Dynamic charging method for lithium batteries based on user behavior and environmental conditions

Abstract

The present invention discloses a method for dynamic charging of lithium batteries based on user behavior and environmental conditions, and relates to the technical field of dynamic charging of lithium batteries, and includes the following steps: for high-frequency users, during the dynamic charging process of the lithium battery, firstly, real-time charging data is collected from all cells in the current user's charging process to establish a first data set. The present invention accurately identifies the performance status of the cells through real-time data analysis and deep learning prediction, divides them into normal charging units and low-performance charging units, and adopts targeted charging strategies. The low-performance units are charged with fixed low power to reduce the risk of heating and delay aging. The normal units realize dynamic power allocation through fuzzy logic control, and preferentially allocate higher charging power to high-performance cells, thereby improving charging efficiency and shortening charging time. It effectively solves the problems of energy waste and safety hazards in traditional synchronous charging, and significantly improves charging speed, life and overall reliability.

Description

Technical Field

[0001] The present invention relates to the technical field of dynamic charging of lithium batteries, and particularly to a method for dynamically charging lithium batteries based on user behavior and environmental conditions. Background Art

[0002] Dynamic charging of lithium batteries based on user behavior and environmental conditions is an intelligent charging technology that dynamically adjusts the charging strategy by collecting real-time user usage habits (such as daily usage frequency, charging time period) and environmental conditions (such as temperature, humidity, and air pressure). The system analyzes the user's device usage patterns and environmental changes to intelligently allocate charging power and charging time, so as to extend the battery life, optimize the charging efficiency, and reduce the loss risk caused by overcharging or fast charging. In high-temperature or low-temperature environments, the system automatically reduces the charging power, and before the peak period of user demand, it preferentially replenishes the power to ensure the maximum battery life of the device.

[0003] During the dynamic charging process of lithium batteries, for users with high-frequency usage, the prior art usually adopts a fast charging method to quickly replenish electrical energy and synchronously charges all battery cells with a fixed charging power (to ensure synchronous management). Over time, even if the design parameters of each battery cell in the lithium battery are exactly the same, due to aging degree, production errors, and environmental differences, the performance of the battery cells will gradually become inconsistent in actual use. During synchronous charging, the performance of some battery cells may become worse, while the system still maintains the same charging power to charge all battery cells, which will cause a significant decrease in the charging efficiency of the battery cells with poor performance, and more energy will be converted into heat during the charging process, and electrical energy cannot be efficiently stored. At the same time, the battery cells with better performance may not be charged to the maximum because the overall charging efficiency of the system is dragged down by the battery cells with poor performance. In this case, the charging time is extended and the user's waiting time increases, seriously affecting the user experience. In addition, fixed-power synchronous charging will cause more electrical energy waste, accelerate battery aging, and affect the overall life and safety of the battery. Therefore, in the case of obvious performance differences, continuing to use the fixed-power synchronous charging method not only reduces the charging efficiency and user experience, but also increases unnecessary electrical energy loss and potential safety hazards of the battery.

[0004] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure, and therefore it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0005] The purpose of the present invention is to provide a dynamic charging method for lithium batteries based on user behavior and environmental conditions. Through real-time data analysis and deep learning prediction, the performance status of the battery cells can be accurately identified, and the battery cells can be divided into normal charging units and low-performance charging units. A targeted charging strategy is adopted. The low-performance units are charged with fixed low power to reduce the risk of heating and delay aging. The normal units are dynamically allocated with fuzzy logic control, and higher charging power is allocated to high-performance batteries first, thereby improving charging efficiency and shortening charging time. This effectively solves the energy waste and safety hazards in traditional synchronous charging, and significantly improves the charging speed, life and overall reliability, so as to solve the problems in the above-mentioned background technology.

[0006] In order to achieve the above purpose, the present invention provides the following technical solution: A lithium battery dynamic charging method based on user behavior and environmental conditions, comprising the following steps:

[0007] For high-frequency users, during the dynamic charging process of lithium batteries, first collect real-time charging data from all cells in the current user's charging process to establish a first data set. At the same time, call the factory charging data of each cell to establish a second data set as a benchmark;

[0008] Compare and analyze the first data set with the second data set, extract key features reflecting the dynamic changes in the charging performance of the battery cell from the compared data, analyze the extracted features in the monitoring window, and quantify the extracted dynamic change features;

[0009] The quantified feature data is input into the pre-trained deep learning model to predict the current state and future performance of the battery cell;

[0010] Based on the prediction results of the deep learning model, all cells are divided into normal charging units and low-performance charging units;

[0011] For low-performance charging units, set a fixed low charging power for real-time charging to reduce the risk of heating and delay further aging of the battery cells. During the charging process, the battery cells corresponding to the low-performance charging units are marked and stored in the database for subsequent maintenance or replacement plans;

[0012] For the cells corresponding to the normal charging units, fuzzy logic control is used to dynamically allocate the charging power in real time based on the charging performance change characteristics of the cells. During the dynamic power allocation process, higher charging power is preferentially allocated to high-performance cells to maximize the overall charging efficiency of the battery pack.

[0013] Preferably, the identification criteria for high-frequency users are based on the frequency of use of the device, the charge and discharge cycle, and the battery consumption characteristics and behavior patterns.

