Method, system and equipment for detecting lithium precipitation of lithium battery and storage medium

Abstract

The invention belongs to the field of battery detection, and relates to a detection method, system and device for lithium precipitation of a lithium battery and a storage medium, and the method comprises the steps: obtaining voltage time sequence data of the lithium battery in a standing process after the lithium battery is charged to a full charge state; smoothing the voltage time sequence data, and performing differential calculation on the smoothed voltage time sequence data to generate voltage change rate characteristics of the lithium battery; the voltage change rate features are matched with features in a lithium precipitation feature library, a lithium precipitation judgment result is generated according to a matching result, and the lithium precipitation feature library is used for storing voltage change rate feature data representing different lithium precipitation risks; and outputting a lithium precipitation judgment result. The method can better adapt to the lithium separation detection requirements of the lithium battery in different scenes, and meets the requirements for timeliness and accuracy of lithium separation detection of the lithium battery as much as possible.

Description

Technical Field

[0001] This application relates to the technical field of battery testing, and in particular to a method, system, device and storage medium for detecting lithium plating in lithium batteries. Background Technology

[0002] With the rapid development of new energy technologies, lithium batteries, with their advantages of high energy density and long cycle life, are widely used in electric vehicles, energy storage devices, portable electronic terminals, and other fields. During long-term charge-discharge cycles, lithium plating can occur, causing lithium dendrites to gradually grow and potentially puncture the separator, leading to safety hazards such as internal short circuits and thermal runaway. This also results in battery capacity decay and reduced driving range. Therefore, accurate detection of lithium plating in lithium batteries is crucial for ensuring safe operation and extending their lifespan.

[0003] Currently, the industry commonly uses a mechanism-based detection scheme for lithium plating detection in lithium batteries. This scheme requires first constructing a mathematical model describing the internal reaction of the battery through a large number of physicochemical experiments, and then combining the voltage and current data during the battery charging and discharging process to deduce whether there are signs of lithium plating inside the battery by solving differential equations or computational fluid dynamics simulations; or using the internal resistance measurement method, which continuously monitors the changes in the battery's internal resistance and judges the risk of lithium plating based on abnormal increases in internal resistance.

[0004] However, detection schemes based on mechanistic models rely on in-depth physicochemical knowledge and massive experimental data for model construction. Furthermore, due to uncertainties such as material differences and varying degrees of aging in actual battery systems, the accuracy and applicability of the models are easily limited, making it difficult to adapt to the lithium battery plating detection needs in different scenarios. On the other hand, the internal resistance measurement method is easily affected by external factors such as battery charge / discharge rate and ambient temperature fluctuations, resulting in poor stability of the detection results. It cannot accurately identify early slight lithium plating states and cannot meet the timeliness and accuracy requirements of lithium battery plating detection. Summary of the Invention

[0005] This application provides a method, system, device, and storage medium for detecting lithium plating in lithium batteries, aiming to better adapt to the lithium plating detection needs in different scenarios and to meet the timeliness and accuracy requirements of lithium plating detection as much as possible.

[0006] To address the aforementioned technical problems, this application provides the following technical solutions: A method for detecting lithium plating in lithium batteries, comprising: After the lithium battery is fully charged, the voltage timing data of the lithium battery during the resting process is acquired. The voltage time series data is smoothed, and the smoothed voltage time series data is differentially calculated to generate the voltage change rate characteristics of the lithium battery. The voltage change rate feature is matched with features in the lithium plating feature library, and a lithium plating judgment result is generated based on the matching result. The lithium plating feature library is used to store voltage change rate feature data characterizing different lithium plating risks. Output the lithium plating determination result.

[0007] On the other hand, the embodiments of this application also provide the following technical solutions: A lithium battery lithium plating detection system, comprising: The data acquisition module is used to acquire the voltage timing data of the lithium battery during the resting process after the lithium battery has been fully charged. The feature extraction module is used to smooth the voltage time series data and perform differential calculation on the smoothed voltage time series data to generate the voltage change rate feature of the lithium battery. The lithium plating judgment module is used to match the voltage change rate feature with the features in the lithium plating feature library, and generate a lithium plating judgment result based on the matching result. The lithium plating feature library is used to store voltage change rate feature data characterizing different lithium plating risks. The early warning output module is used to output the lithium plating judgment result.

[0008] On the other hand, a computer device is provided, the device including a processor and a memory, the memory storing at least one instruction or at least one program, the at least one instruction or the at least one program being loaded and executed by the processor to implement the lithium plating detection method of the lithium battery as described above.

[0009] On the other hand, a computer-readable storage medium is provided, wherein at least one instruction or at least one program is stored therein, the at least one instruction or the at least one program being loaded and executed by a processor to implement the lithium plating detection method of the lithium battery as described above.

[0010] On the other hand, a computer program product or computer program is provided, which includes computer instructions that, when executed by a processor, implement the lithium plating detection method for lithium batteries as described above.

[0011] The beneficial effects of this application are as follows: 1. This application obtains voltage time-series data of a fully charged lithium battery during its resting process, directly using the actual voltage change after the battery has been fully charged and left to rest as the basis for detection. This eliminates the need to construct complex mechanistic models describing the internal reactions of the battery, thus avoiding reliance on specialized knowledge and massive amounts of experimental data. Furthermore, by combining voltage change rate characteristics with a lithium plating feature library, the application generates judgment results using the pre-stored correspondence between voltage change rate characteristics and lithium plating risk in the feature library, without requiring complex calculations to deduce the internal state of the battery. Compared to mechanistic model solutions, this application does not require adjusting model parameters due to individual battery differences or changing scenarios. It only needs to adapt the voltage change patterns under different scenarios through a feature library, resulting in higher applicability and better suitability for lithium battery testing needs in different scenarios such as electric vehicles and energy storage devices.

[0012] 2. This application selects the "resting stage after the lithium battery is fully charged" for data collection. During this stage, the battery has no charging or discharging current, and the influence of ambient temperature on voltage is more stable under resting conditions. This avoids abnormal data fluctuations caused by charging rate fluctuations or real-time charging and discharging processes. Compared to the internal resistance measurement method, which collects data during charging and discharging, external factors have less interference with the detection data. Furthermore, smoothing the voltage time-series data eliminates random fluctuations, and differential calculations generate voltage change rate features, further reducing the impact of random interference on feature extraction. Compared to the internal resistance measurement method, which is susceptible to external factors, the detection data from this application is more stable, providing more reliable basic data for subsequent lithium plating determination.

[0013] 3. This application generates voltage change rate features through "smoothing processing + differential calculation." The voltage change rate can more sensitively reflect subtle voltage decay anomalies caused by early lithium plating within the battery—compared to internal resistance parameters, changes in voltage time-series data are more significant in the early lithium plating stage, and differential calculation can amplify this subtle trend. Furthermore, by matching the voltage change rate features with a lithium plating feature library, it can more accurately identify characteristic signals corresponding to early, slight lithium plating based on the voltage change rate features corresponding to different lithium plating risks in the feature library. Compared to the limitation of internal resistance measurement methods in identifying early, slight lithium plating, this application has a stronger ability to identify early lithium plating, enabling earlier detection of lithium plating risks and providing more timely early warning support for the safe operation of lithium batteries. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1A schematic diagram of the overall process of a lithium battery lithium plating detection method provided in an embodiment of this application; Figure 2 A system block diagram of a lithium battery lithium plating detection system provided in this application embodiment; Figure 3 This is a hardware structure block diagram of an electronic device for detecting lithium plating in lithium batteries, as provided in an embodiment of this application. Detailed Implementation

[0016] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0017] This application provides a method for detecting lithium plating in lithium batteries, referring to... Figure 1 ,include: Step 1: After the lithium battery is fully charged, acquire the voltage timing data of the lithium battery during the resting process; Step 2: Smooth the voltage time series data and perform differential calculation on the smoothed voltage time series data to generate the voltage change rate characteristics of the lithium battery. Step 3: Match the voltage change rate feature with the features in the lithium plating feature library, and generate a lithium plating judgment result based on the matching result. The lithium plating feature library is used to store voltage change rate feature data that characterize different lithium plating risks. Step 4: Output the lithium plating determination result.

[0018] Specifically, preferably, the implementation process of step 1 is as follows: First, determine the criteria for determining the full charge state of the lithium battery. The current change during the charging process of the lithium battery is the core reference. When the lithium battery enters the constant voltage (CV) stage and the charging current in this stage drops to the set cutoff current threshold, the lithium battery is determined to be fully charged. The setting of the cutoff current threshold needs to be determined in combination with the nominal capacity of the lithium battery and the application scenario requirements. For example, for a power battery with a nominal capacity of 5000 mAh, the cutoff current threshold can be set to 300 mAh. This setting can avoid misjudging the full charge state due to insufficient charging as much as possible, and also prevent damage to the battery caused by overcharging. Secondly, after determining that the lithium battery has reached full charge, the voltage data acquisition process is triggered, and the pre-deployed voltage sampling module (such as a high-precision analog-to-digital converter chip) is started to continuously collect the voltage data of the lithium battery during the resting process according to the set sampling interval. The setting of the sampling interval needs to be related to the environmental conditions of the lithium battery. When the ambient temperature is lower than the preset temperature threshold (such as 10 degrees Celsius) or the charging current during the previous charging process of the lithium battery is greater than the preset current threshold (such as the current value corresponding to 1.5 times the nominal capacity), a smaller sampling interval (such as 5 seconds / time) is used to capture the rapid change trend of voltage after low temperature or high rate charging. When the ambient temperature is higher than or equal to the preset temperature threshold and the charging current is less than or equal to the preset current threshold, the default sampling interval (such as 10 seconds / time) is used to reduce data redundancy while ensuring data validity. The duration of data acquisition needs to cover the critical stage of voltage relaxation after the lithium battery is fully charged. It is usually set to continuously collect for 30 minutes (i.e., 1800 seconds) to obtain complete voltage decay process data. Finally, the continuously collected, timestamped voltage data is arranged in chronological order of collection time to form voltage time-series data of the lithium battery during the resting process. This voltage time-series data contains complete information on the voltage change of the lithium battery over time after it is fully charged, providing a basic data source for subsequent data processing.

