Battery overvoltage prediction and control method and device, and energy storage equipment

Abstract

The application provides a battery overvoltage prediction and control method and device and energy storage equipment, and relates to the technical field of battery safety management. The battery overvoltage prediction and control method comprises the following steps: collecting first time sequence data of a battery module and second time sequence data of a fuse, wherein the first time sequence data comprises a current single cell voltage and a current total voltage, and the second time sequence data comprises a voltage difference across the fuse; inputting the first time sequence data into a pre-constructed voltage prediction model to obtain a voltage change value of each battery cell unit after a future preset time length, and calculating a single cell predicted voltage and a total predicted voltage; identifying an operating state of the energy storage equipment, and generating a first control instruction for a switching device or a second control instruction for the fuse based on the operating state of the energy storage equipment. In the technical scheme of the application, the overvoltage trend of the battery is predicted in advance, so that damage to the battery cell unit in the initial stage of overvoltage is effectively avoided, and the battery attenuation and fault risk are reduced.

Description

Technical Field

[0001] This invention relates to the field of battery safety management technology, and more specifically, to a battery overvoltage prediction and control method, device, and energy storage device. Background Technology

[0002] Overvoltage faults are one of the core hidden dangers restricting the safe and stable operation of energy storage batteries. If the overvoltage state of the battery is not identified and effectively handled in time, it is very easy to cause the battery cells to bulge, which can lead to thermal runaway and fire, posing a significant safety hazard.

[0003] In related technologies, overvoltage protection for energy storage batteries is relatively lagging. It usually provides passive protection after the energy storage battery experiences overvoltage. Power is only triggered when the measured voltage of the battery cell or battery module reaches the preset overvoltage threshold. This makes it difficult to avoid damage to the energy storage battery in the early stages of overvoltage and the risk of accidental fuse blowing. Summary of the Invention

[0004] One objective of this invention is to provide a battery overvoltage prediction and control method, which can effectively avoid damage to battery cells in the early stage of overvoltage by predicting the overvoltage trend in advance, thereby reducing battery degradation and failure risk.

[0005] Another object of the present invention is to provide a battery overvoltage prediction and control device.

[0006] Another object of the present invention is to provide an energy storage device.

[0007] To achieve the above objectives, the first aspect of the present invention provides a battery overvoltage prediction and control method, applied to an energy storage device. The energy storage device includes a battery management system and a battery module. The charging and discharging circuit of the battery module is equipped with a switching device and a fuse, both of which are communicatively connected to the battery management system. The battery overvoltage prediction and control method includes: collecting first time-series data of the battery module and second time-series data of the fuse, wherein the first time-series data includes the current individual cell voltage of each battery cell in the battery module and the current total voltage of the battery module, and the second time-series data includes the voltage difference across the fuse; inputting the first time-series data into a pre-constructed voltage prediction model to obtain the voltage change value of each battery cell after a preset time period; calculating the predicted individual cell voltage of each battery cell and the total predicted voltage of the battery module based on the current individual cell voltage and the voltage change value; identifying the operating state of the energy storage device based on the predicted individual cell voltage, the total predicted voltage, the current total voltage, and the voltage difference; and generating a first control command for the switching device or a second control command for the fuse based on the operating state of the energy storage device.

[0008] This invention aims to provide a battery overvoltage prediction and control method. Based on a voltage prediction model, it predicts the battery voltage (voltage of battery cells and voltage of battery modules) for a preset time period in the future, realizes the early prediction of battery overvoltage trend, replaces the traditional passive protection method, effectively avoids damage to battery cells in the early stage of overvoltage, and reduces battery degradation and failure risk.

[0009] In addition, by combining the predicted voltage (including individual predicted voltage and total predicted voltage), the current total voltage and the voltage difference across the fuse, the operating status of the equipment can be comprehensively identified. This can accurately distinguish various operating conditions and whether they are abnormal, improve the accuracy of status identification, and significantly reduce problems such as fuses blowing falsely or malfunctioning.

[0010] In some technical solutions, optionally, the first time-series data also includes charging current value, discharging current value, and temperature value; inputting the first time-series data into a pre-built voltage prediction model to obtain the voltage change value of each battery cell after a preset time period includes: inputting the charging current value, discharging current value, and temperature value into a pre-built voltage prediction model to obtain the voltage change value of each battery cell after a preset time period; wherein, under charging conditions, the voltage change value is positive, and the voltage change value is positively correlated with the charging current value and also positively correlated with the temperature value; under discharging conditions, the voltage change value is negative, and the absolute value of the voltage change value is positively correlated with the discharging current value and also positively correlated with the temperature value.

[0011] This technical solution introduces multi-dimensional characteristic parameters such as charging current, discharging current, and temperature as input parameters, breaking through the limitations of relying solely on voltage data for prediction. It fully aligns with the actual electrochemical characteristics and operating condition changes of the battery module, effectively improving the fitting accuracy and generalization ability of the voltage prediction model.

[0012] Furthermore, by clearly distinguishing the positive and negative attributes of voltage changes under charging and discharging conditions, and defining the correlation between voltage changes and charging current, discharging current, and temperature, the voltage prediction process is made more consistent with the actual working mechanism of the battery module, and the prediction results are more in line with the actual operating trend.

[0013] In some technical solutions, optionally, based on the current individual cell voltage and voltage change value, the predicted voltage of each battery cell and the total predicted voltage of the battery module are calculated, including: calculating the predicted voltage of each battery cell based on the current individual cell voltage and voltage change value, wherein the predicted voltage of each individual cell is equal to the sum of the current individual cell voltage and the voltage change value; and calculating the total predicted voltage of the battery module based on the predicted voltage of each battery cell, wherein the total predicted voltage is equal to the sum of the predicted voltages of all battery cells.

[0014] In this technical solution, the predicted voltage of a single cell is calculated by superimposing the current single cell voltage with the voltage change value. The calculation logic is simple and can accurately characterize the voltage change trend of the battery cell in the future.

[0015] Simultaneously, it outputs prediction results in two dimensions: single-cell predicted voltage and total predicted voltage. This enables accurate identification of potential overvoltage hazards in battery cells and also allows for prediction of the overall voltage status of the battery module, providing data support for subsequent multi-dimensional operational status identification.

[0016] In some technical solutions, optionally, the operating status of the energy storage device is identified based on the predicted voltage of a single cell, the total predicted voltage, the current total voltage, and the voltage difference. This includes: when the predicted voltage of a single cell is within a first preset voltage range and the total predicted voltage is within a second preset voltage range, the operating status is identified as normal; when the predicted voltage of a single cell is not within the first preset voltage range, if the current total voltage is within the second preset voltage range and the voltage difference is within a preset voltage difference range, the operating status is identified as sampling abnormal; when the predicted voltage of a single cell is greater than or equal to a first preset overvoltage threshold and / or the total predicted voltage is greater than or equal to a second preset overvoltage threshold, if the voltage difference is within a preset voltage difference range, the operating status is identified as overvoltage abnormal.

[0017] This technical solution integrates multi-dimensional information such as predicted voltage, current total voltage, and voltage difference for comprehensive judgment. It can accurately distinguish between normal state, sampled abnormal state, and actual overvoltage abnormal state under complex operating conditions, achieving a comprehensive and accurate characterization of the battery's operating state. This fundamentally improves the accuracy and reliability of state identification and enhances the safety control level of the battery management system.

[0018] In some technical solutions, optionally, the first control command includes a closing control command and a disconnection control command; based on the operating state of the energy storage device, a first control command for the switching device or a second control command for the fuse is generated, including: generating a closing control command for the switching device when the operating state is normal; generating a disconnection control command for the switching device when the operating state is an abnormal sampling state; and generating a disconnection control command for the switching device when the operating state is an abnormal overvoltage state.

