment humidity.

[0056] During the operation of energy storage equipment, such as Figure 3 As shown, the data acquisition process includes the following steps:

[0057] (1) Battery information is collected through a three-layer architecture of the Battery Management System (BMS), including battery cluster voltage, battery cluster temperature, battery pack voltage, battery pack temperature, battery cell voltage, battery cell temperature, and battery SOC. The entire data is then transmitted to the energy storage system management platform via the Battery Management System (BMS).

[0058] (2) During the operation of the energy storage system, the PCS device will collect data such as the corresponding charging and discharging power. This data will then be transmitted to the energy storage system management platform through the PCS controller.

[0059] (3) There are also some auxiliary battery-optimized devices in the energy storage system, such as liquid coolers, dehumidifiers, etc. The temperature and humidity of these devices will be transmitted to the energy storage system management platform.

[0060] (4) Data is integrated in the energy storage system management platform and stored in the database in real time.

[0061] like Figure 4 As shown, the process of using AI models for health status assessment includes the following steps:

[0062] (1) Basic process of battery health detection: including data acquisition, preprocessing, feature extraction, model training, and health status assessment. Data acquisition refers to the data that the energy storage system stores in the database in real time during operation, which can be directly retrieved.

[0063] (2) Data analysis method: Machine learning model (LSTM) is used to predict battery health status.

[0064] (3) Visual display method: Battery health test report, real-time parameter monitoring, and chart display can make battery information more intuitive to observe.

[0065] like Figure 5 As shown, the process of analyzing a battery based on its health includes the following steps:

[0066] (1) When the energy storage system is in operation, all data acquisition, data analysis and data storage begin to run.

[0067] (2) The collected data will be analyzed by AI large model according to different battery capacities, and the battery health will be obtained after the analysis by AI large model.

[0068] (3) Determine whether immediate shutdown and manual intervention are needed based on battery health status, or automatically adjust energy distribution to increase battery life.

[0069] (4) If the result obtained by the AI ​​large model analysis is not optimal when the equipment is charged and discharged for the first time in the factory, the battery cells will be replaced immediately in the factory to ensure the maximum performance of the energy storage system.

[0070] The pre-shipment testing includes the following steps:

[0071] (1) The system collects multi-dimensional time-series data for each cell during the factory testing phase, including but not limited to: electrochemical data: battery voltage, battery temperature, SOC, etc.; environmental data: ambient temperature and temperature rise curve during testing; operating condition data: performance under different power levels.

[0072] (2) The above-mentioned fused data is analyzed using an AI large model (deep neural network). The model is trained with massive historical cell data (full life cycle data from good to failure) and can predict the potential lifespan (SOH), decay rate and failure risk of the cell after long-term cycling from short-term factory test data, rather than just the current health status.

[0073] Overall health score: The model outputs a comprehensive health status score, which is obtained by combining potential lifespan, decay rate and failure risk.

[0074] (3) Intelligent cell grouping and adaptation strategy based on clustering and consensus algorithms: The system not only evaluates individual cells, but also performs an overall analysis of all cells to be grouped into a battery pack. Using clustering algorithms (such as K-means, hierarchical clustering), based on predicted health, battery aging trend indicators and other characteristics, cells with highly consistent performance are automatically grouped into the same battery cluster or battery pack to achieve "superior matching" and maximize system consistency. A "goodness threshold" is set to automatically identify and mark "problem cells" that do not meet the health standards or whose performance characteristics differ too much from the group (i.e., "incompatible").

[0075] In the aforementioned AI model-based energy storage system management method, testing is conducted at the cell manufacturing stage to evaluate the health status of the cells and predict their aging trend indicators. Cells with low health status scores are directly eliminated, avoiding the "weakest link" problem from the outset and significantly reducing the probability of safety accidents caused by cell inconsistencies or hidden defects in the energy storage system. Based on the first health status score (SOH), aging trend indicators, and internal resistance of healthy cells, a clustering algorithm is executed to group healthy cells belonging to the same cluster into the same battery pack during assembly. The battery pack is then tested as a whole to predict its second health status score (SOH) and second aging trend indicators, and clustered to group packs belonging to the same range into battery clusters. Each battery cluster is used in a single energy storage device. This "optimal matching" intelligent grouping strategy ensures high consistency of the cells in the groups, effectively slowing down the rate at which premature degradation of individual cells drags down the overall system lifespan, thus improving the overall lifespan of the battery clusters. Meanwhile, during the operation of the battery cluster, the operating data of the battery cluster is also monitored, and the health scores of the cells and battery packs in the battery cluster are predicted. Based on this, energy allocation is carried out to improve the overall lifespan of the battery cluster.

