Storage battery charging control system and control method thereof

Abstract

The invention discloses a storage battery charging control system and a control method thereof, and relates to the technical field of storage battery charging, and the control method comprises the following steps: firstly, comprehensively obtaining multi-modal sensing data from a storage battery; preprocessing the obtained original data, and extracting key features beneficial to identification of lithium ion migration blocking from the preprocessed data under a detection window; and inputting the extracted features into a pre-learned deep learning model, and carrying out intelligent assessment on the lithium ion migration blocking risk in the storage battery charging process through the deep learning model. Multi-modal sensing data and a deep learning model are utilized, the lithium ion migration blocking risk is accurately identified, an electrochemical polarization rapid change factor and an electrode dynamics non-equilibrium factor are extracted to generate a deposition index and classify the deposition index, a charging strategy is dynamically adjusted, high-current charging is carried out during normal migration, and current is reduced during blocked migration. Lithium deposition and dendritic crystal risks are reduced, efficiency and safety are improved, and the service life of the battery is prolonged.

Description

Technical Field

[0001] The present invention relates to the technical field of battery charging, and in particular to a battery charging control system and a control method thereof. Background Art

[0002] Battery charging control refers to the process of dynamically adjusting the charging current and voltage by monitoring key parameters such as the battery's voltage, current, temperature, and internal resistance, combined with the charging curve characteristics and actual usage requirements, to ensure that the battery is in a safe, stable, and efficient state during the charging process. This process usually uses intelligent control technology, such as a charging method based on a constant current-constant voltage (CC-CV) strategy, or an optimization algorithm combined with temperature compensation and state prediction, to avoid battery performance degradation, overheating, or safety hazards caused by overcharging or undercharging, thereby extending the battery's service life and improving energy utilization efficiency.

[0003] The prior art has the following deficiencies:

[0004] Prior art battery charging usually adopts a strategy of precharging with a small current and then entering a constant current charge (high current to quickly replenish the battery power) to balance charging efficiency and battery safety while protecting battery performance. However, when the battery enters the constant current charging stage, the high current may cause insufficient lithium ion migration rate, and some lithium ions cannot be embedded in the negative electrode graphite structure, but are deposited as metallic lithium on the negative electrode surface, thereby forming unstable metallic lithium dendrites. The deposition of metallic lithium consumes reversible lithium ions and forms an unstable solid electrolyte interface film (SEI) on its surface, further accelerating the loss of lithium ions, resulting in rapid decay of battery capacity, reduced energy density and decreased output power. In addition, dendrites have directional growth and may pierce the diaphragm and contact the positive electrode to form an internal short circuit. Internal short circuits can cause local overheating, generate a lot of heat and trigger electrolyte decomposition, release flammable gases, and aggravate thermal runaway. If thermal runaway cannot be controlled, it may cause the battery to catch fire or explode, seriously endangering equipment safety and personal health, and posing huge safety hazards.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The purpose of the present invention is to provide a battery charging control system and a control method thereof, which can accurately identify the risk of blocked lithium ion migration in the battery through multimodal sensing data and deep learning models. Features are extracted based on the electrochemical polarization rapid change factor and the electrode kinetic non-equilibrium factor, and the lithium ion deposition index is generated and classified. The charging strategy is dynamically adjusted according to the classification results, and high current constant current charging is performed during normal migration, and the current is reduced during blocked migration, thereby reducing the risk of lithium deposition and dendrites, improving charging efficiency and safety, and extending battery life, so as to solve the problems in the above-mentioned background technology.

[0007] In order to achieve the above object, the present invention provides the following technical solution: a battery charging control method, comprising the following steps:

[0008] First, multimodal sensor data is comprehensively acquired from the battery;

[0009] Preprocessing the acquired raw data, and extracting key features that are helpful for identifying lithium ion migration obstruction from the preprocessed data within the detection window;

[0010] The extracted features are input into a pre-learned deep learning model, and the risk of lithium ion migration obstruction during battery charging is intelligently evaluated through the deep learning model;

[0011] According to the evaluation results of the deep learning model, the lithium ion migration state is divided into normal lithium ion migration and blocked lithium ion migration;

[0012] For normal lithium ion migration, a constant current charging strategy is maintained to quickly replenish the battery power through high current to ensure charging efficiency and time optimization;

[0013] For the hindered migration of lithium ions, based on the evaluation results of the deep learning model, the charging current is dynamically reduced based on the high current of constant current charging to mitigate the risk of lithium ion deposition and dendrite formation.

[0014] Preferably, the dynamic change rate of polarization voltage and the degree of symmetry of the spatial distribution of the negative electrode electrochemical reaction rate during the battery charging process are extracted from the preprocessed data. After acquisition, the dynamic change rate of polarization voltage and the degree of symmetry of the spatial distribution of the negative electrode electrochemical reaction rate are analyzed under the detection window to generate an electrochemical polarization rapid change factor and an electrode kinetic non-equilibrium factor respectively. The electrochemical polarization rapid change factor quantifies the dynamic change rate of polarization voltage with time during the battery charging process, reflects the aggravation of polarization phenomenon during lithium ion migration, and indirectly reveals the risk of obstruction of lithium ion migration; the electrode kinetic non-equilibrium factor quantifies the degree of asymmetry of the spatial distribution of the electrochemical reaction rate on the negative electrode surface, reflects the degree of uneven distribution of lithium ions during the negative electrode embedding process, thereby evaluating the local deposition trend caused by obstruction of lithium ion migration.

[0015] Preferably, in the detection window, the specific steps of generating the electrochemical polarization rapid change factor after analyzing the dynamic change rate of the polarization voltage are as follows:

[0016] In the detection window, the voltage data during the charging process is collected in real time, and the polarization voltage is extracted. The calculation expression is as follows:

[0017] V p (t)=V(t)-I(t)·R s

[0018] , where V p (t) is the polarization voltage, V(t) is the total voltage, i.e. the total voltage value of the battery measured at time point t, I(t) is the charging current of the battery at time point t, R s is the internal resistance of the battery;

[0019] By polarization voltage V p (t) Calculate the dynamic change rate of polarization voltage. The calculation expression is as follows:

[0020]

[0021] , where R p (t) is the dynamic change rate of polarization voltage, Δt is the time interval;

[0022] In order to capture the microscopic details of voltage changes, high-order partial derivatives are introduced to enhance the sensitivity of dynamic response. The calculation expression is as follows:

[0023]

[0024] , where is the rate of dynamic change of polarization voltage enhancement, α is the sensitivity weight coefficient, which is used to adjust the contribution weight of the second-order derivative to the rate;

[0025] To further quantify the severity of the dynamic change, an amplitude weighting factor is introduced to reflect the nonlinear fluctuation characteristics of the rate. The calculation expression is as follows:

[0026]

[0027] , where W p (t) is the weighting factor, β is the amplitude adjustment coefficient, and γ is the nonlinear index;

[0028] Combined with polarization voltage to enhance the dynamic change rate and weighting factor W p (t), construct the electrochemical polarization rapid change index, and the calculation expression is as follows:

[0029]

[0030] , where EPDVF is the electrochemical polarization rapid change factor, sin(ω·t+φ) is the harmonic modulation function, ω is the angular frequency, φ is the initial phase, and T is the total duration of the detection window.

