Lithium battery overcharge protection control method and system

Abstract

The invention relates to the technical field of lithium batteries, and provides a lithium battery overcharge protection control method and system, and the method comprises the steps: obtaining a voltage value, a voltage change rate, a temperature value and a temperature change rate of each battery cell in a lithium battery, carrying out the interval consistency correction, constructing a battery cell state data set and a change trend feature set, carrying out the state trend analysis, and obtaining a state data set; obtaining the current risk grade and the composite risk factor value of each battery cell, matching the composite risk factor value with the current risk grade to obtain a charging adjustment instruction set, and adjusting a charging control module of the lithium battery according to the charging adjustment instruction set. Through trend analysis, temperature rise change in a future charging period is accurately simulated, and critical proximity and abnormal probability are evaluated in combination with a temperature rise safety interval, so that early warning is realized, real-time control of charging current limiting and voltage amplitude limiting is realized, and the problem of protection delay or excessive protection in a complex battery environment in the prior art is solved.

Description

Technical Field

[0001] The present application relates to the technical field of lithium batteries, and in particular to a lithium battery overcharge protection control method and system. Background Art

[0002] With the rapid development of new energy vehicles, portable electronic devices, and energy storage systems, lithium-ion batteries are widely used due to their high energy density, long life, and excellent cycle performance. To meet the dual demands of charging safety and efficiency in different application scenarios, the role of the lithium battery management system (BMS) in battery operation is becoming increasingly critical. In particular, the accurate identification and prevention of overcharge risks during the charging process has become one of the core technologies for improving battery safety.

[0003] Among relevant technical approaches, lithium battery overcharge protection methods primarily include setting a maximum charging voltage threshold, current cutoff protection, temperature protection strategies, and SOC (State of Charge) predictive control. Most systems monitor the voltage, temperature, and SOC of individual cells in real time. When any parameter exceeds a preset threshold, current limiting or charging interruption mechanisms are triggered to prevent overcharging of the cells from causing thermal runaway, swelling, or performance degradation. These methods are highly effective in achieving basic overcharge protection and can ensure the basic operational safety of the battery system.

[0004] While the aforementioned technical solutions can initially achieve overcharge risk control by setting static safety thresholds and combining them with SOC estimation, they still have certain limitations in practical applications. For example, when batteries are exposed to complex temperature gradients, large variations in charging rates, or inconsistent aging conditions, traditional protection strategies based on single parameter thresholds are prone to misjudgments or delayed responses, making it difficult to accurately identify subtle differences and trending anomalies between cells, leading to protection hysteresis or overprotection. Summary of the Invention

[0005] In order to improve the existing technology in complex battery environment, there are problems of protection hysteresis or over-protection, the present application provides a lithium battery overcharge protection control method and system.

[0006] The present invention provides a lithium battery overcharge protection control method, comprising: obtaining a voltage value, a voltage change rate, a temperature value, and a temperature change rate of each battery cell in a lithium battery, performing interval consistency correction on the voltage value, the temperature value, the voltage change rate, and the temperature change rate to construct a battery cell state data set and a change trend feature set; performing state trend analysis on the battery cell state data set and the change trend feature set to obtain a current risk level and a short-term state prediction value of each battery cell; inputting the short-term state prediction value and the battery cell state data set into a preset temperature rise prediction model to obtain a critical proximity, an overcharge risk factor, and an abnormal trigger probability of each battery cell, performing dual-threshold judgment on the abnormal trigger probability and the critical proximity to generate a temperature rise abnormality mark set; performing weighted fusion on the temperature rise abnormality mark set and the overcharge risk factor to obtain a composite risk factor value, matching the composite risk factor value with the current risk level to obtain a charging adjustment instruction set, and adjusting the charging control module of the lithium battery according to the charging adjustment instruction set.

[0007] As a preferred solution, the step of obtaining the voltage value, voltage change rate, temperature value and temperature change rate of each battery cell in the lithium battery, and performing interval consistency correction on the voltage value, temperature value, voltage change rate and temperature change rate to construct a battery cell state data set and a change trend feature set includes: obtaining voltage sampling data and temperature sampling data of each battery cell in the lithium battery, performing noise filtering and normalization on the voltage sampling data and the temperature sampling data to obtain a voltage standardized sequence and a temperature standardized sequence; obtaining the voltage change rate and temperature change rate of each battery cell, performing interval consistency correction on the voltage change rate, the temperature change rate, the voltage standardized sequence and the temperature standardized sequence to obtain a state vector set, comparing the state vector set with preset operating condition characteristic parameters, eliminating out-of-limit abnormal samples, and generating a valid battery cell data window; constructing a fused state vector based on the voltage standardized sequence and temperature standardized sequence in the valid battery cell data window, and combining the fused state vector with the voltage change rate and temperature change rate in the valid battery cell data window to construct a battery cell state data set and a change trend feature set.

[0008] As a preferred solution, the step of performing state trend analysis on the cell state data set and the change trend feature set to obtain the current risk level and short-term state prediction value of each cell includes: segmenting the cell state data set and the change trend feature set into time windows to obtain a number of time series state segments, calculating the joint trend gradient value of the voltage change rate and the temperature change rate in all the time series state segments, and analyzing the state gradient factor and mutation coefficient of each cell based on the joint trend gradient value; predicting the short-term trend of the corresponding cell through the state gradient factor and the mutation coefficient to obtain a short-term voltage prediction value and a short-term temperature prediction value, and constructing a short-term state prediction value using the short-term voltage prediction value and the short-term temperature prediction value; performing interval cross-comparison on the short-term voltage prediction value and the short-term temperature prediction value to obtain a prediction interval risk deviation and a degree of overcharge proximity; fusing the state gradient factor, the prediction interval risk deviation and the degree of overcharge proximity to generate a primary state risk factor and a disturbance stability coefficient; calculating a risk weight score based on the primary state risk factor and the disturbance stability coefficient, and generating a current risk level based on the risk weight score.

[0009] As a preferred solution, the steps of segmenting the cell state data set and the change trend feature set into time windows to obtain a number of time series state segments, calculating the joint trend gradient value of the voltage change rate and the temperature change rate in all the time series state segments, and analyzing the state gradient factor and mutation coefficient of each cell based on the joint trend gradient value include: dividing the cell state data set and the change trend feature set into a plurality of time series state segments according to a preset time window, smoothing and filtering the voltage data and temperature data in each of the time series state segments to obtain a smoothed voltage sequence and a smoothed temperature sequence; calculating the change rate curves of the smoothed voltage sequence and the smoothed temperature sequence, extracting the voltage change rate and temperature change rate of each segment according to the change rate curve, jointly analyzing the voltage change rate and the temperature change rate of each time series state segment to obtain a joint trend gradient value; and analyzing the change trend of each cell using the joint trend gradient value to obtain a state gradient factor and a mutation coefficient.

[0010] As a preferred solution, the short-term state prediction value and the battery cell state data set are input into a preset temperature rise prediction model to obtain the critical proximity, overcharge risk factor and abnormal trigger probability of each battery cell, and a dual-threshold judgment is performed on the abnormal trigger probability and the critical proximity to generate a temperature rise abnormal mark set. The steps include: inputting the short-term state prediction value and the battery cell state data set into a preset temperature rise prediction model to obtain a temperature rise prediction curve and a dynamic temperature rise increase value for each battery cell in a future charging cycle, comparing a preset temperature rise safety interval with the temperature rise prediction curve and the dynamic temperature rise increase value, respectively, to obtain a temperature rise out-of-bounds amplitude and a critical proximity; calculating a trend correction factor using the dynamic temperature rise increase value and the current risk level, calculating an overcharge risk weight based on the trend correction factor, fusing the temperature rise out-of-bounds amplitude and the overcharge risk weight to obtain an overcharge risk factor and an abnormal trigger probability; performing a dual-threshold judgment based on the abnormal trigger probability and the critical proximity to generate a temperature rise abnormal mark set.