[0019] In this technical solution, the first control command is refined into closing control command and opening control command. The command types are clearly defined and the control logic is well-organized, which facilitates the battery management system to accurately execute the corresponding actions. Differentiated control strategies are configured for normal state, sampling abnormal state, and overvoltage abnormal state to achieve precise control by state and level, avoiding the problems of over-protection or under-protection caused by uniform control.

[0020] In some technical solutions, the second control command may optionally include a fuse control command; based on the operating state of the energy storage device, a first control command for the switching device or a second control command for the fuse is generated, and the second control command for the fuse is also generated, so that the fuse is in a fuse-free state when the operating state is an overvoltage abnormal state.

[0021] In this technical solution, by setting a progressive control logic of "if the current still exceeds the limit after the switching device disconnection command is executed, the fuse will be triggered to blow", a graded protection mechanism of "normal protection, failure verification, and ultimate fuse blowing" is formed to ensure that under abnormal overvoltage conditions, the circuit can be effectively cut off regardless of whether the switching device is working normally, thus avoiding protection failure.

[0022] In some technical solutions, optionally, the battery overvoltage prediction and control method further includes: after generating a first control command for the switching device or a second control command for the fuse based on the operating state of the energy storage device, issuing a first reminder message when the operating state is in a sampling abnormal state; and issuing a second reminder message when the operating state is in an overvoltage abnormal state.

[0023] This technical solution distinguishes between two scenarios: sampling anomalies and overpressure anomalies, issuing different alert messages accordingly to accurately guide staff in troubleshooting. The first alert focuses on sampling faults, facilitating quick location and resolution of sampling deviation issues; the second alert focuses on overpressure risks, reminding staff to promptly address overpressure faults and check equipment status, improving problem-solving efficiency.

[0024] In some technical solutions, optionally, the battery overvoltage prediction and control method further includes: after generating a first control command for the switching device or a second control command for the fuse based on the operating status of the energy storage device, issuing a third reminder message when the fuse is in a blown state.

[0025] In this technical solution, the third reminder message accurately focuses on the fuse's blown state, clearly conveying the core message that "the circuit has been cut off and the fault needs to be investigated." This not only helps staff quickly locate the root cause of the fault but also provides clear guidance for fault handling and system recovery, reducing troubleshooting time.

[0026] A second aspect of the present invention provides a battery overvoltage prediction and control device, comprising: a data acquisition unit for acquiring first time-series data of a battery module and second time-series data of a fuse, wherein the first time-series data includes the current individual cell voltage of each battery cell in the battery module and the current total voltage of the battery module, and the second time-series data includes the voltage difference across the fuse; a voltage prediction unit for inputting the first time-series data into a pre-built voltage prediction model to obtain the voltage change value of each battery cell after a preset time period, and calculating the individual cell predicted voltage of each battery cell and the total predicted voltage of the battery module based on the current individual cell voltage and the voltage change value; and an identification and control unit for identifying the operating state of an energy storage device based on the individual cell predicted voltage, the total predicted voltage, the current total voltage, and the voltage difference, and generating a first control command for a switching device or a second control command for a fuse based on the operating state of the energy storage device.

[0027] The present invention aims to provide a battery overvoltage prediction and control device, which predicts the battery voltage (voltage of battery cells and voltage of battery modules) for a preset time period based on a voltage prediction model, so as to realize the early prediction of battery overvoltage trend, replace the traditional passive protection method, effectively avoid damage to battery cells in the early stage of overvoltage, and reduce battery degradation and failure risk.

[0028] In addition, by combining the predicted voltage (including individual predicted voltage and total predicted voltage), the current total voltage and the voltage difference across the fuse, the operating status of the equipment can be comprehensively identified. This can accurately distinguish various operating conditions and whether they are abnormal, improve the accuracy of status identification, and significantly reduce problems such as fuses blowing falsely or malfunctioning.

[0029] A third aspect of the present invention provides an energy storage device, including a battery management system and a battery module. The charging and discharging circuit of the battery module is provided with a switching device and a fuse, both of which are communicatively connected to the battery management system. The battery management system is used to execute the steps of the battery overvoltage prediction and control method of any of the technical solutions in the first aspect.

[0030] Energy storage devices have the beneficial effects of any of the technical solutions in the first aspect, which will not be elaborated here.

[0031] Additional aspects and advantages of the technical solutions of the present invention will become apparent in the following description or may be learned by practice of the invention. Attached Figure Description

[0032] Figure 1 A flowchart of a battery overvoltage prediction and control method according to an embodiment of the present invention is shown;

[0033] Figure 2A flowchart of a battery overvoltage prediction and control method according to another embodiment of the present invention is shown;

[0034] Figure 3 A flowchart of a battery overvoltage prediction and control method according to another embodiment of the present invention is shown;

[0035] Figure 4 A flowchart of a battery overvoltage prediction and control method according to another embodiment of the present invention is shown;

[0036] Figure 5 A flowchart of a battery overvoltage prediction and control method according to another embodiment of the present invention is shown;

[0037] Figure 6 A flowchart of a battery overvoltage prediction and control method according to another embodiment of the present invention is shown;

[0038] Figure 7 A flowchart of a battery overvoltage prediction and control method according to another embodiment of the present invention is shown;

[0039] Figure 8 A flowchart of a battery overvoltage prediction and control method according to another embodiment of the present invention is shown;

[0040] Figure 9 A structural block diagram of a battery overvoltage prediction and control device according to an embodiment of the present invention is shown;

[0041] Figure 10 A structural block diagram of an energy storage device according to an embodiment of the present invention is shown.

[0042] The attached figures are labeled as follows:

[0043] 200: Battery overvoltage prediction and control device; 210: Data acquisition unit; 220: Voltage prediction unit; 230: Identification and control unit; 300: Energy storage device; 310: Battery management system; 320: Battery module; 321: Battery cell; 330: Charging and discharging circuit; 340: Switching device; 350: Fuse. Detailed Implementation

[0044] To better understand the above-described objectives, features, and advantages of the embodiments of the present invention, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments of the present invention and the features thereof can be combined with each other.

[0045] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, embodiments of the invention may be implemented in other ways different from those described herein. Therefore, the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0046] Overvoltage faults are one of the core hidden dangers restricting the safe and stable operation of energy storage batteries. If the overvoltage state of the battery is not identified and effectively handled in time, it is very easy to cause the battery cells to bulge, which can lead to thermal runaway and fire, posing a significant safety hazard.

[0047] In related technologies, overvoltage protection for energy storage batteries is relatively lagging. It usually provides passive protection after the energy storage battery experiences overvoltage. Power is only triggered when the measured voltage of the battery cell or battery module reaches the preset overvoltage threshold. This makes it difficult to avoid damage to the energy storage battery in the early stages of overvoltage and the risk of accidental fuse blowing.

[0048] This invention aims to provide a battery overvoltage prediction and control method, device, and energy storage equipment. Based on a voltage prediction model, it predicts the battery voltage (voltage of battery cells and voltage of battery modules) for a preset time period in the future, realizes the early prediction of battery overvoltage trend, replaces the traditional passive protection method, effectively avoids damage to battery cells in the early stage of overvoltage, and reduces battery degradation and failure risk.

[0049] In addition, by combining the predicted voltage (including individual predicted voltage and total predicted voltage), the current total voltage and the voltage difference across the fuse, the operating status of the equipment can be comprehensively identified. This can accurately distinguish various operating conditions and whether they are abnormal, improve the accuracy of status identification, and significantly reduce problems such as fuses blowing falsely or malfunctioning.

[0050] In related technologies, fuses (FUSEs) are key actuators for battery overvoltage protection, and accurate determination of their melting state is a crucial prerequisite for the Battery Management System (BMS) to implement safe control. However, in actual operation, abnormal conditions such as sampling line breakage, poor contact, and data acquisition drift frequently occur, which can easily lead to misjudgments of the fuse's melting state by the BMS. This can result in safety risks such as false power outages, false alarms, or failure to disconnect power in a timely manner, seriously affecting the operational stability and safety of the energy storage system.

