Active equalization control method for power battery

Abstract

The invention discloses a power battery active equalization control method, which comprises the steps of extracting characteristic information of a battery pack, generating quantitative state change trend description, and obtaining potential unbalanced risk point distribution; analyzing operation constraint conditions related to load changes, generating trigger weights of unbalanced risks, screening high-priority risk points, generating a preliminary equalization path planning scheme, calculating equalization operation current intensity of each single battery in different time periods in combination with battery pack state data collected in real time, and generating an equalization operation schedule; obtaining feedback state data in real-time execution, and comparing the feedback state data with the predicted risk point to obtain a final control decision instruction; and for the final control decision instruction, acquiring executed battery pack state update data, analyzing the consistency of the battery pack state update data and a target equalization state, judging whether an equalization path or time arrangement needs to be finely adjusted, and generating a continuously optimized equalization control scheme. Accurate prediction and balance control of the state of the battery pack are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of power battery management, and in particular to a power battery active balancing control method. Background Art

[0002] Battery balancing control is a core research area in the field of new energy vehicles and energy storage systems. Its importance is self-evident, directly impacting the lifespan, safety, and overall efficiency of battery packs. With the acceleration of electrification, ensuring the consistency of the state of each individual cell in a battery pack has become a key issue for the industry. Battery balancing control not only affects the stable operation of the system but also plays an irreplaceable role in improving energy utilization and reducing maintenance costs.

[0003] However, many current balancing control methods still have significant limitations. Traditional strategies often rely on passive responses to real-time conditions and lack foresight into future operating conditions. This leads to inaccurate control decisions and can easily lead to over-balancing or under-balancing. This short-sighted control approach struggles to adapt to dynamic changes in battery status under complex operating conditions, thus compromising overall system efficiency.

[0004] Against this backdrop, the field faces several pressing challenges. The most pressing issue is accurately understanding the battery pack's evolving state over time. Due to a lack of advance prediction of load changes, control strategies can often only be adjusted remedially after a problem occurs. This issue further necessitates another technical challenge: how to rationally plan balancing paths and timing based on predictions. Without an optimized plan in place, balancing operations may be triggered frequently or at inappropriate times, leading to energy waste and increased system burden. These two interrelated issues, the former being the foundation and the latter the deeper layer, directly determine the intelligence and efficiency of balancing control.

[0005] Therefore, how to predict the evolution of battery status by building a dynamic model and optimize the balancing path and timing based on this has become a key issue in improving the active balancing control effect of power batteries. Summary of the Invention

[0006] In order to solve the technical problems existing in the above-mentioned prior art, the present invention proposes a method for active balancing control of a power battery to improve the battery life.

[0007] To achieve the above objectives, the present invention provides a power battery active balancing control method, comprising:

[0008] Obtain target single-cell battery status data, extract characteristic information of the battery pack, generate a quantitative description of state change trends, and based on the description of state change trends, simulate the state evolution of each single-cell battery over a period of time in the future to obtain the distribution of potential imbalance risk points. The description of state variable trends includes voltage deviation trends, temperature distribution changes, and state of charge deviation trends.

[0009] Based on the distribution of potential imbalance risk points, the operating constraints related to load changes are analyzed to generate trigger weights for imbalance risks. Based on the trigger weights, high-priority risk points are screened and a preliminary balancing path planning scheme is generated. Furthermore, the balancing operation current intensity of each single battery in different time periods is calculated in combination with the real-time collected battery pack status data to generate a balancing operation schedule.

[0010] Processing the balancing operation schedule to generate an optimized balancing time schedule, and obtaining feedback status data during real-time execution based on the optimized balancing time schedule, and obtaining a final control decision instruction by comparing the data with the predicted risk points;

[0011] For the control decision instruction, the battery pack status update data after execution is obtained, and its consistency with the target balancing state is analyzed to generate a continuously optimized balancing control plan.

[0012] Compared with the prior art, the present invention has the following advantages and technical effects:

[0013] The present invention generates a description of state change trends by extracting characteristic information from historical data. Combining load patterns and real-time operating conditions, a long-short-term memory network is used to predict future state evolution and obtain the potential imbalance risk distribution. Operational constraints are analyzed, key factors are extracted, and risk trigger weights are generated. High-priority risk points are screened and a preliminary balancing path is formulated. Dynamic adjustments are made based on real-time status, the balancing operation current intensity is calculated, and an operation schedule is generated. Timing is optimized considering energy loss, and control instructions are generated after verification of effectiveness. Consistency is analyzed after execution, and the balancing scheme is continuously optimized. The present invention achieves accurate prediction and balancing control of battery pack status, thereby improving battery service life and system reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0015] Figure 1 This is a flow chart of a power battery active balancing control method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0016] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0017] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0018] This embodiment proposes a power battery active balancing control method, such as Figure 1 ,include:

[0019] Obtain target single-cell battery status data, extract characteristic information of the battery pack, generate a quantitative description of state change trends, and based on the description of state change trends, simulate the state evolution of each single-cell battery over a period of time in the future to obtain the distribution of potential imbalance risk points. The description of state variable trends includes voltage deviation trends, temperature distribution changes, and state of charge deviation trends.