[0011] As a preferred solution, the step of calculating a trend correction factor using the dynamic temperature rise increase value and the current risk level, calculating an overcharge risk weight based on the trend correction factor, and fusing the temperature rise out-of-bounds amplitude with the overcharge risk weight to obtain an overcharge risk factor and an abnormal trigger probability includes: calculating a trend correction factor based on the dynamic temperature rise increase value and the current risk level, weightedly fusing the trend correction factor with the current risk factor to obtain an overcharge risk weight; wherein the current risk factor refers to a comprehensive judgment result of the current temperature, charging current and voltage of the battery cell; calculating an overcharge risk factor using the overcharge risk weight and the temperature rise out-of-bounds amplitude in combination with the temperature change rate of the battery cell; fusing the overcharge risk factor with the temperature rise out-of-bounds amplitude, and calculating the abnormal trigger probability based on the fusion result.

[0012] As a preferred solution, the temperature rise abnormality mark set and the overcharge risk factor are weightedly fused to obtain a composite risk factor value, the composite risk factor value is matched with the current risk level to obtain a charging adjustment instruction set, and the charging control module of the lithium battery is adjusted according to the charging adjustment instruction set. The steps include: fusing the temperature rise abnormality mark set and the overcharge risk factor according to a preset weighting function to obtain a composite risk factor value for each battery cell, matching the composite risk factor value with the current risk level to obtain a risk response priority; selecting the corresponding current adjustment coefficient in a preset charging control library based on the risk response priority, combining the composite risk factor value with the current adjustment coefficient, and calculating a dynamic current limiting value and a voltage limiting parameter; generating a charging adjustment instruction set according to the dynamic current limiting value and the voltage limiting parameter, and adjusting the charging control module of the lithium battery using the charging adjustment instruction set.

[0013] The present application also provides a lithium battery overcharge protection control system, comprising: an acquisition module, configured to acquire the voltage value, voltage change rate, temperature value, and temperature change rate of each battery cell in the lithium battery, and perform interval consistency correction on the voltage value, temperature value, voltage change rate, and temperature change rate to construct a battery cell state data set and a change trend feature set; an analysis module, configured to perform state trend analysis on the battery cell state data set and the change trend feature set to obtain a current risk level and a short-term state prediction value of each battery cell; a determination module, configured to input the short-term state prediction value and the battery cell state data set into a preset temperature rise prediction model to obtain a critical proximity, an overcharge risk factor, and an abnormal trigger probability of each battery cell, perform a dual-threshold determination on the abnormal trigger probability and the critical proximity, and generate a temperature rise abnormality mark set; and a matching module, configured to perform weighted fusion on the temperature rise abnormality mark set and the overcharge risk factor to obtain a composite risk factor value, match the composite risk factor value with the current risk level, obtain a charging adjustment instruction set, and adjust the charging control module of the lithium battery according to the charging adjustment instruction set.

[0014] Compared with the existing technology, the present application has the following beneficial effects: rapid and accurate protection. By introducing the standardization and consistency correction mechanism of voltage values, temperature values ​​and their change rates, data alignment between different battery cells is achieved, thereby improving the accuracy of state analysis; in the state trend analysis process, by introducing multi-dimensional features such as joint trend gradient value, state gradient factor and mutation coefficient, and combining sliding window and prediction algorithm, the state evolution trend and potential risks of the battery cell are effectively identified; in the temperature rise prediction process, short-term state prediction value and historical state data are input into the multi-layer neural network to accurately simulate the temperature rise changes in the future charging cycle, and the critical proximity and abnormal probability are evaluated in combination with the temperature rise safety interval, so as to achieve early warning. In the final control decision, by integrating the temperature rise abnormality mark set and the overcharge risk factor, a composite risk factor model with dynamic response capability is formed to achieve real-time control of charging current limiting and voltage limiting, thereby improving the existing technology in a complex battery environment. The problem of protection hysteresis or over-protection exists. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0016] The structures, proportions, sizes, etc. depicted in the drawings of this specification are only used to match the contents disclosed in the specification so as to facilitate understanding and reading by persons familiar with this technology. They are not intended to limit the conditions under which the present invention can be implemented and therefore have no substantive technical significance. Any structural modifications, changes in proportional relationships, or adjustments in size should still fall within the scope of the technical contents disclosed in the present invention without affecting the effects and objectives that can be achieved by the present invention.

[0017] Figure 1 1 is a flow chart of a lithium battery overcharge protection control method provided by an embodiment of the present invention; Figure 2 The figure is a schematic block diagram of the structure of a lithium battery overcharge protection control system provided by an embodiment of the present invention.

[0018] Description of reference numerals: 10. Lithium battery overcharge protection control system; 11. Acquisition module; 12. Analysis module; 13. Judgment module; 14. Matching module. DETAILED DESCRIPTION

[0019] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0020] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may be decomposed, combined, or partially merged, so the actual execution order may vary depending on the actual situation.

[0021] It should also be understood that the terms used in this specification are for the purpose of describing specific embodiments only and are not intended to limit the present application. As used in this specification and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0022] It should be further understood that the term "and / or" used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0023] The technical solution of the present invention will be further described below with reference to the accompanying drawings and through specific implementation methods.

[0024] Example 1: like Figure 1 As shown, the present application provides a lithium battery overcharge protection control method, including steps S100 to S400.

[0025] Step S100: Obtain the voltage value, voltage change rate, temperature value and temperature change rate of each cell in the lithium battery, and perform interval consistency correction on the voltage value, temperature value, voltage change rate and temperature change rate to construct a cell state data set and a change trend feature set.

[0026] In this step, periodic data sampling is performed on all cells in the lithium battery pack, and the real-time voltage and temperature values ​​of each cell are extracted. The difference between adjacent sampling points is calculated to obtain the voltage change rate and temperature change rate. Specifically, the original voltage data and temperature data of each cell are input into a one-dimensional Gaussian filter for noise suppression, and the voltage and temperature values ​​are standardized using the Min-Max normalization method to form a voltage standardization sequence and a temperature standardization sequence. On this basis, the voltage change rate ΔU and the temperature change rate ΔT are calculated, and the calculation formula is as follows: ; ; in, and Respectively represent the voltage and temperature values ​​at the i-th sampling time, is the sampling interval. Next, an interval consistency correction function is used to align the data intervals of different cells, making the voltage change rate, temperature change rate, voltage normalized sequence, and temperature normalized sequence comparable within a unified interval. After removing abnormal data exceeding safety limits, the remaining state vectors constitute a valid cell data window.

[0027] For example, the voltage values ​​of all battery cells are normalized using the set interval standard [0,1], and the boundary of the change rate is set to [-0.05,0.05]. Data exceeding this range will be marked as abnormal samples and removed, thereby improving the accuracy of subsequent model analysis.

[0028] Step S200 : performing state trend analysis on the cell state data set and the change trend feature set to obtain the current risk level and short-term state prediction value of each cell.

[0029] In this step, the cell status dataset and trend feature set are time-segmented, dividing the entire charging process into multiple fixed-length time windows. Joint trend indicators are then calculated for the data within each time window. Specifically, the synergistic trend between the voltage and temperature rates of change is analyzed, and representative trend factors such as rising speed, oscillation frequency, and change direction are extracted.

[0030] Combined with historical state change samples, the change amplitude and mutation characteristics of the trend factor are identified to generate a state mutation mark; then, based on the stability of the change rate and the mutation intensity, the trend stability and potential abnormal tendencies of each battery cell are quantified, thereby predicting its short-term state trend, including phenomena such as temperature rise and voltage approaching the threshold in the short term in the future.