[0031] Preferably, in the detection window, the specific steps of generating the electrode kinetic non-equilibrium factor after analyzing the symmetry degree of the negative electrode electrochemical reaction rate in the spatial distribution are as follows:

[0032] Under the detection window, the local current density of multiple areas on the negative electrode surface is collected. The negative electrode surface is divided into an N×M grid. Each grid node (i, j) represents a local area. The collected local current density is stored in a matrix form as J, J = {J(i, j)}, where J(i, j) represents the local current density of nodes i and j. The spatial gradient and distribution pattern of the local current density data matrix J are analyzed to quantify the degree of asymmetry.

[0033] For the collected J(i,j), calculate its gradient strength in space. The calculation expression is as follows:

[0034]

[0035] , where G(i,j) is the gradient strength;

[0036] The global imbalance of the reaction rate distribution on the entire negative electrode surface is calculated using the gradient strength G(i,j) of each node. The calculation expression is as follows:

[0037]

[0038] , where Q is the global imbalance indicator, k1 is the gradient smoothing parameter, and ω(i,j) is the weight factor;

[0039] Based on the global imbalance index Q, the electrode kinetic imbalance factor is defined and the calculation is expressed as follows:

[0040]

[0041] , where EKSAF is the electrode kinetic non-equilibrium factor, A is the total surface area of ​​the negative electrode, Ω is the spatial area of ​​the negative electrode, and e is the natural base number. is the adjustment coefficient, θ is the sensitivity adjustment parameter, G max is the maximum value of the gradient intensity, and γ is the threshold adjustment parameter.

[0042] Preferably, the extracted electrochemical polarization rapid change factor characteristics and electrode kinetic non-equilibrium factor characteristics are input into a pre-learned deep learning model, and a lithium ion deposition index is generated based on the deep learning model. The lithium ion deposition index is used to intelligently evaluate the risk of lithium ion migration obstruction during battery charging.

[0043] Preferably, the generated lithium ion deposition index is compared with a preset lithium ion deposition index reference threshold value to divide the lithium ion migration state, and the division steps are as follows:

[0044] If the lithium ion deposition index is greater than or equal to a preset lithium ion deposition index reference threshold, the lithium ion migration state is classified as normal lithium ion migration;

[0045] If the lithium ion deposition index is less than a preset lithium ion deposition index reference threshold, the lithium ion migration state is classified as lithium ion hindered migration.

[0046] Preferably, for lithium ion blocked migration, based on the deep learning model evaluation results, the charging current is dynamically reduced based on the constant current charging high current to mitigate the risk of lithium ion deposition and dendrite generation. The specific steps are as follows:

[0047] After the risk of lithium ion blocked migration is detected, the dynamic adjustment factor of the charging current is calculated based on the deviation between the lithium ion deposition index LIDSI value and the reference threshold of the lithium ion deposition index to determine the magnitude of the reduction in the charging current. The calculation expression is as follows:

[0048]

[0049] , where H adj is the dynamic adjustment factor of the charging current, k is the basic adjustment coefficient, Y1 is the temperature correction coefficient, Y2 is the internal resistance correction coefficient, ΔT is the current temperature change, LIDSI ref is the reference threshold of lithium ion deposition index, T max is the maximum safe temperature allowed by the battery, R init is the internal resistance correction factor;

[0050] According to the calculated dynamic adjustment factor H adj , dynamically adjust the high current of constant current charging to determine the new charging current. The calculation expression is as follows:

[0051]

[0052] , where I new is the dynamically adjusted charging current, I max is the maximum allowable value of charging current, I baseis the basic high current of constant current charging, G is the voltage response decreasing coefficient, which controls the proportional coefficient of the current decreasing amplitude when the voltage rises rapidly, dV / dt is the voltage change rate, SOC current is the current state of charge;

[0053] During the process of adjusting the charging current, the lithium ion deposition index LIDSI value is continuously monitored and the evaluation results are updated in real time. If the subsequent detection shows that the lithium ion deposition index LIDSI drops below the reference threshold LIDSI ref , then the charging current is gradually restored, and the expression is as follows:

[0054]

[0055] , where I′ new is the charging current after dynamic recovery, μ is the recovery rate coefficient, is the hyperbolic tangent function term, T current is the current temperature, δ is the temperature driving coefficient, which reflects the influence of the current battery temperature on the recovery of the charging current.

[0056] A battery charging control system includes a multimodal data acquisition module, a data preprocessing and feature extraction module, a deep learning evaluation module, a migration state classification module, a constant current charging optimization module and a dynamic charging adjustment module:

[0057] The multimodal data acquisition module firstly obtains the multimodal sensor data from the battery in an all-round way;

[0058] The data preprocessing and feature extraction module preprocesses the acquired raw data and extracts key features that are helpful for identifying lithium ion migration obstruction from the preprocessed data within the detection window;

[0059] The deep learning evaluation module inputs the extracted features into a pre-learned deep learning model, and uses the deep learning model to intelligently evaluate the risk of lithium ion migration obstruction during battery charging;

[0060] The migration state classification module divides the lithium ion migration state into normal lithium ion migration and blocked lithium ion migration according to the evaluation results of the deep learning model;

[0061] Constant current charging optimization module, for normal lithium ion migration, maintains constant current charging strategy, quickly replenishes battery power through high current, and ensures charging efficiency and time optimization;

[0062] The dynamic charging adjustment module dynamically reduces the charging current based on the evaluation results of the deep learning model and the high current of constant current charging to mitigate the risk of lithium ion deposition and dendrite formation when lithium ion migration is blocked.

[0063] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0064] Through the comprehensive acquisition of multimodal sensing data and the introduction of deep learning models, the present invention can accurately identify the risk of obstructed lithium ion migration during battery charging. Based on the innovative feature extraction method of electrochemical polarization rapid change factor and electrode kinetic non-equilibrium factor, the scheme can capture the dynamic change rate of polarization voltage and the degree of imbalance of negative electrode reaction rate, thereby quantifying the possibility of obstructed lithium migration and generating a lithium ion deposition index. After comparing the lithium ion deposition index with the reference threshold, the intelligent classification of normal and obstructed lithium ion migration is achieved, avoiding the uncertainty of traditional empirical threshold judgment. This intelligent evaluation method can not only provide early warning of the risk of lithium deposition and dendrite formation, but also provide a scientific basis for subsequent charging strategies, greatly improving the safety and reliability of the charging process.