[0019] Preferably, in the specific technical implementation of step 2, the voltage time series data generated in step 1 is first smoothed: taking each voltage sampling point in the voltage time series data as the processing object, a moving average algorithm is used, and a fixed number of sampling points are set as a sliding window (such as 5 consecutive sampling points forming a sliding window). The arithmetic average of all voltage sampling values ​​in the sliding window is calculated to obtain the smoothed voltage value corresponding to the window. According to the time order from morning to evening, the sliding window is moved sequentially along the voltage time series data, and the above averaging calculation operation is repeated for the voltage sampling values ​​in each window. The smoothed voltage values ​​corresponding to all windows are arranged in time order to generate smoothed voltage time series data. This processing can effectively remove abnormal fluctuation data points in the voltage time series data caused by sampling noise or instantaneous interference, and reduce the error of subsequent differential calculation. Secondly, differential calculations are performed on the smoothed voltage time-series data: Taking the smoothed voltage time-series data as the processing object, the central difference method is used to select any target sampling point in the data (except for the first and last sampling points). The smoothed voltage value and timestamp corresponding to the two adjacent sampling points before and after the target sampling point are extracted. The difference between the smoothed voltage values ​​of the adjacent sampling points before and after the target sampling point is calculated, and the difference between the timestamps of the adjacent sampling points before and after the target sampling point (i.e., the time interval) is also calculated. The ratio of the voltage difference to the time interval is used as the voltage change rate of the target sampling point. For the first and last sampling points, the forward difference or backward difference method is used to supplement the calculation of their voltage change rate (e.g., backward difference is used for the first sampling point, and forward difference is used for the last sampling point) to ensure that each sampling point corresponds to a voltage change rate value as much as possible. Finally, the voltage change rates corresponding to all sampling points are arranged in chronological order to generate the voltage change rate feature of the lithium battery. This voltage change rate feature can amplify the subtle change trend of the lithium battery voltage after full charge, providing core feature basis for subsequent lithium plating risk assessment.

[0020] Preferably, in the specific implementation of step 3, a lithium plating feature library is first constructed: based on lithium plating experimental data of lithium batteries under different usage conditions, lithium batteries with different ambient temperatures (e.g., -15 degrees Celsius, -5 degrees Celsius, 10 degrees Celsius, 25 degrees Celsius), different cycle numbers (e.g., 1 week, 5 weeks, 10 weeks, 15 weeks), and different degrees of lithium plating (e.g., no plating, slight plating, moderate plating, severe plating, violent plating) are selected as experimental samples. Steps 1 and 2 are performed on each experimental sample to obtain the voltage change rate characteristics corresponding to each sample, and this is further verified by disassembly... The method of observing battery electrodes records the actual lithium plating state of each sample (including electrode color state, such as golden yellow, dark yellow, blackish yellow, black spots, etc.). The voltage change rate characteristics of each experimental sample are associated with the corresponding lithium plating state (including lithium plating risk level, such as low risk, medium risk, high risk) and stored to form a lithium plating feature library. Each entry in the lithium plating feature library contains the correspondence between "voltage change rate characteristics - ambient temperature - number of cycles - lithium plating risk level - electrode color state", so as to ensure the multi-dimensionality and accuracy of feature matching as much as possible. Secondly, the voltage change rate feature of the lithium battery generated in step 2 is matched with the features in the lithium plating feature library: First, the ambient temperature and cycle number information of the lithium battery to be tested are extracted. From the lithium plating feature library, a feature subset that is consistent with or similar to the ambient temperature and cycle number of the lithium battery to be tested is selected to narrow the matching range and reduce the matching complexity. Then, a feature similarity calculation method (such as Euclidean distance method) is used to calculate the similarity between the voltage change rate feature of the lithium battery to be tested and the voltage change rate feature corresponding to each entry in the feature subset. The entry with the highest similarity is selected as the matching result. If the highest similarity value is greater than the preset similarity threshold (such as 0.85), it is determined that the lithium plating state of the lithium battery to be tested is consistent with the lithium plating state corresponding to the entry. If the highest similarity value is less than or equal to the preset similarity threshold, it is determined that the lithium plating state of the lithium battery to be tested is an undefined state, and further data needs to be supplemented before rematching. Finally, based on the above matching results, a lithium plating judgment result is generated. The lithium plating judgment result must clearly include the lithium plating risk level of the lithium battery to be tested (such as low risk, medium risk, high risk) and the corresponding electrode color state prediction information, so as to provide specific content for subsequent output results.

[0021] Preferably, the specific implementation process of step 4 is as follows: First, taking the lithium plating judgment result generated in step 3 as the processing object, the information in the lithium plating judgment result is structured and processed. The lithium plating risk level and electrode color state prediction information are organized according to a preset format (such as "current lithium plating risk level of lithium battery: XX; predicted electrode color state: XX") to form a standardized result output text, so as to ensure that the output information is clear and easy to understand, and is easy for different users (such as maintenance personnel and system controllers) to understand. Secondly, based on the application scenario, the corresponding result output method is selected: If applied to a local monitoring scenario of a lithium battery management system (BMS), the standardized result output text is sent to the local display module (such as an instrument panel display) for real-time display to on-site maintenance personnel; if applied to a remote monitoring scenario, the standardized result output text is sent to the remote monitoring platform via a communication module (such as a wireless communication module), along with information such as the device identifier and testing timestamp of the lithium battery to be tested, facilitating centralized management and statistical analysis of the lithium plating status of multiple lithium battery devices by the remote platform; if applied to an automatic control scenario, the lithium plating risk level in the lithium plating judgment result is converted into a corresponding control signal (such as a "no action" signal for low risk, a "warning" signal for medium risk, and a "stop and check" signal for high risk), and sent to the lithium battery's charging and discharging control module to provide a basis for subsequent charging strategy adjustments. Finally, after outputting the lithium plating judgment result, the key data in this detection process (including the voltage timing data in step 1, the voltage change rate characteristics in step 2, the matching process record in step 3, and the output result in step 4) are stored. The storage path is set to a preset database or local storage unit, which facilitates subsequent traceability of the detection process, optimization of the lithium plating feature library, and provides data support for the full life cycle health management of lithium batteries.

[0022] Optionally, step 1 includes: step 1.1, when the lithium battery charging enters the constant voltage stage and the charging current drops to the set cutoff current, it is determined that the lithium battery has reached a fully charged state. Step 1.2: After the lithium battery reaches full charge, the data acquisition module is triggered. The data acquisition module collects the voltage and current data of the lithium battery according to the set sampling interval and continues for the set resting time to obtain voltage timing data and current timing data.

[0023] Specifically, preferably, the specific implementation process of step 1.1 is as follows: First, determine the criteria for judging the lithium battery charging stage. Taking the voltage and current change trends during the lithium battery charging process as the core reference, collect the charging voltage data and charging current data of the lithium battery in real time. By comparing the charging voltage data with the nominal cutoff voltage of the lithium battery (i.e., the full-charge voltage value designed for the lithium battery, for example, the common nominal cutoff voltage of ternary lithium batteries is 4.2 volts), when the charging voltage data reaches the nominal cutoff voltage and remains stable thereafter and no longer increases with charging time, it is determined that the lithium battery charging has entered the constant voltage (CV) stage. The determination of this stage can avoid the error of the traditional stage that only relies on current change to judge the stage, because voltage stability is the essential characteristic of the constant voltage stage, and can more reliably distinguish between the constant current stage and the constant voltage stage. Secondly, the rules for determining the cutoff current should be set. The cutoff current should be set based on the nominal capacity of the lithium battery (in milliampere-hours, mAh) and the safety requirements of the application scenario. It is usually taken as a fixed percentage of the nominal capacity (such as 5%-10%). For example, for a lithium storage battery with a nominal capacity of 6000 mAh, the cutoff current can be set to 300 mAh (i.e. 5% of the nominal capacity). This setting can ensure that the lithium battery is basically fully charged as much as possible (avoiding false full charge judgment due to insufficient charging) and prevent battery damage caused by overcharging. At the same time, the cutoff current should be stored as a preset threshold in the system as a benchmark for subsequent current judgment. Finally, the charging current data during the constant voltage stage is continuously monitored. When the charging current data drops to the preset cutoff current threshold and remains stable (i.e., the charging current data does not exceed the cutoff current threshold for multiple consecutive sampling cycles, for example, 5 consecutive sampling cycles, each sampling cycle is 1 second), a result is generated to determine the full charge status of the lithium battery based on the dual conditions of "constant voltage stage + current compliance". This result is more accurate than the traditional single current or voltage judgment method and can effectively avoid false full charge judgments caused by fluctuations in the charging environment (such as unstable voltage).

[0024] Preferably, in the specific technical implementation of step 1.2, the startup logic of the data acquisition module is first triggered, using the lithium battery full-charge status determination result generated in step 1.1 as the trigger signal. Upon receiving the determination result, a startup command is automatically sent to the data acquisition module. The data acquisition module includes a voltage sampling unit and a current sampling unit. The voltage sampling unit uses a high-precision analog-to-digital converter (ADC) chip, and the current sampling unit uses a high-precision sampling resistor (such as an alloy resistor with an accuracy of ±1%). Both need to be calibrated in advance to ensure the consistency of the sampled data as much as possible. This module design differs from the traditional scheme that only collects voltage data. The simultaneous collection of current data can be used for subsequent data integrity verification (such as determining whether there is abnormal leakage during the static stage), improving the reliability of detection. Secondly, a dynamic adjustment rule for the sampling interval is set. The initial value of the sampling interval (default sampling interval) is based on the typical operating environment of the lithium battery (e.g., 10 seconds / time). Two adjustment parameters are introduced: ambient temperature and historical charging current. Real-time acquisition of ambient temperature data (obtained through a temperature sensor deployed near the battery) and charging current data prior to step 1.1 (i.e., the charging current data in the final stage before full charge) are used. When the ambient temperature data is lower than a preset temperature threshold (e.g., 10 degrees Celsius) or the historical charging current data is greater than a preset current threshold (e.g., the current value corresponding to 1.5 times the nominal capacity), the sampling interval is adjusted accordingly. When the battery capacity is 6000 mAh and the preset current threshold is 9 amps, the sampling interval is adjusted to a smaller sampling interval (e.g., 5 seconds / time). This is because the lithium battery voltage decays faster after low temperature or high rate charging, requiring more frequent sampling to capture subtle changes. When the ambient temperature data is higher than or equal to the preset temperature threshold and the historical charging current data is less than or equal to the preset current threshold, the default sampling interval is maintained. This dynamic adjustment rule can avoid the problems of "insufficient data in low temperature / high rate scenarios" or "data redundancy in normal scenarios" caused by the traditional fixed sampling interval, balancing data validity and acquisition efficiency. Then, the criteria for determining the resting time are set. The resting time needs to cover the critical stage of voltage relaxation after the lithium battery is fully charged (i.e., the complete process of voltage transition from full charge voltage to stable voltage). It is usually set to a fixed duration (e.g., 1800 seconds, or 30 minutes) based on a large amount of experimental data, so as to ensure that the complete voltage decay trend can be collected as much as possible. Within the resting time, the data acquisition module continuously collects the voltage and current data of the lithium battery according to the dynamically adjusted sampling interval, and adds a corresponding timestamp (accurate to the second) to each set of collected voltage and current data.Finally, voltage time-series data and current time-series data are generated. The voltage data with timestamps collected during the resting period are arranged in chronological order to form voltage time-series data; the current data with timestamps collected during the same period are arranged in chronological order to form current time-series data. These two types of time-series data need to be stored together (i.e., voltage data and current data with the same timestamp correspond to the same storage entry), providing a complete and related basic data source for the data processing in step 2. At the same time, the current time-series data can be used to determine whether there are any abnormalities during the resting period (e.g., if the current data is continuously greater than 0 and exceeds the normal leakage current range, it can be determined that the data acquisition is abnormal), ensuring that subsequent processing is based on a reliable data foundation as much as possible.