[0031] For example, if the temperature of a battery cell continues to rise by more than 3°C / min in the last three windows and the voltage change rate gradually accelerates, the system determines that it has a short-term overcharging risk tendency and predicts that its temperature will exceed the set safety threshold within 10 minutes. At the same time, its risk level is assessed as "medium to high" to guide subsequent charging control strategies.

[0032] Step S300: Input the short-term state prediction value and the battery cell state data set into a preset temperature rise prediction model to obtain the critical proximity, overcharge risk factor and abnormal trigger probability of each battery cell, perform dual threshold judgment on the abnormal trigger probability and critical proximity, and generate a temperature rise abnormality mark set.

[0033] In this step, the predicted short-term state values ​​(such as predicted temperature and voltage) are input into a pre-trained temperature rise prediction model along with the current cell state dataset. Specifically, this model, based on a deep regression neural network structure, maps historical charging data to actual temperature rise responses to predict temperature rise trends over the next charging cycle. During the prediction process, the system outputs the temperature rise rate, the proximity to the critical temperature point, and the probability of exceeding the temperature rise safety range within a specific time period in the future. This information is then used to determine whether the cell is at risk of overheating or reaching a critical state.

[0034] For example, if a battery cell is predicted to reach 60°C within 5 minutes, and the upper safety limit is 65°C, its critical proximity is high. If the predicted temperature rise trend deviates significantly from the normal range of variation, and the model predicts that its abnormal trigger probability is greater than 70%, the system will mark it as "temperature rise abnormality" based on the pre-set dual-threshold judgment mechanism and add it to the temperature rise abnormality mark set.

[0035] Step S400: weightedly fuse the temperature rise abnormality mark set and the overcharge risk factor to obtain a composite risk factor value, match the composite risk factor value with the current risk level, obtain a charging adjustment instruction set, and adjust the charging control module of the lithium battery according to the charging adjustment instruction set.

[0036] In this step, the composite risk factor value for each battery cell is calculated based on the weighted relationship between the temperature rise anomaly flag set and the overcharge risk factor. A specific risk response strategy is then formulated based on the current risk level. Specifically, different weighting coefficients are set, with the urgency of the temperature rise flag as the dominant weight. Multi-dimensional parameters of the overcharge risk factor, such as voltage trends and historical temperature rise responses, are integrated to produce a more comprehensive risk assessment indicator. The system then matches the composite risk factor value to a preset charging adjustment strategy library based on the level range (e.g., "high risk," "medium risk," or "low risk"), selecting the corresponding current adjustment coefficient, voltage control strategy, and other factors to form a charging adjustment instruction set.

[0037] For example, if the composite risk factor value of a battery cell is high and the corresponding level is "high risk", the system will match the strongest current limiting instruction in the strategy library, reduce the charging current to 50% of the original value, and start the dynamic temperature control mechanism. If necessary, it will interrupt the independent charging path of the battery cell to ensure the safety of the overall system.

[0038] In this embodiment, the voltage value, voltage change rate, temperature value, and temperature change rate of each cell in the lithium battery are obtained, and interval consistency correction is performed on the voltage value, temperature value, voltage change rate, and temperature change rate to construct a cell state data set and a change trend feature set. Then, a state trend analysis is performed on the cell state data set and the change trend feature set to obtain the current risk level and short-term state prediction value of each cell. Subsequently, the short-term state prediction value and the cell state data set are input into a preset temperature rise prediction model to obtain the critical proximity, overcharge risk factor, and abnormal trigger probability of each cell, and a dual-threshold judgment is performed on the abnormal trigger probability and the critical proximity to generate a temperature rise abnormality mark set. Then, the temperature rise abnormality mark set is weightedly fused with the overcharge risk factor to obtain a composite risk factor value, and the composite risk factor value is matched with the current risk level to obtain a charging adjustment instruction set. Finally, the charging control module of the lithium battery is adjusted according to the charging adjustment instruction set to achieve dynamic protection control. By constructing a cell state data set and a change trend feature set, a comprehensive quantitative expression of the cell operating state can be achieved, which can improve the accuracy of state identification. Based on the current risk level and short-term state prediction value generated by state trend analysis, abnormal trends can be predicted in advance, improving the foresight of response. Combined with the preset temperature rise prediction model for multi-dimensional prediction, it can dynamically calculate the critical proximity, overcharge risk factor and abnormal trigger probability, and improve the robustness of abnormality detection through a dual threshold mechanism. By weightedly fusing the temperature rise abnormality marker set with the overcharge risk factor, a composite risk factor value is formed, and matching it with the current risk level to generate a charging adjustment instruction set, a refined and differentiated charging control strategy is implemented. The above method can significantly enhance the overcharge protection capability of lithium batteries in complex environments, improve system safety and cell life, and reduce the misjudgment rate and the probability of overprotection.

[0039] By introducing a standardization and consistency correction mechanism for voltage, temperature, and their rates of change, data alignment between different battery cells is achieved, thereby improving the accuracy of state analysis. In the state trend analysis process, multi-dimensional features such as the joint trend gradient value, state gradient factor, and mutation coefficient are introduced, combined with a sliding window and prediction algorithm to effectively identify cell state evolution trends and potential risks. In the temperature rise prediction process, short-term state prediction values ​​and historical state data are input into a multi-layer neural network to accurately simulate temperature rise changes during future charging cycles. The critical proximity and anomaly probability are assessed based on the temperature rise safety interval, enabling early warning. In the final control decision, a composite risk factor model with dynamic response capabilities is formed by integrating the temperature rise anomaly marker set with the overcharge risk factor, enabling real-time control of charging current and voltage limiting. The entire solution not only improves charging system safety but also exhibits strong adaptability and generalization capabilities. It maintains high identification and control accuracy, especially under nonlinear state disturbances. This improves existing technologies that suffer from protection hysteresis or over-protection in complex battery environments.

[0040] Example 2: In step S100, the voltage value, voltage change rate, temperature value, and temperature change rate of each cell in the lithium battery are obtained, and interval consistency correction is performed on the voltage value, temperature value, voltage change rate, and temperature change rate to construct a cell state data set and a change trend feature set, which specifically includes: The voltage sampling data and temperature sampling data of each cell in the lithium battery are obtained, and the voltage sampling data and the temperature sampling data are subjected to noise filtering and normalization processing to obtain a voltage normalized sequence and a temperature normalized sequence.

[0041] This approach involves introducing a data cleaning and standardization mechanism during the sampling phase. Specifically, the multi-channel voltage and temperature acquisition circuit embedded in the lithium battery management system is first used to collect raw voltage and temperature data sequences for each battery cell in real time at a fixed sampling frequency (e.g., 10Hz). To eliminate data noise caused by sensor fluctuations and external interference, a sliding average filter is used to smooth the raw voltage and temperature data. The filter window size can be set to 5 to 7 sampling points. Subsequently, the filtered voltage and temperature sequences are numerically compressed using the maximum and minimum normalization method, normalizing them to the [0, 1] range. This unifies the scale of the features across all dimensions and facilitates subsequent model processing.

[0042] For example, if the raw voltage data collected for a battery cell over the past 10 seconds is 3.85, 3.86, 3.84, 3.87, and 3.86 volts, then after processing it with an averaging filter with a sliding window of 3, the smoothed sequence is 3.85, 3.85, and 3.857 volts. Assuming the maximum voltage of the entire set of samples is 4.2V and the minimum voltage is 3.0V, the corresponding normalization result is (3.85-3.0) / (4.2-3.0) = 0.708. After this processing, the raw voltage and temperature sampling data are converted into voltage-normalized and temperature-normalized sequences, respectively, for subsequent analysis.