[0065] The present invention dynamically adjusts the charging process through a differentiated strategy based on the assessed lithium ion migration state. Under the normal lithium ion migration state, the constant current charging strategy is maintained to quickly replenish the battery power with high current to optimize the charging efficiency and time; under the lithium ion blocked migration state, the charging current is dynamically reduced to significantly reduce the risk of lithium deposition and dendrite formation, avoiding safety hazards caused by internal short circuits and thermal runaway. This real-time adjustment charging strategy based on the evaluation results of the deep learning model not only protects the internal structure of the battery and extends the battery life, but also takes into account the maximization of charging efficiency. Compared with the traditional fixed charging method, this method achieves a dynamic balance between the efficiency and safety of the charging process, providing a strong guarantee for the long-term and stable use of the battery. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0067] Figure 1 The present invention is a method flow chart of a battery charging control method.

[0068] Figure 2 The module diagram of a battery charging control system of the present invention is shown in FIG. DETAILED DESCRIPTION

[0069] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of the present disclosure will be more comprehensive and complete, and the concept of the example embodiments will be fully conveyed to those skilled in the art.

[0070] The present invention provides Figure 1 A battery charging control method shown includes the following steps:

[0071] First, multimodal sensor data is comprehensively acquired from the battery;

[0072] Acquiring multimodal sensing data from batteries is usually achieved through high-precision sensors integrated into the battery management system (BMS), which can monitor the key operating parameters of the battery in real time, including voltage, current, temperature, and internal resistance. In addition, more dimensional data can be obtained through additional dedicated equipment, such as using acoustic sensors to detect abnormal sounds inside the battery, using infrared thermal imagers to monitor the heat distribution on the battery surface, or using dedicated electrochemical monitoring modules to analyze polarization voltage and electrochemical reaction characteristics. All of these sensors are linked to the BMS to transmit the real-time collected data to the data processing module, providing high-quality raw data input for subsequent analysis.

[0073] Preprocessing the acquired raw data, and extracting key features that are helpful for identifying lithium ion migration obstruction from the preprocessed data within the detection window;

[0074] The purpose of preprocessing the acquired raw data is to improve the quality and availability of the data and lay a solid foundation for subsequent feature extraction and analysis. Through preprocessing, the noise in the data (such as sensor measurement errors and environmental interference) can be effectively removed, missing data can be completed, and the integrity and consistency of the data can be ensured. At the same time, the data range can be adjusted through normalization and standardization to avoid analysis deviations caused by different dimensions. In addition, preprocessing can also highlight key trends and dynamic changes in the data, such as eliminating random fluctuations through smoothing filtering or differential analysis, making the extracted features more representative, thereby improving the risk assessment ability and prediction accuracy of subsequent machine learning models.

[0075] The dynamic change rate of polarization voltage and the degree of symmetry of the spatial distribution of the negative electrode electrochemical reaction rate during the battery charging process are extracted from the preprocessed data. After acquisition, the dynamic change rate of polarization voltage and the degree of symmetry of the spatial distribution of the negative electrode electrochemical reaction rate are analyzed under the detection window to generate the electrochemical polarization rapid change factor and the electrode kinetic non-equilibrium factor respectively. The electrochemical polarization rapid change factor reflects the aggravation of the polarization phenomenon during the lithium ion migration process by quantifying the dynamic change rate of polarization voltage with time during the battery charging process, and indirectly reveals the risk of lithium ion migration being blocked; the electrode kinetic non-equilibrium factor reflects the degree of uneven distribution of lithium ions during the negative electrode embedding process by quantifying the degree of asymmetry of the electrochemical reaction rate on the negative electrode surface, thereby evaluating the local deposition trend caused by the blockage of lithium ion migration;

[0076] The abnormal increase in the dynamic change rate of the polarization voltage indicates that the migration rate of lithium ions is insufficient, which is usually caused by the fact that the lithium insertion rate of the negative electrode material cannot keep up with the demand of the charging current. When the charging current is too high or other factors cause the migration rate of lithium ions from the positive electrode to the negative electrode to be slow, the lithium ion concentration on the surface of the negative electrode accumulates rapidly, and the polarization voltage increases abnormally. The increase in polarization voltage reflects the increased resistance to the electrochemical reaction on the surface of the negative electrode, indicating that lithium ions cannot be quickly embedded in the graphite structure, thereby depositing on the surface of the negative electrode to form metallic lithium. In addition, the deposited metallic lithium is unstable, which will further induce an increase in local current density, aggravate the non-uniformity of the reaction, and cause the metallic lithium to grow in the form of dendrites. These dendrites are directional, and once they pierce the diaphragm, they may cause internal short circuits and thermal runaway. Therefore, the abnormal increase in the dynamic change rate of the polarization voltage is an early signal of obstructed lithium ion migration and the formation of dendrites by the deposition of metallic lithium. Capturing this phenomenon can provide a key basis for dynamically adjusting the charging strategy and avoid further safety hazards.

[0077] The specific steps of generating the electrochemical polarization rate change factor after analyzing the dynamic change rate of polarization voltage in the detection window are as follows:

[0078] In the detection window, the voltage data during the charging process is collected in real time, and the polarization voltage is extracted. The calculation expression is as follows:

[0079] V p (t)=V(t)-I(t)·R s

[0080] , where V p (t) is the polarization voltage, V(t) is the total voltage, i.e. the total voltage value of the battery measured at time point t, I(t) is the charging current of the battery at time point t, R s is the internal resistance of the battery;

[0081] By polarization voltage V p(t) Calculate the dynamic change rate of polarization voltage. The calculation expression is as follows:

[0082]

[0083] , where R p (t) is the dynamic change rate of polarization voltage, which indicates the instantaneous rate at which polarization voltage changes with time, and Δt is the time interval;

[0084] In order to capture the microscopic details of voltage changes, high-order partial derivatives are introduced to enhance the sensitivity of dynamic response. The calculation expression is as follows:

[0085]

[0086] , where is the rate of dynamic change of polarization voltage enhancement, α is the sensitivity weight coefficient, which is used to adjust the contribution weight of the second-order derivative to the rate;

[0087] This step calculates the dynamic rate over time, enhancing the ability to perceive abnormal lithium ion migration.

