A battery management system with active monitoring function

Abstract

The application relates to the technical field of battery management, and discloses a battery management system with an active monitoring function, which comprises a processor, a battery state monitoring module, an active balancing control module, a performance evaluation module and a fault early warning module. The battery state monitoring module collects various parameters of the battery at different stages and generates an abnormal signal by using a dynamic threshold algorithm. The active balancing control module adjusts a balancing strategy based on a fuzzy logic algorithm, and a fuzzy rule base can be updated. The performance evaluation module tracks and evaluates a balancing process, analyzes balancing efficiency and quality, and optimizes the balancing strategy accordingly. The fault early warning module displays and warns abnormal signals in stages, and the processor is further connected to an aging feature extraction module and a thermal runaway prediction module, which realize battery life prediction and thermal runaway early warning, respectively. The application realizes comprehensive battery monitoring, efficient balancing control, accurate performance evaluation and reliable fault early warning, and improves the safety and service life of the battery.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of battery management, in particular to a battery management system with active monitoring function. BACKGROUND

[0002] In today's energy field, batteries as important energy storage devices are widely used in electric vehicles, energy storage systems, portable electronic devices and other fields. With the increasing demand for battery performance and safety in these application scenarios, the importance of battery management system (BMS) is increasingly prominent.

[0003] The traditional battery management system mostly uses fixed threshold monitoring method when monitoring the battery state. For example, when monitoring the battery voltage, only a fixed upper and lower threshold is set, and when the battery voltage exceeds this range, it is judged as abnormal. However, the normal working parameter range of the battery will change in different working stages (charging, discharging, standing) and different use environments. Taking electric vehicles as an example, when charging in low temperature environment, the normal charging voltage range of the battery will be different from that in normal temperature environment. The fixed threshold monitoring method cannot adapt to this change, which is easy to cause misjudgment or omission, cannot find the potential problems of the battery in time, and affects the safety and service life of the battery.

[0004] In terms of battery balancing, the traditional balancing strategy is often simple, usually based on fixed voltage difference for balancing control. This way does not fully consider the influence of other factors such as battery current and temperature on balancing effect. In actual application, the imbalance of the battery is not only reflected in the voltage difference, but also closely related to the internal resistance and self-discharge rate of the battery. Moreover, different battery packs have different balancing needs in different use stages. For example, the balancing needs of new battery packs and aged battery packs are quite different. The simple fixed balancing strategy cannot dynamically adjust the balancing strategy according to the real-time state of the battery, resulting in low balancing efficiency and unable to effectively improve the overall performance of the battery pack.

[0005] For battery aging prediction and life evaluation, the traditional method mainly relies on empirical formula or simple statistical model. These methods often only consider part of the parameters of the battery, such as cycle number or capacity attenuation, ignoring the factors such as battery resistance change and temperature history which have important influence on battery life. This makes the accuracy of battery life prediction poor, and cannot provide reliable battery replacement or maintenance suggestions for users. In the field of electric vehicles, if the battery life cannot be accurately predicted, it may lead to insufficient power during driving, affecting user experience, and even causing safety problems.

[0006] In terms of fault early warning, the traditional battery management system has imperfect early warning function, which can only give simple prompt for some obvious faults, and lacks hierarchical processing and accurate early warning of faults. When potential fault hazards occur in the battery, the user cannot be timely and accurately informed of the severity of the fault and the possible influence range, so the user cannot take corresponding measures in advance, thereby increasing the risk of serious faults of the battery.

[0007] With the continuous expansion of battery application scenarios and the rapid development of battery technology, the limitations of traditional battery management systems are increasingly obvious. In order to meet the requirements of safety, reliability and high performance of modern battery applications, it is urgent to develop a system that can comprehensively, real-time and accurately monitor and manage the battery, which is also the research background and starting point of the present application. SUMMARY

[0008] The present application aims to provide a battery management system with active monitoring function to solve the problems raised in the background art.

[0009] To achieve the above-mentioned purpose, the present application provides the following technical solution: a battery management system with active monitoring function, the system comprising a processor, a battery state monitoring module, an active balancing control module, a performance evaluation module and a fault early warning module;

[0010] The battery state monitoring module monitors the working state of the battery pack in real time, and the working state of the battery pack includes the charging stage, the discharging stage and the standing stage. The single cell voltage, charging current and ambient temperature of the battery are collected in the charging stage, the single cell voltage, discharging current and internal temperature of the battery are collected in the discharging stage, and the open circuit voltage and self-discharge rate of the battery are collected in the standing stage.