[0025] Optionally, in step 1.2, the sampling interval duration is adjusted according to the charging current of the lithium battery or the ambient temperature.

[0026] Optionally, when dynamically adjusting the data sampling interval, if the ambient temperature is lower than the preset temperature threshold or the charging current is greater than the preset current threshold, a smaller sampling interval is used for data acquisition. If the ambient temperature is higher than or equal to the preset temperature threshold and the charging current is less than or equal to the preset current threshold, the default sampling interval is used for data acquisition to generate voltage timing data and current timing data under the corresponding sampling interval.

[0027] Specifically, preferably, the implementation process of step 1.3 is as follows: First, determine the input parameters required for adjusting the sampling interval, using the charging current data before the lithium battery is fully charged and the ambient temperature data during the resting stage as the core input parameters; wherein, the charging current data before the lithium battery is fully charged needs to be extracted from the charging process monitoring data in step 1.1, specifically the average charging current value after the lithium battery enters the constant voltage (CV) stage until it reaches the set cutoff current, denoted as I_avg. This current value can reflect the charging intensity of the lithium battery before it is fully charged, avoiding judgment deviations caused by instantaneous current fluctuations; the ambient temperature data during the resting stage needs to be collected in real time by a temperature sensor deployed near the lithium battery casing, denoted as T_env. The collection frequency is set to be consistent with the initial default sampling interval, so as to ensure that the temperature data can reflect the temperature state of the environment in which the lithium battery is located in real time, providing accurate environmental parameter basis for subsequent adjustment logic.

[0028] Secondly, determine the rules for setting the preset temperature threshold and the preset current threshold. The preset temperature threshold (denoted as T_th) needs to be set based on the experimental data of the lithium plating risk and voltage change characteristics of lithium batteries. The core logic is as follows: when the ambient temperature is lower than this threshold, the ion migration rate of the lithium battery electrolyte decreases, and the voltage relaxation speed after full charge accelerates. Subtle voltage changes are more likely to be missed. Therefore, it is necessary to capture key features by reducing the sampling interval. For example, for ternary lithium batteries or lithium iron phosphate batteries, the preset temperature threshold T_th can be set to 10 degrees Celsius. This value is obtained based on a large number of lithium battery voltage decay experiments in low-temperature environments and can effectively cover the low-temperature lithium plating risk range in most scenarios. The preset current threshold (denoted as I_th) needs to be related to the nominal capacity of the lithium battery (denoted as C, unit: milliampere-hour, mAh), and is usually set as a fixed multiple (such as 1.5 times) of the current value corresponding to the nominal capacity, that is, I_th = 1.5 × C (unit: milliampere, mA). The reason for this setting is that when the charging current exceeds this threshold, the lithium battery is prone to increased voltage fluctuation amplitude after full charge due to the polarization effect, and the fixed sampling interval may not be able to completely record the details of voltage changes. For example, for a lithium battery with a nominal capacity of 5000 milliampere-hours, its preset current threshold I_th can be set to 7500 milliamperes (i.e., 1.5 × 5000 milliamperes). The preset temperature threshold T_th and the preset current threshold I_th need to be stored as parameters that can be called by the system and support personalized configuration based on different battery types (such as ternary lithium, lithium iron phosphate) to avoid the problem of poor adaptability of traditional fixed thresholds.

[0029] Then, execute the dynamic adjustment logic of the sampling interval, taking the collected ambient temperature data T_env and the average charging current data I_avg before full charge as the processing objects, and first compare T_env with the preset temperature threshold T_th, and at the same time compare I_avg with the preset current threshold I_th: If T_env < T_th, or I_avg > I_th, when either condition is met, it is determined that the voltage change rate of the lithium battery in the current scenario is relatively fast, and a smaller sampling interval (denoted as Δt1) needs to be used; the value of the smaller sampling interval Δt1 should be set as small as possible to ensure that it can completely capture the subtle changes in the voltage relaxation process. Usually, it is set to 1 / 2 of the default sampling interval. For example, when the default sampling interval is 10 seconds, the smaller sampling interval Δt1 is set to 5 seconds. This ratio is determined based on the requirement of extracting voltage change rate characteristics, which can not only ensure that the data density meets the subsequent differential calculation requirements but also avoid data redundancy caused by over-dense sampling. If T_env ≥ T_th and I_avg ≤ I_th, when both conditions are met, it is determined that the voltage change rate of the lithium battery in the current scenario is gentle, and the default sampling interval (denoted as Δt0) can be used. The default sampling interval Δt0 needs to be set based on the voltage decay period of lithium batteries in a normal environment, such as 10 seconds. This interval can reduce the data storage and processing pressure while ensuring the data validity.

[0030] Finally, voltage and current time-series data are generated based on the adjusted sampling interval. Using a defined sampling interval (Δt1 or Δt0), the data acquisition module (including voltage and current sampling units) is controlled to continuously collect data at that interval: within a set rest period (e.g., 1800 seconds), the lithium battery voltage and current data are collected once every Δt1 or Δt0, and a unique timestamp (accurate to the second) is added to each set of data; all collected voltage data are arranged in chronological order according to their timestamps to form voltage time-series data for the corresponding sampling interval; all voltage data collected simultaneously are then... The streaming data are arranged in the same timestamp order to form current time series data under the corresponding sampling interval; the voltage time series data and the current time series data need to be stored in association with each other according to the timestamp, so as to ensure that the validity of the voltage data can be verified by the current data during the subsequent data processing in step 2 (such as outlier removal and smoothing). For example, if the current data increases abnormally during the resting stage, the corresponding voltage data can be determined to be invalid, avoiding the data reliability risk caused by traditionally only collecting voltage data. At this point, the dynamic adjustment of the sampling interval and data acquisition are completed, and the generated voltage time series data and current time series data are directly used as input for step 2.

[0031] This application's solution differs from the traditional fixed sampling interval design. By adjusting the dual parameters of "ambient temperature + charging current", it more accurately matches the voltage change characteristics of lithium batteries under different scenarios. This not only solves the problem of insufficient data at fixed sampling intervals in low-temperature or high-rate charging scenarios, but also avoids data redundancy waste in conventional scenarios as much as possible. At the same time, through clear threshold setting rules and interval ratio design, it enables different technicians to obtain consistent results when implementing it.

[0032] Optionally, the smoothing process for the voltage timing data in step 2 includes: Step 2.1: Use an outlier removal method based on the standard deviation criterion to remove data points that exceed the set range from the voltage time series data to generate valid voltage data; Step 2.2: Use the moving average algorithm to smooth the effective voltage data to generate smoothed voltage time series data; Step 2 involves differential calculation of the smoothed voltage timing data, including: Step 2.3: The central difference method is used to calculate the smoothed voltage time series data to generate voltage change rate characteristics.

[0033] Specifically, preferably, the implementation process of step 2.1 is as follows: First, determine the basic data required for outlier removal based on the standard deviation criterion, using the voltage time series data generated in step 1.2 as the processing object. This voltage time series data contains multiple voltage sampling points arranged in chronological order (denoted as V1, V2, ..., V...).n (where n is the total number of sampling points) and their corresponding timestamps, the values ​​of all voltage sampling points need to be extracted first to form a voltage sampling value set (denoted as S_V={V1,V2,…,V...). n This set is the core basis for subsequent calculation of standard deviation and identification of outliers, avoiding calculation bias caused by the inclusion of timestamp information. Secondly, the statistical parameters of the voltage sample value set are calculated. The arithmetic mean of the voltage sample value set S_V is calculated to obtain the voltage average value (denoted as μ_V). This average value reflects the overall level of the voltage time series data. Then, based on the voltage average value μ_V, the sum of squares of the deviations of each voltage sampling point V_i from μ_V is calculated, thus obtaining the voltage standard deviation (denoted as σ_V). The standard deviation σ_V quantifies the dispersion of the voltage sample values. The greater the dispersion, the more likely there are outlier data points. For example, if the average value μ_V of the voltage time series data is 4.15 volts and the standard deviation σ_V is 0.005 volts, these two parameters can be used to define the normal data range. Then, the outlier judgment range is set. The core logic based on the standard deviation criterion is that normal data points are usually distributed within the range of "mean ± k times standard deviation" (k is a constant). In this application, the value of k is set to 3, that is, the outlier judgment range is [μ_V-3σ_V,μ_V+3σ_V]. The reason for this setting is that in statistics, a range of 3 times the standard deviation can cover about 99.7% of normal data, which can effectively eliminate extreme outliers caused by sampling noise and instantaneous electromagnetic interference (such as sudden voltage jumps), while avoiding the accidental deletion of normal, minor voltage fluctuation data. Finally, an outlier removal operation is performed, comparing each voltage sampling point V_i in the voltage sampling value set S_V with the outlier judgment range [μ_V-3σ_V, μ_V+3σ_V]. If V_i is within the range, it is determined to be a normal voltage data point, and its value and corresponding timestamp are retained. If V_i is outside the range, it is determined to be an outlier voltage data point, and the data point and its corresponding timestamp are removed. All retained normal voltage data points are rearranged in their original chronological order to form valid voltage data. This valid voltage data eliminates the interference of extreme outliers on subsequent smoothing processing, allowing subsequent steps to proceed based on a reliable data foundation.