[0076] In one embodiment, the test data of the battery cell includes: battery cell voltage, battery cell current, ambient temperature, ambient humidity, charge / discharge power, and capacity;

[0077] The test data for the battery pack include: battery pack voltage, battery pack current, ambient temperature, ambient humidity, and charging / discharging power.

[0078] The battery cluster operating data includes: cell voltage, cell current, battery pack voltage, battery pack current, ambient temperature, ambient humidity, charging and discharging power, and battery SOC.

[0079] In one embodiment, the battery aging trend indicators include: potential lifespan, degradation rate, and failure risk; the State of Health (SOH) score is obtained by comprehensively considering the potential lifespan, degradation rate, and failure risk.

[0080] In one embodiment, the construction of the AI ​​prediction model includes:

[0081] Collect historical full-lifecycle test data for battery cells, battery packs, and battery clusters from new products to failure, and establish a training sample set. The full-lifecycle test data includes cell test data, battery pack test data, and battery cluster test data. The cell test data includes: cell current, cell voltage, capacity, ambient temperature, ambient humidity, charge / discharge power, and the corresponding actual potential lifespan, degradation rate, and failure risk. The battery pack test data includes: battery pack current, battery pack voltage, capacity, ambient temperature, ambient humidity, charge / discharge power, and the corresponding actual potential lifespan, degradation rate, and failure risk. The battery cluster test data includes: battery cluster current, battery cluster voltage, capacity, ambient temperature, ambient humidity, charge / discharge power, and the corresponding actual potential lifespan, degradation rate, and failure risk.

[0082] Feature extraction is performed based on the training sample set to form a model input feature set. The model feature set is then input into the deep neural network-based model to obtain an AI prediction model.

[0083] In one embodiment, using the first health status score (SOH) of the healthy battery cell and the battery cell aging trend index as clustering features, the clustering algorithm further includes:

[0084] When the total number of healthy cells in the same cluster is less than a predetermined value, the healthy cells in that cluster are removed.

[0085] In one embodiment, the energy distribution adjustment command includes: increasing the depth of charge / discharge for battery packs with a high State of Health (SOH) score, and reducing power or bypassing battery packs with a low SOH score.

[0086] In one embodiment, the human intervention information includes uploading fault codes and issuing maintenance work orders.

[0087] In one embodiment, an energy storage system management system based on AI model analysis is provided, the system comprising:

[0088] Cell testing module: used to acquire test data of the cells during the testing phase before the cells leave the factory, forming a cell factory test dataset; input the cell factory test dataset into a pre-built AI prediction model, and output the current first health status score (SOH) and the first cell aging trend index;

[0089] Cell assembly module: used to remove cells whose current first health status score (SOH) is less than a certain threshold, and record the remaining cells as healthy cells; using the first health status score (SOH) of the healthy cells and the cell aging trend index as clustering features, execute a clustering algorithm so that healthy cells falling into the same cluster are assembled into the same battery pack during assembly;

[0090] Battery pack testing module: used to acquire test data of multiple battery packs, form a battery pack factory test dataset, input the battery pack factory test dataset into a pre-built first AI prediction model, and output the current second health status score (SOH) and the second battery pack aging trend index.

[0091] Battery pack assembly module: used to perform a clustering algorithm with the second health status score (SOH) and the aging trend index of the battery pack as clustering features, so that healthy battery packs falling into the same cluster are assembled into the same battery cluster during assembly; any one of the battery clusters is used for a single energy storage device;

[0092] Battery operation monitoring module: used to collect real-time operating data of the battery cluster during the operation of the energy storage device, input the data into the AI ​​model, output the second state of health (SOH) score of all battery packs and individual cells in the battery cluster, and trigger energy distribution adjustment instructions or human intervention information based on this, and send them to the battery management system.