[0014] Preferably, real-time charging data is collected from all battery cells during the charging process of the current user to establish a first data set. The specific steps are as follows:

[0015] During the dynamic charging process of the lithium battery, first collect charging data from each battery cell in real time;

[0016] After the data collection is completed, preprocess the original data to ensure the accuracy, continuity, and consistency of the data;

[0017] After the data cleaning and preprocessing are completed, classify and structure the real-time charging data of all battery cells to finally form a first data set. The data set is constructed in a hierarchical storage manner, that is, multi-dimensional classification is performed according to time, device, and battery cell number for subsequent dynamic analysis and feature extraction. At the same time, compare and correlate the real-time data with the historical data to establish a time series model of the charging data to better capture the change trend of the battery cell performance.

[0018] Preferably, extract the key features reflecting the dynamic changes in the charging performance of the battery cell. Among them, the extracted key features include the growth ratio of the internal resistance of the battery cell compared to the initial internal resistance after multiple charge and discharge cycles and the offset of the charging time in the same battery charge interval compared to when it left the factory. Under the monitoring window, analyze the growth ratio of the internal resistance of the battery cell compared to the initial internal resistance after multiple charge and discharge cycles and the offset of the charging time in the same battery charge interval compared to when it left the factory, and generate an internal resistance growth quantization value and a charging time drift quantization value respectively. Quantify the degree of increase in the internal resistance of the battery cell after multiple charge and discharge cycles through the internal resistance growth quantization value, reflecting the decrease in the electrochemical activity and the reduction in the energy conversion efficiency of the battery cell. Quantify the offset of the charging time in the same battery charge interval through the charging time drift quantization value, reflecting the impact on the charging speed during the charging process.

[0019] Preferably, after obtaining the quantified internal resistance growth quantization value and charging time drift quantization value, input the quantified internal resistance growth quantization value and charging time drift quantization value into a pre-trained deep learning model to generate a performance degradation index, and predict the current state and future performance of the battery cell through the performance degradation index.

[0020] Preferably, compare and analyze the performance degradation index generated when predicting the current state and future performance of the battery cell under the monitoring window with a pre-set performance degradation index reference threshold, and classify the battery cells in the lithium battery. The specific classification steps are as follows:

[0021] If the performance degradation index is less than the performance degradation index reference threshold, then classify the battery cell as a normal charging unit. The battery cells in the normal unit have a stable state and can store energy efficiently.

[0022] If the performance degradation index is greater than or equal to the performance degradation index reference threshold, the battery cell is classified as a low-performance charging unit, and there are obvious degradations or abnormalities in the battery cells in the low-performance unit.

[0023] Preferably, under the monitoring window, the specific steps for analyzing the growth ratio of the internal resistance of the battery cell after multiple charge and discharge cycles compared to the initial internal resistance to generate an internal resistance growth quantization value are as follows:

[0024] Within the monitoring window, collect the real-time internal resistance data of each battery cell during multiple charge and discharge cycles , and call its initial internal resistance at the time of factory as the reference value, where represents the internal resistance of the battery cell measured at the t-th charging cycle, is the initial internal resistance of the battery cell;

[0025] To quantify the increase in the internal resistance after multiple charge and discharge cycles, calculate the internal resistance growth ratio for each cycle. The formula is as follows: , where in the formula, is the internal resistance growth ratio at the t-th cycle, represents the internal resistance of the battery cell measured at the (t - 1)-th charging cycle, is the correction term for the influence of the ambient temperature on the internal resistance growth, where is the ambient temperature at the t-th cycle, is the ambient reference temperature;

[0026] Calculate the cumulative internal resistance growth quantization value within the monitoring window according to the internal resistance growth ratio of each cycle. The calculation expression of the internal resistance growth quantization value is: , where in the formula, is the internal resistance growth quantization value, which is used to quantify the overall internal resistance change trend of the battery cell within the monitoring window. n is the total number of charge and discharge cycles within the monitoring window, is the weight coefficient, which is used to reflect the importance difference between cycles, is the periodic correction term, which captures the fluctuation characteristics of the internal resistance over time.

[0027] Preferably, under the monitoring window, the specific steps for analyzing the offset of the charging time compared to the factory time in the same state of charge range to generate a charging time drift quantization value are as follows:

[0028] During the dynamic charging process of the lithium battery, define and extract the charging time offset in the same state of charge range. Assume that the charging process of each battery cell can be divided into multiple fixed state of charge ranges. For each range, record the actual charging time during the current charging process, and compare it with the charging time of the battery cell at the factory in the same state of charge range to calculate the charging time offset. The calculation expression is: , where in the formula, and respectively represent the start point and the end point of the power range, represents the actual time required to charge from the start point to the end point during the current charging cycle, is the time required to charge from the start point to the end point when the battery cell leaves the factory, is the charging time offset;

[0029] After obtaining the charging time offsets for each power range, the next step is to calculate the charging time drift rate. The drift rate is defined as the ratio of the charging time offset for each range to the charging time when leaving the factory, representing the degree of change in charging performance. The calculation expression is: , where, is the charging time drift rate from the start point to the end point , used to measure the charging time offset degree for each power range;

[0030] In order to comprehensively quantify the charging time drift quantization value during the entire charging process, a non-linear weighting function is used to weight and calculate the charging time drift rates for all power ranges, so that the contribution of the charging time drift rate of the power range to the charging time drift quantization value has an exponential growth relationship with the drift amplitude. The calculation expression is: , where, is the charging time drift quantization value, comprehensively quantifying the severity of time drift during the entire charging process, is the charging time drift rate for the i-th power range from the start point to the end point , and are non-linear weighting coefficients, controlling the amplification degree of the charging time drift quantization value, controlling the weight influence of the large charging time drift rate on the overall charging time drift quantization value, and m represents the total number of power ranges within the monitoring window.