[0088] To further quantify the severity of the dynamic change, an amplitude weighting factor is introduced to reflect the nonlinear fluctuation characteristics of the rate. The calculation expression is as follows:

[0089]

[0090] , where W p (t) is a weighting factor, which is used to describe the nonlinear degree of the polarization voltage rate change, amplify the impact of drastic changes, and weaken the weight of smooth changes on subsequent calculations. β is the amplitude adjustment coefficient, which controls the amplification effect of the rate amplitude on the weighting factor. γ is a nonlinear exponent, which affects the absolute value of the rate. Perform nonlinear amplification;

[0091] By weighting factor W p (t), a higher weight can be given to abnormal changes in subsequent index calculations, further enhancing its sensitivity to the obstruction of lithium ion migration.

[0092] Combined with polarization voltage to enhance the dynamic change rate and weighting factor W p (t), construct the electrochemical polarization rapid change index, and the calculation expression is as follows:

[0093]

[0094] , where EPDVF is the electrochemical polarization rapid change factor, sin(ω·t+φ) is the harmonic modulation function, which describes the periodicity or oscillation characteristics of the polarization voltage, ω is the angular frequency, which is related to the charging dynamic characteristics of the battery. The higher the frequency, the more frequent the polarization voltage changes of the battery system, φ is the initial phase, which describes the initial state of the signal and can be used to align the starting point of the periodic fluctuation to ensure the continuity and accuracy of the calculation, and T is the total duration of the detection window.

[0095] Through the above steps, the rate, nonlinear fluctuation amplitude and periodic characteristics are comprehensively considered, and the severity of the obstruction of lithium ion migration can be accurately quantified.

[0096] The angular frequency ω represents the oscillation rate of the sine wave, which is defined as the number of radians completed per second by the fluctuation, and the unit is radians per second (rad / s). During the battery charging process, the angular frequency reflects the frequency characteristics of the polarization voltage change. A high angular frequency indicates that the voltage changes quickly and frequently, which may be related to high-frequency interference or fast electrochemical reactions in the system. The introduction of angular frequency can help capture and quantify the characteristics of periodic fluctuations in the lithium ion migration process.

[0097] The initial phase φ represents the initial position of the sine wave at time t=0, in radians (rad). It determines the specific position of the starting point of the signal in a cycle. In the battery system, the initial phase can be used to align the initial states of different signals, ensure the synchronous analysis of periodic changes, and help to more accurately describe and compare dynamic behaviors.

[0098] Under the detection window, the larger the value of the electrochemical polarization rapid change factor generated after analyzing the dynamic change rate of the polarization voltage, the more severe the polarization voltage change, reflecting that the negative electrode material's lithium insertion capacity is insufficient and the lithium ion migration rate cannot keep up with the high current charging demand. At this time, some lithium ions cannot be embedded in the negative electrode graphite structure, but are deposited as metallic lithium on the negative electrode surface, forming unstable metallic lithium dendrites, which may cause safety hazards. On the contrary, when the value of the electrochemical polarization rapid change factor is small, it means that the polarization voltage changes steadily, the negative electrode material can effectively insert lithium, and the lithium ion migration is smooth and evenly distributed, indicating that the lithium ion migration during the charging process is normal.

[0099] The asymmetry in the spatial distribution of the electrochemical reaction rate of the negative electrode indicates that the migration rate of lithium ions is insufficient, and some lithium ions cannot be embedded in the graphite structure of the negative electrode, but are deposited as metallic lithium on the surface of the negative electrode, thereby forming unstable metallic lithium dendrites. This is because the asymmetry of the reaction rate of the negative electrode usually stems from the imbalance of local lithium ion supply and demand. Under high current charging conditions, the diffusion rate of lithium ions from the electrolyte to the negative electrode may not be able to meet the lithium insertion requirements of various regions of the negative electrode, resulting in a significant increase in the current density in some regions of the negative electrode, while other regions are relatively low. The reaction in the area with higher current density is active, and the degree of polarization is aggravated, so that lithium ions cannot be smoothly embedded in the graphite lattice, and instead precipitate as metallic lithium on the surface. These metallic lithiums are usually deposited in the form of irregular dendrites. As charging continues, the dendrites continue to grow, further aggravating the concentration of local current density and forming a vicious circle. Ultimately, the metallic lithium dendrites may pierce the diaphragm, causing an internal short circuit, leading to serious safety risks. Therefore, the asymmetric distribution of the reaction rate of the negative electrode is not only a manifestation of the obstruction of lithium ion migration, but also an early signal of lithium deposition and dendrite formation.

[0100] The specific steps of generating the electrode kinetic non-equilibrium factor after analyzing the symmetry of the spatial distribution of the negative electrode electrochemical reaction rate within the detection window are as follows:

[0101] Under the detection window, the local current density of multiple areas on the negative electrode surface is collected. These data reflect the insertion rate of lithium ions at different positions. The negative electrode surface is divided into an N×M grid. Each grid node (i, j) represents a local area. The collected local current density is stored in matrix form as J, J = {J(i, j)}, where J(i, j) represents the local current density of nodes i and j. The data matrix J is the basis for generating the electrode kinetic non-equilibrium index. The degree of asymmetry is quantified by analyzing the spatial gradient and distribution pattern of the local current density data matrix J.

[0102] For the collected J(i,j), calculate its gradient strength in space. The calculation expression is as follows:

[0103]

[0104] , where G(i,j) is the gradient strength, which represents the local electrochemical reaction rate gradient strength at node (i,j), and measures the severity of the spatial variation of the local current density J(i,j);

[0105] Regions with higher gradient intensity mean that the local reaction rate varies dramatically and lithium ion migration may be hindered.

[0106] The global imbalance of the reaction rate distribution on the entire negative electrode surface is calculated using the gradient strength G(i,j) of each node. The calculation expression is as follows:

[0107]

[0108] , where Q is the global imbalance index, which indicates the overall imbalance of the electrochemical reaction rate distribution on the negative electrode surface. It combines the asymmetry of the gradient strength and weight of all local regions. The higher the Q, the more uneven the distribution, reflecting the aggravated imbalance of the negative electrode reaction, indicating that lithium migration is blocked or the risk of local lithium deposition increases. k1 is the gradient smoothing parameter, which is used to control the weight influence of the gradient strength G(i,j) on the global imbalance Q. ω(i,j) is the weight factor, which is the weight factor at the grid node, reflecting the contribution of the region to the overall distribution imbalance.

[0109] The exponential function is used to amplify the contribution of regions with larger gradient intensity, making the imbalance index more sensitive to local anomalies.