[0051] In the technical solution of this invention, by collecting the voltage difference across the fuse and combining it with the predicted voltage and the current total voltage for comprehensive status identification, the invention can accurately distinguish between sampling anomalies and the actual fuse failure state, fundamentally solving the problem of misjudgment caused by operating conditions such as broken sampling lines, poor contact, or data acquisition drift. Based on different operating states, corresponding control commands are generated, which can significantly reduce the probability of false power outages and false alarms, ensuring timely execution of protective actions in the event of actual overvoltage or fuse failure, and eliminating safety hazards caused by untimely power outages.

[0052] In related technologies, the representation of battery operating status is limited by a single dimension and the coverage of operating conditions is incomplete. Under complex operating conditions, it is easy to misjudge the status and affect the reliability of the battery protection system by failing to accurately distinguish between different state types such as normal operation, sampling abnormality and actual overvoltage.

[0053] In the technical solution of this invention, when the predicted voltage of a single cell is within a first preset voltage range and the total predicted voltage is within a second preset voltage range, the operating state is determined to be normal. When the predicted voltage of a single cell is not within the first preset voltage range, the current total voltage is within the second preset voltage range, and the voltage difference is within a preset voltage difference range, the operating state is determined to be a sampling abnormal state. When the predicted voltage of a single cell is greater than or equal to a first preset overvoltage threshold, and / or the total predicted voltage is greater than or equal to a second preset overvoltage threshold, if the voltage difference is within a preset voltage difference range, the operating state is determined to be an overvoltage abnormal state. When the operating state is an overvoltage abnormal state, if after generating a disconnection control command, the current value of the charging and discharging circuit is detected to be greater than or equal to a preset current threshold, the operating state is determined to still be an overvoltage abnormal state, and a fuse-breaking control command is generated to make the fuse in a fuse-breaking state.

[0054] By integrating multi-dimensional information such as predicted voltage, current total voltage, and voltage difference for comprehensive judgment, it can accurately distinguish between normal state, sampled abnormal state, and actual overvoltage abnormal state under complex operating conditions, achieving a comprehensive and accurate characterization of battery operating status, fundamentally improving the accuracy and reliability of state identification, and enhancing the safety control level of the battery management system.

[0055] The following reference Figures 1 to 10 This invention describes a battery overvoltage prediction and control method, apparatus, and energy storage device provided according to some embodiments of the present invention.

[0056] In one embodiment of the present invention, such as Figure 10 As shown, the energy storage device 300 includes a battery management system 310 and a battery module 320. The charging and discharging circuit 330 of the battery module 320 is equipped with a switching device 340 and a fuse 350. Both the switching device 340 and the fuse 350 are communicatively connected to the battery management system 310.

[0057] Optionally, the battery module 320 is composed of multiple battery cells 321 connected in series and / or in parallel. A switching device 340 and a fuse 350 are connected in series on the charging / discharging circuit 330 of the battery module 320. The switching device 340 is a controllable switching element for the charging / discharging circuit 330, and the fuse 350 is an overcurrent and ultimate safety protection element for the charging / discharging circuit 330. The control terminal of the switching device 340 is communicatively connected to the control output terminal of the battery management system 310 (BMS), and the status detection terminal of the fuse 350 is communicatively connected to the signal acquisition terminal of the battery management system 310, so as to realize the on / off control of the switching device 340 by the battery management system 310, and the real-time acquisition and detection of the operating status of the fuse 350.

[0058] In one embodiment of the present invention, the battery overvoltage prediction and control method is applied to the energy storage device 300.

[0059] like Figure 1 As shown, the battery overvoltage prediction and control method includes:

[0060] S102, collect the first timing data of the battery module and the second timing data of the fuse. The first timing data includes the current individual voltage of each battery cell in the battery module and the current total voltage of the battery module. The second timing data includes the voltage difference across the fuse.

[0061] The first time-series data is used to characterize the real-time operating status of the battery module. This data includes, but is not limited to, the current individual cell voltage of each battery cell in the battery module and the current total voltage of the battery module. The second time-series data is used to characterize the operating status of the fuse, specifically the real-time voltage difference across the fuse, to distinguish between sampling anomalies and the actual fuse failure state.

[0062] Optionally, a data acquisition module can collect multi-dimensional time-series data of the battery module and fuses in real time at a frequency of 1Hz, covering the entire life cycle of the battery module (0 to 3000 cycles) and all operating conditions (temperature range of -20℃ to 50℃, charge / discharge rate of 0.1C to 2C, covering normal charging and fast charging, etc.). The collected data is processed to remove transient glitches in voltage and current signals to ensure the accuracy of the collected data.

[0063] It should be noted that "C" in 0.1C and 2C is the unit symbol for charge / discharge rate. The larger the value, the faster the charging or discharging speed.

[0064] Optionally, the first time-series data may also include the number of battery cells, real-time charging and discharging current (including the value and direction of charging and discharging current), and real-time temperature data of the battery module.

[0065] S104, input the first time series data into the pre-built voltage prediction model to obtain the voltage change value of each battery cell after a preset time period in the future. Based on the current individual cell voltage and voltage change value, calculate the individual predicted voltage of each battery cell and the total predicted voltage of the battery module.

[0066] The collected time-series data is input into a pre-built and trained voltage prediction model. The voltage prediction model is used to predict the voltage change trend over time to obtain the voltage change value of each battery cell after a preset time period. Then, based on the current cell voltage, the predicted voltage of each battery cell is calculated in combination with the voltage change value. The total predicted voltage of the battery module is further calculated based on the predicted voltage of each cell, so as to realize the early prediction of the future overvoltage trend of the battery.

[0067] Optionally, the voltage prediction model is an AI (Artificial Intelligence) prediction model. The voltage prediction model uses an LSTM (Long Short-Term Memory) model, which can effectively process the time-series data of the battery module and accurately predict future voltage change trends.

[0068] During model training, preprocessed multi-dimensional time-series data is used as training samples. The model parameters are optimized through repeated iterations, enabling the model (voltage prediction model) to accurately predict the single-cell predicted voltage and total predicted voltage at three key time nodes in the future: 3s, 5s, and 10s. This allows for early prediction of battery overvoltage trends and lays the foundation for subsequent anomaly identification and decision control.

[0069] S106 identifies the operating status of the energy storage device based on the individual predicted voltage, total predicted voltage, current total voltage, and voltage difference, and generates a first control command for the switching device or a second control command for the fuse based on the operating status of the energy storage device.

[0070] By comprehensively utilizing the predicted voltage of individual cells, the total predicted voltage, the current total voltage, and the voltage difference across the fuse, the operating status of the energy storage device is comprehensively identified from multiple dimensions, accurately distinguishing between normal status, sampling abnormal status, and overvoltage abnormal status. Based on the identified operating status, corresponding control commands are adaptively generated, namely, a first control command for controlling the on / off state of switching devices, or a second control command for controlling the fuse to perform fuse protection, thereby achieving hierarchical, precise, and proactive safety protection control.

[0071] Optionally, the predicted voltage of a single cell, the total predicted voltage, the current total voltage, and the voltage difference are input into a pre-built AI classification model to obtain the operating status of the energy storage device. Optionally, the AI ​​classification model adopts a lightweight neural network structure, adapted to the operating requirements of the MCU (Microcontroller Unit) of the battery management system. During training, real-time data transmitted by the data acquisition module and voltage prediction data output by the voltage prediction model are fused as input features. Through sample labeling and model iteration, the AI ​​classification model can stably output four distinct battery operating states (operating states of the energy storage device): normal state, sampling abnormal state, overvoltage abnormal state, and fuse failure state, achieving accurate identification and differentiation of battery operating abnormality types.