[0031] Preferably, for the battery cells corresponding to the normal charging unit, fuzzy logic control is used to dynamically allocate the charging power in real time. The specific steps are as follows:

[0032] Calculate the charging priority according to the relationship between the performance degradation index of each battery cell and the performance degradation index reference threshold. The calculation expression is: , where, is the charging priority of the j-th battery cell, and the higher the value, the higher the priority, is the performance degradation index of the j-th battery cell, reflecting the current state of the battery cell, is the performance degradation index reference threshold;

[0033] After evaluating the priorities, fuzzy logic is used to determine the charging power distribution weights for each cell. The fuzzy logic rules combine the performance priorities of the cells with the actual requirements, and the formula is as follows: , is the charging weight for the j-th cell, which determines the power ratio to be allocated. is the fuzzy logic control parameter, which is used to amplify the charging weight of the high-priority cells. N is the total number of cells in the normal charging unit. represents the charging priority of the k-th cell;

[0034] According to the charging weights of each cell, the actual charging power is dynamically allocated. The total power is the maximum charging power of the battery pack, and the dynamic allocation formula is as follows: , where is the charging power allocated to the j-th cell, is the total available charging power of the battery pack.

[0035] In the above technical solution, the technical effects and advantages provided by the present invention are as follows:

[0036] Through real-time data analysis and deep learning prediction, the present invention accurately identifies the performance status of the cells, divides the cells into normal charging units and low-performance charging units, and adopts targeted charging strategies to significantly improve the charging efficiency. For low-performance units, a fixed low-power charging is adopted to reduce the risk of overheating and avoid the aggravation of cell degradation caused by high power, thereby effectively delaying cell aging. For normal charging units, the dynamic allocation of charging power is achieved through fuzzy logic control, and higher charging power is preferentially allocated to high-performance cells to maximize the charging efficiency, shorten the charging time, and improve the user experience. Secondly, by specially marking the low-performance cells and storing them in the database, accurate basis is provided for subsequent maintenance and replacement, ensuring the long-term stability and safety of the battery pack, effectively solving the problems of energy waste and safety hazards in traditional synchronous charging, significantly improving the charging speed, life, and overall reliability of the battery pack, and providing a more efficient and safe charging experience for high-frequency users. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0038] Figure 1 is the method flow chart of the lithium battery dynamic charging method based on user behavior and environmental conditions of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0039] Example embodiments will now be described more fully with reference to the accompanying drawings. However, the example embodiments can be implemented in various forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the example embodiments to those skilled in the art.

[0040] The present invention provides a Figure 1 lithium battery dynamic charging method based on user behavior and environmental conditions as shown below, including the following steps:

[0041] For users with high-frequency usage, during the dynamic charging process of the lithium battery, first collect real-time charging data from all the battery cells during the current user's charging process to establish a first data set. At the same time, call the charging data of each battery cell at the time of factory to establish a second data set as a benchmark;

[0042] The criteria for identifying high-frequency usage users are usually based on the usage frequency of the device, the charge-discharge cycle, and the consumption characteristics and behavior patterns of the battery. Specifically, high-frequency usage users refer to users who frequently use the device daily or multiple times and have a large power consumption. Their manifestations include:

[0043] 1. The average daily usage duration exceeds a set threshold (such as 8 hours);

[0044] 2. The charging frequency is relatively high, such as charging once a day or every 1-2 days;

[0045] 3. The average charge-discharge cycle times of the device are much higher than other user groups within a certain period of time (such as exceeding 30-50 cycles per month);

[0046] 4. The proportion of deep discharge is relatively high, often discharging the battery to less than 20% before charging;

[0047] 5. The operating scenarios of the device are complex, such as a mobile phone being used for frequent calls, video conferences, or an electric vehicle for long-distance and high-frequency driving.

[0048] Through these metrics, the system can identify high-frequency usage users, thereby adjusting the charging strategy accordingly to ensure efficient battery management and optimized user experience.

[0049] Collect real-time charging data from all the battery cells during the current user's charging process to establish a first data set. The specific steps are as follows:

[0050] During the dynamic charging process of a lithium battery, the system first needs to collect charging data from each battery cell in real time. This process is achieved through multiple sensors installed in the Battery Management System (BMS). These sensors can monitor and record the key parameters of each battery cell, including voltage, current, temperature, internal resistance, etc. These data are usually collected at a high frequency to ensure that the changes in the state during the charging process can be reflected in real time. To ensure the accuracy of the data, the system needs to use the self-calibration mechanism of the sensors to correct possible errors in real time, ensuring that the data reflects the true state of the battery cell.

[0051] After the data collection is completed, the system will preprocess these raw data. This process includes noise filtering, outlier handling, and time alignment of the sensor data. Since the battery cells may be affected by external interferences during operation, such as electromagnetic interference or accidental failures of the sensors, the raw data collected may contain abnormal fluctuations or deviations. The system uses preset data cleaning algorithms to eliminate illogical abnormal data and ensure the accuracy, continuity, and consistency of the data. In addition, since the data collection frequencies of different sensors may be different, the system needs to synchronize the data in time to ensure that the data of each battery cell can be compared within the same time window. This step is crucial because the relationships between parameters such as voltage, current, and temperature during the charging process need to be captured at accurate time points.

[0052] After the data cleaning and preprocessing are completed, the system will classify and structure the real-time charging data of all battery cells to finally form the first data set. The construction of this data set takes into account the individual characteristics and real-time states of the battery cells, including the capacities, internal resistances, voltage stabilities, etc. of different battery cells, and groups and records them according to the battery cell numbers. The data set is constructed using a hierarchical storage method, that is, multi-dimensional classification is carried out according to time, device, and battery cell number for subsequent dynamic analysis and feature extraction. In addition, the system will also compare and correlate the real-time data with the historical data to establish a time series model of the charging data, so as to better capture the change trend of the battery cell performance. After the establishment is completed, the data will be securely stored in the local database or the cloud, and a regular backup and update mechanism will be set up to ensure the long-term availability and security of the data.