[0011] The battery state monitoring module generates voltage abnormal signal, current abnormal signal or temperature abnormal signal through dynamic threshold algorithm, and sends the abnormal signal to the active balancing control module and the fault early warning module through the processor. The fault early warning module displays the abnormal signal in a hierarchical manner and triggers the warning level, and the active balancing control module dynamically adjusts the balancing strategy of the battery pack based on fuzzy logic algorithm when receiving the abnormal signal.

[0012] The performance evaluation module tracks and evaluates the balancing process of the active balancing control module, records the duration from the time when the abnormal signal is triggered to the time when the balancing is completed, and if the duration exceeds the preset balancing duration threshold, generates a balancing timeout mark and updates the balancing failure times, otherwise updates the balancing success times.

[0013] Preferably, the specific operation process of the dynamic threshold algorithm is as follows:

[0014] In the charging phase, the instantaneous change rate of the single cell voltage, the fluctuation amplitude of the charging current and the gradient value of the ambient temperature are calculated, the instantaneous change rate is subtracted from the mean value of the historical charging data to obtain a voltage dynamic deviation value, and the current dynamic deviation value and the temperature dynamic deviation value are obtained in the same way; if any dynamic deviation value exceeds the corresponding adaptive threshold range, the corresponding abnormal signal is generated;

[0015] In the discharging phase, the number of consecutive drops of the single cell voltage, the harmonic distortion rate of the discharging current and the time domain variance of the internal temperature are counted by a sliding window, the number of consecutive drops is compared with a preset safety number threshold, and if it exceeds, a voltage abnormal signal is generated, and in the same way, a current or temperature abnormal signal is generated based on the harmonic distortion rate and the time domain variance;

[0016] In the static phase, the decay slope of the open circuit voltage and the correlation coefficient of the self-discharge rate are calculated, and if the correlation coefficient is lower than a preset correlation threshold, a self-discharge abnormal signal is generated.

[0017] Preferably, the processor is in communication connection with the aging feature extraction module, the aging feature extraction module collects the cycle number, the maximum capacity attenuation rate and the internal resistance growth rate of the battery pack, normalizes the cycle number to obtain a cycle aging coefficient, and marks the product of the maximum capacity attenuation rate and the internal resistance growth rate as a composite aging index;

[0018] The cycle aging coefficient and the composite aging index are input into the aging prediction model based on the radial basis function neural network, and the remaining life prediction value of the battery pack is output; if the remaining life prediction value is lower than a preset life threshold, a life warning signal is generated and sent to the fault warning module by the processor.

[0019] Preferably, the training process of the aging prediction model comprises:

[0020] The cycle number, the capacity attenuation curve, the internal resistance change curve and the corresponding actual remaining life value of the historical battery data set are collected;

[0021] The second derivative of the capacity attenuation curve and the first derivative of the internal resistance change curve are convoluted to extract aging acceleration features;

[0022] The cycle number, the aging acceleration features and the actual remaining life value are used as training samples, and the weight parameters of the radial basis function neural network are adjusted by the particle swarm optimization algorithm until the prediction error converges within a preset accuracy range.

[0023] Preferably, the fuzzy logic algorithm of the active balancing control module specifically comprises:

[0024] The input variables are defined as voltage deviation level, current deviation level and temperature deviation level, and the output variables are balancing priority and balancing power;

[0025] A fuzzy rule base is established, and the rule form is: if the voltage deviation level is high and the temperature deviation level is medium, then the balancing priority is extremely high and the balancing power is full;

[0026] The fuzzy output is converted into accurate balancing instructions through defuzzification calculation, and balancing energy is dynamically allocated to the target battery monomer.

[0027] Preferably, the updating mechanism of the fuzzy rule base is as follows:

[0028] The actual balancing efficiency in the balancing process is collected, and the ratio of the theoretical efficiency of the balancing instruction to the actual efficiency is calculated;

[0029] If the ratio is lower than the preset efficiency threshold for three consecutive times, the adaptive adjustment of the rule base is triggered, and the membership function of the fuzzy rule is re-optimized by using the genetic algorithm.

[0030] Preferably, the processor is in communication connection with a thermal runaway prediction module, and the specific operation process of the thermal runaway prediction module includes:

[0031] During the charging or discharging stage, the surface temperature distribution, gas release rate and internal pressure change of the battery monomer are collected in real time;

[0032] The surface temperature distribution is subjected to Fourier transform to extract high-frequency component energy, and the derivative of the gas release rate and the internal pressure change is coupled to obtain a thermal runaway risk index;

[0033] If the thermal runaway risk index exceeds the dynamically adjusted critical threshold, a thermal runaway warning signal is generated and an emergency disconnection mechanism is triggered.