[0072] Optionally, an AI fusion decision-making model is constructed and trained. The operating status of the energy storage device is input into the pre-constructed AI fusion decision-making model to generate a first control command for the switching devices or a second control command for the fuse. During training, the voltage prediction results of the voltage prediction model, the state recognition results of the AI ​​classification model, various preset threshold parameters (overvoltage threshold, differential voltage threshold, etc.), and the real-time state parameters of the fuse circuit are used as inputs. By constructing multi-dimensional decision logic and optimizing the decision algorithm, a comprehensive judgment on the battery operating status (the operating status of the energy storage device) is achieved.

[0073] If the AI ​​classification model outputs a sampling anomaly, and simultaneously detects that the current total voltage of the battery module is within the normal range (second preset voltage range), the voltage difference across the fuse does not fluctuate abnormally (voltage difference is within the preset voltage difference range), and the battery charging current remains normal and stable (without sudden increases or decreases), then it is determined that the sampling line is abnormal (including sampling line breakage, poor contact, and acquisition drift). At this time, the AI ​​fusion decision model outputs a sampling line alarm signal and cuts off the charging and discharging circuit in advance to prevent subsequent misjudgments caused by sampling anomalies, which helps ensure equipment safety.

[0074] If the predicted voltage of a single cell or the total predicted voltage within a preset time period (e.g., 3s, 5s, 10s) output by the voltage prediction model exceeds a preset overvoltage threshold (the predicted voltage of a single cell is greater than or equal to the first preset overvoltage threshold, and / or the total predicted voltage is greater than or equal to the second preset overvoltage threshold), and the state identification result output by the AI ​​classification model is an overvoltage abnormal state, and at the same time, no abnormal fluctuation is detected in the voltage difference across the fuse (the voltage difference is within the preset voltage difference range), then it is determined that the battery module has a real overvoltage risk (the operating state is an overvoltage abnormal state). At this time, the AI ​​fusion decision model outputs an overvoltage warning signal and triggers a control command to cut off the charging and discharging circuit in advance to prevent the overvoltage from further aggravating.

[0075] If the AI ​​fusion decision model has triggered the command to cut off the charging and discharging circuit in advance, and detects that the current value of the charging and discharging circuit is greater than or equal to the preset current threshold, and the battery module still has an overvoltage risk (voltage continues to exceed the standard), the system will automatically trigger the fuse's blowing mechanism to completely cut off the high-voltage circuit. When the detected voltage difference is greater than the preset voltage difference threshold, the AI ​​fusion decision model determines that the fuse has blown due to overvoltage and outputs a fuse alarm signal to remind staff to handle it in time.

[0076] It should be noted that the battery management system integrates all functions such as data acquisition, data transmission, signal communication, command execution, and fault alarm, ensuring smooth data interaction and efficient command execution between modules. After the platform is built, the trained voltage prediction model, AI classification model, and AI fusion decision model are uniformly deployed to the MCU (Microcontroller Unit) of the BMS (Battery Management System) to complete the integration and debugging of the entire system, ensuring stable model operation, accurate judgment, timely response, and adaptability to actual engineering application scenarios.

[0077] This invention aims to provide a battery overvoltage prediction and control method. Based on a voltage prediction model, it predicts the battery voltage (voltage of battery cells and voltage of battery modules) for a preset time period in the future, realizes the early prediction of battery overvoltage trend, replaces the traditional passive protection method, effectively avoids damage to battery cells in the early stage of overvoltage, and reduces battery degradation and failure risk.

[0078] In addition, by combining the predicted voltage (including individual predicted voltage and total predicted voltage), the current total voltage and the voltage difference across the fuse, the operating status of the equipment can be comprehensively identified. This can accurately distinguish various operating conditions and whether they are abnormal, improve the accuracy of status identification, and significantly reduce problems such as fuses blowing falsely or malfunctioning.

[0079] In some embodiments, optionally, the battery module uses lithium iron phosphate batteries. Raw data of lithium iron phosphate batteries under all energy storage operating conditions are collected, covering scenarios including:

[0080] Peak shaving conditions: from 23:00 at night to 7:00 the next day, constant current charging is performed at a rate of 0.3C (off-peak period); from 10:00 to 14:00 during the day, constant current discharging is performed at a rate of 0.4C (peak period).

[0081] Environmental conditions: The energy storage power station was simulated by adjusting the high and low temperature test chamber, covering the temperature range of -10℃ to 45℃ (including three key nodes: low temperature -10℃, normal temperature 25℃, and high temperature 45℃), and the relative humidity was controlled between 30% and 70%. Battery operation data under different temperature and humidity conditions were collected.

[0082] Aging conditions: The aging process of the battery throughout its entire life cycle was simulated by cyclic charging and discharging, and a total of 0 to 2000 cycles were completed. Data were collected for new batteries (0 cycles), moderately degraded batteries (1000 cycles, 20% capacity reduction), and deeply degraded batteries (2000 cycles, 40% capacity reduction).

[0083] Abnormal simulation conditions: The system simulates three typical scenarios: sampling line breakage, poor contact of sampling line, sampling drift, and fuse failure. The sampling line abnormality simulation includes three typical cases (complete breakage, excessive contact resistance, and ±5% drift of the acquired signal). The fuse failure simulation simulates a battery overvoltage-triggered failure scenario. Feature data under various abnormal conditions are collected for training the AI ​​classification model.

[0084] The data acquisition frequency is set to 1Hz. The acquired parameters (first time-series data) include the current individual cell voltage of each battery cell in the battery module, the number of battery cells, the current total voltage of the battery module, the real-time charging and discharging current (including the values ​​and directions of charging and discharging currents), the real-time voltage difference across the fuse, the real-time temperature of the battery module, and other parameters. Among these, the "other parameters" include: Direct Current Resistance (DCR) and capacity degradation rate.

[0085] It should be noted that DC internal resistance refers to the equivalent internal resistance exhibited by the battery module under DC operating conditions. It reflects the polarization characteristics and aging degree of the battery module. The more cycles the battery module has and the more severe the aging, the greater the DC internal resistance. The DC internal resistance is obtained through a pulse test method with a pulse current of 10A and a pulse duration of 10ms.

[0086] Furthermore, the capacity decay rate is expressed as ΔC / C0, where C0 is the initial capacity of 100 Ah and ΔC is the difference between the current capacity and the initial capacity.

[0087] The collected raw data is preprocessed in the following steps: anti-shake processing is used for 4 acquisition cycles (e.g., 4s) to remove instantaneous spikes in voltage and current signals; high-frequency interference signals are removed by low-pass filtering algorithm (cutoff frequency 10Hz); invalid abnormal data (such as abnormal data where the voltage is not in the range of 2.5V to 4.35V or the current change exceeds ±50A / s) are removed, and finally the effective data is selected for subsequent training, verification and testing of AI models.

[0088] Based on the collected and preprocessed effective data, a voltage prediction model, an AI classification model, and an AI fusion decision model were constructed. Model training and parameter optimization were then performed. The voltage prediction model training process is as follows: An LSTM (Long Short-Term Memory) model was used as the core prediction model, with the following specific configuration: The input layer of the model structure selected 6 key features (e.g., [T_now, dT / dt, U, dU / dt, I, DCR]) covering multiple dimensions such as temperature, voltage, current, and aging; the output layer output the predicted voltage of individual cells and the total predicted voltage of the battery module at three key time nodes in the future: 3s, 5s, and 10s. After the model training was completed, the prediction accuracy was verified through a test set. The results showed that the voltage prediction errors in the future 3s, 5s, and 10s were all less than or equal to 2%, with the prediction error less than or equal to 1.5% under normal temperature conditions and less than or equal to 2% under low temperature (e.g., -10℃) and high temperature (e.g., 45℃) conditions, meeting the requirements for early warning of overvoltage.

[0089] It should be noted that "T_now" represents temperature; "dT / dt" represents the rate of temperature change; "U" represents voltage; "dU / dt" represents the rate of voltage change; "I" represents current; and "DCR" represents DC internal resistance.