[0053] The factory data of each battery cell is usually recorded during the production stage and reflects the initial performance of the battery cell. Through real-time monitoring and recording, the system can build a historical archive of the battery cell state, realizing the continuity and accuracy of the data. The Battery Management System (BMS) in the prior art already supports multiple sensors to collect the key parameters of the battery cells and stores the factory data in the cloud or the local database.

[0054] Compare and analyze the first data set with the second data set, extract the key features reflecting the dynamic changes in the charging performance of the battery cell from the compared data, analyze the extracted features under the monitoring window, and quantify the extracted dynamic change features;

[0055] Input the quantified feature data into a pre-trained deep learning model to predict the current state and future performance of the battery cell;

[0056] Extract the key features reflecting the dynamic changes in the charging performance of the battery cell. Among them, the extracted key features include the growth ratio of the internal resistance of the battery cell after multiple charge and discharge cycles compared with the initial internal resistance, and the offset of the charging time in the same state of charge range compared with the time at the factory. Under the monitoring window, analyze the growth ratio of the internal resistance of the battery cell after multiple charge and discharge cycles compared with the initial internal resistance and the offset of the charging time in the same state of charge range compared with the time at the factory, and generate the internal resistance growth quantization value and the charging time drift quantization value respectively. Quantify the degree of increase in the internal resistance of the battery cell after multiple charge and discharge cycles through the internal resistance growth quantization value, reflecting the decrease in the electrochemical activity and the reduction in the energy conversion efficiency of the battery cell. Quantify the offset of the charging time in the same state of charge range through the charging time drift quantization value, reflecting the impact on the charging speed during the charging process.

[0057] After obtaining the quantified internal resistance growth quantization value and the charging time drift quantization value, input the quantified internal resistance growth quantization value and the charging time drift quantization value into a pre-trained deep learning model to generate a performance degradation index, and predict the current state and future performance of the battery cell through the performance degradation index.

[0058] The pre-trained deep learning model means that before formal application, the model has been fully trained with a large amount of historical data and feature data, enabling it to identify complex patterns and make accurate predictions. In the dynamic charging management of lithium batteries, this deep learning model has undergone multiple rounds of training, can capture the dynamic changes in the charging performance of the battery cell, and predict the current state and future performance. The training process usually involves using supervised learning methods, that is, the deep learning model adjusts its internal parameters through a large amount of labeled data (such as historical charging performance data and battery cell degradation patterns) to minimize the prediction error. As the deep learning model continues to iterate, it can identify potential relationships and trends in multi-dimensional data (such as the internal resistance growth quantization value and the charging time drift quantization value), and learn how to accurately evaluate the state of the battery cell.

[0059] The deep learning model may adopt structures such as long short-term memory networks (LSTM), Transformers, or convolutional neural networks (CNN). These models are good at processing time series data and multi-variable inputs. In the application of lithium batteries, the model not only analyzes the current internal resistance growth quantization value and charging time drift quantization value, but also constructs complex time dynamic relationships by combining historical charging data. This ability enables the deep learning model to identify early performance degradation signals and predict future changes in charging performance. For example, it can predict which battery cells may experience performance bottlenecks in the next charging cycle through an exponential change trend. This prediction function is particularly important for high-frequency users, as it can timely remind the system to take preventive maintenance measures to prevent unexpected failures. Through the introduction of a pre-trained deep learning model, the charging management system has achieved data-driven intelligent management, improving both charging efficiency and ensuring the safety and long life of the battery.

[0060] The growth ratio of the internal resistance of the battery cell compared to the initial internal resistance after multiple charge and discharge cycles is relatively high, indicating that the charging efficiency of the battery cell is deteriorating. The reason is that the increase in the internal resistance of the battery cell will cause more electrical energy to be lost in the form of heat during charging, reducing the effectively stored energy. As the internal resistance increases, the electrochemical reaction activity of the battery cell decreases, and the resistance to energy transfer between the electrodes increases, resulting in a decrease in the charging efficiency of the battery cell. In addition, the increase in internal resistance makes the battery cell more likely to heat up during charging, which not only reduces the charging speed but also accelerates the aging of the battery cell, further deteriorating its performance. Therefore, a high internal resistance growth rate is a clear sign of the deterioration of the charging efficiency of the battery cell.

[0061] The specific steps for analyzing the growth ratio of the internal resistance of the battery cell compared to the initial internal resistance after multiple charge and discharge cycles to generate the internal resistance growth quantization value under the monitoring window are as follows:

[0062] Collect the real-time internal resistance data of each battery cell during multiple charge and discharge cycles within the monitoring window and call its initial internal resistance at the time of factory as the reference value, where represents the internal resistance of the battery cell measured at the t-th charging cycle, is the initial internal resistance of the battery cell, usually recorded during factory testing;

[0063] At this stage, the data structure is a time series, recording the changes in the internal resistance of the battery cell during each charge and discharge cycle.