[0043] The voltage change rate and temperature change rate of each battery cell are obtained, and the voltage change rate, temperature change rate, voltage normalized sequence, and temperature normalized sequence are corrected for interval consistency to obtain a state vector set. The state vector set is compared with the preset operating condition characteristic parameters, and out-of-limit abnormal samples are eliminated to generate a valid battery cell data window.

[0044] By establishing a dynamic trend comparison mechanism and an interval consistency processing mechanism; specifically, firstly, the voltage change rate and temperature change rate within each time interval are calculated based on the voltage standardization sequence and the temperature standardization sequence, and this is achieved by differential calculation (i.e., the current value minus the value at the previous moment divided by the time interval); then, the above four data are integrated into a multi-dimensional state vector, and the characteristic value of each dimension is corrected for consistency according to the preset safe operating range corresponding to the battery cell type and usage scenario. The correction method is: if the value of a certain dimension exceeds its safety upper and lower limits, the state vector is considered abnormal. Subsequently, the corrected state vector set is compared with the set standard operating condition characteristic parameters, such as charge and discharge rate, temperature rise rate, initial SOC state and other parameters, and samples that do not meet the operating condition range are eliminated. A sliding window method is used to construct an effective battery cell data window that can be used for subsequent analysis. The window size can be set to a continuous data segment of 10 to 20 seconds.

[0045] For example, if a cell's current voltage change rate is 0.015V / s and its temperature change rate is 0.4°C / s, combined with a normalized voltage of 0.72 and a temperature of 0.66, a four-dimensional state vector is formed: 0.015, 0.4, 0.72, 0.66. The preset operating parameters include a temperature change rate upper limit of 0.5°C / s and a voltage change rate upper limit of 0.02V / s. If a vector's temperature change rate is 0.6°C / s, it is considered abnormal and removed from the valid data window. The remaining state vectors that meet these conditions sequentially constitute the current valid cell data window.

[0046] A fused state vector is constructed based on the voltage normalized sequence and temperature normalized sequence in the valid cell data window. The fused state vector is combined with the voltage change rate and temperature change rate in the valid cell data window to construct a cell state dataset and a change trend feature set.

[0047] Multidimensional state reconstruction is achieved by constructing a fusion expression model. Specifically, based on the voltage and temperature normalized sequences sampled from each valid cell data window, stability and trend features are extracted by performing mean extraction, fluctuation coefficient calculation, and interval segmentation trend modeling on the two sequences. A fused state vector is then constructed by feature splicing. Simultaneously, the voltage and temperature change rates for the corresponding time periods are extracted from the same data window and subjected to sliding window aggregation to obtain a trend vector representing the rate of change of the cell state. Ultimately, the fused state vector and trend vector are integrated into unified structural data, named the cell state dataset and change trend feature set, respectively, to serve as the input data source for subsequent state trend analysis.

[0048] For example, in a valid cell data window, the normalized voltage sequence is 0.71, 0.72, 0.72, 0.73, and 0.74, and the normalized temperature sequence is 0.65, 0.66, 0.67, 0.68, and 0.69. Their means are 0.724 and 0.67, respectively, with fluctuation coefficients of approximately 0.01 to 0.02, indicating a slowly increasing trend. The fused state vector can be represented as 0.724, 0.67, 0.02, 0.01, and rising. Combined with the average voltage rate of change of 0.013 V / s and the temperature rate of change of 0.35°C / s for this period, these are combined into a six-dimensional input feature structure, which is used to construct the cell state dataset and trend feature set.

[0049] In step S200, the steps of performing state trend analysis on the cell state data set and the change trend feature set to obtain the current risk level and short-term state prediction value of each cell specifically include: The battery cell state dataset and change trend feature set are segmented into time windows to obtain several time series state segments. The joint trend gradient value of the voltage change rate and the temperature change rate in all time series state segments is calculated, and the state gradient factor and mutation coefficient of each battery cell are analyzed based on the joint trend gradient value.

[0050] This approach introduces a sliding time window mechanism and a joint trend modeling approach. Specifically, each cell state dataset is segmented into equally spaced sliding windows along the time dimension, with the window length set to T (e.g., 10 seconds) and the sliding step size set to t (e.g., 2 seconds). This yields several time series state segments. A time series fit is performed on the voltage and temperature change rates within each state segment, and first-order linear regression or polynomial regression is used to calculate the joint trend gradient value, reflecting the slope direction and growth amplitude of the state change within the current segment. The sudden changes in the gradients of adjacent windows are further analyzed, and the gradient difference and fluctuation amplitude are calculated to obtain the state gradient factor and mutation coefficient. The state gradient factor describes the persistence of the trend, and the mutation coefficient characterizes the intensity of the sudden change.

[0051] For example, if the voltage change rate of a battery cell in five consecutive time windows is 0.012, 0.013, 0.016, 0.021, and 0.034 V / s, and the temperature change rate is 0.31, 0.32, 0.38, 0.41, and 0.52°C / s, the calculated joint trend gradient of voltage and temperature within these windows is positive and gradually increasing. Based on the slope growth rate calculation, the state gradient factor is set to 0.78, indicating a gradually increasing trend. The mutation coefficient is 0.46, indicating strong fluctuations and requiring further risk analysis.

[0052] The short-term trend of the corresponding battery cell is predicted by the state gradient factor and the mutation coefficient to obtain the short-term voltage prediction value and the short-term temperature prediction value, and the short-term state prediction value is constructed using the short-term voltage prediction value and the short-term temperature prediction value.

[0053] By employing a weighted time series extrapolation mechanism and a nonlinear trend modeling strategy, specifically using the current cell's state gradient factor as the dominant weight and combining it with the mutation coefficient as a modulating factor, a short-term forecast of voltage and temperature trends within the future prediction timeframe T is generated. Furthermore, a multi-step extrapolation of the normalized voltage and temperature values ​​is performed using prediction models such as exponential smoothing or LSTM neural networks to obtain short-term voltage and temperature forecasts. These two values ​​are then combined to construct a short-term state forecast value, expressed as a two-dimensional vector: .

[0054] For example, if the current normalized voltage value of a battery cell is 0.72, the normalized temperature value is 0.66, the state gradient factor is 0.75, and the mutation coefficient is 0.22, then it is predicted that the voltage will rise to 0.77 and the temperature will rise to 0.71 in the short term. Then its short-term state prediction value is 0.77, 0.71, which will be used to determine whether it has entered the high-risk range.

[0055] The short-term voltage prediction value and the short-term temperature prediction value are cross-compared to obtain the prediction interval risk deviation and overcharge proximity.

[0056] By constructing an interval risk benchmark model and a cross-excursion measurement algorithm, we specifically set upper voltage and temperature limits within the safe operating range (e.g., a voltage limit of 0.85 and a temperature limit of 0.8), and then perform a differential analysis between the predicted value and the interval boundaries. The closer the predicted value is to the upper limit, the higher the risk excursion. Furthermore, if the temperature rises significantly when the predicted voltage approaches the upper limit, it indicates an overcharge trend. The degree of overcharge proximity is then calculated as a reference indicator to quantify how close the abnormal cell state is to the critical point.

[0057] For example, the short-term voltage prediction value of a battery cell is 0.84, and the temperature prediction value is 0.79, which are only 0.01 and 0.01 away from their respective upper limits. The prediction interval risk deviation is recorded as 0.92, and the overcharge proximity is 0.94, indicating that the battery cell will enter an overcharge risk state in a short period of time and requires further response and intervention.

[0058] The state gradient factor, prediction interval risk deviation and overcharge proximity are integrated to generate the primary state risk factor and disturbance stability coefficient.

[0059] By adopting the weighted fusion function and risk disturbance model; specifically, setting the weight coefficient group 、 、 Corresponding to the state gradient factor, risk deviation and overcharge proximity, the calculation formula for the primary state risk factor is constructed as follows: ; in, is the state gradient factor, is the risk bias, The disturbance stability coefficient is jointly evaluated by the mutation coefficient and the trend fluctuation frequency, which represents the system's resistance and recovery ability to disturbances.