[0020] Based on the distribution of potential imbalance risk points, the operating constraints related to load changes are analyzed to generate trigger weights for imbalance risks. Based on the trigger weights, high-priority risk points are screened and a preliminary balancing path planning scheme is generated. Furthermore, the balancing operation current intensity of each single battery in different time periods is calculated in combination with the real-time collected battery pack status data to generate a balancing operation schedule.

[0021] Processing the balancing operation schedule to generate an optimized balancing time schedule, and obtaining feedback status data during real-time execution based on the optimized balancing time schedule, and obtaining a final control decision instruction by comparing the data with the predicted risk points;

[0022] For the control decision instruction, the battery pack status update data after execution is obtained, and its consistency with the target balancing state is analyzed to generate a continuously optimized balancing control plan.

[0023] This embodiment generates a description of state change trends by extracting feature information from historical data. Combining load patterns and real-time operating conditions, a long-short-term memory network is used to predict future state evolution and obtain the potential imbalance risk distribution. Operational constraints are analyzed, key factors are extracted, and risk trigger weights are generated. High-priority risk points are screened and a preliminary balancing path is formulated. Dynamic adjustments are made based on real-time status, the balancing operation current intensity is calculated, and an operation schedule is generated. Timing is optimized considering energy loss, and control instructions are generated after verification of effectiveness. Consistency is analyzed after execution, and the balancing solution is continuously optimized. Accurate prediction and balancing control of battery pack status are achieved, improving battery life and system reliability.

[0024] Furthermore, a quantitative description of the state change trend is generated, including:

[0025] Obtaining voltage, current, temperature, and state of charge data of the target single battery from historical operating data, and performing preliminary processing using a time series model to obtain basic characteristic data of battery operation;

[0026] Performing environmental correction on the basic characteristic data by fusing the ambient temperature parameter and the ambient humidity parameter to obtain corrected characteristic data;

[0027] If the voltage value in the corrected characteristic data exceeds a preset voltage threshold range, anomaly detection is performed on the voltage data to obtain a voltage anomaly data set;

[0028] Extracting the voltage deviation trend based on the voltage anomaly data set and combining it with the corrected characteristic data, and fitting the voltage variation curve using a linear regression algorithm to obtain a quantitative description of the voltage deviation trend;

[0029] The spatial distribution characteristics of the single battery temperature are analyzed through the corrected characteristic data, and the temperature distribution area is divided using a cluster analysis algorithm to obtain a quantitative description of the temperature distribution change;

[0030] Based on the corrected characteristic data, the dynamic change of the state of charge is calculated, and the state of charge deviation is predicted by combining the time series model to obtain a quantitative description of the state of charge deviation trend.

[0031] Specifically, when processing historical operating data for single cells, time series data can be obtained from dimensions such as voltage, current, temperature, and state of charge. Suppose that the voltage data for a battery over the past 24 hours exhibits periodic fluctuations, with an average value of 3.7 volts, and the temperature data varies between 20 and 30 degrees Celsius. Using time series models, such as the moving average method, this data can be smoothed to extract fundamental features and remove noise interference, resulting in a more stable operating trend. This helps ensure the accuracy of subsequent analysis.

[0032] In the environmental correction process, integrating ambient temperature and humidity parameters is particularly important. Assuming an ambient temperature of 25 degrees Celsius and a humidity of 60%, a weighted average algorithm is used to calculate the environmental impact coefficient, perhaps with a temperature weight of 0.6 and a humidity weight of 0.4. This ultimately calibrates the basic characteristic data. The corrected voltage value might be adjusted from 3.7 volts to 3.68 volts, reflecting the actual operating conditions. This correction effectively reduces the impact of external factors on the data and improves analysis reliability.

[0033] If the corrected voltage value exceeds the pre-set voltage threshold range, such as 3.2 to 4.2 volts, anomaly detection is required. Using the isolation forest algorithm, anomalies such as voltage values ​​reaching 4.3 volts at a specific time point are identified, forming a voltage anomaly dataset. This helps to promptly detect potential faults and ensure battery safety.