[0064] To quantify the increase in internal resistance after multiple charge and discharge cycles, calculate the internal resistance growth ratio for each cycle. The formula is as follows: where is the internal resistance growth ratio at the t-th cycle, represents the internal resistance of the battery cell measured at the (t - 1)-th charging cycle, is the correction term for the influence of ambient temperature on internal resistance growth, where is the ambient temperature at the t-th cycle, is the ambient reference temperature;

[0065] This step reflects the non-linear influence of ambient temperature on internal resistance growth through an exponential term, avoiding errors that may be caused by a simple linear relationship. During long-term use, the performance degradation of the battery cell is not only related to the change in internal resistance but also significantly affected by temperature. Therefore, introducing a temperature correction term can improve the accuracy of the model.

[0066] Calculate the cumulative quantified value of internal resistance growth within the monitoring window according to the internal resistance growth ratio of each cycle. The calculation expression of the quantified value of internal resistance growth is: , where is the quantified value of internal resistance growth, used to quantify the overall internal resistance change trend of the battery cell within the monitoring window. n is the total number of charge-discharge cycles within the monitoring window, is the weight coefficient, used to reflect the importance difference between cycles. It is not specifically defined here and can be dynamically changed according to the actual situation to adapt to the performance changes of the battery cell under different time periods or different environmental conditions. Its main purpose is to reflect the importance difference between different cycles in the calculation of the quantified value of internal resistance growth and give a greater weight to the internal resistance change in certain time periods. For example, when the temperature of the battery cell fluctuates greatly or the charge and discharge are frequent in certain cycles, the weight of this cycle can be appropriately increased to more accurately reflect the influence of these cycles on the overall internal resistance change. In practical applications, the weight coefficient may be set according to various factors, such as the historical charging data of the battery, the usage environment, temperature conditions, etc., to ensure that the model can flexibly adapt to various charging scenarios and improve the accuracy and practicality of the quantified value of internal resistance growth. is the periodic correction term, which captures the fluctuating characteristics of internal resistance over time;

[0067] This step generates the quantified value of internal resistance growth by accumulating the internal resistance growth ratio of each cycle and weighting the periodic change. The periodic correction term enables the model to capture the non-linear change of internal resistance at different time points and further improves the accuracy of the quantification result.

[0068] From the quantization value of the internal resistance growth, it can be seen that under the monitoring window, when analyzing the growth ratio of the internal resistance of the battery cell after multiple charge-discharge cycles compared to the initial internal resistance, the larger the performance value of the internal resistance growth quantization value generated, the greater the increase in the internal resistance of the battery cell. This means that the electrochemical reaction efficiency of the battery cell decreases, the energy conversion efficiency during charging becomes worse, resulting in more energy being lost in the form of heat, thus seriously affecting the charging efficiency and the life of the battery cell. On the contrary, if the performance value of the internal resistance growth quantization value is small, it indicates that the increase in the internal resistance of the battery cell is limited, the performance and energy conversion efficiency of the battery cell remain relatively stable, the loss during the charging process is small, and the decrease in the charging efficiency is not obvious. The magnitude of the internal resistance growth quantization value provides a reliable reference for the battery management system, enabling the system to identify the battery cells with more serious degradation and take timely measures to ensure the overall performance and safety of the battery.

[0069] A higher offset of the charging time in the same charge level range compared to when it left the factory indicates that the charging efficiency of this battery cell is deteriorating. This is because within the same charge level range, the battery cell takes longer to be fully charged, meaning that the process of converting electrical energy is no longer as efficient as when it left the factory. The main reason may be that the active materials inside the battery cell have degraded, resulting in a slower movement speed of lithium ions between the electrodes, or side reactions have occurred on the electrode surface, reducing the effective area of the electrode and hindering the normal electrochemical reaction. These phenomena will all lead to a decrease in the reaction efficiency during the charging process, ultimately manifested as an extension of the charging time.

[0070] Under the monitoring window, the specific steps for analyzing the offset of the charging time in the same charge level range compared to when it left the factory to generate the charging time drift quantization value are as follows:

[0071] During the dynamic charging process of the lithium battery, define and extract the charging time offset in the same charge level range. Assume that the charging process of each battery cell can be divided into multiple fixed charge level ranges. For each range, record the actual charging time during the current charging process and compare it with the charging time of the battery cell in the same charge level range when it left the factory to calculate the charging time offset. The calculation formula is: , where and respectively represent the starting point and the ending point of the charge level range, with the unit of mAh or percentage (% charge level), represents the actual time required to charge from the starting point to the ending point during the current charging cycle, is the time required for the battery cell to charge from the starting point to the ending point when it left the factory, is the charging time offset;

[0072] This step quantifies the offset of the charging time within each state of charge range, preparing for subsequent analysis.

[0073] After obtaining the charging time offsets for each state of charge range, the next step is to calculate the charging time drift rate. The drift rate is defined as the ratio of the charging time offset for each range to the charging time at the factory, representing the degree of change in charging performance. The calculation expression is: , where is the charging time from the starting point to the ending point The charging time drift rate, which measures the degree of charging time offset for each state of charge range;

[0074] The charging time drift rate quantifies the degree of change between the actual charging time and the ideal charging time. A higher drift rate indicates a significant prolongation of the charging time, reflecting a significant decrease in the charging efficiency of the battery cell within that state of charge range.

[0075] To comprehensively quantify the charging time drift quantization value throughout the charging process, a non - linear weighting function is used to weight - calculate the charging time drift rates for all state of charge ranges, such that the contribution of the charging time drift rate of the state of charge range to the charging time drift quantization value has an exponential growth relationship with the drift amplitude. The calculation expression is: , where is the charging time drift quantization value, comprehensively quantifying the severity of the time drift throughout the charging process, is the charging time drift rate for the i - th state of charge range from the starting point to the ending point , and are non - linear weighting coefficients, controlling the amplification degree of the charging time drift quantization value, controlling the weight influence of the large charging time drift rate on the overall charging time drift quantization value, and m represents the total number of state of charge ranges within the monitoring window.