[0060] For example, if the state gradient factor of a cell is 0.73, the risk offset is 0.88, and the overcharge proximity is 0.91, and α=0.3, β=0.35, and γ=0.35, then its primary state risk factor is: =0.3×0.73+0.35×0.88+0.35×0.91=0.848. The mutation coefficient is 0.26, the trend fluctuation frequency is 0.41, and the calculated disturbance stability coefficient is 0.67. The risk weight score is calculated based on the primary state risk factor and the disturbance stability coefficient, and the current risk level is generated based on the risk weight score.

[0061] By building a hierarchical scoring mechanism and a risk level mapping model; specifically, setting the risk scoring function: ; in, is the primary state risk factor, is the disturbance stability coefficient, and To control the weight (such as 0.6 and 0.4). After normalization, the scores are mapped to the grade interval: [0, 0.3) for low risk, [0.3, 0.6) for medium risk, [0.6, 0.85) for high risk, and ≥ 0.85 for severe risk.

[0062] For example, if the primary state risk factor of a cell is 0.82 and the disturbance stability coefficient is 0.65, substituting them into the scoring function yields: =0.6×0.82+0.4×(1-0.65)=0.492+0.14=0.63, which is classified as a high risk level; the system ultimately outputs its current risk level as "high" for the control system to subsequently determine whether to perform protective actions. The steps of segmenting the cell state data set and the change trend feature set into time windows to obtain several time series state segments, calculating the joint trend gradient value of the voltage change rate and the temperature change rate in all time series state segments, and analyzing the state gradient factor and mutation coefficient of each cell based on the joint trend gradient value specifically include: The cell state data set and the change trend feature set are divided into multiple time series state segments according to a preset time window. The voltage data and temperature data in each time series state segment are smoothed and filtered to obtain a smoothed voltage sequence and a smoothed temperature sequence.

[0063] By employing a sliding time window mechanism and a multi-order filtering algorithm—specifically, setting the time window length to T seconds and the step size to t seconds—the original voltage and temperature series are time-divided to generate multiple non-overlapping or partially overlapping time series state segments. To remove high-frequency noise components in the data and improve the accuracy of trend calculation, a smoothing filter is applied to each state segment. Either the Savitzky–Golay filter (SG filter) or the exponential moving average (EMA) filter can be used. Fitting is performed on each sampling point in the series, preserving the main trend information while suppressing random disturbances, resulting in smoothed voltage and temperature series, respectively.

[0064] For example, if the time window is set to 10 seconds and the step size is 2 seconds, and a set of cell voltage data with a sampling frequency of 1 Hz is segmented, the original voltage sequence in a window is obtained as follows: 3.78, 3.80, 3.81, 3.82, 3.85, 3.84, 3.83, 3.86, 3.88, 3.90. After SG filtering, the smoothed voltage sequence is: 3.78, 3.795, 3.81, 3.825, 3.84, 3.845, 3.85, 3.865, 3.88, 3.895, the jitter noise is filtered out, making subsequent calculations more stable.

[0065] The change rate curves of the smoothed voltage sequence and the smoothed temperature sequence are calculated, and the voltage change rate and temperature change rate of each segment are extracted according to the change rate curves. The voltage change rate and temperature change rate of each time series state segment are jointly analyzed to obtain the joint trend gradient value.

[0066] Through differential operations and two-dimensional trend modeling, specifically, first-order differencing is performed on the smoothed voltage and temperature series to obtain the instantaneous rate of change at each time point, forming voltage and temperature rate of change curves. Based on this, a joint slope analysis method is used to transform the two rate of change curves into a two-dimensional rate of change vector sequence within each time series segment. Using linear regression, principal component analysis, or vector angle calculation, the main direction of the overall trend is extracted, resulting in the joint trend gradient value for that segment, which serves as a quantitative characteristic indicator of the battery cell's state change during that time period.

[0067] For example, the smoothed voltage change rate of a certain battery cell in a 10-second segment is 0.003, 0.0025, 0.0027, 0.0032, and 0.0031 V / s, and the temperature change rate is 0.06, 0.058, 0.061, 0.065, and 0.067°C / s. By calculating that the vector angle is 12°, the joint trend gradient value can be defined as the sum of the voltage and temperature change rates multiplied by the cosine angle, which is 0.0053×cos(12°)≈0.0052, which is used to characterize the trend direction and intensity.

[0068] The change trend of each battery cell is analyzed using the joint trend gradient value to obtain the state gradient factor and mutation coefficient.

[0069] This approach constructs a trend persistence function and a mutation amplitude detection mechanism. Specifically, the combined trend gradient value sequence of the battery cells in multiple consecutive time series segments is used as input to calculate the stability and continuity of the changing trend. The state gradient factor, defined as the ratio of the average growth rate of the gradient value sequence to the fluctuation amplitude, is used to describe whether the trend is continuously strengthening. The mutation coefficient is calculated based on the gradient transition amplitude and the number of slope mutation points between adjacent segments. It represents the uncertainty and abnormal fluctuation level in the trend. By setting a mutation threshold to detect "turning points", the mutation behavior can be quantified.

[0070] For example, the joint trend gradient values ​​of a certain battery cell in four consecutive time segments are 0.0048, 0.0051, 0.0056, and 0.0073, respectively, showing an overall obvious increasing trend. The calculated average growth rate is 0.00083, the fluctuation standard deviation is 0.00093, and the state gradient factor = 0.00083 / 0.00093≈0.89, indicating good trend sustainability. However, the jump difference between segments 3 and 4 is 0.0017, exceeding the set threshold of 0.0015. In this case, a mutation point is determined to exist, and the mutation coefficient is set to 0.72, indicating that the current battery cell state has changed dramatically.

[0071] In step S300, the short-term state prediction value and the cell state data set are input into a preset temperature rise prediction model to obtain the critical proximity, overcharge risk factor, and abnormal trigger probability of each cell. A dual-threshold determination is performed on the abnormal trigger probability and the critical proximity to generate a temperature rise abnormality mark set. The steps specifically include: The short-term state prediction value and the battery cell state data set are input into the preset temperature rise prediction model to obtain the temperature rise prediction curve and dynamic temperature rise increase value of each battery cell in the future charging cycle. The preset temperature rise safety range is compared with the temperature rise prediction curve and the dynamic temperature rise increase value to obtain the temperature rise out-of-bounds amplitude and critical proximity.

[0072] A temperature rise prediction model based on a deep neural network is constructed. Specifically, a fused multi-input neural network model is used. Inputs include short-term voltage and temperature predictions, the current cell temperature, current, resistance, voltage curve morphology, and other state parameters. The output is the temperature change trend for multiple future time steps, known as the temperature rise prediction curve. Simultaneously, based on the slope trend of the temperature rise prediction curve and the maximum temperature change, a dynamic temperature rise increment is calculated. This value is then compared with a preset temperature rise safety range (e.g., 45°C to 60°C) to determine the extent of the temperature violation and the relative proximity of the cell's current state to the upper temperature rise limit, defined as the critical proximity.

[0073] For example, if the temperature rise prediction curve predicts that the temperature will rise from 43°C to 59°C in the next 10 minutes, the dynamic temperature rise increase value is 16°C. If the temperature rise safety upper limit is 60°C, the temperature rise beyond the limit is 0°C (not beyond the limit), and the critical proximity is (59–43) / (60–43)=0.94, indicating that the temperature is very close to the risk critical point.

[0074] The trend correction factor is calculated using the dynamic temperature rise increase value and the current risk level. The overcharge risk weight is calculated based on the trend correction factor. The temperature rise beyond the limit range and the overcharge risk weight are integrated to obtain the overcharge risk factor and abnormal trigger probability.