[0034] When analyzing voltage deviation trends, we combined the corrected characteristic data with the abnormal data set to extract the trend of voltage increasing over time. Using a linear regression algorithm to fit the change curve, we could quantitatively describe a voltage deviation increase of 0.02 volts per hour. This quantitative result provided data support for fault prediction.

[0035] Based on the spatial temperature distribution characteristics, a cluster analysis algorithm is used to divide the temperatures in different areas of the battery pack into high-temperature and low-temperature zones. Assuming the average temperature in the high-temperature zone is 35 degrees Celsius and the low-temperature zone is 22 degrees Celsius, this quantitative description of the distribution changes can help identify uneven heat dissipation and optimize battery design.

[0036] Furthermore, based on the description of the state change trend, the state evolution of each single battery in the future period is simulated to obtain the distribution of potential imbalance risk points, including:

[0037] Based on the description of the state change trend, combined with the periodic regularity data of the external load and the real-time collected operating condition data, a prediction model based on the long short-term memory network is used to simulate the state evolution of each single battery in the future period of time to obtain a state change sequence;

[0038] If the deviation of a single battery in the state change sequence exceeds a preset deviation threshold, the battery is identified as a high-deviation battery, and the identification of the high-deviation battery and its corresponding time period distribution are obtained;

[0039] According to the identification and time period distribution of the high-deviation battery, the imbalance risk value of each single battery is calculated to obtain the potential imbalance risk point distribution.

[0040] Specifically, the external load periodicity data and operating condition data obtained from the battery management system are the basis for building a state prediction model. External load data may include changes in the power demand of the battery pack during operation, and operating condition data include the real-time voltage, current, and temperature of the battery. For example, the temperature of a single cell reaches 45 degrees Celsius under high load. During data preprocessing, the sliding window averaging method is used to denoise the voltage signal and eliminate abnormal spikes caused by sensor jitter, such as reducing voltage fluctuations from ±0.5 volts to ±0.1 volts. Normalization processing unifies data of different dimensions to the range of 0 to 1 for easy model input. This preprocessing ensures the reliability of the feature data set and provides high-quality input for subsequent model training.

[0041] In one possible implementation, when building a long-short-term memory network model, the input features include the cyclical pattern of the external load, such as a daily peak load of approximately 1000 watts during peak hours, as well as temperature and current data during operating conditions. By memorizing historical sequences, the model captures the dynamic response of the battery under varying loads. During training, seven consecutive days of operating data, including hourly load and temperature changes, can be input. The model learns the subtle changes in the battery's internal resistance as the temperature rises. After training, the model can predict the state change sequence of each single cell over the next 24 hours, such as predicting that a cell's voltage will drop by 0.2 volts after a high load. This predictive capability helps identify potential risks in advance.

[0042] Furthermore, based on the distribution of potential imbalance risk points, the operating constraints related to load changes are analyzed to generate trigger weights for imbalance risks, including:

[0043] Obtain load change data and operating constraints from the battery management system, use data cleaning methods to remove outliers, and obtain a standardized constraint data set;

[0044] Extracting load variation characteristics and operation constraint characteristics based on the standardized constraint data set, determining key influencing factors using a principal component analysis method, and obtaining an influencing factor set;

[0045] A weight calculation model based on logistic regression is constructed through the set of influencing factors, and load change characteristics and operation constraint characteristics are input. The weight calculation model based on logistic regression is trained to obtain the imbalance risk trigger weight and store it in the data set.

[0046] Specifically, load variation data and operating constraints are obtained from the battery management system. Data cleaning methods are used to remove outliers to obtain a standardized constraint dataset. Based on this standardized constraint dataset, load variation characteristics and operating constraint characteristics are extracted. Principal component analysis is used to identify key influencing factors, resulting in an influencing factor set. A weight calculation model based on logistic regression is constructed using this influencing factor set. Load variation characteristics and operating constraint characteristics are input, and the model is trained to obtain the imbalance risk trigger weight.

[0047] If the trigger weight exceeds the preset threshold, the corresponding risk point is identified. Combined with time series data, the probability of occurrence of each risk point is calculated to obtain a probability quantification result. Based on this probability quantification result, time series analysis methods are used to determine the time point distribution corresponding to each risk point, resulting in a time point distribution sequence. Cluster analysis methods are used to group risk points within this time point distribution sequence, resulting in a risk point classification set. Based on this risk point classification set, a dynamic monitoring data set for imbalanced risk is generated and stored in a database, resulting in monitoring data records.