[0110] Based on the global imbalance index Q, the electrode kinetic imbalance factor is defined and the calculation is expressed as follows:

[0111]

[0112] , where EKSAF is the electrode kinetic non-equilibrium factor, A is the total surface area of ​​the negative electrode, Ω is the spatial area of ​​the negative electrode, the two-dimensional spatial distribution area of ​​the negative electrode, usually corresponding to the gridded analysis range, covering all local current densities J(i,j), and e is the natural base number, is the adjustment coefficient, used to control G max The contribution weight of the electrode kinetic non-equilibrium factor EKSAF, θ is the sensitivity adjustment parameter, and controls the exponential function The steepness of the gradient, i.e., the sensitivity to abnormal points of gradient intensity, G max is the maximum value of the gradient intensity, the maximum value of the gradient intensity G(i,j) of all nodes on the negative electrode surface, reflecting the extreme change point in the local reaction rate distribution, and γ is the threshold adjustment parameter used to compare with the actual maximum gradient G max For comparison. If G max >γ, indicating that the gradient strength is abnormally significant, which will significantly amplify the value of the electrode kinetic non-equilibrium factor EKSAF.

[0113] Under the detection window, the larger the value of the electrode kinetic non-equilibrium factor generated after analyzing the degree of symmetry of the spatial distribution of the negative electrode electrochemical reaction rate, the less the lithium ion migration rate is. Some lithium ions cannot be embedded in the negative electrode graphite structure, but are deposited as metallic lithium on the negative electrode surface, thereby forming unstable metallic lithium dendrites. The charging reaction asymmetry index reflects the dynamic changes in the migration and embedding process of lithium ions by quantifying the degree of non-uniformity of the spatial distribution of the negative electrode electrochemical reaction rate. When the index performance value is large, it means that the lithium insertion rate in some areas of the negative electrode surface is significantly lower than that in other areas, resulting in excessive accumulation of local lithium ions, which are eventually reduced to metallic lithium in these high-risk areas. This asymmetric reaction exacerbates the uneven current density on the electrode surface, making it easier for the metal lithium deposition process to induce dendrite formation. Conversely, when the charging reaction asymmetry index performance value is small, it indicates that the negative electrode surface reaction rate is evenly distributed, lithium ions can migrate normally and efficiently embed into the negative electrode graphite structure, and the charging process is in a safe state.

[0114] The extracted features are input into a pre-learned deep learning model, and the risk of lithium ion migration obstruction during battery charging is intelligently evaluated through the deep learning model;

[0115] The extracted electrochemical polarization rapid change factor characteristics and electrode kinetics non-equilibrium factor characteristics are input into a pre-learned deep learning model, and a lithium ion deposition index is generated based on the deep learning model. The lithium ion deposition index is used to intelligently evaluate the risk of lithium ion migration obstruction during battery charging.

[0116] A pre-learned deep learning model refers to an artificial intelligence model that is built and trained in advance to assess the risk of lithium ion migration obstruction during the battery charging process. This model uses a large amount of historical data and experimental annotated data for learning during the training phase to grasp the complex relationship between electrochemical behavior and lithium ion migration patterns. The training data usually includes multimodal sensor data and extracted key features (such as electrochemical polarization rapid change factors and electrode kinetic non-equilibrium factors) and corresponding lithium ion migration state labels (such as normal migration or obstructed migration). The model repeatedly optimizes parameters and learns nonlinear mapping relationships to accurately predict the lithium ion deposition index and quantify the risk of migration obstruction during the charging process. This model has been verified and tuned to efficiently process real-time input data in practical applications and maintain high robustness and prediction accuracy in complex and changing battery operating environments.

[0117] The pre-learned deep learning model has powerful feature extraction and pattern recognition capabilities, and can capture early signs of lithium deposition or migration obstruction trends from input features. Specifically, the model may extract spatial features from the input data, such as local non-balanced distribution, based on a convolutional neural network (CNN); or capture time series features, such as dynamic change rates and historical behavior patterns, based on a long short-term memory network (LSTM). The model learning process includes a lot of training and validation, such as evaluating model performance through cross-validation, and using hyperparameter tuning (such as learning rate, number of layers) to further improve predictive capabilities. Once the model is pre-trained and verified by the test set, its parameters will be frozen and become a pre-learned model. After deployment, the model can infer real-time input data, provide intelligent evaluation results through the lithium ion deposition index, and provide a scientific basis for dynamic charging strategy adjustments, thereby optimizing the safety and efficiency of battery charging.

[0118] The deep learning model is not limited here, and any deep learning model that can perform comprehensive analysis of the electrochemical polarization rapid change factor EPDVF and the electrode kinetic non-equilibrium factor EKSAF to generate the lithium ion deposition index LIDSI is acceptable. In order to realize the technical solution of the present invention, the present invention provides a specific implementation method;

[0119] The formula for generating the lithium ion deposition index LIDSI is as follows:

[0120]

[0121] , where u1 and u2 are the preset proportional coefficients of the electrochemical polarization rapid change factor EPDVF and the electrode kinetic non-equilibrium factor EKSAF, respectively, and u1 and u2 are both greater than 0.

[0122] It can be seen from the lithium ion deposition index that, under the detection window, the larger the electrochemical polarization rapid change factor value generated after analyzing the dynamic change rate of the polarization voltage, and the larger the electrode kinetic non-equilibrium factor value generated after analyzing the degree of symmetry of the spatial distribution of the negative electrode electrochemical reaction rate, the larger the lithium ion deposition index value generated after comprehensive analysis of the extracted key features under the detection window, indicating that the lithium ion migration rate is insufficient, and some lithium ions cannot be embedded in the negative electrode graphite structure, but are deposited as metallic lithium on the negative electrode surface, thereby forming unstable metallic lithium dendrites. On the contrary, it indicates that lithium ions can migrate normally and efficiently embed into the negative electrode graphite structure, and the charging process is in a safe state.

[0123] According to the evaluation results of the deep learning model, the lithium ion migration state is divided into normal lithium ion migration and blocked lithium ion migration;

[0124] The generated lithium ion deposition index is compared with the preset lithium ion deposition index reference threshold value to divide the lithium ion migration state. The division steps are as follows:

[0125] If the lithium ion deposition index is greater than or equal to a preset lithium ion deposition index reference threshold, the lithium ion migration state is classified as normal lithium ion migration;

[0126] If the lithium ion deposition index is less than a preset lithium ion deposition index reference threshold, the lithium ion migration state is classified as lithium ion hindered migration.

[0127] Hindered lithium ion migration means that during the battery charging process, the rate at which lithium ions migrate from the positive electrode to the negative electrode cannot meet the charging current requirements, resulting in some lithium ions being unable to embed into the negative electrode material, and being deposited as metallic lithium on the negative electrode surface. This state usually occurs when the charging current is too large, the negative electrode's lithium embedding capacity is insufficient (such as high SOC), the temperature is low, or the electrode reaction kinetics are limited. The direct consequence of hindered lithium ion migration is an abnormal increase in the polarization voltage on the negative electrode surface, uneven distribution of lithium ions, and the occurrence of side reactions (such as electrolyte decomposition) and lithium deposition.