[0075] This approach employs a temperature-rise-risk mapping function and a weighted fusion mechanism. Specifically, the dynamic temperature rise increment and the current risk level (e.g., medium risk, high risk) are mapped into a trend correction factor. The trend correction factor adjusts the weighting of the temperature change rate on risk assessment. Based on this, a risk weighting function is constructed: overcharge risk weight = trend correction factor × risk level factor, where the risk level factor is mapped based on the risk level score obtained in the previous step (e.g., low = 1, medium = 1.5, high = 2.0). Subsequently, the temperature rise threshold is linearly or nonlinearly fused with the overcharge risk weight, for example, using cosine similarity, fuzzy logic, or normalized product methods, to generate a comprehensive overcharge risk factor. Furthermore, based on the nonlinear variation characteristics of this factor and the temperature rise curve, the probability of future anomalies is estimated to determine the anomaly trigger probability.

[0076] For example, if a battery cell's temperature rise exceeds the threshold by 2°C, the current risk is medium, the trend correction factor is 1.3, and the risk level factor is 1.5, then the overcharge risk weight is 1.95. By combining the normalized product of the temperature rise and the threshold, the overcharge risk factor is 1.95 × (2 / 10) = 0.39. If the abnormality probability model is set to a sigmoid function, the abnormality trigger probability is ≈ sigmoid(0.39) ≈ 0.596.

[0077] A double-threshold judgment is performed based on the abnormal trigger probability and critical proximity to generate a temperature rise abnormality mark set.

[0078] By setting a composite threshold strategy and abnormal state marking rules; specifically, setting a critical proximity determination threshold (such as 0.85) and an abnormal trigger probability determination threshold (such as 0.6), battery cells that meet the "double high" criteria are marked as abnormal temperature rise conditions. The system assesses each battery cell within each prediction cycle. If the current abnormal trigger probability exceeds the abnormal threshold and the critical proximity exceeds the proximity threshold, it is determined to be in a high-risk state of impending overheating. The cell's status mark for the current prediction cycle is included in the temperature rise abnormality mark set for subsequent protection control module invocation.

[0079] For example, if cell A has an abnormal trigger probability of 0.65 and a critical proximity of 0.91, both exceeding the set dual thresholds, the cell's status is marked as "temperature rise abnormal," and its index, predicted time point, and related risk factors are written to the temperature rise abnormality tag set. This tag set is then uploaded to the system control module, which issues protection commands such as current limiting, voltage reduction, and charging pause.

[0080] The steps of calculating the trend correction factor using the dynamic temperature rise increase value and the current risk level, calculating the overcharge risk weight based on the trend correction factor, and fusing the temperature rise beyond-limit range with the overcharge risk weight to obtain the overcharge risk factor and abnormal trigger probability specifically include: The trend correction factor is calculated based on the dynamic temperature rise increase value and the current risk level. The trend correction factor is weighted and integrated with the current risk factor to obtain the overcharge risk weight, which reflects the overcharge risk of the battery cell under specific charging conditions. Among them, the current risk factor refers to the comprehensive judgment result of the current temperature, charging current and voltage of the battery cell.

[0081] By defining a trend correction function, the dynamic temperature rise increase value is mapped to a trend correction factor, and weighted fusion is performed in combination with the current risk factor of the battery cell; specifically, the trend correction factor adopts a linear mapping function, such as: trend correction factor = 1 + ×Dynamic temperature rise increase value; in, The current risk factor is an empirical adjustment factor that reflects the impact of temperature rise changes on risk. The current risk factor is calculated using a multi-dimensional logic judgment model based on the real-time temperature, charging current, and voltage of the battery cell. After normalization, it is converted into a value between 0 and 1. Subsequently, the trend correction factor and the current risk factor are combined through weighted averaging or weighted product. The calculation formula is as follows: Overcharge risk weight = α × trend correction factor + (1-α) × current risk factor; Among them, α is the weight coefficient, which is optimized according to the actual charging environment and historical data.

[0082] For example, assuming the dynamic temperature rise increment is 5°C, =0.1, the trend correction factor is 1.5; the current risk factor is 0.7, the weight coefficient α=0.6, then the overcharge risk weight is: 0.6×1.5+0.4×0.7=0.9+0.28=1.18, which reflects the comprehensive overcharge risk level of the battery cell under the current charging state.

[0083] The overcharge risk factor is calculated using the overcharge risk weight, the temperature rise limit and the temperature change rate of the battery cell. The overcharge risk factor is used to indicate the risk level of overcharging of the battery cell.

[0084] By constructing a temperature change rate function and a risk-weighted model, we first calculate the product of the cell temperature change rate (i.e., the temperature change rate) and the temperature rise threshold to reflect the rapid temperature increase trend. This product is then multiplied by the overcharge risk weight or nonlinearly fused to calculate the comprehensive overcharge risk factor. The calculation formula is as follows: Overcharge risk factor = overcharge risk weight × (temperature change rate, temperature rise beyond the limit); Among them, the function Weighted summation or exponential mapping can be used to ensure that the risk of cells with rapidly rising temperatures is significantly amplified.

[0085] For example, if the temperature change rate is 0.8°C / minute and the temperature rise exceeds the limit by 3°C, the simple product method is: 𝑓=0.8×3=2.4; Combined with the overcharge risk weight of 1.18, the overcharge risk factor is calculated as: 1.18×2.4=2.832; the higher the value, the greater the risk of battery cell overcharge.

[0086] The overcharge risk factor is fused with the temperature rise beyond the limit, and the abnormal trigger probability is calculated based on the fusion result.

[0087] By designing an abnormality trigger probability calculation model, specifically using a Sigmoid function or cumulative distribution function (CDF) to map the fused overcharge risk factor and temperature rise threshold to a probability value between 0 and 1, which serves as the abnormality trigger probability. The fusion method can be weighted sum or product. The example formula is: ; Abnormal trigger probability = ; in, It is an empirical parameter determined by training based on historical data to ensure a smooth and reasonable distribution of abnormal probability.

[0088] For example, assuming the overcharge risk factor is 2.832, the temperature rise margin is 3, and the weight coefficient is =0.7, then: ; set up =1.5, =2.0, then the probability of abnormal triggering is: ; This indicates that the battery cell has a 78.9% probability of triggering an abnormality, which requires system warnings and protective actions.

[0089] In step S400, the temperature rise abnormality flag set and the overcharge risk factor are weightedly fused to obtain a composite risk factor value, the composite risk factor value is matched with the current risk level, and a charging adjustment instruction set is obtained. The steps of adjusting the charging control module of the lithium battery according to the charging adjustment instruction set specifically include: The temperature rise abnormality mark set and the overcharge risk factor are fused according to the preset weighting function to obtain the composite risk factor value of each battery cell. The composite risk factor value is matched with the current risk level to obtain the risk response priority.

[0090] By defining a weighted fusion function, the binary value or probability value of the temperature rise anomaly mark is linearly or nonlinearly fused with the overcharge risk factor. Specifically, the composite risk factor value is calculated in the form of weighted average or product, for example: ; in, and is the weight coefficient, satisfying , and adjust according to historical charging data and risk response requirements. By statistically analyzing the distribution of composite risk factors in historical data, the corresponding risk confidence interval is calculated to characterize the confidence level of risk judgment.

[0091] Assuming that the temperature rise abnormality mark of a battery cell is 0.8 (probability value), the overcharge risk factor is 2.5, and the weight coefficients are 0.4 and 0.6 respectively, the composite risk factor is calculated as: 0.4×0.8+0.6×2.5=0.32+1.5=1.82; The risk confidence interval, derived from historical data analysis, ranges from 1.5 to 2.0. By matching the composite risk factor value with the current risk level threshold (e.g., low, medium, high), the risk response priority for the cell is determined, for example, a "high" priority.