[0048] Furthermore, based on the trigger weights, high-priority risk points are screened and a preliminary balanced path planning solution is generated, including:

[0049] For the imbalanced risk trigger weight, the probability of each risk point is compared with the preset threshold condition to screen out high-priority risk points, and the risk points are grouped using a cluster analysis method based on the high-priority risk points to obtain risk point grouping results;

[0050] Based on the risk point grouping results, a preliminary framework for the equilibrium path was constructed. Combined with the battery identification information, the set of intervention nodes corresponding to each path was determined. The correlation between each node and the equilibrium time was analyzed, and the time series analysis method was used to obtain the recommended equilibrium time node distribution;

[0051] If there is a time conflict in the proposed balanced time node distribution, the time allocation of the intervention nodes is adjusted by priority sorting to determine the final balanced time node;

[0052] The preliminary balanced path planning scheme is generated according to the final balanced time node.

[0053] Specifically, when constructing the preliminary framework for balancing paths, we can combine battery identification information to associate each battery's operating status with risk points. For example, if a battery is identified as A001 and its corresponding risk point falls into the high-risk group, the path planned for it should prioritize load adjustment nodes. This approach ensures targeted path planning.

[0054] When analyzing the correlation between intervention nodes and balancing times, you can use time series data to observe the impact of load changes on balancing results. For example, if an intervention node performs load adjustment at 6:00 PM daily and historical data shows that load fluctuations are minimal during this time, it's recommended to set the balancing time node to that time. This analysis helps optimize time allocation.

[0055] If there's a conflict in the recommended balanced time node distribution, for example, if two intervention nodes are scheduled for 6:00 PM simultaneously, we can prioritize the intervention node for the high-risk group and postpone it to 7:00 PM for the medium-risk group. This adjustment avoids resource competition and ensures timely processing of high-priority risk points.

[0056] When generating balancing route plans for each battery ID, a route can be designed for battery ID A001, including three intervention nodes, with load adjustments performed at 18:00, 20:00, and 22:00, and the plan can be stored in the database. This planning approach ensures a clear intervention plan for each battery.

[0057] Furthermore, generating the balancing operation schedule includes:

[0058] Obtaining the status data of each single cell from the battery pack, preprocessing the data based on voltage, temperature, and power indicators to obtain a standardized battery status data set. Based on the standardized battery status data set and a pre-established set of intervention nodes, a support vector machine model is used to classify and determine the balancing requirements of each single cell, thereby determining a list of single cells requiring balancing operations.

[0059] For the determined list of single cells, the corresponding time segment division is obtained. Combined with the dynamic optimization strategy, the balanced operating current value of each single cell in different time segments is calculated to obtain a preliminary current value allocation plan;

[0060] By using the preliminary balancing path planning scheme, the balancing operation requirements of each single cell are extracted from the preliminary current value distribution scheme. In combination with the balancing path constraint conditions, if the current value of a single cell exceeds a preset current value range, the operating current value is adjusted to generate an optimized current distribution result;

[0061] Based on the optimized current distribution results and the specific requirements of the balancing operation, corresponding balancing operation instructions are generated for each single cell to obtain a detailed operation execution sequence. The balancing operation details of each single cell are extracted from the operation execution sequence, and a structured balancing operation schedule is generated by combining the time segment and the operation current value.

[0062] Furthermore, the balancing operation schedule is processed to generate an optimized balancing time schedule, including:

[0063] Acquire real-time energy loss data for the balancing operation schedule, and perform preliminary cleaning on the data using time series analysis technology to obtain an initial energy loss data set;

[0064] For the initial energy loss data set, a normalization operation is performed on the data using a normalization processing technology to obtain a standardized energy loss data set;

[0065] The standardized energy loss data set is compared with the preset loss threshold one by one. If the energy loss value of a single battery exceeds the preset loss threshold, it is determined that the battery needs to adjust the balancing timing, and a timing adjustment requirement list is generated;

[0066] Based on the timing adjustment requirement list, a dynamic programming algorithm is used to calculate the balanced timing arrangement of each single battery in different time periods to obtain a preliminary timing adjustment plan;

[0067] The initial timing adjustment plan is verified by balancing the timing constraints. If the timing arrangement of a single battery exceeds the constraint range, its timing parameters are readjusted to generate an optimized timing adjustment result.

[0068] The timing arrangement details of each single battery are extracted from the optimized timing adjustment result, and a structured balancing timing schedule is generated using data integration technology to obtain the optimized balancing time arrangement.