[0034] Preferably, in the specific technical implementation of step 2.2, firstly, the core parameters and processing object of the moving average algorithm are determined, taking the effective voltage data generated in step 2.1 as the processing object. The effective voltage data includes normal voltage data points arranged in chronological order (denoted as V'1, V'2, ..., V'_m, where m is the total number of effective data points, m≤n) and corresponding timestamps; the sliding window size (denoted as w) is set. The window size w needs to be determined based on the fluctuation characteristics and smoothing requirements of the voltage time series data. The core logic is: if the window is too large, the voltage change trend will be over-smoothed (loss of key details), and if the window is too small, residual noise cannot be effectively filtered out. In this application, w is set to 5, that is, each sliding window contains 5 consecutive effective voltage data points. This value is derived from a large number of voltage smoothing experiments, which can filter out small noise (such as fluctuations of ±0.001 volts) remaining in the sampling process while retaining the subtle voltage change trend. Secondly, a smoothing calculation for the sliding window is performed. Starting with the first sliding window of effective voltage data (containing V'1, V'2, V'3, V'4, V'5), the arithmetic mean of the five effective voltage data points within the window is calculated to obtain the smoothed voltage value corresponding to the window (denoted as V_s3, corresponding to the timestamp of the middle data point V'3 in the original window). Then, the sliding window is moved one data point backward along the effective voltage data to form a new sliding window (such as V'2, V'3, V'4, V'5, V'6), and the above averaging calculation operation is repeated to obtain the smoothed voltage value corresponding to the new window (denoted as V_s4, corresponding to the timestamp of the middle data point V'4 in the original window). This processing method differs from the traditional design of "the first point of the window corresponds to the smoothed value". By using the method of "the middle point of the window corresponds to the smoothed value", the smoothed data is prevented from having a time offset, and the correspondence between the timestamp and the voltage value is kept as accurate as possible. Finally, the smoothing and supplementation of the first and last data points is processed, specifically for the first two data points (V'1, V'2) and the last two data points (V'_m) in the effective voltage data. -1 Since a complete 5-point sliding window could not be formed, the smoothed voltage values ​​were supplemented using "forward filling" and "backward filling" methods: the smoothed voltage values ​​of the first two data points were taken from the smoothed voltage values ​​of the first complete window (V_s3) and the second complete window (V_s4), respectively; the smoothed voltage values ​​of the last two data points were taken from the smoothed voltage values ​​of the second-to-last complete window and the last complete window, respectively. All smoothed voltage values ​​were arranged in their original time order to generate smoothed voltage time series data. This data eliminated noise interference and completely preserved the trend of voltage change over time, providing high-quality data for subsequent differential calculations.

[0035] Preferably, in a scenario, when implementing step 2.3, firstly, the processing object and basic data extraction of the central difference method are determined. The smoothed voltage time series data generated in step 2.2 is used as the processing object. This data includes smoothed voltage values ​​arranged in chronological order (denoted as V_s1, V_s2, ..., V_s_m, which correspond one-to-one with the timestamps of the effective voltage data) and corresponding timestamps (denoted as t1, t2, ..., t_m). The timestamp t_i corresponding to each smoothed voltage value V_s_i is extracted to form a timestamp set (denoted as S_t={t1,t2,...,t_m}). This set is used to calculate the time interval to ensure the accuracy of the correspondence between "voltage change" and "time change" in the differential calculation as much as possible. Secondly, the difference calculation rules are set. The core of the central difference method is to calculate the rate of change by the difference between the target data point and the adjacent data points. For any target smoothed voltage value V_s_i (2≤i≤m-1, i.e., excluding the first and last data points) in the smoothed voltage time series data, it is necessary to extract the previous adjacent smoothed voltage value V_s(i-1) and the corresponding timestamp t(i-1), and at the same time extract the next adjacent smoothed voltage value V_s(i+1) and the corresponding timestamp t(i+1). The voltage change of the target data point ΔV_i=V_s(i+1)-V_s(i-1) is calculated, and the time change of the target data point Δt_i=t(i+1)-t(i-1) is calculated. By using the "difference between two points", the problem of the traditional "difference between two adjacent points" being easily affected by single-point noise is avoided, thus improving the stability of the rate of change calculation. Then, the voltage change rate is calculated. For each target data point, the ratio of the voltage change ΔV_i to the time change Δt_i is taken as the voltage change rate of that data point (denoted as r_i=ΔV_i / Δt_i). The unit of the voltage change rate r_i is "volts per second". This value can intuitively reflect how fast the voltage changes with time. For example, if r_i is -0.0012 volts per second, it means that the voltage drops by 0.0012 volts per second. This change rate characteristic is the core indicator for subsequent lithium plating judgment.Finally, the rate of change of the first and last data points is supplemented. For the first data point V_s1 in the smoothed voltage time series data, the voltage rate of change is calculated using the "backward difference method": extract the next two adjacent smoothed voltage values ​​V_s2 and V_s3 of V_s1 and their corresponding timestamps t2 and t3, and calculate ΔV1=V_s3-V_s2, Δt1=t3-t2, and the voltage rate of change r1=ΔV1 / Δt1. For the last data point V_s_m, the voltage rate of change is calculated using the "forward difference method": extract the next two adjacent smoothed voltage values ​​V_s1 and V_s2, V_s3, and their corresponding timestamps t2 and t3, and calculate ΔV1=V_s3-V_s2, Δt1=t3-t2, and the voltage rate of change r1=ΔV1 / Δt1. Given voltage values ​​V_s(m-2) and V_s(m-1) and their corresponding timestamps t(m-2) and t(m-1), calculate ΔV_m = V_s(m-1) - V_s(m-2) and Δt_m = t(m-1) - t(m-2), and the voltage change rate r_m = ΔV_m / Δt_m. Arrange the voltage change rates of all data points in their original time order to generate the voltage change rate feature of the lithium battery. This feature fully reflects the voltage change trend of the lithium battery during the resting process after it is fully charged, and directly serves as the core basis for matching with the lithium plating feature library in step 3.

[0036] This application clarifies the selection criteria for "3 times the standard deviation" and the logic for defining the outlier range, avoiding inconsistencies in the removal effect caused by the ambiguity of the k-value in the traditional standard deviation criterion. Step 2.2 adopts a design of "5-point sliding window + smoothing value corresponding to the midpoint" to solve the problems of time offset and smoothing overload in the traditional moving average algorithm. Step 2.3 uses the method of "central difference as the main method and forward and backward difference as supplementary methods" to balance the stability of the rate of change calculation and data integrity. These optimizations make the data processing process more in line with the voltage characteristics of lithium batteries, providing more reliable feature support for subsequent lithium plating judgment.

[0037] Optionally, step 3 includes: step 3.1, when matching the voltage change rate feature with the features in the lithium plating feature library, if the voltage change rate feature meets the preset inflection point judgment condition, the voltage change rate feature is determined to be a potential lithium plating feature; Step 3.2: Match the potential lithium plating features with the features in the lithium plating feature library, determine the lithium plating risk level based on the matching results, and generate a lithium plating judgment result.

[0038] Optionally, the preset inflection point determination condition is that the absolute value of the voltage change rate characteristic is greater than a preset threshold, and the duration of this state reaches a preset duration. If this condition is met, it is determined to be a potential lithium plating characteristic.

[0039] Specifically, preferably, the specific implementation process of step 3.1 is as follows: First, determine the constituent elements of the preset inflection point judgment condition, taking the voltage change rate feature generated in step 2.3 as the processing object. The voltage change rate feature includes multiple voltage change rate values ​​(denoted as r1, r2, ..., r_m, where m is the total number of data points) arranged in time sequence and the corresponding timestamps. The preset inflection point judgment condition must simultaneously include two core elements: "voltage change rate absolute value threshold" and "duration threshold". The reason for combining the two is that: a single voltage change rate threshold is easy to misjudge instantaneous fluctuations as lithium plating inflection points, and a single duration threshold cannot distinguish between minute changes and effective lithium plating features. It is necessary to ensure the accuracy of the judgment as much as possible through dual conditions. Secondly, a threshold for the absolute value of the voltage change rate (denoted as |r_th|) is set. This threshold needs to be determined based on the "maximum voltage change rate in the non-lithium-plating state" in lithium battery lithium plating experiments. The core logic is that when the absolute value of the voltage change rate exceeds this threshold, there is a high probability that abnormal voltage changes caused by lithium plating will occur inside the lithium battery. For example, by analyzing the voltage change rate characteristics of a large number of non-lithium-plating lithium batteries, |r_th| can be set to 0.001 volts / second. This value can effectively distinguish between normal voltage relaxation (the absolute value of the change rate is usually less than 0.001 volts / second) and abnormal changes caused by lithium plating. Then, a duration threshold (denoted as t_th) is set. This threshold needs to be determined based on the stability requirements of the lithium plating characteristics to avoid judging short-term interference as a valid inflection point. For example, combined with the sampling interval of the voltage change rate characteristics (such as 5 seconds or 10 seconds), t_th can be set to 50 seconds. That is, the absolute value of the voltage change rate must exceed |r_th| in multiple consecutive sampling periods to be judged as a valid inflection point. This setting can filter out short-term abnormal changes caused by instantaneous electromagnetic interference. Finally, the potential lithium plating characteristic determination operation is performed. Taking each voltage change rate value r_i and its corresponding timestamp in the voltage change rate characteristic as the processing object, the absolute value of r_i is checked point by point to see if it is greater than |r_th|. If the absolute value of r_i is greater than |r_th|, the timestamp of that data point is recorded as the inflection point start time. The process continues to check subsequent voltage change rate values ​​r(i+1), r(i+2), ..., until the absolute value of r_j is less than or equal to |r_th|. The timestamp of that data point is recorded as the inflection point end time. The inflection point end time is calculated. The difference between the time and the start time is used to obtain the duration t_d of the abnormal change. If t_d is greater than t_th, the voltage change rate feature within this time period is determined to be a potential lithium plating feature. At the same time, the time range, maximum and minimum voltage change rate corresponding to this feature are extracted to form details of the potential lithium plating feature. If t_d is less than or equal to t_th, it is determined to be invalid interference and is not included in the subsequent matching process. Through this judgment process, local features suspected of lithium plating are screened from the complete voltage change rate features, laying the foundation for accurate matching in step 3.2.