[0076] From the charging time drift quantization value, it can be seen that under the monitoring window, the larger the performance value of the charging time drift quantization value generated after analyzing the offset of the charging time for the same state of charge range compared to the factory, the more severe the efficiency decline of the battery cell during charging, that is, the battery cell requires significantly more time to complete charging within the same state of charge range. This may reflect the degradation of the internal materials of the battery cell, the reduction of activity, or the deterioration of the electrochemical reaction efficiency. Conversely, the smaller the value of the charging time drift quantization value, the smaller the change in the charging time compared to the factory, the better the performance of the battery cell is maintained, and the decline in charging efficiency is not obvious. Therefore, the charging time drift quantization value is an important indicator for quantifying the dynamic change of the battery cell's charging performance, and can intuitively judge the current state of the battery cell and the health level of its charging efficiency.

[0077] The deep learning model is not specifically limited herein, and it can be used to implement the quantization value of the internal resistance increase and the quantization value of the charging time drift to perform comprehensive analysis to generate a performance degradation index Any deep learning model that can do this is acceptable. To implement the technical solution of the present invention, the present invention provides a specific implementation method; the performance degradation index The calculation formula for generation is: , where in the formula, , are respectively the preset proportionality coefficients of the quantization value of the internal resistance increase and the quantization value of the charging time drift , and , are both greater than 0.

[0078] It can be seen from the performance degradation index that under the monitoring window, the larger the value of the performance value of the quantization value of the internal resistance increase generated by analyzing the growth ratio of the internal resistance of the battery cell compared to the initial internal resistance after multiple charge and discharge cycles, and the larger the value of the performance value of the quantization value of the charging time drift generated by analyzing the offset of the charging time in the same power range compared to when it left the factory, that is, the larger the value of the performance degradation index generated when predicting the current state and future performance of the battery cell under the monitoring window, the more serious the performance degradation of the battery cell, and vice versa, it indicates that the performance of the battery cell is maintained better.

[0079] According to the prediction results of the deep learning model, all battery cells are divided into normal charging units and low-performance charging units;

[0080] Compare and analyze the performance degradation index generated when predicting the current state and future performance of the battery cell under the monitoring window with a preset performance degradation index reference threshold, and divide the battery cells in the lithium battery. The specific division steps are as follows:

[0081] If the performance degradation index is less than the performance degradation index reference threshold, then divide the battery cell into a normal charging unit. The battery cells in the normal unit are in a stable state and can store energy efficiently.

[0082] If the performance degradation index is greater than or equal to the performance degradation index reference threshold, then divide the battery cell into a low-performance charging unit. There are obvious degradations or abnormalities in the battery cells in the low-performance unit.

[0083] This kind of unit division helps the system to adopt different charging strategies for different types of battery cells to ensure that the normal unit and the low-performance unit do not interfere with each other. The accuracy of state recognition is further enhanced through deep learning prediction.

[0084] For low-performance charging units, a fixed low charging power is set for real-time charging to reduce the risk of overheating and delay the further aging of the battery cells. During the charging process, the battery cells corresponding to the low-performance charging units are specially marked and stored in a database for subsequent maintenance or replacement planning.

[0085] For the battery cells corresponding to the identified low-performance charging units, specific charging strategies are adopted to ensure safety and delay the degradation of the battery cells. During the charging process, the battery cells of these low-performance charging units are set to a fixed low charging power to slow down the charging speed, thereby reducing the risk of overheating. Since low-performance battery cells are more likely to overheat under high-power charging, this low-power charging method not only reduces the risk of temperature rise but also effectively avoids the phenomenon of accelerated aging and extends the remaining life of the battery cells. At the same time, the system will specially mark the battery cells of low-performance units and record their status information, charging history, etc. in the database to form a complete data file. These data can be used in the future to judge the aging degree of the battery cells, predict the remaining life, and provide a basis for the maintenance plan. The marking information can help maintenance personnel quickly identify the battery cells that need further inspection or replacement, avoiding the accumulation of faults and causing greater problems. Through these measures, not only the charging safety is effectively guaranteed, but also detailed records are provided for future replacement and maintenance, realizing the intelligence and maintainability of the charging process.

[0086] For the battery cells corresponding to normal charging units, based on the charging performance change characteristics of the battery cells, fuzzy logic control is used to dynamically allocate the charging power in real time. During the dynamic power allocation process, higher charging power is preferentially allocated to high-performance battery cells to maximize the overall charging efficiency of the battery pack.

[0087] For the battery cells corresponding to normal charging units, fuzzy logic control is used to dynamically allocate the charging power in real time. The specific steps are as follows:

[0088] The charging priority is calculated according to the relationship between the performance degradation index of each battery cell and the reference threshold of the performance degradation index. The calculation formula is: , where is the charging priority of the j-th battery cell, and the higher the value, the higher the priority. is the performance degradation index of the j-th battery cell, reflecting the current state of the battery cell. is the reference threshold of the performance degradation index, used to judge whether the battery cell is in an efficient state.

[0089] The charging priority is obtained by comparing the current performance with the ideal state. The battery cell with a higher charging priority requires more charging power to be allocated to maximize the efficiency.

[0090] After evaluating the priorities, fuzzy logic is used to determine the charging power allocation weights for each cell. The fuzzy logic rules combine the performance priorities of the cells with the actual demands, and the formula is as follows: , is the charging weight of the j-th cell, which determines the power ratio to be allocated, is the fuzzy logic control parameter, which is used to amplify the charging weight of the high-priority cells. N is the total number of cells in the normal charging unit, represents the charging priority of the k-th cell;

[0091] This step applies exponential weighting to the priorities, ensuring that high-performance cells receive higher weights while the weights of low-priority cells are suppressed, achieving preferential allocation of charging resources.