[0069] Specifically, when collecting energy loss data during battery pack operation in real time, the energy loss value of each individual battery can be obtained through a high-precision sensor interface. Sensors are typically deployed at key nodes in the battery pack, collecting parameters such as voltage, current, and temperature. Combined with an energy loss calculation model, this data is used to obtain loss data for each battery. For example, suppose a battery pack contains 10 individual batteries, and the collected energy loss values ​​are 0.5%, 0.7%, 1.2%, and so on. Batteries that exceed the preset threshold of 0.8% are marked as abnormal. This collection method ensures high real-time and accuracy of data, laying the foundation for subsequent analysis.

[0070] In one possible implementation, time series analysis technology is used to clean the initial energy loss data. A sliding window method is used to remove outliers and fill in missing data. For example, if the loss value of a battery at a certain point in time suddenly jumps to 10%, which is significantly deviated from the normal range, the cleaning process will interpolate the previous and subsequent data to ensure the continuity of the data set. The cleaned data is more suitable for subsequent standardization processing and improves the reliability of analysis. Specifically, the standardization processing technology normalizes the energy loss data to eliminate dimensional differences. For example, the loss values ​​of all single cells are mapped to a range of 0 to 1. Assuming that the original loss of a battery is 1.2%, it is 0.6 after normalization. This processing facilitates data comparison between different batteries and provides a unified standard for threshold comparison.

[0071] Furthermore, the final control decision instruction is obtained, including:

[0072] By collecting the execution feedback data during the system operation in real time, a preliminary analysis is conducted on the matching degree between the equilibrium time and the time arrangement to obtain the initial deviation assessment results;

[0073] Based on the initial deviation assessment results, historical data records related to the predicted risk are obtained, and a support vector machine model is used to compare the real-time data with the historical data to determine the distribution of potential risk points;

[0074] If the distribution of potential risk points exceeds the preset distribution range, the equilibrium path will be adjusted based on the risk comparison results, a path correction plan will be generated, and the optimal path adjustment direction will be determined;

[0075] According to the path correction scheme, combined with real-time data and the current value of the switch state, the dynamic adjustment range of the intensity parameter is calculated to obtain the preliminary instruction content of the control decision;

[0076] Through the preliminary instruction content, combined with the closed-loop mechanism of feedback verification, a secondary check is performed on the matching of the time arrangement and the balance time, the final decision instruction parameters are determined, and the final control decision instruction is obtained.

[0077] Specifically, by collecting real-time feedback data from the system during operation, the sensor network can continuously monitor the battery pack's operating status and obtain data on the matching of balancing times with the actual schedule. For example, suppose a battery pack's balancing time is preset to 2 hours, but actual operational feedback data shows that some cells overheat after only 1.5 hours. This deviation will be recorded as the initial deviation assessment result for subsequent analysis.

[0078] For example, by combining the initial deviation assessment results with historical data records, the database can be used to retrieve risk records from similar operating conditions over the past month. Suppose historical data shows that in high-temperature environments, when the battery pack's balancing time deviation exceeds 20%, the risk points are concentrated in overheating and short circuits. Using a support vector machine model, feature comparisons are performed between real-time and historical data to identify the distribution of risk points that may arise during current operation. For example, if the temperature of a single battery cell remains above 45 degrees Celsius, it may become a potential risk point.

[0079] If the distribution of risk points exceeds the preset threshold, for example, if the temperature threshold is set at 40 degrees Celsius and the actual value reaches 45 degrees Celsius, the balancing path needs to be adjusted. One possible implementation method is to generate a path correction solution by reducing the charging rate of the individual battery or shortening its balancing time. Assuming that the charging rate is reduced from 5A to 3A after adjustment, the optimal path adjustment direction is initially determined to be to reduce energy input and protect battery life.

[0080] For example, when calculating the dynamic adjustment range of the intensity parameter based on the path correction scheme, real-time data and switch status can be combined. Assuming the switch is currently on and the real-time current value is 4A, the dynamic adjustment range might be set between 3A and 5A. The initial instruction content will be based on this range and recommend adjusting the current to 3.5A to balance efficiency and safety.

[0081] A secondary verification of the initial command content allows for a closed-loop mechanism to reconfirm the compatibility of the timing and balancing time. Assuming the adjusted balancing time is 1.8 hours, and actual operational feedback shows a deviation of 5%, the final decision command parameters can be set at 1.8 hours of balancing time and 3.5 amps of current, ensuring operational stability.