[0040] Preferably, in the specific technical implementation of step 3.2, firstly, the structure and data source of the lithium plating feature library are clarified. The lithium plating feature library needs to be constructed based on the lithium plating experimental data of lithium batteries in multiple scenarios in the early stage. Each feature entry in the library contains four core fields: "potential lithium plating feature template", "corresponding lithium plating risk level", "related environmental parameters", and "electrode state reference". Among them, the "potential lithium plating feature template" is the core basis for matching with the potential lithium plating features generated in step 3.1, including the feature duration range, the extreme range of voltage change rate, and the change trend (such as continuous decrease, fluctuating decrease); the "lithium plating risk level" is divided into three levels: low, medium, and high, which correspond to different voltage change rate feature intensities (such as low risk corresponding to voltage). The absolute value of the rate of change is 0.001-0.005 volts / second, with medium risk corresponding to 0.005-0.01 volts / second and high risk corresponding to greater than 0.01 volts / second. The "Associated Environmental Parameters" record the experimental environment temperature and charging rate corresponding to the feature template, ensuring that the matching can be combined with the actual environmental conditions of the current lithium battery as much as possible. The "Electrode State Reference" records the electrode color observed in the experiment (e.g., golden yellow for no precipitation, dark yellow for slight precipitation, and blackish yellow for severe precipitation), which provides an intuitive reference for generating lithium plating judgment results later. This feature library structure differs from the traditional design that only stores a single feature value. It improves the accuracy of matching through multi-field association and avoids misjudgment caused by environmental differences. Secondly, a matching process between potential lithium plating features and the feature library is executed. Taking the potential lithium plating features generated in step 3.1 and the current environmental parameters of the lithium battery (such as the ambient temperature T_env during the resting stage and the charging rate before full charge) as the processing objects, feature entries whose "related environmental parameters" are consistent with or similar to the current environmental parameters are first screened from the lithium plating feature library to form a candidate feature subset, narrowing the matching range (for example, if the current ambient temperature is -5 degrees Celsius, then feature entries in the library with ambient temperatures between -10 and 0 degrees Celsius are screened). Then, a feature similarity calculation method (optimized Euclidean distance method) is used to compare the three dimensions of the potential lithium plating features, namely "duration", "extreme value of voltage change rate", and "change trend", with each entry in the candidate feature subset. Similarity calculation is performed on the "potential lithium plating feature templates": For example, the differences between the duration t_d of the potential lithium plating feature and the template duration t_template, the differences between the maximum voltage change rate r_max and the template maximum value r_template_max, and the differences between the minimum voltage change rate r_min and the template minimum value r_template_min are assigned weights (duration weight 0.4, maximum value weight 0.3, minimum value weight 0.3), and the weighted Euclidean distance is calculated. The smaller the distance, the higher the similarity. A similarity threshold is set (e.g., 0.8), and the entry with the smallest weighted Euclidean distance in the candidate feature subset that is less than the similarity threshold is selected as the optimal matching result.Then, the lithium plating risk level is determined based on the optimal matching result. If an optimal matching result exists, the "lithium plating risk level" and "electrode state reference" of that item are extracted. If no optimal matching result exists (i.e., the weighted Euclidean distance of all candidate items is greater than the similarity threshold), the current potential lithium plating feature is determined to be an unrecorded feature, temporarily classified as medium risk, and the feature and corresponding environmental parameters are recorded in the unverified area of ​​the lithium plating feature library for subsequent feature library updates. Finally, a lithium plating judgment result is generated. The determined lithium plating risk level, electrode state reference, time range of potential lithium plating features, voltage change rate extreme value, and other information are integrated according to a preset format to form a structured lithium plating judgment result (e.g., "Lithium plating risk level: medium risk; inferred electrode state: dark yellow (slight precipitation); duration of abnormal feature: 60 seconds; voltage change rate range: -0.006 to -0.008 volts / second"). This result is directly used as the core content of the output of step 4 and provides a basis for adjusting the charging strategy if necessary.

[0041] In this application, the dual judgment condition of "absolute value threshold of voltage change rate + duration threshold" in step 3.1 solves the problem of easy misjudgment by the traditional single threshold and improves the accuracy of potential lithium plating feature screening. The multi-field association design of the lithium plating feature library and the optimized weighted Euclidean distance matching algorithm in step 3.2 can avoid the deviation caused by the traditional matching ignoring environmental differences and only looking at a single feature value through environmental parameter screening and multi-dimensional feature weighted calculation. At the same time, it supports the dynamic updating of the feature library and enhances the adaptability of the solution to different lithium battery types and different usage scenarios.

[0042] Optionally, the detection method also includes: step 5, determining the corresponding risk level based on the lithium plating judgment result, and generating a charging strategy adjustment instruction based on the risk level.

[0043] Optionally, step 5 includes: when determining the corresponding risk level based on the lithium plating judgment result, dividing multiple risk levels according to the numerical range of the voltage change rate characteristic, determining the corresponding risk level based on the numerical range to which the voltage change rate characteristic belongs, and then generating corresponding charging rate adjustment instructions and charging curve adjustment instructions according to different risk levels as charging strategy adjustment instructions.

[0044] Specifically, preferably, the implementation process of step 51 is as follows: First, determine the core basis and input data for risk level classification. Take the lithium plating judgment result generated in step 3.2 as the processing object. This lithium plating judgment result includes key information such as "voltage change rate characteristics" and "electrode state reference". Among them, the voltage change rate extreme value (including the maximum value r_max and the minimum value r_min, both in volts / second) in the "voltage change rate characteristics" is the core basis for classifying risk levels. The reason for choosing the voltage change rate extreme value is that the more severe the lithium plating, the more significant the abnormal voltage change during the static stage after the lithium battery is fully charged, and the larger the absolute value of the voltage change rate extreme value, which can directly reflect the strength of the lithium plating risk. At the same time, the ambient temperature data T_env collected in step 1.2 needs to be extracted as an auxiliary reference to avoid risk misjudgment caused by the difference in voltage change rate benchmarks at different temperatures.

[0045] Secondly, the risk level classification rules are set. Based on a large amount of lithium battery lithium plating experimental data, the risk level is divided into multiple levels (usually three levels: Level 1, Level 2, and Level 3) according to the absolute value range of the extreme values ​​of the voltage change rate. Each level corresponds to a specific numerical range, and the boundary of the range needs to be verified by the correspondence between "degree of lithium plating" and "voltage change rate". Level 1 risk corresponds to "slight lithium plating risk", and the absolute value range of its extreme voltage change rate is set to "greater than the preset inflection point threshold |r_th| and less than or equal to the first critical value |r1|". For example, when |r_th| is 0.001 volts / second, |r1| can be set to 0.007 volts / second. Within this range... Lithium battery electrodes typically appear dark yellow (with slight precipitation) and show no obvious lithium dendrite growth. Level 2 risk corresponds to "general lithium plating risk," where the absolute value range of the extreme voltage change rate is set to "greater than the first critical value |r1| and less than or equal to the second critical value |r2|." For example, if |r2| is set to 0.011 volts / second, the electrodes within this range are mostly blackish-yellow (with general precipitation) and may contain a small number of fine lithium dendrites. Level 3 risk corresponds to "severe lithium plating risk," where the absolute value range of the extreme voltage change rate is set to "greater than the second critical value |r2|." For example, in the range above |r2|, the electrodes often show black pitting (severe precipitation), and lithium dendrite growth is already quite obvious, posing a risk of puncturing the separator. Meanwhile, for different ambient temperatures T_env, the numerical range needs to be slightly modified: when T_env is lower than the preset temperature threshold T_th (e.g., 10 degrees Celsius), because the voltage change rate benchmark is slightly higher at low temperatures, the critical values ​​(|r1|, |r2|) are increased by 5%-10% to avoid underestimating the risk level in low-temperature scenarios; when T_env is higher than or equal to T_th, the critical values ​​remain unchanged. This temperature correction design differs from the traditional fixed numerical range division method and solves the problem of risk judgment deviation at different temperatures.

[0046] Then, based on the extreme value of the voltage change rate, the specific risk level is determined. Taking the extreme value of the voltage change rate in the lithium plating judgment result (the larger absolute value between r_max and r_min, denoted as |r_extreme|) as the processing object, |r_extreme| is compared with the numerical range corresponding to the risk level: if |r_extreme| is within the numerical range of level 1 risk, and the requirement of this range is still met after environmental temperature correction, the current lithium battery is determined to be at level 1 risk; if |r_extreme| is within the numerical range of level 2 risk, it is determined to be at level 2 risk; if |r_extreme| is within the numerical range of level 3 risk, it is determined to be at level 3 risk. After the determination, a "risk level result" is generated. This result must clearly indicate the risk level name, the corresponding extreme value range of the voltage change rate, and the basis for environmental temperature correction, so as to ensure the pertinence of subsequent charging strategy adjustments as much as possible.

[0047] Next, charging rate adjustment instructions corresponding to the risk level are generated. The core logic of charging rate adjustment is that "the higher the risk level, the lower the charging rate needs to be to slow down lithium dendrite growth." Each risk level corresponds to a unique maximum allowable charging rate (the charging rate is usually based on the nominal capacity C of the lithium battery, such as 1C, which represents the current required to fully charge the nominal capacity in one hour): For Level 1 risk, since the risk of lithium plating is low, there is no need to significantly adjust the charging rate. The maximum allowable charging rate is set to "not higher than the original default charging rate" (e.g., if the original default rate is 1.5C, keep it unchanged at 1.5C). Only a "charging rate maintenance instruction" is generated to avoid excessive adjustment affecting charging efficiency. For Level 2 risk, the charging rate needs to be moderately reduced. The maximum allowable charging rate is set to "not higher than 1C". A "charging rate reduced to 1C instruction" is generated to slow down lithium dendrite growth by reducing the polarization effect during charging. For Level 3 risk, the charging rate needs to be significantly reduced. The maximum allowable charging rate is set to "not higher than 0.5C". A "charging rate reduced to 0.5C instruction" is generated to minimize the factors promoting lithium plating during charging. At the same time, the charging rate adjustment command needs to be associated with the nominal capacity C of the lithium battery and the specific current value should be clearly marked (e.g., for a lithium battery with a nominal capacity of 5000 mAh, 1C corresponds to a current of 5000 mAh) to ensure that the charging module can execute it directly as much as possible.