[0092] Based on the risk response priority, the corresponding current adjustment coefficient in the preset charging control library is selected, the composite risk factor value is combined with the current adjustment coefficient, and the dynamic current limit value and voltage limit parameters are calculated.

[0093] The charging strategy library is preset by mapping the risk response priority. Specifically, the charging strategy library contains multiple sets of current adjustment coefficients and voltage limit parameters corresponding to different risk levels and priorities. According to the current composite risk factor, the corresponding current adjustment coefficient is selected, and the dynamic current limit value and voltage limit parameters are calculated using the following formula: ; ; in, and are the nominal current and voltage, respectively, It is an adjustment coefficient and is pre-set by the charging control library.

[0094] For example, the nominal charging current of a lithium battery is 2A, the composite risk factor is 1.82, and the coefficient is 0.3, then the dynamic current limit value is: 2×(1-0.3×1.82)=2×(1-0.546)=2×0.454=0.908A; The corresponding voltage limit parameters are also calculated using a similar method to ensure charging safety.

[0095] A charging adjustment instruction set is generated according to the dynamic current limit value and the voltage limit parameters, and the charging control module of the lithium battery is adjusted using the charging adjustment instruction set.

[0096] The calculated dynamic current limit and voltage limit parameters are converted into a charging adjustment instruction set through the charging control protocol. Specifically, the instruction set includes current limit commands, voltage limit commands, and risk warning flags, and its format complies with the communication specifications of the charging management system. Based on this instruction set, the system adjusts the operating status of the charging module in real time to ensure that the battery cell charging current and voltage remain within a safe range.

[0097] For example, the generated instruction set includes "limiting the charging current to 0.9A" and "setting the voltage upper limit to 4.1V". After receiving the instructions, the charging control module automatically adjusts the power switch and voltage regulator to reduce the charging rate and avoid overcharging and abnormal temperature rise of the battery cells.

[0098] In this embodiment, by segmenting the battery cell state data set and the change trend feature set into time windows, extracting the smoothed series of voltage and temperature and their change rates, and calculating the joint trend gradient value, the state gradient factor and mutation coefficient of each battery cell are obtained, thereby achieving an accurate characterization of the battery cell state trend; by inputting a preset temperature rise prediction model based on the short-term state prediction value and historical data, the battery cell temperature rise curve and increase are dynamically predicted, and the temperature rise safety interval is combined to evaluate the temperature rise out-of-bounds amplitude and critical proximity, thereby effectively identifying the abnormal temperature rise risk; by using the dynamic temperature rise increase value and the current risk level to calculate the trend correction factor, the risk weight and temperature rise out-of-bounds amplitude are integrated to determine the temperature rise risk. degree, quantify the overcharge risk factor and abnormal trigger probability, and realize the dynamic assessment of overcharge risk; in the risk fusion stage, the temperature rise abnormality mark and the overcharge risk factor are fused according to the weighted function to generate a composite risk factor, and matched with the current risk level to form a risk response priority to guide the adjustment of the charging control strategy; based on the risk response priority, combined with the current adjustment coefficient and voltage limit parameters of the preset charging control library, the dynamic current limit value and voltage limit value are calculated, and finally a charging adjustment instruction set is generated to adjust the lithium battery charging control module in real time to ensure that the battery cell is charged within a safe range, reduce the risk of overcharge and temperature rise abnormalities, and improve the safety and reliability of battery charging. This embodiment effectively realizes the precise monitoring and intelligent adjustment of the battery cell charging status through a multi-dimensional and multi-level risk prediction and fusion mechanism, combined with a dynamic model and control strategy, and significantly improves the overall safety performance and life management level of the lithium battery system.

[0099] Example 3: like Figure 2 As shown, the present application also provides a lithium battery overcharge protection control system 10, including an acquisition module 11, an analysis module 12, a determination module 13 and a matching module 14.

[0100] The acquisition module 11 is mainly used to obtain the voltage value, voltage change rate, temperature value and temperature change rate of each cell in the lithium battery, and perform interval consistency correction on the voltage value, temperature value, voltage change rate and temperature change rate to construct a cell status data set and a change trend feature set.

[0101] The analysis module 12 is mainly used to perform state trend analysis on the battery cell state data set and the change trend feature set to obtain the current risk level and short-term state prediction value of each battery cell.

[0102] The determination module 13 is mainly used to input the short-term state prediction value and the battery cell state data set into the preset temperature rise prediction model, obtain the critical proximity, overcharge risk factor and abnormal trigger probability of each battery cell, perform dual threshold judgment on the abnormal trigger probability and critical proximity, and generate a temperature rise abnormality mark set.

[0103] The matching module 14 is mainly used to perform weighted fusion of the temperature rise abnormality mark set and the overcharge risk factor to obtain a composite risk factor value, match the composite risk factor value with the current risk level, obtain a charging adjustment instruction set, and adjust the charging control module of the lithium battery according to the charging adjustment instruction set.

[0104] In this embodiment, the acquisition module 11 collects the voltage value, voltage change rate, temperature value, and temperature change rate of each lithium battery cell in real time, and performs interval consistency correction on the above data to ensure the accuracy and comparability of the data, thereby constructing a comprehensive battery cell status data set and change trend feature set. Specifically, through multi-dimensional data fusion and preprocessing, the reliable foundation for subsequent analysis is improved. The analysis module 12 performs a detailed status trend analysis on the status data set and change trend feature set, calculates the current risk level and short-term status prediction value of each battery cell, effectively captures the dynamic trend of battery cell status changes, and enhances the accuracy of risk prediction. The judgment module 13 inputs the short-term status prediction value and real-time status data into the preset temperature rise prediction model, accurately predicts the temperature rise changes of each battery cell in the future charging cycle, and obtains the critical proximity, overcharge risk factor, and abnormal trigger probability. The dual-threshold judgment mechanism is used to generate a temperature rise abnormality mark set to promptly identify potential abnormal risks. Matching module 14 combines the temperature rise anomaly marker set with the overcharge risk factor according to a weighted function to obtain a composite risk factor value. This is then matched with the current risk level to generate a reasonable set of charging adjustment instructions, enabling dynamic adjustment of the lithium battery charging control module, effectively reducing overcharge risks and improving system safety and stability. Through modular design and multi-dimensional data integration, the overall solution achieves accurate identification and dynamic protection against lithium battery overcharge risks, significantly enhancing the safety assurance capabilities of the charging process.

[0105] It should be noted that, those skilled in the art will clearly understand that, for the sake of convenience and brevity of description, the specific working processes of the above-described system and each module can refer to the corresponding processes in the aforementioned embodiment 1 and will not be repeated here.

[0106] The structures, proportions, sizes, etc. depicted in the drawings of this specification are only used to match the contents disclosed in the specification so as to facilitate understanding and reading by persons familiar with this technology. They are not intended to limit the conditions under which the present invention can be implemented and therefore have no substantive technical significance. Any structural modifications, changes in proportional relationships, or adjustments in size should still fall within the scope of the technical contents disclosed in the present invention without affecting the effects and objectives that can be achieved by the present invention.

[0107] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A lithium battery overcharge protection control method, characterized in that: include: Obtaining the voltage value, voltage change rate, temperature value, and temperature change rate of each cell in the lithium battery, and performing interval consistency correction on the voltage value, the temperature value, the voltage change rate, and the temperature change rate to construct a cell state data set and a change trend feature set; Performing a state trend analysis on the battery cell state data set and the change trend feature set to obtain a current risk level and a short-term state prediction value of each battery cell; Inputting the short-term state prediction value and the battery cell state data set into a preset temperature rise prediction model to obtain the critical proximity, overcharge risk factor and abnormal trigger probability of each battery cell, performing dual threshold determination on the abnormal trigger probability and the critical proximity, and generating a temperature rise abnormality mark set; The temperature rise abnormality mark set and the overcharge risk factor are weightedly fused to obtain a composite risk factor value, the composite risk factor value is matched with the current risk level to obtain a charging adjustment instruction set, and the charging control module of the lithium battery is adjusted according to the charging adjustment instruction set.