[0092] According to the charging weights of each cell, the actual charging power is dynamically allocated. The total power is the maximum charging power of the battery pack, and the dynamic allocation formula is as follows: , where is the charging power allocated to the j-th cell, is the total available charging power of the battery pack;

[0093] Through this allocation formula, the actual charging power of each cell is dynamically adjusted based on its charging priority weight to ensure optimal allocation of resources, enabling high-efficiency cells to receive higher power support and improving the overall charging efficiency.

[0094] Adopting fuzzy logic control to perform real-time dynamic allocation of the charging power of the cells in the normal charging unit and preferentially allocating higher charging power to high-performance cells can effectively improve the overall charging efficiency of the battery pack. This strategy brings multiple benefits:

[0095] First of all, the charging resources can be flexibly allocated according to the state of the cells, avoiding wasting too much power on cells with lower charging efficiency, thereby reducing energy loss; secondly, by preferentially charging high-performance cells, the total charging time can be shortened, improving the user experience; finally, this intelligent power management helps to extend the life of the cells because low-performance cells receive low-power charging, and the heating and aging speeds are slowed down, and the stability and safety of the entire battery pack are also improved.

[0096] Therefore, this dynamic allocation strategy not only improves the charging efficiency but also takes into account safety and life extension.

[0097] Through real-time data analysis and deep learning prediction, the present invention accurately identifies the performance status of battery cells, divides the battery cells into normal charging units and low-performance charging units, and adopts targeted charging strategies to significantly improve the charging efficiency. For low-performance units, fixed low-power charging is adopted to reduce the risk of overheating and avoid high power exacerbating the degradation of battery cells, thus effectively delaying the aging of battery cells. For normal charging units, dynamic allocation of charging power is achieved through fuzzy logic control, preferentially allocating higher charging power to high-performance battery cells to maximize the charging efficiency, shorten the charging time, and improve the user experience. At the same time, the system specially marks the low-performance battery cells and stores them in the database, providing accurate basis for subsequent maintenance and replacement to ensure the long-term stability and safety of the battery pack. This solution effectively solves the problems of energy waste and potential safety hazards in traditional synchronous charging, significantly improves the charging speed, lifespan, and overall reliability of the battery pack, and provides a more efficient and safe charging experience for high-frequency users.

[0098] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the actual situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0099] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claimed rights.

[0100] Only some exemplary embodiments of the present invention have been described by way of illustration above. Undoubtedly, for those of ordinary skill in the art, the described embodiments can be modified in various different ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the protection scope of the claims of the present invention.

Claims

1. A dynamic charging method for lithium batteries based on user behavior and environmental conditions, characterized in that: The following steps are involved: For high-frequency users, during the dynamic charging process of lithium batteries, real-time charging data is first collected from all cells in the current user's charging process to establish a first data set. At the same time, the factory charging data of each cell is called to establish a second data set as a benchmark. Compare and analyze the first data set with the second data set, extract key features reflecting the dynamic changes of the charging performance of the battery cell from the compared data, analyze the extracted features under the monitoring window, and quantify the extracted dynamic change features; Extract key features that reflect the dynamic changes in the charging performance of the battery cell. The extracted key features include the growth rate of the battery cell internal resistance after multiple charge and discharge cycles compared to the initial internal resistance and the offset of the charging time in the same power range compared to the factory time. Input the quantified feature data into a pre-trained deep learning model to predict the current state and future performance of the battery cell; Based on the prediction results of the deep learning model, all cells are divided into normal charging units and low-performance charging units; For low-performance charging units, a fixed low charging power is set for real-time charging to reduce the risk of heating and delay further aging of the battery cells. During the charging process, the battery cells corresponding to the low-performance charging units are marked and stored in the database for subsequent maintenance or replacement plans; For the cells corresponding to the normal charging units, fuzzy logic control is used to dynamically allocate the charging power in real time based on the charging performance change characteristics of the cells. During the dynamic power allocation process, higher charging power is preferentially allocated to high-performance cells to maximize the overall charging efficiency of the battery pack. Under the monitoring window, the growth ratio of the internal resistance of the battery cell after multiple charge and discharge cycles compared to the initial internal resistance and the offset of the charging time in the same power range compared to the factory are analyzed, and the internal resistance growth quantification value and the charging time drift quantification value are generated respectively. The internal resistance growth quantification value is used to quantify the degree of increase in the internal resistance of the battery cell after multiple charge and discharge cycles, reflecting the decrease in the electrochemical activity and energy conversion efficiency of the battery cell. The charging time drift quantification value is used to quantify the offset of the charging time in the same power range, reflecting the impact on the charging speed during the charging process; After obtaining the quantized internal resistance growth quantization value and the charging time drift quantization value, the quantized internal resistance growth quantization value and the charging time drift quantization value are input into a pre-trained deep learning model to generate a performance degradation index, and the current state and future performance of the battery cell are predicted through the performance degradation index; The performance degradation index generated when predicting the current state and future performance of the battery cell under the monitoring window is compared and analyzed with the pre-set performance degradation index reference threshold, and the battery cells in the lithium battery are divided. The specific division steps are as follows: If the performance degradation index is less than the performance degradation index reference threshold, the battery cell is classified as a normal charging unit, and the battery cell in the normal unit is in a stable state and can store energy efficiently; If the performance degradation index is greater than or equal to the performance degradation index reference threshold, the battery cell is classified as a low-performance charging unit, and the battery cell in the low-performance unit has obvious degradation or abnormality; For the cells corresponding to the normal charging units, fuzzy logic control is used to dynamically allocate the charging power in real time. The specific steps are as follows: The charging priority is calculated based on the relationship between the performance degradation index of each battery cell and the performance degradation index reference threshold. The calculation expression is: , where is the charging priority of the jth battery cell. The higher the value, the higher the priority. is the performance degradation index of the jth battery cell, reflecting the current state of the battery cell. is the reference threshold of the performance degradation index; After evaluating the priority, fuzzy logic is used to determine the charging power allocation weight for each battery cell. The fuzzy logic rule combines the performance priority of the battery cell with the actual demand. The formula is as follows: , is the charging weight of the jth battery cell, which determines the proportion of power allocated. is a fuzzy logic control parameter used to amplify the charging weight of high-priority cells. N is the total number of normal charging unit cells. Indicates the charging priority of the kth battery cell; The actual charging power is dynamically allocated according to the charging weight of each battery cell. The total power is the maximum charging power of the battery pack. The dynamic allocation formula is as follows: , where is the charging power allocated to the jth battery cell, is the total available charging power of the battery pack.