[0048] Finally, a charging curve adjustment instruction corresponding to the risk level is generated. The charging curve adjustment needs to be coordinated with the charging rate adjustment, and differentiated charging curve types are designed for different risk levels: Under Level 1 risk, the charging curve remains the original default curve (usually a constant current / constant voltage curve, i.e., first charging at a fixed current to the nominal cutoff voltage, then charging at a fixed voltage to the cutoff current), generating a "charging curve maintenance instruction"; Under Level 2 risk, the charging curve is adjusted to a "segmented constant current / constant voltage curve," i.e., adding 1-2 current steps to the original constant current stage (e.g., first charging at 1C to 3.9V). First, the battery is charged at 0.8C to 4.1 volts, and finally charged at constant voltage to the cutoff current. By gradually reducing the current, the polarization of the later charging stages is reduced, generating a "charging curve adjustment to segmented constant current and constant voltage instruction". Under level three risk, the segmented constant current and constant voltage curve is further optimized by increasing the number of current steps (e.g., 3-4 steps) and the current reduction at each step is larger. At the same time, the cutoff voltage of the constant voltage stage is appropriately reduced (e.g., from 4.2 volts to 4.15 volts) to reduce the risk of overcharging, generating a "charging curve adjustment to multi-segment constant current and constant voltage and reduced cutoff voltage instruction". The "charging rate adjustment instruction" and the "charging curve adjustment instruction" are integrated to form the final charging strategy adjustment instruction. This instruction must include information such as "risk level basis", "specific adjustment parameters" and "execution time (effective when the next charging starts)" and is sent directly to the lithium battery charging control module to ensure that the charging strategy is adjusted accurately according to the risk level as much as possible. At the same time, the instruction is stored in association with the risk level result for subsequent charging effect tracking and strategy optimization.

[0049] In this application, the risk level classification combines "voltage change rate extreme value + ambient temperature correction" to solve the bias problem of traditional risk classification based on a single parameter; the charging strategy adjustment adopts a two-dimensional design of "rate adjustment + curve adjustment" instead of the traditional single rate adjustment, which can more comprehensively mitigate lithium plating; each instruction is associated with specific parameters (such as clear current value and voltage value), avoiding the execution deviation caused by the ambiguity of traditional instruction parameters. These designs make the charging strategy adjustment more accurate.

[0050] Optionally, step 3 includes: step 31, based on the temperature-corrected voltage change rate features, using a multi-scale feature fusion algorithm to extract the instantaneous fluctuation features within a short time window and the global trend features within a long time window, respectively, to generate dual-scale voltage change rate features; Step 32: Based on the physical law of voltage decay during the static stage of lithium battery, assign dynamic weights to instantaneous fluctuation features and global trend features (the weight of global trend features is higher than that of instantaneous fluctuation features in the early stage of lithium plating, and the weight of instantaneous fluctuation features gradually increases in the middle and late stages of lithium plating). Perform weighted fusion processing on the dual-scale voltage change rate features to generate weighted fused voltage change rate features. Step 33: Perform hierarchical similarity calculation between the weighted fusion voltage change rate feature and the features in the lithium plating feature library (first match the global trend, then verify the instantaneous fluctuation), and generate the lithium plating judgment result based on the hierarchical matching result.

[0051] Specifically, preferably, the specific implementation process of step 31 is as follows: First, clarify the adaptation and modification direction of the multi-scale feature fusion algorithm, taking the voltage change rate feature after temperature correction as the processing object. This feature includes voltage change rate values ​​arranged in time order and corresponding timestamps. The multi-scale feature fusion algorithm originates from the feature extraction technology in the field of image. Its original function is to capture the local details and global contours of the image. This application makes targeted modifications to adapt it to the one-dimensional feature extraction scenario of lithium battery voltage time series data. Unlike the two-dimensional window division in the field of image, this application adopts a one-dimensional time window design to solve the contradiction of dimensional adaptation between time series data and spatial data. Secondly, a dual-scale time window division rule is set. The short time window is used to capture instantaneous fluctuation characteristics. Its length is determined based on the response period of the instantaneous fluctuation of the lithium battery voltage. It needs to cover 3-5 consecutive sampling points to ensure that sudden voltage change rate jumps can be recorded as completely as possible. The long time window is used to extract global trend characteristics. Its length is determined based on the voltage relaxation period of the lithium battery after it is fully charged and is in a resting state. It needs to cover 1 / 5-1 / 3 of the resting time to ensure that the overall decay or rise trend of the voltage change rate can be reflected as much as possible. For example, when the resting time is 30 minutes, the short time window can be set to include 5 sampling points (50 seconds when the sampling interval is 10 seconds), and the long time window can be set to include 360 ​​sampling points (1 / 3 of the 60-minute resting time). Then, dual-scale feature extraction is performed. The temperature-corrected voltage change rate feature is slid across short time windows, and the extreme values, variance, and slope of the voltage change rate within each window are calculated. These parameters are then integrated to generate instantaneous fluctuation features. Simultaneously, the feature is slid across long time windows, and the mean, linear fitting coefficient, and trend direction (rising / falling / stable) of the voltage change rate within each window are calculated. These parameters are then integrated to generate global trend features. Finally, the instantaneous fluctuation features and global trend features are correlated one-to-one by timestamp to form a dual-scale voltage change rate feature. This feature retains both local abnormal fluctuation information and overall trend information, providing a multi-dimensional data foundation for subsequent weighted fusion.

[0052] Preferably, in the specific technical implementation of step 32, firstly, a correlation model between the physical law of voltage decay and weight allocation is established, based on the physical law of voltage decay during the static stage of lithium battery. This law indicates that in the early stage of lithium plating, the amount of lithium ion deposition inside the lithium battery is small, and the voltage change is mainly slow relaxation. The global trend feature can more accurately reflect the initial state of lithium plating. In the middle and late stages of lithium plating, the accelerated growth of lithium dendrites leads to frequent voltage fluctuations, and the instantaneous fluctuation feature is more sensitive to the degree of lithium plating. Based on this law, a dynamic weight allocation function is designed. The function input is the trend strength of the global trend feature (represented by the absolute value of the linear fitting coefficient) and the fluctuation amplitude of the instantaneous fluctuation feature (represented by the variance) in the dual-scale voltage change rate feature. The output is the weight of the instantaneous fluctuation feature and the weight of the global trend feature, and the sum of the two weights is always 1. Secondly, determine the initial weight values ​​and adjustment rules. In the early stage of lithium plating (when the global trend strength is greater than the preset trend threshold and the instantaneous fluctuation amplitude is less than the preset fluctuation threshold), the initial weight of the global trend feature is set higher than that of the instantaneous fluctuation feature, for example, a global trend feature weight of 0.7 and an instantaneous fluctuation feature weight of 0.3. As the degree of lithium plating deepens, when the instantaneous fluctuation amplitude exceeds the preset fluctuation threshold and the global trend strength begins to weaken, the instantaneous fluctuation feature weight is gradually increased and the global trend feature weight is decreased in a fixed step size until a stable state of 0.7 for the instantaneous fluctuation feature and 0.3 for the global trend feature is reached in the middle and late stages of lithium plating. The adjustment step size needs to be determined based on the sampling frequency of the voltage change rate feature to ensure that the weight change is synchronized with the change in the degree of lithium plating as much as possible. Then, a weighted fusion process is performed, taking the instantaneous fluctuation feature and the global trend feature from the dual-scale voltage change rate feature as the processing objects. Each parameter value of the instantaneous fluctuation feature is multiplied by its corresponding instantaneous fluctuation feature weight, and each parameter value of the global trend feature is multiplied by its corresponding global trend feature weight. The two sets of weighted parameter values ​​are then superimposed along the same dimension. For example, the extreme values ​​of the weighted instantaneous fluctuation feature are added to the mean of the weighted global trend feature, and the variance of the weighted instantaneous fluctuation feature is added to the linear fitting coefficient of the weighted global trend feature. Finally, all the superimposed parameters are arranged in timestamp order to generate the weighted fused voltage change rate feature. This feature adapts to the differences in feature sensitivity at different lithium plating stages through dynamic weights, solving the problem of early lithium plating missed detection or mid-to-late-stage lithium plating misdetection caused by traditional fixed-weight fusion.

[0053] Preferably, in the specific implementation of step 33, firstly, a hierarchical similarity calculation framework is constructed, taking the weighted fusion of voltage change rate features and lithium plating feature library as the processing objects. Each feature entry in the lithium plating feature library contains a global trend template and an instantaneous fluctuation template, and is stored according to environmental temperature and lithium plating risk level. The hierarchical similarity calculation framework is divided into two layers: the first layer is the global trend matching layer, and the second layer is the instantaneous fluctuation verification layer. The two layers are executed sequentially, and the matching range of the latter layer is limited by the matching result of the former layer, avoiding the problem of excessively large matching range and low efficiency caused by traditional single similarity calculation. Secondly, global trend matching is performed to extract global trend-related parameters (mean, linear fitting coefficient, trend direction) from the weighted fusion voltage change rate features. Feature entries consistent with the current ambient temperature and trend direction are selected from the lithium plating feature library to form a global trend candidate set. The similarity between the global trend parameters of the weighted fusion voltage change rate features and the global trend template of each entry in the global trend candidate set is calculated (using the cosine similarity algorithm). A global trend similarity threshold is set, and feature entries with similarity higher than the threshold are retained to generate a global trend matching pass set. This process can quickly eliminate feature entries that do not conform to the overall trend and narrow the scope of subsequent verification. Then, instantaneous fluctuation verification is performed. Instantaneous fluctuation-related parameters (extreme values, variance, and slope of change) are extracted from the weighted fused voltage change rate features. The similarity between these parameters and the global trend is calculated by matching the instantaneous fluctuation templates of each entry in the set (using the Euclidean distance algorithm). An instantaneous fluctuation similarity threshold is set, and feature entries with similarity higher than the threshold are retained to generate the final matching set. If the final matching set is empty, it is determined that the lithium plating state is unclear. If the final matching set contains multiple feature entries, the entry with the highest similarity is selected as the optimal matching result. Finally, the corresponding lithium plating risk level and electrode state reference information are extracted based on the optimal matching result. This information is integrated to generate a lithium plating judgment result. This lithium plating judgment result not only meets the global trend consistency requirement but also ensures the local feature matching degree as much as possible through instantaneous fluctuation verification, significantly improving the reliability of the lithium plating judgment. It directly serves as the basis for the output result of step 4 and the charging strategy adjustment in step 5, forming a closed loop of technical logic.