[0128] When lithium ion migration is blocked, metal lithium deposition may further develop into dendrite growth, eventually leading to diaphragm puncture and internal short circuit, significantly increasing the risk of thermal runaway. This state will also accelerate the abnormal formation of solid electrolyte interface film (SEI), leading to increased internal resistance and battery capacity decay. Obstructed lithium ion migration is a state that needs to be monitored and controlled during battery charging. Once migration obstruction is detected, timely measures should be taken (such as reducing the charging current or switching the charging mode) to reduce the risk of lithium deposition and dendrite formation, thereby protecting the battery safety and extending its service life.

[0129] Normal lithium ion migration means that during the battery charging process, lithium ions can smoothly migrate from the positive electrode to the negative electrode and evenly embed into the graphite structure or other lithium-intercalated materials of the negative electrode to complete the electrochemical energy storage reaction. In this state, the kinetic conditions of lithium ion migration (such as diffusion rate, electrode reaction rate) match the charging current requirements, the lithium ion concentration in the electrolyte is evenly distributed, the negative electrode polarization is low, and there is no excessive lithium metal deposition or side reactions on the surface. At this time, the battery charging efficiency is high, the electrochemical reaction is stable, the increase in heat and internal resistance is small, and it will not pose a threat to battery performance or safety.

[0130] For normal lithium ion migration, a constant current charging strategy is maintained to quickly replenish the battery power through high current to ensure charging efficiency and time optimization;

[0131] For the normal migration state of lithium ions, the purpose of maintaining a constant current charging strategy and quickly replenishing the battery power through high current is to maximize the charging efficiency and shorten the charging time during the charging process, while making full use of the lithium insertion capacity of the negative electrode material to ensure that the lithium ions can be stably and evenly embedded in the graphite structure. In this state, the electrochemical reaction rate of the negative electrode can match the high current charging demand, and will not cause the risk of lithium deposition or dendrite formation, thereby achieving a higher energy replenishment speed and better charging performance while ensuring battery safety, meeting the needs of fast charging of batteries.

[0132] For lithium ion migration obstruction, based on the deep learning model evaluation results, the charging current is dynamically reduced based on the constant current charging high current to mitigate the risk of lithium ion deposition and dendrite formation;

[0133] For lithium ion blocked migration, based on the deep learning model evaluation results, the specific steps to dynamically reduce the charging current based on constant current charging high current to mitigate the risk of lithium ion deposition and dendrite formation are as follows:

[0134] After the risk of lithium ion blocked migration is detected, the dynamic adjustment factor of the charging current is calculated based on the deviation between the lithium ion deposition index LIDSI value and the reference threshold of the lithium ion deposition index to determine the magnitude of the reduction in the charging current. The calculation expression is as follows:

[0135]

[0136] , where H adj is the dynamic adjustment factor of the charging current, with a value range of [0,1], where 0 means completely stopping charging, 1 means keeping the high current of constant current charging unchanged, k is the basic adjustment coefficient, Y1 is the temperature correction coefficient, which reflects the effect of temperature rise on the adjustment factor during charging, Y2 is the internal resistance correction coefficient, which is used to correct the effect of internal resistance change caused by lithium deposition on the adjustment factor, ΔT is the current temperature change, LIDSI ref is the reference threshold of lithium ion deposition index, T max is the maximum safe temperature allowed by the battery, R init is the internal resistance correction coefficient, which indicates the weight of the influence of the internal resistance change on the adjustment factor;

[0137] The basic adjustment factor k is a key parameter in the dynamic adjustment formula of charging current, which is used to control the sensitivity of the adjustment factor to the risk of lithium deposition. It determines the value of the adjustment factor when the lithium ion deposition index LIDSI exceeds the reference threshold LIDSI. refThe magnitude of the reduction in charging current when the battery is charged. Specifically, the larger the value of the basic adjustment coefficient, the more sensitive the system is to the risk of lithium deposition, and the faster the charging current will be reduced to slow down the trend of lithium deposition and dendrite formation; on the contrary, if the coefficient is smaller, the adjustment range will be more moderate, which is suitable for situations with lower risks. Its role is to balance battery safety and charging efficiency by flexibly adjusting sensitivity, so that the system can adapt to different charging environments and lithium deposition risk levels, thereby achieving precise dynamic control of charging current.

[0138] When the lithium ion deposition index LIDSI exceeds the reference threshold by a large margin, the charging current dynamic adjustment factor, H adj This indicates that the charging current needs to be significantly reduced to mitigate the risk of lithium deposition.

[0139] According to the calculated dynamic adjustment factor H adj , dynamically adjust the high current of constant current charging to determine the new charging current. The calculation expression is as follows:

[0140]

[0141] , where I new is the dynamically adjusted charging current, I max It is the maximum allowable charging current, usually 1C or 2C of the rated capacity of the battery (such as 1A or 2A). base It is the basic high current of constant current charging, generally 0.5C to 1C. G is the voltage response decreasing coefficient, which controls the proportional coefficient of the current decreasing amplitude when the voltage rises rapidly. dV / dt is the voltage change rate, the change of voltage per unit time, which reflects the dynamic trend of voltage response during charging. SOC current is the current state of charge;

[0142] Through this adjustment mechanism, H adj Gradually tending to 0.5 (or lower), significantly reducing the charging current, ensuring that the lithium ion migration rate matches the current demand, and avoiding lithium deposition and dendrite formation.

[0143] During the process of adjusting the charging current, the lithium ion deposition index LIDSI value is continuously monitored and the evaluation results are updated in real time. If the subsequent detection shows that the lithium ion deposition index LIDSI drops below the reference threshold LIDSI ref (i.e. LIDSI>LIDSI ref ), the charging current is gradually restored, and the expression is as follows:

[0144]

[0145] , where I′ newis the charging current after dynamic recovery, μ is the recovery rate coefficient, which determines the speed and sensitivity of the charging current to recover to the basic high current. is the hyperbolic tangent function term, which is used to smooth LIDSI and LIDSI ref The difference reflects the degree of reduction in lithium deposition risk, T current is the current temperature, δ is the temperature driving coefficient, which reflects the influence of the current battery temperature on the recovery of the charging current.

[0146] This closed-loop regulation process enables the charging system to adaptively adjust to dynamic risk changes, reducing the risk of lithium deposition and dendrite formation while maximizing charging efficiency when conditions permit.