[0082] When generating specific execution signals, the final decision-making instruction parameters can be converted into executable signals for the control system. For example, a current value of 3.5A and a time of 1.8 hours can be converted into electrical signals and output to the controller for real-time control. If the feedback data still shows deviations after execution, such as the temperature remaining above 42°C, the path and parameters are repeatedly adjusted until the deviation converges to within the threshold range. This continuous optimization approach can effectively improve the stability and safety of battery pack operation.

[0083] Furthermore, for the control decision instruction, obtaining the battery pack status update data after execution, analyzing its consistency with the target balancing state, and generating a continuously optimized balancing control scheme, including:

[0084] Obtain battery status data after executing the command, collect real-time operating information of the battery pack through system sensors, record key indicators such as voltage and current, and obtain a preliminary status data set;

[0085] Based on the preliminary status data set, a pre-established analysis model is used to compare the matching degree between the battery status and the target balance to determine whether there is any deviation and obtain the matching analysis result;

[0086] If the matching analysis results show that the deviation exceeds the preset deviation range, a path adjustment suggestion is generated based on the deviation situation. The deviation distribution characteristics are extracted through the information processing link to determine the adjustment direction;

[0087] Based on the route adjustment suggestions and the current configuration of the schedule, a new time allocation plan is calculated. By referring to historical operation data, the optimized time parameters are determined;

[0088] Through the optimized time parameters, a continuously optimized control scheme is generated, which is dynamically updated according to the real-time changes in the battery status to obtain the adjusted control instruction content;

[0089] Generate specific execution signals based on the adjusted control instruction content, process information based on the operating environment of the battery pack, and determine the final signal output form;

[0090] The final signal output is used to update the operating status of the battery pack, continuously collect feedback data, and perform cyclic verification on the matching degree between the feedback data and the target balance to obtain continuously optimized operating results.

[0091] Specifically, when acquiring battery status data after executing instructions, system sensors can collect real-time operating information about the battery pack. Regarding key indicators such as voltage and current, for example, suppose a battery pack has a rated voltage of 3.7V, the collected real-time voltage value is 3.5V, and the current value is 1.2A, lower than the expected value of 1.5A. This data forms a preliminary status dataset, providing a foundation for subsequent analysis. This collection method can promptly reflect the operating status of the battery pack, laying the data foundation for subsequent matching analysis.

[0092] For example, when comparing the initial status data set for accuracy, a pre-established analysis model can be used to analyze the difference between the battery status and the target balance. Assuming the target balance voltage should be stable at 3.7V, but the actual value is lower, this indicates a possible uneven charge distribution issue. Model comparison determines the deviation is 0.2V, exceeding the preset threshold of 0.1V and requiring further adjustment. This comparison method helps quickly identify problem points and provides a basis for subsequent path adjustments.

[0093] When generating path adjustment recommendations for deviations exceeding a threshold, information processing can be used to extract deviation distribution characteristics. For example, if a battery cell's voltage is consistently low, this could be due to capacity decay from long-term use. Adjustments could be made to reduce the discharge load of that cell. By analyzing historical operating data, the recommended approach is to increase the proportion of charging time. This approach allows for targeted recommendations tailored to specific issues, ensuring the rationality of the adjustment.

[0094] When calculating a new time allocation based on the schedule, assuming the original time configuration is 2 hours for charging and 1 hour for discharging, based on deviations and historical data, the optimization is to 2.5 hours for charging and 0.5 hours for discharging to increase the recovery speed of low-voltage cells. This optimized time parameter better matches the actual needs of the battery pack and extends its service life. When generating a continuously optimized control plan, the control instructions can be dynamically updated based on real-time changes. For example, if the battery pack voltage gradually approaches 3.7V during operation but still fluctuates slightly, the control plan can fine-tune the charging intensity to reduce the fluctuation. This dynamic update method adapts to changes in battery status and ensures operational stability.

[0095] When generating a specific execution signal and determining the final output form, the signal can be adjusted to a pulse output according to the operating environment of the battery pack to reduce the impact on the battery. Assuming the ambient temperature is 25 degrees, the signal output form is selected as a low-frequency pulse to protect the battery while ensuring the effect. This method can improve the adaptability of the execution signal and reduce operational risks. When updating the operating status of the battery pack and continuously collecting feedback data, assuming that the adjusted voltage value is stable at 3.65V, close to the target value, it is found through cyclic verification that the matching degree has met the standard. This continuous feedback mechanism can ensure that the battery pack operates in the best state, reduce deviation accumulation, and improve overall efficiency.