[0054] In traditional lithium battery lithium plating detection, feature extraction often employs single-time-scale analysis, which is susceptible to interference from local instantaneous noise or misses global trend information, resulting in low feature recognition accuracy. Furthermore, the matching process is often a single-dimensional comparison, failing to account for the feature differences at different lithium plating stages, leading to missed detections in early lithium plating or misjudgments in mid-to-late stages. The broad matching range also affects judgment efficiency. However, step 31 of this application utilizes a multi-scale feature fusion algorithm to simultaneously extract the instantaneous fluctuation features of the voltage change rate after temperature correction within a short time window and the global trend features within a long time window. The resulting dual-scale voltage change rate features retain both local abnormal fluctuation information and encompass the overall trend, effectively overcoming the limitations of traditional single-scale feature extraction and laying the foundation for accurate matching. Step 32, based on the physical laws of voltage decay during the lithium battery's resting stage, assigns dynamic weights to the dual-scale voltage change rate features and performs weighted fusion, overcoming the bottleneck of traditional fixed-weight fusion's inability to adapt to the differences in feature sensitivity at different lithium plating stages, thus enabling weighted fusion... The voltage change rate feature can better match the actual state of different lithium plating stages; Step 33 adopts a hierarchical similarity calculation method of first global trend matching and then instantaneous fluctuation verification. First, a set of candidate features with similar trends is selected from the lithium plating feature library, and then local feature verification is performed. This not only narrows the matching range and improves the matching efficiency, but also reduces the error of single-dimensional matching through double verification. Compared with the traditional indiscriminate global matching, it greatly improves the reliability of lithium plating judgment. In particular, it can better deal with specific scenarios where the voltage changes are complex and the characteristics of different stages of lithium plating are different during the static process of lithium battery, and effectively reduces the judgment bias problem that traditional processing methods are prone to in this scenario.

[0055] Optionally, step 5 includes: step 51, based on the lithium plating judgment result, the cycle life data of the lithium battery and the current health status data, constructing a dual-objective optimization model of lithium plating risk and charging efficiency, and solving the optimal charging parameter range under different risk levels through the model predictive control algorithm to generate the initial charging parameter range; Step 52: Based on the initial charging parameter range, and combined with the type of lithium battery (power type / energy storage type) and the charging time requirements of the application scenario, the optimal charging parameter range is constrained and corrected (power type lithium batteries prioritize ensuring the lower limit of charging efficiency, and energy storage type lithium batteries prioritize reducing the upper limit of lithium plating risk) to generate the constrained and corrected charging parameter range. Step 53: Convert the constrained charging parameter range into specific charging rate adjustment instructions and charging curve adjustment instructions. Verify the execution effect of the instructions through real-time feedback data from the battery management system, and dynamically fine-tune the parameters to generate the final charging strategy adjustment instructions.

[0056] Preferably, the specific implementation process of step 51 is as follows: First, the core input and construction logic of the dual-objective optimization model are clarified, using the lithium plating judgment result, the cycle life data of the lithium battery, and the current health status data as joint input parameters. The lithium plating judgment result provides core information on the current lithium plating risk level and voltage change rate characteristics. The cycle life data reflects the lithium battery's used cycle and remaining life potential. The current health status data includes key indicators such as internal resistance and capacity decay rate. The three together constitute the input basis of the model, solving the optimization one-sidedness problem caused by traditional single-parameter modeling. Second, the input parameters are standardized. The risk level in the lithium plating judgment result is converted into a quantitative risk coefficient (e.g., a low coefficient for level 1 risk and a high coefficient for level 3 risk). The cycle life data is converted into the remaining life ratio (the reciprocal of the ratio of the number of cycles already completed to the total number of designed cycles). The internal resistance and capacity decay rate in the current health status data are scaled to the [0,1] interval according to a uniform ratio to generate a standardized set of input parameters, ensuring that parameters of different dimensions can participate in unified calculations as much as possible. Then, a hierarchical structure of a dual-objective optimization model for lithium plating risk and charging efficiency is constructed. This model includes an objective function layer, a constraint layer, and a solution layer. The objective function layer sets two core objectives: one is to minimize the lithium plating risk (using standardized risk coefficients and health status data as variables to construct a linear objective function), and the other is to maximize charging efficiency (using charging rate and charging time as variables to construct a nonlinear objective function). The constraint layer clarifies the parameter boundaries, including the upper limit of the charging rate (set based on the lithium battery type and health status), the charging voltage range (not exceeding the nominal cutoff voltage), and temperature constraints (the temperature during charging does not exceed the safety threshold). The solution layer adopts an optimized model predictive control algorithm, adapting the rolling optimization window of the traditional model predictive control algorithm—corresponding the window length to the lithium battery charging stage (e.g., a short window length for the constant current stage and a long window length for the constant voltage stage), avoiding the dynamic response lag problem caused by the traditional fixed window. Finally, the model solving process is executed. Using the standardized input parameter set as input, the dual objective function is solved through the optimized model predictive control algorithm to obtain the charging rate range, charging curve slope range, and constant voltage range that meet the constraints under different lithium plating risk levels. These ranges are integrated to generate the initial charging parameter range, which ensures that the lithium plating risk is within a controllable range while also taking into account a reasonable level of charging efficiency.

[0057] Preferably, in the specific technical implementation of step 52, firstly, the core basis for scenario-based constraint correction is clarified. Taking the initial charging parameter range as the processing object, and combining the type of lithium battery (power type / energy storage type) and the charging timeliness requirements of the application scenario, a constraint correction rule base is established. Power type lithium batteries are mostly used in electric vehicles, portable devices and other scenarios, with high charging timeliness requirements, and the lower limit of charging efficiency should be prioritized. Energy storage type lithium batteries are mostly used in grid energy storage, home energy storage and other scenarios, with higher priority for safety and lifespan, and the upper limit of lithium plating risk should be prioritized. This rule base solves the problem that traditional unified parameters cannot be adapted to different scenarios. Secondly, differentiated constraint correction strategies are formulated. For power lithium batteries: the lower limit of the charging rate in the initial charging parameter range is extracted and raised to a preset efficiency guarantee threshold (based on the nominal charging efficiency of power lithium batteries), while the upper limit of lithium plating risk is appropriately relaxed (not exceeding a fixed proportion of the upper limit of the initial range) to ensure that the charging efficiency is not lower than the user's core needs as much as possible. For energy storage lithium batteries: the upper limit of lithium plating risk in the initial charging parameter range is extracted and lowered to a preset safety threshold (based on the long-term cycle life requirements of energy storage lithium batteries), while the upper limit of the charging rate can be appropriately reduced to sacrifice some non-core charging efficiency in exchange for a lower lithium plating risk. For example, the lower limit of the charging rate for power lithium batteries can be raised from 0.8C to 1.0C, and the voltage change rate threshold corresponding to lithium plating risk for energy storage lithium batteries can be lowered from 0.011 volts / second to 0.010 volts / second. Then, a constraint correction operation is performed, comparing each parameter range in the initial charging parameter range with the thresholds in the differentiated constraint correction strategy. Parameter ranges exceeding the threshold range are truncated and adjusted: the lower limit of the charging rate range for power lithium batteries is corrected according to the efficiency guarantee threshold, while other parameter ranges are retained in their initial range; the upper limit of the lithium plating risk-related parameter range for energy storage lithium batteries is corrected according to the safety threshold, and the charging rate range is adjusted accordingly. Finally, a constraint-corrected charging parameter range is generated. This range inherits the optimization foundation of the initial charging parameter range and adapts to the core needs of different types of lithium batteries through scenario-based constraints, ensuring the charging strategy's relevance and practicality as much as possible.

[0058] Preferably, in the specific implementation of step 53, firstly, the conversion of parameter ranges into control commands is completed. Taking the constraint-corrected charging parameter range as the processing object, the charging rate range is converted into specific charging rate adjustment commands (clarifying the minimum allowable charging rate and the maximum allowable charging rate), and the charging curve slope range and the constant voltage stage voltage range are converted into specific charging curve adjustment commands (clarifying the current change slope in the constant current stage and the target voltage value in the constant voltage stage). For example, when the constraint-corrected charging rate range is 0.5C-1.0C, the charging rate adjustment command is set to "maintain the rate between 0.5C and 1.0C during the charging process, and prioritize the execution of the initial rate"; the charging curve adjustment command is set to "the current slope in the constant current stage increases by 0.1C / minute, and the target voltage in the constant voltage stage is 98%-100% of the nominal cutoff voltage". Secondly, a real-time feedback verification mechanism is established. The battery management system collects key data during the charging process in real time (including real-time charging rate, battery temperature, voltage change rate, and health status data). This real-time feedback data is compared with the constrained and corrected charging parameter range and the lithium plating judgment results. If the real-time voltage change rate exceeds the threshold range corresponding to the lithium plating risk, it indicates an increased lithium plating risk, and the charging rate needs to be reduced. If the real-time charging time exceeds the timeliness requirement threshold of the application scenario, it indicates insufficient charging efficiency, and the charging rate needs to be increased within a safe range. If the real-time battery temperature exceeds the safe threshold, adjustments need to be paused and the current parameters maintained until the temperature drops. Then, dynamic fine-tuning is performed. Based on the real-time feedback verification results, the parameters in the charging rate adjustment command and the charging curve adjustment command are corrected according to a preset fine-tuning step size: when the lithium plating risk increases, the charging rate is reduced by a fixed step size, while simultaneously reducing the slope of the charging curve; when the charging efficiency is insufficient, the charging rate is increased by a fixed step size, while maintaining a stable charging curve slope; when the temperature is abnormal, fine-tuning continues in the original direction after the temperature recovers. Finally, the final charging strategy adjustment instruction is generated, integrating the dynamically fine-tuned charging rate adjustment instruction with the charging curve adjustment instruction. The execution order, triggering conditions, and termination conditions of the instructions are clearly defined (such as termination when charging to full capacity or when an abnormality occurs). This instruction, through a closed-loop process of "conversion-verification-fine-tuning," ensures that the execution effect is consistent with expectations as much as possible. This not only solves the problem that traditional fixed instructions cannot cope with dynamic changes during the charging process, but also ensures a balance between lithium plating risk and charging efficiency through real-time feedback. It directly affects the charging control module of the lithium battery, completing the optimized output of the entire charging strategy.