[0147] For the lithium ion blocked migration stage, the strategy of dynamically reducing the high current of constant current charging based on the evaluation results of the deep learning model plays a key role. Its core purpose is to prevent lithium deposition and dendrite formation through precise current regulation after identifying the risk of lithium ion migration, so as to ensure the safety of the charging process and the stability of battery performance. When lithium ion blocked migration occurs, the lithium ion insertion rate of the negative electrode is lower than the current demand, and the lithium ions that fail to be embedded will be deposited on the surface of the negative electrode to form metal lithium deposition, which will induce unstable lithium dendrites. The growth of these dendrites will further pierce the diaphragm, causing internal short circuit, thermal runaway and even battery fire. By dynamically reducing the charging current, the pressure on the negative electrode can be relieved, and the lithium ion migration rate can be re-matched with the current demand, thereby avoiding the risk of lithium deposition and dendrite formation. In addition, this step makes full use of the real-time evaluation results of the deep learning model on the risk of lithium deposition, quantifies the risk level according to the lithium ion deposition index LIDSI, and flexibly adjusts the charging current to achieve precise and timely intervention. This regulation not only protects the safety of the battery, but also extends the battery life and improves the charging efficiency by dynamically adapting to the lithium ion migration state. After the risk is eliminated, the system can gradually resume high current charging to ensure the optimization of charging time. This intelligent dynamic adjustment strategy is an important breakthrough in battery charging management technology, balancing charging efficiency and safety, and is a key step in efficient and intelligent charging.

[0148] Through the comprehensive acquisition of multimodal sensing data and the introduction of deep learning models, the present invention can accurately identify the risk of obstructed lithium ion migration during battery charging. Based on the innovative feature extraction method of electrochemical polarization rapid change factor and electrode kinetic non-equilibrium factor, the scheme can capture the dynamic change rate of polarization voltage and the degree of imbalance of negative electrode reaction rate, thereby quantifying the possibility of obstructed lithium migration and generating a lithium ion deposition index. After comparing the lithium ion deposition index with the reference threshold, the intelligent classification of normal and obstructed lithium ion migration is achieved, avoiding the uncertainty of traditional empirical threshold judgment. This intelligent evaluation method can not only provide early warning of the risk of lithium deposition and dendrite formation, but also provide a scientific basis for subsequent charging strategies, greatly improving the safety and reliability of the charging process.

[0149] The present invention dynamically adjusts the charging process through a differentiated strategy based on the assessed lithium ion migration state. Under the normal lithium ion migration state, the constant current charging strategy is maintained to quickly replenish the battery power with high current to optimize the charging efficiency and time; under the lithium ion blocked migration state, the charging current is dynamically reduced to significantly reduce the risk of lithium deposition and dendrite formation, avoiding safety hazards caused by internal short circuits and thermal runaway. This real-time adjustment charging strategy based on the evaluation results of the deep learning model not only protects the internal structure of the battery and extends the battery life, but also takes into account the maximization of charging efficiency. Compared with the traditional fixed charging method, this method achieves a dynamic balance between the efficiency and safety of the charging process, providing a strong guarantee for the long-term and stable use of the battery.

[0150] The present invention provides Figure 2 A battery charging control method shown includes a multimodal data acquisition module, a data preprocessing and feature extraction module, a deep learning evaluation module, a migration state classification module, a constant current charging optimization module and a dynamic charging adjustment module:

[0151] The multimodal data acquisition module firstly obtains the multimodal sensor data from the battery in an all-round way;

[0152] The data preprocessing and feature extraction module preprocesses the acquired raw data and extracts key features that are helpful for identifying lithium ion migration obstruction from the preprocessed data within the detection window;

[0153] The deep learning evaluation module inputs the extracted features into a pre-learned deep learning model, and uses the deep learning model to intelligently evaluate the risk of lithium ion migration obstruction during battery charging;

[0154] The migration state classification module divides the lithium ion migration state into normal lithium ion migration and blocked lithium ion migration according to the evaluation results of the deep learning model;

[0155] Constant current charging optimization module, for normal lithium ion migration, maintains constant current charging strategy, quickly replenishes battery power through high current, and ensures charging efficiency and time optimization;

[0156] The dynamic charging adjustment module dynamically reduces the charging current based on the evaluation results of the deep learning model and the high current of constant current charging to mitigate the risk of lithium ion deposition and dendrite formation when lithium ion migration is blocked.

[0157] A battery charging control method provided in an embodiment of the present invention is implemented by the above-mentioned battery charging control system. The specific method and process of a battery charging control system are detailed in the embodiment of the above-mentioned battery charging control method, which will not be repeated here.

[0158] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A battery charging control method, characterized in that: The following steps are involved: First, multimodal sensor data is comprehensively acquired from the battery; Preprocessing the acquired raw data, and extracting key features that are helpful for identifying lithium ion migration obstruction from the preprocessed data within the detection window; The extracted features are input into a pre-learned deep learning model, and the risk of lithium ion migration obstruction during battery charging is intelligently evaluated through the deep learning model; According to the evaluation results of the deep learning model, the lithium ion migration state is divided into normal lithium ion migration and blocked lithium ion migration; For normal lithium ion migration, a constant current charging strategy is maintained to quickly replenish the battery power through high current to ensure charging efficiency and time optimization; For the hindered migration of lithium ions, based on the evaluation results of the deep learning model, the charging current is dynamically reduced based on the high current of constant current charging to mitigate the risk of lithium ion deposition and dendrite formation.

2. A battery charging control method according to claim 1, characterized in that: Extract the dynamic change rate of polarization voltage and the symmetry degree of negative electrode electrochemical reaction rate in spatial distribution during battery charging from the preprocessed data. After acquisition, analyze the dynamic change rate of polarization voltage and the symmetry degree of negative electrode electrochemical reaction rate in spatial distribution under the detection window to generate electrochemical polarization rapid change factor and electrode kinetic non-equilibrium factor respectively. The electrochemical polarization rapid change factor reflects the aggravation degree of polarization phenomenon in lithium ion migration process by quantifying the dynamic change rate of polarization voltage with time during battery charging process, and indirectly reveals the risk of lithium ion migration being blocked. The electrode kinetic non-equilibrium factor reflects the degree of uneven distribution of lithium ions during the negative electrode embedding process by quantifying the degree of asymmetry in the spatial distribution of the electrochemical reaction rate on the negative electrode surface, thereby evaluating the local deposition trend caused by the obstruction of lithium ion migration.