[0093] The advantages of this application include the following:

[0094] 1. Significantly enhances the inherent safety of energy storage systems: Cells with potential thermal runaway risks and early signs of degradation are precisely eliminated before leaving the factory, avoiding the "weakest link" cells in the "barrel effect" from the source, and greatly reducing the probability of safety accidents caused by inconsistencies or hidden defects in energy storage systems.

[0095] 2. Significantly extends the overall lifespan of the battery system: Through a smart grouping strategy of "optimal matching," the high consistency of the grouped cells is ensured, effectively slowing down the rate at which premature degradation of individual cells drags down the lifespan of the entire system. The overall system lifespan can be increased by 15% - 30%, directly improving the product's return on investment.

[0096] 3. Maximize the utilization and performance of battery assets: Highly consistent battery packs can release more available capacity (because there is no need to reserve too much redundancy for the worst cells), while the system can operate in a better charge and discharge range, thus improving energy efficiency.

[0097] 4. Significantly reduce the total lifecycle maintenance costs:

[0098] (1) Reduced after-sales maintenance: Avoids high costs such as on-site recalls and replacements caused by batch battery cell problems.

[0099] (2) Reduce the complexity of operation and maintenance: For systems in good condition at the factory, the balancing pressure of the battery management system (BMS) is small and the frequency of operation and maintenance intervention is low.

[0100] (3) Predictive maintenance foundation: The provided "battery health digital passport" provides accurate initial data for subsequent predictive maintenance, making operation and maintenance more intelligent and efficient.

[0101] 5. Improved production efficiency and product quality control: The platform automates and intelligently processes cell screening and grouping, reducing reliance on manual experience and increasing production line efficiency. Simultaneously, it forms a robust quality firewall, ensuring controllable and traceable product quality, significantly enhancing brand reputation and market competitiveness.

[0102] For specific limitations regarding the AI ​​model-based energy storage system management system, please refer to the limitations of the AI ​​model-based energy storage system management method described above, which will not be repeated here. The various modules in the aforementioned AI model-based energy storage system management system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0103] In one embodiment, a computer device, which may be a server, is provided. The computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage medium. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements the aforementioned AI model-based energy storage system management method.

[0104] In one embodiment, a computer-readable storage medium is also provided, on which a computer program is stored relating to all or part of the processes in the methods of the above embodiments.