[0096] The above are merely preferred embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A power battery active balancing control method, characterized in that: include: Obtain target single-cell battery status data, extract characteristic information of the battery pack, generate a quantitative description of state change trends, and based on the description of state change trends, simulate the state evolution of each single-cell battery over a period of time in the future to obtain the distribution of potential imbalance risk points. The description of state variable trends includes voltage deviation trends, temperature distribution changes, and state of charge deviation trends. Based on the distribution of potential imbalance risk points, the operating constraints related to load changes are analyzed to generate trigger weights for imbalance risks. Based on the trigger weights, high-priority risk points are screened and a preliminary balancing path planning scheme is generated. Furthermore, the balancing operation current intensity of each single battery in different time periods is calculated in combination with the real-time collected battery pack status data to generate a balancing operation schedule. Processing the balancing operation schedule to generate an optimized balancing time schedule, and obtaining feedback status data during real-time execution based on the optimized balancing time schedule, and obtaining a final control decision instruction by comparing the data with the predicted risk points; For the final control decision instruction, the battery pack status update data after execution is obtained, and its consistency with the target balancing state is analyzed to generate a continuously optimized balancing control scheme.

2. The active balancing control method for power batteries according to claim 1, characterized in that: Generate a quantitative description of state change trends, including: Obtaining voltage, current, temperature, and state of charge data of the target single battery from historical operating data, and performing preliminary processing using a time series model to obtain basic characteristic data of battery operation; Performing environmental correction on the basic characteristic data by fusing the ambient temperature parameter and the ambient humidity parameter to obtain corrected characteristic data; If the voltage value in the corrected characteristic data exceeds a preset voltage threshold range, anomaly detection is performed on the voltage data to obtain a voltage anomaly data set; Extracting the voltage deviation trend based on the voltage anomaly data set and combining it with the corrected characteristic data, and fitting the voltage variation curve using a linear regression algorithm to obtain a quantitative description of the voltage deviation trend; The spatial distribution characteristics of the single battery temperature are analyzed through the corrected characteristic data, and the temperature distribution area is divided using a cluster analysis algorithm to obtain a quantitative description of the temperature distribution change; Based on the corrected characteristic data, the dynamic change of the state of charge is calculated, and the state of charge deviation is predicted by combining the time series model to obtain a quantitative description of the state of charge deviation trend.

3. The active balancing control method for power batteries according to claim 1, characterized in that: Based on the state change trend description, the state evolution of each single battery in the future period is simulated to obtain the distribution of potential imbalance risk points, including: Based on the description of the state change trend, combined with the periodic regularity data of the external load and the real-time collected operating condition data, a prediction model based on the long short-term memory network is used to simulate the state evolution of each single battery in the future period of time to obtain a state change sequence; If the deviation of a single battery in the state change sequence exceeds a preset deviation threshold, the battery is identified as a high-deviation battery, and the identification of the high-deviation battery and its corresponding time period distribution are obtained; According to the identification and time period distribution of the high-deviation battery, the imbalance risk value of each single battery is calculated to obtain the potential imbalance risk point distribution.

4. The active balancing control method for power batteries according to claim 1, characterized in that: Based on the distribution of potential imbalance risk points, the operating constraints related to load changes are analyzed to generate trigger weights for imbalance risks, including: Obtain load change data and operating constraints from the battery management system, use data cleaning methods to remove outliers, and obtain a standardized constraint data set; Extracting load variation characteristics and operation constraint characteristics based on the standardized constraint data set, determining key influencing factors using a principal component analysis method, and obtaining an influencing factor set; A weight calculation model based on logistic regression is constructed through the set of influencing factors, and load change characteristics and operation constraint characteristics are input. The weight calculation model based on logistic regression is trained to obtain the imbalance risk trigger weight and store it in the data set.

5. The active balancing control method for power batteries according to claim 4, characterized in that: Based on the imbalance risk trigger weights, high-priority risk points are screened and a preliminary balanced path planning solution is generated, including: For the imbalanced risk trigger weight, the probability of each risk point is compared with the preset threshold condition to screen out high-priority risk points, and the risk points are grouped using a cluster analysis method based on the high-priority risk points to obtain risk point grouping results; Based on the risk point grouping results, a preliminary framework for the equilibrium path was constructed. Combined with the battery identification information, the set of intervention nodes corresponding to each path was determined. The correlation between each node and the equilibrium time was analyzed, and the time series analysis method was used to obtain the recommended equilibrium time node distribution; If there is a time conflict in the proposed balanced time node distribution, the time allocation of the intervention nodes is adjusted by priority sorting to determine the final balanced time node; The preliminary balanced path planning scheme is generated according to the final balanced time node.