[0090] The training process of the AI ​​classification model is as follows: The input layer of the model structure selects key features, covering the predicted voltage (individual predicted voltage and predicted total voltage), current individual voltage, current total voltage, voltage change rate (dU / dt), voltage difference across the fuse, charging and discharging current, DC internal resistance (DCR), sampling line voltage, and real-time temperature output by the AI ​​classification model, to achieve multi-dimensional feature fusion; The output layer outputs four distinct state classification results: normal state, sampling abnormal state, overvoltage abnormal state, and fuse blown state; After the model training is completed, the classification accuracy is verified through a test set. The results show that the overall anomaly recognition accuracy is greater than or equal to 98.5%, of which the recognition accuracy of sampling abnormal state is greater than or equal to 99% (covering three scenarios: sampling line breakage, poor contact, and acquisition drift), the recognition accuracy of fuse blown state is greater than or equal to 99.5%, and the recognition accuracy of overvoltage abnormal state is ≥98%. This method can accurately distinguish various abnormal states and meet the needs of anomaly recognition.

[0091] The AI-integrated decision-making model serves as the core decision-making unit. Input parameters include: voltage prediction results from the voltage prediction model, state classification results from the AI ​​classification model, and real-time state parameters of the fuse circuit. Sample data from various operating conditions are used to verify the accuracy of the decision-making logic, ensuring precise execution of the three core logics: real overvoltage judgment, sampling line anomaly judgment, and fuse status judgment.

[0092] The above technical solutions can be implemented and validated under all operating conditions and scenarios, with a focus on testing and verifying the three core functions: overvoltage prediction, anomaly identification, and fuse status determination. Simultaneously, the overall stability and reliability of the system are verified. Based on AI-driven prediction, the technical solution effectively reduces the need for manual intervention, enabling intelligent empowerment of the entire process of overvoltage prediction, anomaly identification, and decision control, significantly improving the automation level and protection reliability of battery module safety management.

[0093] By predicting battery overvoltage trends 3 to 10 seconds in advance using a voltage prediction model, proactive warnings can be achieved. This can trigger operations such as power reduction and early warning in advance, avoiding damage to the battery module from overvoltage. At the same time, it reduces the probability of fuses blowing falsely, which is beneficial to improving the service life of the battery module.

[0094] In some embodiments, the first timing data may optionally include charging current value, discharging current value, and temperature value.

[0095] like Figure 2 As shown, the first time-series data is input into a pre-built voltage prediction model to obtain the voltage change value of each battery cell after a preset time period, including:

[0096] S1042 inputs the charging current value, discharging current value, and temperature value into the pre-built voltage prediction model to obtain the voltage change value of each battery cell after a preset time period.

[0097] In the charging condition, the voltage change value is positive, and the voltage change value is positively correlated with the charging current value and the temperature value. In the discharging condition, the voltage change value is negative, and the absolute value of the voltage change value is positively correlated with the discharging current value and the temperature value.

[0098] Optionally, the collected charging current value, discharging current value, and temperature value are input into the trained voltage prediction model. Based on the voltage prediction model's ability to fit the time-series correlation of multiple parameters, the voltage change value corresponding to each battery cell after a preset time period is output.

[0099] It should be noted that under charging conditions, the battery voltage generally shows an upward trend, with positive voltage changes. Under these conditions, the voltage change is positively correlated with both the charging current and temperature. Under discharging conditions, the battery voltage generally shows a downward trend, with negative voltage changes. Under these conditions, the absolute value of the voltage change is positively correlated with both the discharging current and temperature.

[0100] By introducing multi-dimensional characteristic parameters such as charging current, discharging current, and temperature as input parameters, the limitations of relying solely on voltage data for prediction are overcome. This fully aligns with the actual electrochemical characteristics and operating condition changes of the battery module, effectively improving the fitting accuracy and generalization ability of the voltage prediction model.

[0101] Furthermore, by clearly distinguishing the positive and negative attributes of voltage changes under charging and discharging conditions, and defining the correlation between voltage changes and charging current, discharging current, and temperature, the voltage prediction process is made more consistent with the actual working mechanism of the battery module, and the prediction results are more in line with the actual operating trend.

[0102] In some embodiments, optionally, such as Figure 3 As shown, based on the current individual cell voltage and voltage change value, the predicted voltage of each battery cell and the total predicted voltage of the battery module are calculated, including:

[0103] S1044, based on the current cell voltage and voltage change value, calculates the predicted cell voltage of each battery cell, where the predicted cell voltage is equal to the sum of the current cell voltage and the voltage change value.

[0104] Optionally, the predicted voltage of each battery cell under a preset future time period is calculated by superimposing the current single cell voltage collected in real time as the reference value and the voltage change value predicted by the voltage prediction model.

[0105] S1046, based on the predicted voltage of each individual battery cell, calculate the total predicted voltage of the battery module, where the total predicted voltage is equal to the sum of the predicted voltages of all individual battery cells.

[0106] Optionally, after obtaining the predicted voltage of each individual battery cell, the predicted voltages of all individual battery cells are summed to obtain the total predicted voltage of the entire battery module.

[0107] Optionally, the first time-series data also includes the number of battery cells. Based on the predicted voltage of each individual battery cell and the number of battery cells, the total predicted voltage of the battery module is calculated (by summing).

[0108] The total predicted voltage of the battery module is obtained by summing the predicted voltages of each individual battery cell, ensuring logical consistency and data correlation between the individual cell predictions and the overall module prediction.

[0109] The predicted voltage of a single cell is calculated by superimposing the current single cell voltage with the voltage change value. The calculation logic is simple and can accurately characterize the voltage change trend of the battery cell in the future.

[0110] Simultaneously, it outputs prediction results in two dimensions: single-cell predicted voltage and total predicted voltage. This enables accurate identification of potential overvoltage hazards in battery cells and also allows for prediction of the overall voltage status of the battery module, providing data support for subsequent multi-dimensional operational status identification.

[0111] In some embodiments, optionally, such as Figure 4 As shown, based on the predicted voltage of individual cells, the total predicted voltage, the current total voltage, and the voltage difference, the operating status of the energy storage device is identified, including:

[0112] S1062, when the predicted voltage of a single unit is within the first preset voltage range and the total predicted voltage is within the second preset voltage range, the operating state is identified as normal.

[0113] When the predicted voltage of each individual battery cell falls within the first preset voltage range and the total predicted voltage of the battery module is within the second preset voltage range, it indicates that the future voltage trends of the battery cells and the battery module are within a safe and reasonable range, with no overvoltage risk, and the energy storage device is judged to be in normal operating condition.

[0114] S1064, when the predicted voltage of a single cell is not within the first preset voltage range, if the current total voltage is within the second preset voltage range and the voltage difference is within the preset voltage difference range, then the operating state is identified as an abnormal sampling state.

[0115] When the predicted voltage of a single cell deviates from the first preset voltage range and shows an abnormal deviation trend, if the current total voltage of the battery module is still maintained within the normal range of the second preset voltage range, and the voltage difference across the fuse is stable within the preset voltage difference range, it indicates that there is no real overvoltage abnormality in the actual operating condition of the battery module. The abnormal voltage phenomenon is caused by sampling faults such as sampling line breakage, poor contact, and acquisition drift. The operating state of the energy storage device is determined to be an abnormal sampling state.

[0116] Optionally, if the status identification result indicates an abnormal sampling state, and the current total voltage of the battery module is detected to be within the normal range (second preset voltage range), the voltage difference across the fuse does not fluctuate abnormally (voltage difference is within the preset voltage difference range), and the battery charging current remains normal and stable (without sudden increases or decreases), then it is determined that the sampling line is abnormal (including sampling line breakage, poor contact, and sampling drift). If an abnormality is determined in the sampling line, an alarm signal is output, and the charging / discharging circuit is disconnected in advance to prevent subsequent misjudgments caused by the sampling abnormality, thus helping to ensure equipment safety.