2. The method for dynamic charging of lithium batteries based on user behavior and environmental conditions according to claim 1, characterized in that: The criteria for identifying high-frequency users are based on the frequency of device usage, charge and discharge cycles, and battery consumption characteristics and behavior patterns.

3. The method for dynamic charging of lithium batteries based on user behavior and environmental conditions according to claim 1, characterized in that: Collect real-time charging data from all cells in the current user's charging process to establish a first data set. The specific steps are as follows: During the dynamic charging process of lithium batteries, charging data is first collected from each battery cell in real time; After completing data collection, pre-process the raw data to ensure the accuracy, continuity and consistency of the data; After data cleaning and preprocessing, the real-time charging data of all battery cells will be classified and structured to form the first data set. The data set is constructed in a hierarchical storage manner, that is, multi-dimensional classification by time, equipment and battery cell number to facilitate subsequent dynamic analysis and feature extraction. At the same time, the real-time data is compared and associated with the historical data to establish a time series model of the charging data to better capture the changing trend of battery cell performance.

4. The method for dynamic charging of lithium batteries based on user behavior and environmental conditions according to claim 1, characterized in that: Under the monitoring window, the specific steps for analyzing the growth ratio of the cell internal resistance after multiple charge and discharge cycles compared to the initial internal resistance to generate the internal resistance growth quantitative value are as follows: Collect real-time internal resistance data of each battery cell during multiple charge and discharge cycles within the monitoring window , and call its initial internal resistance at the factory as the reference value, where represents the internal resistance of the battery cell measured at the tth charging cycle, is the initial internal resistance of the battery cell; In order to quantify the increase in internal resistance after multiple charge and discharge cycles, the internal resistance growth rate of each cycle is calculated as follows: , where is the internal resistance growth rate of the tth cycle, represents the internal resistance of the battery cell measured at the t-1th charging cycle, is the correction term for the internal resistance growth due to ambient temperature, where is the ambient temperature at the tth cycle, is the ambient reference temperature; The accumulated internal resistance growth quantitative value in the monitoring window is calculated according to the internal resistance growth ratio of each cycle. The calculation expression of the internal resistance growth quantitative value is: , where is the quantitative value of internal resistance growth, which is used to quantify the overall internal resistance change trend of the battery cell within the monitoring window. n is the total number of charge and discharge cycles within the monitoring window. is the weight coefficient, which is used to reflect the importance difference between cycles. is a periodic correction term that captures the fluctuation characteristics of internal resistance over time.

5. The method for dynamic charging of lithium batteries based on user behavior and environmental conditions according to claim 1, characterized in that: In the monitoring window, the specific steps for analyzing the offset of the charging time in the same power range compared with the factory time to generate the charging time drift quantification value are as follows: In the dynamic charging process of lithium batteries, the charging time offset of the same power interval is defined and extracted. Assume that the charging process of each battery cell can be divided into multiple fixed power intervals. For each interval, record the actual charging time in the current charging process and compare it with the charging time of the battery cell in the same power interval when it leaves the factory. Calculate the charging time offset. The calculation expression is: , where and They represent the starting point and end point of the power range respectively. Indicates that the current charging cycle starts from the starting point Charging to the end The actual time required, The battery cell is from the starting point when it leaves the factory. Charging to the end The time required, is the charging time offset; After obtaining the charging time offset of each power interval, the next step is to calculate the charging time drift rate. The drift rate is defined as the ratio of the charging time offset of each interval to the charging time at the factory, which indicates the degree of change in charging performance. The calculation expression is: , where From the starting point Charging to the end Charging time drift rate, used to measure the degree of charging time deviation in each power interval; In order to comprehensively quantify the charging time drift value in the entire charging process, a nonlinear weighting function is used to weight the charging time drift rate of all power intervals, so that the contribution of the charging time drift rate of the power interval to the charging time drift quantification value is exponentially related to the drift amplitude. The calculation expression is: , where It is the quantitative value of charging time drift, which comprehensively quantifies the severity of time drift during the entire charging process. For the i-th power interval from the starting point Charging to the end The charge time drift rate, and Nonlinear weighting coefficients, Controls the amplification degree of the charge time drift quantization value, Control the weighted influence of the large charging time drift rate on the quantized value of the overall charging time drift, and m represents the total number of power intervals in the monitoring window.

Images