6. The power battery active balancing control method according to claim 5, characterized in that: Generating the balancing operation schedule includes: Obtaining the status data of each single cell from the battery pack, preprocessing the data based on voltage, temperature, and power indicators to obtain a standardized battery status data set. Based on the standardized battery status data set and a pre-established set of intervention nodes, a support vector machine model is used to classify and determine the balancing requirements of each single cell, thereby determining a list of single cells requiring balancing operations. For the determined list of single cells, the corresponding time segment division is obtained. Combined with the dynamic optimization strategy, the balanced operating current value of each single cell in different time segments is calculated to obtain a preliminary current value allocation plan; By using the preliminary balancing path planning scheme, the balancing operation requirements of each single cell are extracted from the preliminary current value distribution scheme. In combination with the balancing path constraint conditions, if the current value of a single cell exceeds a preset current value range, the operating current value is adjusted to generate an optimized current distribution result; Based on the optimized current distribution results and the specific requirements of the balancing operation, corresponding balancing operation instructions are generated for each single cell to obtain a detailed operation execution sequence. The balancing operation details of each single cell are extracted from the operation execution sequence, and a structured balancing operation schedule is generated by combining the time segment and the operation current value.

7. The power battery active balancing control method according to claim 1, characterized in that: Processing the balancing operation schedule to generate an optimized balancing time schedule includes: Acquire real-time energy loss data for the balancing operation schedule, and perform preliminary cleaning on the data using time series analysis technology to obtain an initial energy loss data set; For the initial energy loss data set, a normalization operation is performed on the data using a normalization processing technology to obtain a standardized energy loss data set; The standardized energy loss data set is compared with the preset loss threshold one by one. If the energy loss value of a single battery exceeds the preset loss threshold, it is determined that the battery needs to adjust the balancing timing, and a timing adjustment requirement list is generated; Based on the timing adjustment requirement list, a dynamic programming algorithm is used to calculate the balanced timing arrangement of each single battery in different time periods to obtain a preliminary timing adjustment plan; The initial timing adjustment plan is verified by balancing the timing constraints. If the timing arrangement of a single battery exceeds the constraint range, its timing parameters are readjusted to generate an optimized timing adjustment result. The timing arrangement details of each single battery are extracted from the optimized timing adjustment result, and a structured balancing timing schedule is generated using data integration technology to obtain the optimized balancing time arrangement.

8. The power battery active balancing control method according to claim 7, characterized in that: Obtain final control decision instructions, including: By collecting the execution feedback data during the system operation in real time, a preliminary analysis is conducted on the matching degree between the equilibrium time and the time arrangement to obtain the initial deviation assessment results; Based on the initial deviation assessment results, historical data records related to the predicted risk are obtained, and a support vector machine model is used to compare the real-time data with the historical data to determine the distribution of potential risk points; If the distribution of potential risk points exceeds the preset distribution range, the equilibrium path will be adjusted based on the risk comparison results, a path correction plan will be generated, and the optimal path adjustment direction will be determined; According to the path correction plan, combined with real-time data and the current value of the switch state, the dynamic adjustment range of the intensity parameter is calculated to obtain the preliminary instruction content of the control decision; Through the preliminary instruction content, combined with the closed-loop mechanism of feedback verification, a secondary check is performed on the matching of the time arrangement and the balance time, the final decision instruction parameters are determined, and the final control decision instruction is obtained.

9. The power battery active balancing control method according to claim 1, characterized in that: For the final control decision instruction, obtain the battery pack status update data after execution, analyze its consistency with the target balancing state, and generate a continuously optimized balancing control plan, including: Obtain battery status data after executing the command, collect real-time operating information of the battery pack through system sensors, record key indicators such as voltage and current, and obtain a preliminary status data set; Based on the preliminary status data set, a pre-established analysis model is used to compare the matching degree between the battery status and the target balance to determine whether there is any deviation and obtain the matching analysis result; If the matching analysis results show that the deviation exceeds the preset deviation range, a path adjustment suggestion is generated based on the deviation situation. The deviation distribution characteristics are extracted through the information processing link to determine the adjustment direction; Based on the route adjustment suggestions and the current configuration of the schedule, a new time allocation plan is calculated. By referring to historical operation data, the optimized time parameters are determined; Through the optimized time parameters, a continuously optimized control scheme is generated, which is dynamically updated according to the real-time changes in the battery status to obtain the adjusted control instruction content; Generate specific execution signals based on the adjusted control instruction content, process information based on the operating environment of the battery pack, and determine the final signal output form; The final signal output is used to update the operating status of the battery pack, continuously collect feedback data, and perform cyclic verification on the matching degree between the feedback data and the target balance to obtain continuously optimized operating results.

Images