[0059] Traditional lithium battery charging strategies are often adjusted based on fixed parameters set according to a single lithium plating risk level, without taking into account the battery's remaining lifespan potential and differences in health status. This results in low strategy adaptability and a lack of targeted design for different types of batteries and application scenarios. It is difficult to balance the suppression of lithium plating risk and the demand for charging efficiency. At the same time, there is no real-time feedback adjustment mechanism during the charging process, which cannot cope with the dynamically changing battery status. In step 51 of this application, a dual-objective optimization model is constructed by combining the lithium plating judgment result, cycle life data, and current health status data. The optimal charging parameter range is solved by the optimized model predictive control algorithm. The generated initial charging parameter range can more comprehensively take into account both lithium plating risk and charging efficiency compared with the parameter range derived by traditional single parameters. Step 52, based on the initial charging parameter range, performs constraint corrections in combination with the charging time requirements of lithium battery type and application scenario. For power lithium batteries, priority is given to ensuring the lower limit of charging efficiency, and for energy storage lithium batteries, priority is given to reducing the upper limit of lithium plating risk. This solves the problem that traditional unified charging parameters cannot adapt to the core needs of different scenarios. Step 53 transforms the constraint-corrected charging parameter range into specific instructions, and verifies the execution effect and dynamically fine-tunes it through real-time feedback data from the battery management system. Compared with traditional fixed instructions, the generated final charging strategy adjustment instructions can more flexibly respond to the dynamic changes in battery state during charging. This reduces the possibility of increased lithium plating risk and meets the charging time requirements of different scenarios. In particular, it is suitable for specific scenarios where the lithium plating state of lithium batteries is complex after resting and the application scenarios are highly differentiated. It effectively reduces the problems of strategy rigidity and poor effect that traditional processing methods are prone to in this scenario.

[0060] In one embodiment, this application also provides a lithium battery lithium plating detection system, referring to... Figure 2 ,include: The data acquisition module is used to acquire the voltage timing data of the lithium battery during the resting process after the lithium battery has been fully charged. The feature extraction module is used to smooth the voltage time series data and perform differential calculations on the smoothed voltage time series data to generate the voltage change rate features of the lithium battery. The lithium plating judgment module is used to match the voltage change rate feature with the features in the lithium plating feature library, and generate the lithium plating judgment result based on the matching result. The lithium plating feature library is used to store voltage change rate feature data that characterize different lithium plating risks. The early warning output module is used to output the lithium plating judgment result.

[0061] Specifically, after the lithium battery is fully charged, the data acquisition module acquires the voltage timing data of the lithium battery during the resting process (equivalent to current and voltage data acquisition); then, the feature extraction module (equivalent to the MCU1 / CPU feature extraction module) smooths the voltage timing data and performs differential calculation on the smoothed voltage timing data to generate the voltage change rate feature of the lithium battery; next, the lithium plating judgment module matches the voltage change rate feature with the features in the lithium plating feature library and generates a lithium plating judgment result based on the matching result. The lithium plating feature library is used to store voltage change rate feature data characterizing different lithium plating risks; finally, the warning output module outputs the lithium plating judgment result.

[0062] In some embodiments, battery full charge status detection can be performed by a full charge determination system, which can use any existing power detection system. Of course, after the warning output module outputs the lithium plating determination result, the strategy action module also determines the corresponding risk level based on the lithium plating determination result, and generates a charging strategy adjustment instruction based on the risk level, sending it to the MCU2 strategy action module. The dashboard display module can then display the lithium plating determination result.

[0063] This application provides a lithium battery lithium plating detection device. The device can be a terminal or a server, including a processor and a memory. The memory stores at least one instruction or at least one program. The at least one instruction or at least one program is loaded and executed by the processor to implement the lithium battery lithium plating detection method provided in the above method embodiments.

[0064] The memory can be used to store software programs and modules. The processor executes various functional applications and detects lithium plating in lithium batteries by running the software programs and modules stored in the memory. The memory mainly includes a program storage area and a data storage area. The program storage area can store the operating system, application programs required for the functions, etc.; the data storage area can store data created according to the use of the device, etc. In addition, the memory can include high-speed random access memory, and can also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device. Accordingly, the memory can also include a memory controller to provide the processor with access to the memory.

[0065] The methods and embodiments provided in this application can be executed in electronic devices such as mobile terminals, computer terminals, servers, or similar computing devices. Figure 3 This is a hardware structure block diagram of an electronic device for detecting lithium plating in lithium batteries, as provided in an embodiment of this application. Figure 3As shown, the electronic device 900 can vary significantly due to differences in configuration or performance. It may include one or more central processing units (CPUs) 910 (CPUs 910 may include, but are not limited to, microprocessors such as MCUs or programmable logic devices such as FPGAs), a memory 930 for storing data, and one or more storage media 920 (e.g., one or more mass storage devices) for storing application programs 923 or data 922. The memory 930 and storage media 920 may be temporary or persistent storage. The program stored in the storage media 920 may include one or more modules, each module may include a series of instruction operations on the electronic device. Furthermore, the CPU 910 may be configured to communicate with the storage media 920 and execute the series of instruction operations in the storage media 920 on the electronic device 900. Electronic device 900 may also include one or more power supplies 960, one or more wired or wireless network interfaces 950, one or more input / output interfaces 940, and / or one or more operating systems 921, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0066] The input / output interface 940 can be used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the electronic device 900. In one example, the input / output interface 940 includes a network interface controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the input / output interface 940 may be a radio frequency (RF) module for wireless communication with the Internet.

[0067] Embodiments of this application also provide a computer-readable storage medium, which can be disposed in an electronic device to store at least one instruction or at least one program related to implementing a lithium battery lithium plating detection method in the method embodiment. The at least one instruction or the at least one program is loaded and executed by the processor to implement the lithium battery lithium plating detection method provided in the above method embodiment.

[0068] Optionally, in this embodiment, the storage medium may be located at at least one of the multiple network servers in a computer network. Optionally, in this embodiment, the storage medium may include, but is not limited to, various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0069] According to one aspect of this application, a computer program product or computer program is provided, comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the methods provided in the various alternative implementations described above.

[0070] It should be understood that this application is not limited to the processes and structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A method for detecting lithium plating in lithium batteries, characterized in that, include: After the lithium battery is fully charged, the voltage timing data of the lithium battery during the resting process is acquired. The voltage time series data is smoothed, and the smoothed voltage time series data is differentially calculated to generate the voltage change rate characteristics of the lithium battery. The voltage change rate feature is matched with features in the lithium plating feature library, and a lithium plating judgment result is generated based on the matching result. The lithium plating feature library is used to store voltage change rate feature data characterizing different lithium plating risks. Output the lithium plating determination result.

2. The detection method according to claim 1, characterized in that, The step of acquiring voltage timing data of the lithium battery during the resting process after the lithium battery has reached full charge includes: When the charging of the lithium battery enters the constant voltage stage and the charging current drops to the set cutoff current, it is determined that the lithium battery has reached a fully charged state. After the lithium battery reaches full charge, the data acquisition module is triggered. The data acquisition module collects the voltage and current data of the lithium battery at a set sampling interval and continues for a set resting time to obtain the voltage timing data and current timing data.

3. The detection method according to claim 2, characterized in that, The duration of the sampling interval is adjusted according to the charging current of the lithium battery or the ambient temperature.

4. The detection method according to claim 3, characterized in that, When dynamically adjusting the data sampling interval, if the ambient temperature is lower than the preset temperature threshold or the charging current is greater than the preset current threshold, a smaller sampling interval is used for data acquisition. If the ambient temperature is higher than or equal to the preset temperature threshold and the charging current is less than or equal to the preset current threshold, the default sampling interval is used for data acquisition to generate voltage timing data and current timing data under the corresponding sampling interval.

5. The detection method according to claim 1, characterized in that, The smoothing process for the voltage timing data includes: An outlier removal method based on the standard deviation criterion is used to remove data points that exceed the set range from the voltage time series data in order to generate effective voltage data. The effective voltage data is smoothed using a moving average algorithm to generate smoothed voltage time series data. The differential calculation of the smoothed voltage time series data includes: The voltage time series data after smoothing is calculated using the central difference method to generate the voltage change rate feature.

6. The detection method according to claim 1, characterized in that, The process of matching the voltage change rate feature with features in the lithium plating feature library and generating a lithium plating judgment result based on the matching result includes: When matching the voltage change rate feature with features in the lithium plating feature library, if the voltage change rate feature meets the preset inflection point determination condition, the voltage change rate feature is determined to be a potential lithium plating feature. The potential lithium plating features are matched with features in the lithium plating feature library, and the lithium plating risk level is determined based on the matching results to generate the lithium plating judgment result.

7. The detection method according to claim 6, characterized in that, The preset inflection point determination condition is that the absolute value of the voltage change rate characteristic is greater than a preset threshold, and the duration of this state reaches a preset duration. If this condition is met, it is determined to be a potential lithium plating characteristic.

8. The detection method according to claim 1, characterized in that, Also includes: The corresponding risk level is determined based on the lithium plating assessment result, and a charging strategy adjustment instruction is generated based on the risk level.

9. The detection method according to claim 8, characterized in that, The step of determining the corresponding risk level based on the lithium plating judgment result and generating a charging strategy adjustment instruction based on the risk level includes: When determining the corresponding risk level based on the lithium plating judgment result, multiple risk levels are divided according to the numerical range of the voltage change rate characteristic. The corresponding risk level is determined based on the numerical range to which the voltage change rate characteristic belongs. Then, corresponding charging rate adjustment instructions and charging curve adjustment instructions are generated according to different risk levels, which serve as charging strategy adjustment instructions.

10. A detection system for lithium plating in lithium batteries, characterized in that, include: The data acquisition module is used to acquire the voltage timing data of the lithium battery during the resting process after the lithium battery has been fully charged. The feature extraction module is used to smooth the voltage time series data and perform differential calculation on the smoothed voltage time series data to generate the voltage change rate feature of the lithium battery. The lithium plating judgment module is used to match the voltage change rate feature with the features in the lithium plating feature library, and generate a lithium plating judgment result based on the matching result. The lithium plating feature library is used to store voltage change rate feature data characterizing different lithium plating risks. The early warning output module is used to output the lithium plating judgment result.

11. A computer device, characterized in that, The device includes a processor and a memory, the memory storing at least one instruction or at least one program, the at least one instruction or the at least one program being loaded and executed by the processor to implement the lithium battery lithium plating detection method as described in any one of claims 1-9.

12. A computer program product or computer program, characterized in that, The computer program product or computer program includes computer instructions that, when executed by a processor, implement the lithium plating detection method for lithium batteries as described in any one of claims 1-9.

Images