3. A battery charging control method according to claim 2, characterized in that: The specific steps of generating the electrochemical polarization rate change factor after analyzing the dynamic change rate of polarization voltage in the detection window are as follows: In the detection window, the voltage data during the charging process is collected in real time, and the polarization voltage is extracted. The calculation expression is as follows: V p (t)=V(t)-I(t)·R s Where V p (t) is the polarization voltage, V(t) is the total voltage, i.e. the total voltage value of the battery measured at time point t, I(t) is the charging current of the battery at time point t, R s is the internal resistance of the battery; By polarization voltage V p (t) Calculate the dynamic change rate of polarization voltage. The calculation expression is as follows: In the formula, R p (t) is the dynamic change rate of polarization voltage, Δt is the time interval; In order to capture the microscopic details of voltage changes, high-order partial derivatives are introduced to enhance the sensitivity of dynamic response. The calculation expression is as follows: In the formula, is the rate of dynamic change of polarization voltage enhancement, α is the sensitivity weight coefficient, which is used to adjust the contribution weight of the second-order derivative to the rate; To further quantify the severity of the dynamic change, an amplitude weighting factor is introduced to reflect the nonlinear fluctuation characteristics of the rate. The calculation expression is as follows: Where W p (t) is the weighting factor, β is the amplitude adjustment coefficient, and γ is the nonlinear index; Combined with polarization voltage to enhance the dynamic change rate and weighting factor W p (t), construct the electrochemical polarization rapid change index, and the calculation expression is as follows: Where EPDVF is the electrochemical polarization rapid change factor, sin(ω·t+φ) is the harmonic modulation function, ω is the angular frequency, φ is the initial phase, and T is the total duration of the detection window.

4. A battery charging control method according to claim 2, characterized in that: The specific steps of generating the electrode kinetic non-equilibrium factor after analyzing the symmetry of the spatial distribution of the electrochemical reaction rate of the negative electrode within the detection window are as follows: Under the detection window, the local current density of multiple areas on the negative electrode surface is collected. The negative electrode surface is divided into an N×M grid. Each grid node (i, j) represents a local area. The collected local current density is stored in a matrix form as J, J = {J(i, j)}, where J(i, j) represents the local current density of nodes i and j. The spatial gradient and distribution pattern of the local current density data matrix J are analyzed to quantify the degree of asymmetry. For the collected J(i,j), calculate its gradient strength in space. The calculation expression is as follows: Where G(i,j) is the gradient strength; The global imbalance of the reaction rate distribution on the entire negative electrode surface is calculated using the gradient strength G(i,j) of each node. The calculation expression is as follows: Where Q is the global imbalance indicator, k1 is the gradient smoothing parameter, and ω(i,j) is the weight factor; Based on the global imbalance index Q, the electrode kinetic imbalance factor is defined and the calculation is expressed as follows: Where EKSAF is the electrode kinetic non-equilibrium factor, A is the total surface area of ​​the negative electrode, Ω is the spatial area of ​​the negative electrode, and e is the natural base. is the adjustment coefficient, θ is the sensitivity adjustment parameter, G max is the maximum value of the gradient intensity, and γ is the threshold adjustment parameter.

5. A battery charging control method according to claim 2, characterized in that: The extracted electrochemical polarization rapid change factor characteristics and electrode kinetics non-equilibrium factor characteristics are input into a pre-learned deep learning model, and a lithium ion deposition index is generated based on the deep learning model. The lithium ion deposition index is used to intelligently evaluate the risk of lithium ion migration obstruction during battery charging.

6. A battery charging control method according to claim 5, characterized in that: The generated lithium ion deposition index is compared with the preset lithium ion deposition index reference threshold value to divide the lithium ion migration state. The division steps are as follows: If the lithium ion deposition index is greater than or equal to a preset lithium ion deposition index reference threshold, the lithium ion migration state is classified as normal lithium ion migration; If the lithium ion deposition index is less than a preset lithium ion deposition index reference threshold, the lithium ion migration state is classified as lithium ion hindered migration.

7. A battery charging control method according to claim 6, characterized in that: For lithium ion blocked migration, based on the deep learning model evaluation results, the specific steps to dynamically reduce the charging current based on constant current charging high current to mitigate the risk of lithium ion deposition and dendrite formation are as follows: After the risk of lithium ion blocked migration is detected, the dynamic adjustment factor of the charging current is calculated based on the deviation between the lithium ion deposition index LIDSI value and the reference threshold of the lithium ion deposition index to determine the magnitude of the reduction in the charging current. The calculation expression is as follows: In the formula, H adj is the dynamic adjustment factor of the charging current, k is the basic adjustment coefficient, Y1 is the temperature correction coefficient, Y2 is the internal resistance correction coefficient, ΔT is the current temperature change, LIDSI ref is the reference threshold of lithium ion deposition index, T max is the maximum safe temperature allowed by the battery, R init is the internal resistance correction factor; According to the calculated dynamic adjustment factor H adj , dynamically adjust the high current of constant current charging to determine the new charging current. The calculation expression is as follows: In the formula, I new is the dynamically adjusted charging current, I max is the maximum allowable value of charging current, I base is the basic high current of constant current charging, G is the voltage response decreasing coefficient, which controls the proportional coefficient of the current decreasing amplitude when the voltage rises rapidly, DV / dt is the voltage change rate, SOC current is the current state of charge; During the process of adjusting the charging current, the lithium ion deposition index LIDSI value is continuously monitored and the evaluation results are updated in real time. If the subsequent detection shows that the lithium ion deposition index LIDSI drops below the reference threshold LIDSI ref , then the charging current is gradually restored, and the expression is as follows: In the formula, I′ new is the charging current after dynamic recovery, μ is the recovery rate coefficient, is the hyperbolic tangent function term, T current is the current temperature, δ is the temperature driving coefficient, which reflects the influence of the current battery temperature on the recovery of the charging current.

8. A battery charging control system, used to implement a battery charging control method as described in any one of claims 1 to 7, characterized in that: It includes multimodal data acquisition module, data preprocessing and feature extraction module, deep learning evaluation module, migration state classification module, constant current charging optimization module and dynamic charging adjustment module: The multimodal data acquisition module firstly obtains the multimodal sensor data from the battery in an all-round way; The data preprocessing and feature extraction module preprocesses the acquired raw data and extracts key features that are helpful for identifying lithium ion migration obstruction from the preprocessed data within the detection window; The deep learning evaluation module inputs the extracted features into a pre-learned deep learning model, and uses the deep learning model to intelligently evaluate the risk of lithium ion migration obstruction during battery charging; The migration state classification module divides the lithium ion migration state into normal lithium ion migration and blocked lithium ion migration according to the evaluation results of the deep learning model; Constant current charging optimization module, for normal lithium ion migration, maintains constant current charging strategy, quickly replenishes battery power through high current, and ensures charging efficiency and time optimization; The dynamic charging adjustment module dynamically reduces the charging current based on the evaluation results of the deep learning model and the high current of constant current charging to mitigate the risk of lithium ion deposition and dendrite formation when lithium ion migration is blocked.

Images