[0117] S1066, when the predicted voltage of a single unit is greater than or equal to the first preset overvoltage threshold, and / or the total predicted voltage is greater than or equal to the second preset overvoltage threshold, if the voltage difference is within the preset voltage difference range, the operating state is identified as an overvoltage abnormal state.

[0118] When the predicted voltage of a single cell is greater than or equal to the first preset overvoltage threshold, and / or the total predicted voltage of the battery module is greater than or equal to the second preset overvoltage threshold, and the voltage difference across the fuse is within the preset voltage difference range, indicating that the fuse has not blown, it is determined that the battery cell or battery module has experienced or will experience a substantial overvoltage trend, and the operating state of the energy storage device is an overvoltage abnormal state.

[0119] Optionally, if the predicted voltage of a single cell or the total predicted voltage within a preset future time period (e.g., 3s, 5s, 10s) output by the voltage prediction model is greater than a preset overvoltage threshold (the predicted voltage of a single cell is greater than or equal to a first preset overvoltage threshold, and / or the total predicted voltage is greater than or equal to a second preset overvoltage threshold), and the state identification result is an overvoltage abnormal state, and at the same time, no abnormal fluctuation is detected in the voltage difference across the fuse (the voltage difference is within a preset voltage difference range), then it is determined that the battery module has a real overvoltage risk (the operating state is an overvoltage abnormal state). In the case of a real overvoltage risk in the battery module, an overvoltage warning signal is output, and a control command to cut off the charging and discharging circuit in advance is triggered to prevent the overvoltage from further aggravating.

[0120] By integrating multi-dimensional information such as predicted voltage, current total voltage, and voltage difference for comprehensive judgment, it can accurately distinguish between normal state, sampled abnormal state, and actual overvoltage abnormal state under complex operating conditions, achieving a comprehensive and accurate characterization of battery operating status, fundamentally improving the accuracy and reliability of state identification, and enhancing the safety control level of the battery management system.

[0121] In some embodiments, the first control command may optionally include a closing control command and a disconnection control command.

[0122] like Figure 5 As shown, based on the operating status of the energy storage device, a first control command is generated for the switching device, or a second control command is generated for the fuse, including:

[0123] S1067 generates a closing control command for the switching device when the operating state is normal.

[0124] When the energy storage device is in normal condition, it indicates that the voltage trend of the battery cell and battery module is stable and there is no risk of overvoltage. The circuit operation is safe and reliable. At this time, a closing control command is generated for the switching device to maintain the normal conduction of the charging and discharging circuit and ensure the continuous and stable operation of the energy storage device.

[0125] S1068 generates a disconnection control command for the switching device when the operating state is a sampling abnormal state.

[0126] When an energy storage device is in an abnormal sampling state, it indicates that the abnormal voltage phenomenon is caused by sampling faults such as sampling disconnection, poor contact, or sampling drift, and is not a true overvoltage of the battery. However, the abnormal sampling will lead to the distortion of subsequent state judgment and the existence of uncertainty risk in control. At this time, a disconnection control command is generated for the switching device to cut off the charging and discharging circuit in time and avoid safety hazards caused by sampling misjudgment.

[0127] S1069 generates a disconnection control command for the switching device when the operating state is an overvoltage abnormality.

[0128] When an energy storage device is in an overvoltage abnormal state, it indicates that the battery cell or battery module has or will have a substantial overvoltage trend. If not dealt with in time, it may cause safety accidents such as battery bulging and thermal runaway. At this time, a disconnection control command is generated for the switching device to quickly cut off the charging and discharging circuit and curb the further expansion of the overvoltage fault from the source.

[0129] The first control command is refined into closing control commands and opening control commands, with clear command type classification and well-organized control logic, facilitating the battery management system to accurately execute corresponding actions. Differentiated control strategies are configured for normal state, sampling abnormal state, and overvoltage abnormal state to achieve precise control based on state and level, avoiding over-protection or under-protection problems caused by uniform control.

[0130] In some embodiments, the second control command may optionally include a circuit breaker control command.

[0131] like Figure 6 As shown, based on the operating status of the energy storage device, generating a first control command for the switching device or a second control command for the fuse also includes:

[0132] S1070 When the operating state is an overvoltage abnormal state, if after generating the disconnection control command, the current value of the charging and discharging circuit is detected to be greater than or equal to the preset current threshold, then a fuse-breaking control command is generated to make the fuse in the fuse-breaking state.

[0133] When an energy storage device is identified as being in an overvoltage abnormal state, the battery management system (BMS) first generates a disconnection control command for the switching devices, instructing them to quickly cut off the charging and discharging circuit to curb the spread of the overvoltage fault and protect the battery module and related components. Simultaneously, the BMS monitors the current value of the charging and discharging circuit in real time to verify the effectiveness of the disconnection execution of the switching devices. If the detected current value of the charging and discharging circuit is still greater than or equal to a preset current threshold, it indicates a fault in the switching devices (such as contact adhesion, control failure, etc.) that prevents effective circuit disconnection. In this case, the BMS immediately generates a fuse-breaking control command, triggering the fuse to perform a melting action, thus completely cutting off the charging and discharging circuit and achieving ultimate safety protection under overvoltage abnormal conditions.

[0134] By setting a progressive control logic that triggers the fuse to blow if the current still exceeds the limit after the switching device disconnects the command, a graded protection mechanism of "normal protection, failure verification, and ultimate fuse blowing" is formed to ensure that the circuit can be effectively cut off under abnormal overvoltage conditions, regardless of whether the switching device is working normally, thus avoiding protection failure.

[0135] In some embodiments, optionally, such as Figure 7 As shown, after S106 (identifying the operating status of the energy storage device based on the predicted voltage of a single cell, the total predicted voltage, the current total voltage, and the voltage difference, and generating a first control command for the switching device or a second control command for the fuse based on the operating status of the energy storage device), the battery overvoltage prediction and control method further includes:

[0136] S1082, when the running status is in a sampling abnormal state, issues the first reminder message.

[0137] If the operating status of the energy storage device is determined to be an abnormal sampling state (voltage abnormality caused by sampling failure, not actual overvoltage), the first alert message will be issued to prompt the staff to promptly check the sampling equipment for faults and calibrate the sampling accuracy to avoid subsequent control errors due to sampling deviations.

[0138] S1084, when the operating status is an overvoltage abnormality, issues a second warning message.

[0139] If the operating status of the energy storage device is determined to be an overvoltage abnormality, a second alert message will be issued to promptly remind staff to pay attention to the overvoltage fault and the operating status of the equipment, so as to facilitate rapid response and timely handling of potential safety hazards, and ensure that the entire control process forms a closed loop.

[0140] By differentiating between sampling anomalies and overpressure anomalies, different alert messages are issued accordingly, accurately guiding staff to troubleshoot problems. The first alert focuses on sampling faults, facilitating quick location and resolution of sampling deviation issues; the second alert focuses on overpressure risks, reminding staff to promptly address overpressure faults and check equipment status, improving problem-solving efficiency.

[0141] In some embodiments, optionally, such as Figure 8 As shown, after S106 (identifying the operating status of the energy storage device based on the predicted voltage of a single cell, the total predicted voltage, the current total voltage, and the voltage difference, and generating a first control command for the switching device or a second control command for the fuse based on the operating status of the energy storage device), the battery overvoltage prediction and control method further includes:

[0142] S1086, when the fuse is in a blown state, issue a third reminder message.

[0143] When the system detects that the fuse is in a blown state (i.e., the fuse has executed its blowing action and completed the circuit disconnection protection), it automatically issues a third reminder message. This third reminder message can be simultaneously pushed to the battery management system's backend, the field control terminal, and the mobile devices of relevant personnel. It clearly informs them that the fuse has triggered its blown protection and is currently in a blown state, while also clearly indicating the background conditions that triggered the blown fuse (e.g., overvoltage trigger, current exceeding limit trigger). This allows personnel to quickly grasp the circuit status, promptly investigate the root cause of overvoltage, and handle related faults. After the fault is resolved, targeted follow-up operations such as fuse reset and system restart can be performed, ensuring that the entire protection process is closed-loop and controllable.