2. The lithium battery overcharge protection control method according to claim 1, characterized in that: The step of obtaining the voltage value, voltage change rate, temperature value, and temperature change rate of each cell in the lithium battery, and performing interval consistency correction on the voltage value, the temperature value, the voltage change rate, and the temperature change rate to construct a cell state data set and a change trend feature set includes: Acquire voltage sampling data and temperature sampling data of each cell in the lithium battery, perform noise filtering and normalization processing on the voltage sampling data and the temperature sampling data, and obtain a voltage normalized sequence and a temperature normalized sequence; Obtain the voltage change rate and temperature change rate of each battery cell, perform interval consistency correction on the voltage change rate, the temperature change rate, the voltage normalized sequence, and the temperature normalized sequence to obtain a state vector set, compare the state vector set with preset operating condition characteristic parameters, eliminate out-of-limit abnormal samples, and generate a valid battery cell data window; A fusion state vector is constructed based on the voltage normalized sequence and the temperature normalized sequence in the valid cell data window, and the fusion state vector is combined with the voltage change rate and the temperature change rate in the valid cell data window to construct a cell state data set and a change trend feature set.

3. The lithium battery overcharge protection control method according to claim 1, characterized in that: The step of performing state trend analysis on the battery cell state data set and the change trend feature set to obtain the current risk level and short-term state prediction value of each battery cell includes: Segmenting the cell state data set and the change trend feature set into time windows to obtain a plurality of time series state segments, calculating a joint trend gradient value of the voltage change rate and the temperature change rate in all the time series state segments, and analyzing a state gradient factor and a mutation coefficient of each cell based on the joint trend gradient value; Predicting the short-term trend of the corresponding battery cell by using the state gradient factor and the mutation coefficient to obtain a short-term voltage prediction value and a short-term temperature prediction value, and constructing a short-term state prediction value using the short-term voltage prediction value and the short-term temperature prediction value; Performing interval cross-comparison on the short-term voltage prediction value and the short-term temperature prediction value to obtain a prediction interval risk deviation and an overcharge proximity degree; fusing the state gradient factor, the prediction interval risk deviation, and the overcharge proximity to generate a primary state risk factor and a disturbance stability coefficient; A risk weight score is calculated based on the primary state risk factor and the disturbance stability coefficient, and a current risk level is generated based on the risk weight score.

4. The lithium battery overcharge protection control method according to claim 3, characterized in that: The step of segmenting the cell state data set and the change trend feature set into time windows to obtain a plurality of time series state segments, calculating the joint trend gradient value of the voltage change rate and the temperature change rate in all the time series state segments, and analyzing the state gradient factor and mutation coefficient of each cell based on the joint trend gradient value includes: Dividing the cell state data set and the change trend feature set into a plurality of time series state segments according to a preset time window, and performing smoothing filtering on the voltage data and temperature data in each of the time series state segments to obtain a smoothed voltage sequence and a smoothed temperature sequence; Calculating the rate of change curves of the smoothed voltage sequence and the smoothed temperature sequence, extracting the voltage change rate and the temperature change rate of each segment according to the rate of change curves, and jointly analyzing the voltage change rate and the temperature change rate of each time series state segment to obtain a joint trend gradient value; The combined trend gradient value is used to analyze the change trend of each battery cell to obtain a state gradient factor and a mutation coefficient.

5. The lithium battery overcharge protection control method according to claim 1, characterized in that: The step of inputting the short-term state prediction value and the battery cell state data set into a preset temperature rise prediction model to obtain the critical proximity, overcharge risk factor, and abnormal trigger probability of each battery cell, performing dual-threshold determination on the abnormal trigger probability and the critical proximity, and generating a temperature rise abnormality mark set includes: Inputting the short-term state prediction value and the battery cell state data set into a preset temperature rise prediction model to obtain a temperature rise prediction curve and a dynamic temperature rise increase value for each battery cell in a future charging cycle, and comparing a preset temperature rise safety range with the temperature rise prediction curve and the dynamic temperature rise increase value to obtain a temperature rise out-of-bounds amplitude and a critical proximity; Calculating a trend correction factor using the dynamic temperature rise increment value and the current risk level, calculating an overcharge risk weight based on the trend correction factor, and fusing the temperature rise beyond-limit magnitude with the overcharge risk weight to obtain an overcharge risk factor and an abnormality trigger probability; A dual-threshold determination is performed based on the abnormal trigger probability and the critical proximity to generate a temperature rise abnormality mark set.

6. The lithium battery overcharge protection control method according to claim 5, characterized in that: The step of calculating a trend correction factor using the dynamic temperature rise increase value and the current risk level, calculating an overcharge risk weight based on the trend correction factor, and fusing the temperature rise exceeding limit amplitude with the overcharge risk weight to obtain an overcharge risk factor and an abnormality trigger probability includes: Calculating a trend correction factor based on the dynamic temperature rise increase value and the current risk level, and weighting and fusing the trend correction factor with the current risk factor to obtain an overcharge risk weight; wherein the current risk factor refers to a comprehensive judgment result of the current temperature, charging current, and voltage of the battery cell; An overcharge risk factor is calculated using the overcharge risk weight and the temperature rise exceeding the limit in combination with the temperature change rate of the battery cell; The overcharge risk factor is fused with the temperature rise exceeding the limit, and the abnormality triggering probability is calculated based on the fusion result.

7. The lithium battery overcharge protection control method according to claim 1, characterized in that: The steps of weighting and fusing the temperature rise abnormality mark set and the overcharge risk factor to obtain a composite risk factor value, matching the composite risk factor value with the current risk level to obtain a charging adjustment instruction set, and adjusting the charging control module of the lithium battery according to the charging adjustment instruction set include: fusing the temperature rise abnormality mark set and the overcharge risk factor according to a preset weighting function to obtain a composite risk factor value for each battery cell, and matching the composite risk factor value with the current risk level to obtain a risk response priority; Selecting a corresponding current adjustment coefficient from a preset charging control library based on the risk response priority, combining the composite risk factor value with the current adjustment coefficient, and calculating a dynamic current limit value and a voltage limit parameter; A charging adjustment instruction set is generated according to the dynamic current limit value and the voltage limit parameter, and the charging control module of the lithium battery is adjusted using the charging adjustment instruction set.

8. A lithium battery overcharge protection control system, characterized in that: include: An acquisition module is used to obtain the voltage value, voltage change rate, temperature value and temperature change rate of each cell in the lithium battery, and perform interval consistency correction on the voltage value, the temperature value, the voltage change rate and the temperature change rate to construct a cell state data set and a change trend feature set; An analysis module, configured to perform a state trend analysis on the cell state data set and the change trend feature set to obtain a current risk level and a short-term state prediction value of each cell; a determination module, configured to input the short-term state prediction value and the cell state data set into a preset temperature rise prediction model, obtain a critical proximity, an overcharge risk factor, and an abnormal trigger probability for each cell, perform a dual-threshold determination on the abnormal trigger probability and the critical proximity, and generate a temperature rise abnormality flag set; A matching module is used to perform weighted fusion of the temperature rise abnormality mark set and the overcharge risk factor to obtain a composite risk factor value, match the composite risk factor value with the current risk level, obtain a charging adjustment instruction set, and adjust the charging control module of the lithium battery according to the charging adjustment instruction set.

Images