[0105] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical storage, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0106] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0107] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. An energy storage system management method based on AI model analysis, characterized by, The method comprises: In the test stage before the battery cell leaves the factory, test data of the battery cell is acquired to form a battery cell factory test data set; the battery cell factory data test set is input into a pre-constructed AI prediction model to output a current first state of health score SOH and a first battery cell aging trend index; The battery cell whose current first state of health score SOH is less than a certain threshold is removed, and the remaining battery cell is recorded as a healthy battery cell; the first state of health score SOH of the healthy battery cell and the battery cell aging trend index are taken as clustering features, and a clustering algorithm is executed, so that the healthy battery cells falling into the same cluster are assembled into the same battery pack during assembly; Test data of a plurality of battery packs is acquired to form a battery pack factory test data set, and the battery pack factory test data set is input into a pre-constructed first AI prediction model to output a current second state of health score SOH and a second battery pack aging trend index; The second state of health score SOH of the battery pack and the second battery pack aging trend index are taken as clustering features, and a clustering algorithm is executed, so that the healthy battery packs falling into the same cluster are assembled into the same battery cluster during assembly; any battery cluster is used for a single energy storage device; In the operation stage of the energy storage device, operation data in the battery cluster are collected in real time and input into an AI model to output the second state of health score SOH of all battery packs and single battery cells in the battery cluster, and the second state of health score SOH is used to trigger energy distribution adjustment instructions or human intervention information and is sent to a battery management system. 2.The AI model analysis-based energy storage system management method of claim 1, wherein, The test data of the battery cell comprises a battery cell voltage, a battery cell current, an ambient temperature, an ambient humidity, a charging and discharging power and a capacity; The test data of the battery pack comprises a battery pack voltage, a battery pack current, an ambient temperature, an ambient humidity and a charging and discharging power; The battery cluster operation data comprise a battery cell voltage, a battery cell current, a battery pack voltage, a battery pack current, an ambient temperature, an ambient humidity, a charging and discharging power and a battery SOC. 3.The AI model analysis-based energy storage system management method of claim 2, wherein, The battery aging trend index comprises a potential life, a decay rate and a failure risk; and the state of health score SOH is obtained by comprehensively considering the potential life, the decay rate and the failure risk. 4.The AI model analysis-based energy storage system management method of claim 1, wherein, The construction of the AI prediction model comprises: Full-life test data of historical battery cells, battery packs and battery clusters from new products to failure are collected to establish a training sample set; the full-life test data comprises battery cell test data, battery pack test data and battery cluster test data; the battery cell test data comprises a battery cell current, a battery cell voltage, a capacity, an ambient temperature, an ambient humidity, a charging and discharging power and corresponding real potential life, decay rate and failure risk; the battery pack test data comprises a battery pack current, a battery pack voltage, a capacity, an ambient temperature, an ambient humidity, a charging and discharging power and corresponding real potential life, decay rate and failure risk; and the battery cluster test data comprises a battery cluster current, a battery cluster voltage, a capacity, an ambient temperature, an ambient humidity, a charging and discharging power and corresponding real potential life, decay rate and failure risk; According to the training sample set, feature extraction is performed to form a model input feature set, the model feature set is input into the deep neural network-based model, and an AI prediction model is obtained. 5.The AI model analysis-based energy storage system management method of claim 1, wherein, The first state of health score SOH of the healthy battery cell and the battery cell aging trend indicator are used as clustering features, and a clustering algorithm is executed. When the number of healthy battery cells in the same cluster is less than a predetermined value, the healthy battery cells in the cluster are removed. 6.The AI model analysis-based energy storage system management method of claim 1, wherein, The energy distribution adjustment instruction includes increasing the charge and discharge depth of the battery pack with a high second state of health score SOH and reducing the power or bypassing the battery pack with a low second state of health score SOH. 7.The AI model analysis-based energy storage system management method of claim 1, wherein, The human intervention information includes uploading a fault code and issuing a maintenance work order.

8. An energy storage system management system based on AI model analysis, characterized by, The system comprises: A battery cell test module is configured to obtain test data of the battery cell during a test stage before the battery cell is shipped, form a battery cell shipment test data set, input the battery cell shipment test data set into a pre-constructed AI prediction model, and output a current first state of health score SOH and a first battery cell aging trend indicator. A battery cell grouping module is configured to remove the battery cell with a first state of health score SOH less than a certain threshold, and record the remaining battery cells as healthy battery cells; use the first state of health score SOH of the healthy battery cells and the battery cell aging trend indicator as clustering features, and execute a clustering algorithm to group the healthy battery cells falling into the same cluster into the same battery pack during assembly. A battery pack test module is configured to obtain test data of a plurality of battery packs, form a battery pack shipment test data set, input the battery pack shipment test data set into a pre-constructed first AI prediction model, and output a current second state of health score SOH and a second battery pack aging trend indicator. A battery pack grouping module is configured to use the second state of health score SOH of the battery pack and the second battery pack aging trend indicator as clustering features, execute a clustering algorithm, and group the healthy battery packs falling into the same cluster into the same battery cluster during assembly; and any battery cluster is used for a single energy storage device. A battery operation monitoring module is configured to collect operation data in the battery cluster in real time during the operation stage of the energy storage device, input the operation data into an AI model, output the second state of health score SOH of all battery packs and single battery cells in the battery cluster, trigger an energy distribution adjustment instruction or human intervention information based on the second state of health score SOH, and issue the energy distribution adjustment instruction or human intervention information to a battery management system. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-8 when the computer program is executed by the processor. The processor executes the computer program to implement the steps of the method of any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method of any one of claims 1 to 7.

Images