[0144] The third reminder message precisely focuses on the fuse's blown state, clearly conveying the core message that "the circuit has been cut off and the fault needs to be investigated." This not only helps staff quickly locate the root cause of the fault but also provides clear guidance for fault handling and system recovery, reducing troubleshooting time.

[0145] In one embodiment of the present invention, such as Figure 9 As shown, the battery overvoltage prediction and control device 200 includes a data acquisition unit 210, a voltage prediction unit 220, and an identification and control unit 230.

[0146] The data acquisition unit 210 is used to acquire the first timing data of the battery module 320 and the second timing data of the fuse 350. The first timing data includes the current single cell voltage of each battery cell 321 in the battery module 320 and the current total voltage of the battery module 320. The second timing data includes the voltage difference across the fuse 350.

[0147] The first time-series data is used to characterize the real-time operating status of the battery module 320. This first time-series data includes, but is not limited to, the current individual cell voltage of each battery cell 321 in the battery module 320 and the current total voltage of the battery module 320. The second time-series data is used to characterize the operating status of the fuse 350, specifically the real-time voltage difference across the fuse 350, to distinguish between sampling anomalies and the actual blowing state of the fuse 350.

[0148] Optionally, a data acquisition module can be used to collect multi-dimensional time-series data of the battery module 320 and fuse 350 in real time at a frequency of 1Hz, covering the entire life cycle of the battery module 320 (0 to 3000 cycles) and all operating conditions (temperature range of -20℃ to 50℃, charge / discharge rate of 0.1C to 2C, covering normal charging and fast charging, etc.). The collected data is processed to remove transient glitches in the voltage and current signals to ensure the accuracy of the collected data.

[0149] It should be noted that "C" in 0.1C and 2C is the unit symbol for charge / discharge rate. The larger the value, the faster the charging or discharging speed.

[0150] Optionally, the first timing data also includes the number of battery cells 321, real-time charging and discharging current (including the value and direction of charging current and discharging current), and real-time temperature data of battery module 320.

[0151] The voltage prediction unit 220 is used to input the first time series data into the pre-built voltage prediction model to obtain the voltage change value of each battery cell 321 after a preset time period in the future. Based on the current single cell voltage and voltage change value, the single cell predicted voltage of each battery cell 321 and the total predicted voltage of the battery module 320 are calculated.

[0152] The collected first time-series data is input into a pre-built and trained voltage prediction model. The voltage prediction model performs time-series prediction of voltage change trends to obtain the voltage change value of each battery cell 321 after a preset time period. Then, based on the current single cell voltage, the voltage change value is combined to calculate the single cell predicted voltage corresponding to each battery cell 321. Based on the single cell predicted voltage, the total predicted voltage of the battery module 320 is further calculated, realizing the early prediction of the battery's future overvoltage trend.

[0153] Optionally, the voltage prediction model is an AI (Artificial Intelligence) prediction model. The voltage prediction model uses an LSTM (Long Short-Term Memory) model, which can effectively process the time-series data of battery module 320 and accurately predict future voltage change trends.

[0154] During model training, preprocessed multi-dimensional time-series data is used as training samples. The model parameters are optimized through repeated iterations, enabling the model (voltage prediction model) to accurately predict the single-cell predicted voltage and total predicted voltage at three key time nodes in the future: 3s, 5s, and 10s. This allows for early prediction of battery overvoltage trends and lays the foundation for subsequent anomaly identification and decision control.

[0155] The identification and control unit 230 is used to identify the operating status of the energy storage device 300 based on the individual predicted voltage, the total predicted voltage, the current total voltage, and the voltage difference, and to generate a first control command for the switching device 340 or a second control command for the fuse 350 based on the operating status of the energy storage device 300.

[0156] By comprehensively utilizing the predicted voltage of individual cells, the total predicted voltage, the current total voltage, and the voltage difference across the fuse 350, the operating status of the energy storage device 300 is comprehensively identified from multiple dimensions, accurately distinguishing between normal status, sampling abnormal status, and overvoltage abnormal status. Based on the identified operating status, corresponding control commands are adaptively generated, namely, a first control command for controlling the on / off state of the switching device 340, or a second control command for controlling the fuse 350 to perform fuse protection, thereby achieving hierarchical, precise, and proactive safety protection control.

[0157] Optionally, the predicted voltage of a single cell, the total predicted voltage, the current total voltage, and the voltage difference are input into a pre-built AI classification model to obtain the operating status of the energy storage device 300. Optionally, the AI ​​classification model adopts a lightweight neural network structure to adapt to the operating requirements of the MCU (Microcontroller Unit) of the battery management system 310. During training, the real-time data transmitted by the data acquisition module and the voltage prediction data output by the voltage prediction model are fused as input features. Through sample labeling and model iteration, the AI ​​classification model can stably output four distinct battery operating states (operating states of the energy storage device 300): normal state, sampling abnormal state, overvoltage abnormal state, and fuse failure state, achieving accurate identification and differentiation of battery operating abnormality types.

[0158] Optionally, an AI fusion decision-making model is constructed and trained. The operating status of the energy storage device 300 is input into the pre-constructed AI fusion decision-making model to generate a first control command for the switching device 340 or a second control command for the fuse 350. During training, the voltage prediction results of the voltage prediction model, the state recognition results of the AI ​​classification model, various preset threshold parameters (overvoltage threshold, differential voltage threshold, etc.), and the real-time state parameters of the fuse 350 circuit are used as inputs. By constructing multi-dimensional decision logic and optimizing the decision algorithm, a comprehensive judgment of the battery operating status (the operating status of the energy storage device 300) is achieved.

[0159] If the AI ​​classification model outputs a sampling anomaly, and simultaneously detects that the current total voltage of the battery module 320 is within the normal range (second preset voltage range), the voltage difference across the fuse 350 shows no abnormal fluctuations (voltage difference is within the preset voltage difference range), and the battery charging current remains normal and stable (without sudden increases or decreases), then it is determined that the sampling line is abnormal (including sampling line breakage, poor contact, and acquisition drift). At this time, the AI ​​fusion decision model outputs a sampling line alarm signal and prematurely cuts off the charging and discharging circuit 330 to prevent subsequent misjudgments caused by sampling anomalies, which helps ensure equipment safety.

[0160] If the predicted voltage of a single cell or the total predicted voltage within a preset time period (e.g., 3s, 5s, 10s) output by the voltage prediction model is greater than a preset overvoltage threshold (the predicted voltage of a single cell is greater than or equal to the first preset overvoltage threshold, and / or the total predicted voltage is greater than or equal to the second preset overvoltage threshold), and the state identification result output by the AI ​​classification model is an overvoltage abnormal state, and at the same time, no abnormal fluctuation is detected in the voltage difference across the fuse 350 (the voltage difference is within the preset voltage difference range), then it is determined that the battery module 320 has a real overvoltage risk (the operating state is an overvoltage abnormal state). At this time, the AI ​​fusion decision model outputs an overvoltage warning signal and triggers a control command to cut off the charging and discharging circuit in advance to prevent the overvoltage from further aggravating.

[0161] If the AI ​​fusion decision model has triggered the instruction to cut off the charging and discharging circuit 330 in advance, and detects that the current value of the charging and discharging circuit is greater than or equal to the preset current threshold, and the battery module 320 still has an overvoltage risk (voltage continues to exceed the standard), the system will automatically trigger the fuse 350's fuse-breaking mechanism to completely cut off the high-voltage circuit. When the detected voltage difference is greater than the preset voltage difference threshold, the AI ​​fusion decision model determines that the fuse 350 has blown due to overvoltage and outputs a fuse-breaking alarm signal to remind the staff to handle it in time.

[0162] It should be noted that the Battery Management System 310 integrates full-function functions such as data acquisition, data transmission, signal communication, command execution, and fault alarm, ensuring smooth data interaction and efficient command execution between modules. After the platform is built, the trained voltage prediction model, AI classification model, and AI fusion decision model are uniformly deployed to the MCU (Microcontroller Unit) of the BMS (Battery Management System 310) to complete the integration and debugging of the entire system, ensuring stable model operation, accurate judgment, timely response, and adaptability to actual engineering application scenarios.

[0163] The present invention aims to provide a battery overvoltage prediction and control device 200, which predicts the battery voltage (voltage of battery cell 321 and voltage of battery module 320) for a preset time period based on a voltage prediction model, so as to realize the early prediction of battery overvoltage trend, replace the traditional passive protection method, effectively avoid damage to battery cell 321 in the early stage of overvoltage, and reduce battery degradation and failure risk.

[0164] In addition, by combining the predicted voltage (including individual predicted voltage and total predicted voltage), the current total voltage and the voltage difference across the fuse 350, the operating status of the equipment can be comprehensively identified. This can accurately distinguish various operating conditions and whether they are abnormal, improve the accuracy of status identification, and significantly reduce problems such as false fuse 350 blowing or false protection.

[0165] In one embodiment of the present invention, such as Figure 10 As shown, the energy storage device 300 includes a battery management system 310 and a battery module 320. The charging / discharging circuit 330 of the battery module 320 is equipped with a switching device 340 and a fuse 350, both of which are communicatively connected to the battery management system 310. The battery management system 310 is used to execute the steps of the battery overvoltage prediction and control method in any of the above embodiments.

[0166] The energy storage device 300 has the beneficial effects of any of the above embodiments, which will not be repeated here.

[0167] In this invention, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance; the term "multiple" refers to two or more unless otherwise explicitly defined. The terms "install," "connect," "link," and "fix" should be interpreted broadly. For example, "connect" can be a fixed connection, a detachable connection, or an integral connection; "link" can be a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0168] In the description of this invention, it should be understood that the terms "upper," "lower," "left," "right," "front," "rear," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or unit referred to must have a specific orientation or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0169] In the description of this specification, the terms "one embodiment," "some embodiments," "specific embodiment," etc., refer to a specific feature, structure, material, or characteristic described in connection with that embodiment or example, which is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0170] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A battery overvoltage prediction and control method, characterized by, This invention is applied to energy storage devices, which include a battery management system and a battery module. The charging and discharging circuit of the battery module is equipped with a switching device and a fuse, and both the switching device and the fuse are communicatively connected to the battery management system. The battery overvoltage prediction and control method includes: The first timing data of the battery module and the second timing data of the fuse are collected. The first timing data includes the current individual voltage of each battery cell in the battery module and the current total voltage of the battery module. The second timing data includes the voltage difference across the fuse. The first time-series data is input into a pre-built voltage prediction model to obtain the voltage change value of each battery cell after a preset time period. Based on the current single cell voltage and the voltage change value, the single cell predicted voltage of each battery cell and the total predicted voltage of the battery module are calculated. Based on the predicted voltage of the individual cell, the total predicted voltage, the current total voltage, and the voltage difference, the operating state of the energy storage device is identified, and based on the operating state of the energy storage device, a first control command for the switching device or a second control command for the fuse is generated.

2. The battery overvoltage prediction and control method of claim 1, wherein The first timing data also includes charging current value, discharging current value, and temperature value; The step of inputting the first time-series data into a pre-built voltage prediction model to obtain the voltage change value of each battery cell after a preset time period includes: The charging current value, the discharging current value, and the temperature value are input into the pre-built voltage prediction model to obtain the voltage change value of each battery cell after the preset time period in the future; Specifically, during charging, the voltage change value is positive, and the voltage change value is positively correlated with the charging current value and the temperature value; during discharging, the voltage change value is negative, and the absolute value of the voltage change value is positively correlated with the discharging current value and the temperature value.

3. The battery overvoltage prediction and control method of claim 1, wherein The calculation of the predicted voltage of each battery cell and the total predicted voltage of the battery module based on the current individual cell voltage and the voltage change value includes: Based on the current cell voltage and the voltage change value, the predicted cell voltage of each battery cell is calculated, wherein the predicted cell voltage is equal to the sum of the current cell voltage and the voltage change value; Based on the predicted voltage of each individual battery cell, the total predicted voltage of the battery module is calculated, wherein the total predicted voltage is equal to the sum of the predicted voltages of all the individual battery cells.

4. The battery overvoltage prediction and control method according to any one of claims 1 to 3, characterized by, The step of identifying the operating status of the energy storage device based on the single-unit predicted voltage, the total predicted voltage, the current total voltage, and the voltage difference includes: When the predicted voltage of a single unit is within a first preset voltage range and the total predicted voltage is within a second preset voltage range, the operating state is identified as normal. When the predicted voltage of a single unit is not within the first preset voltage range, if the current total voltage is within the second preset voltage range and the voltage difference is within the preset voltage difference range, then the operating state is identified as a sampling abnormal state. When the predicted voltage of a single unit is greater than or equal to a first preset overvoltage threshold, and / or the total predicted voltage is greater than or equal to a second preset overvoltage threshold, if the voltage difference is within the preset voltage difference range, the operating state is identified as an overvoltage abnormal state.

5. The battery overvoltage prediction and control method according to claim 4, characterized in that, The first control command includes a closing control command and a closing control command; The step of generating a first control command for the switching device or a second control command for the fuse based on the operating status of the energy storage device includes: When the operating state is the normal state, the closing control command for the switching device is generated; When the operating state is the sampling abnormal state, the disconnection control command for the switching device is generated; When the operating state is the overvoltage abnormal state, the disconnection control command for the switching device is generated.

6. The battery overvoltage prediction and control method of claim 5, wherein The second control command includes a fuse control command; The step of generating a first control command for the switching device or a second control command for the fuse based on the operating status of the energy storage device further includes: When the operating state is the overvoltage abnormal state, if after generating the disconnection control command, the current value of the charging and discharging circuit is detected to be greater than or equal to a preset current threshold, then the fuse-breaking control command is generated to put the fuse in the fuse-breaking state.

7. The battery overvoltage prediction and control method according to claim 5, characterized in that, Also includes: After generating a first control command for the switching device or a second control command for the fuse based on the operating state of the energy storage device, a first reminder message is issued if the operating state is the sampling abnormal state; and a second reminder message is issued if the operating state is the overvoltage abnormal state.

8. The battery overvoltage prediction and control method according to claim 6, characterized in that, Also includes: After generating a first control command for the switching device or a second control command for the fuse based on the operating state of the energy storage device, a third reminder message is issued if the fuse is in a blown state.

9. A battery overvoltage prediction and control device, characterized by, include: The data acquisition unit is used to acquire first timing data of the battery module and second timing data of the fuse. The first timing data includes the current individual voltage of each battery cell in the battery module and the current total voltage of the battery module. The second timing data includes the voltage difference across the fuse. The voltage prediction unit is used to input the first time-series data into a pre-built voltage prediction model to obtain the voltage change value of each battery cell after a preset time period in the future. Based on the current single cell voltage and the voltage change value, the unit calculates the single cell predicted voltage of each battery cell and the total predicted voltage of the battery module. The identification and control unit is used to identify the operating status of the energy storage device based on the single-cell predicted voltage, the total predicted voltage, the current total voltage, and the voltage difference, and to generate a first control command for the switching device or a second control command for the fuse based on the operating status of the energy storage device.

10. An energy storage device, characterized by, It includes a battery management system and a battery module. The charging and discharging circuit of the battery module is equipped with a switching device and a fuse. The switching device and the fuse are both communicatively connected to the battery management system. The battery management system is used to perform the steps of the battery overvoltage prediction and control method as described in any one of claims 1 to 8.

Images