[0034] Preferably, the calculation method of the dynamically adjusted critical threshold is:

[0035] According to the historical working temperature of the battery pack and the current environmental humidity, a temperature-humidity coupling coefficient is calculated;

[0036] The basic critical threshold and the temperature-humidity coupling coefficient are weighted and summed to obtain a real-time updated dynamic critical threshold.

[0037] Preferably, the performance evaluation module further includes a balancing efficiency analysis submodule, and the specific analysis process is as follows:

[0038] The ratio of the number of balancing successes to the number of balancing failures in a preset period is counted, and is marked as a balancing efficiency coefficient;

[0039] The energy loss rate, balancing time and voltage convergence speed of each balancing process are extracted, and principal component analysis is performed on the three to obtain a balancing quality score;

[0040] If the equalization efficiency coefficient is lower than the first threshold value or the equalization quality score is lower than the second threshold value, an equalization strategy optimization instruction is generated and fed back to the active equalization control module.

[0041] Preferably, the specific steps of the principal component analysis include:

[0042] The energy loss rate, equalization duration and voltage convergence speed are standardized;

[0043] Eigenvalues and eigenvectors of the covariance matrix are calculated, and the first two principal components with a cumulative contribution rate exceeding 85% are selected;

[0044] The principal component scores are mapped to a two-dimensional space, the equalization quality grades are divided by the K-means clustering algorithm, and the corresponding equalization quality scores are output.

[0045] Compared with the prior art, the present application has the following beneficial effects:

[0046] The battery management system of the present application collects different key parameters such as single cell voltage, charging current and ambient temperature in the charging stage, single cell voltage, discharging current and internal temperature in the discharging stage, etc. through the battery state monitoring module in the charging, discharging and standing stages of the battery pack. The dynamic threshold algorithm can accurately determine whether the battery state is abnormal according to the characteristics of different working stages of the battery. Compared with the traditional fixed threshold monitoring method, the dynamic threshold algorithm fully considers the dynamic changes of the battery working parameters, greatly improves the accuracy of monitoring, reduces the misjudgment and omission, and provides strong support for timely discovery of potential problems of the battery.

[0047] The active equalization control module establishes a fuzzy rule base by taking the voltage deviation level, current deviation level and temperature deviation level as input variables, and taking the equalization priority and equalization power as output variables based on the fuzzy logic algorithm. This control method can comprehensively consider the influence of various factors on battery equalization, dynamically adjust the equalization strategy according to the real-time state of the battery, and improve the equalization efficiency. For example, when the battery voltage deviation is large and the temperature is in the medium range, the system will automatically increase the equalization priority and increase the equalization power, so that the battery pack can reach the equalization state faster. At the same time, the fuzzy rule base has an updating mechanism, which re-optimizes the membership function by using the genetic algorithm according to the actual equalization efficiency in the equalization process, further improves the equalization effect, and prolongs the service life of the battery pack.

[0048] The aging feature extraction module collects key parameters such as the cycle number, maximum capacity attenuation rate and internal resistance growth rate of the battery pack, and obtains a cycle aging coefficient and a composite aging index through normalization processing and composite operation. The aging prediction model based on a radial basis function neural network takes these parameters as inputs and can accurately predict the remaining life of the battery pack. Compared with traditional methods that rely on empirical formulas or simple statistical models, the present application considers more factors that affect battery life and has higher prediction accuracy. This helps users plan battery replacement or maintenance in advance, avoids equipment failure or safety hazards caused by battery life problems, and reduces use costs.

[0049] The fault warning module displays the abnormal signals sent by the battery state monitoring module in a hierarchical manner and triggers a warning level, enabling the user to clearly understand the severity of the battery fault and take appropriate measures in a timely manner. The thermal runaway prediction module collects key parameters such as the surface temperature distribution of the battery monomer, the gas release rate and the internal pressure change during the charging or discharging stage, and obtains a thermal runaway risk index through Fourier transform and coupling calculation. When the thermal runaway risk index exceeds the dynamically adjusted critical threshold, a thermal runaway warning signal is generated and an emergency disconnection mechanism is triggered, effectively preventing thermal runaway accidents and ensuring the safe operation of the battery system.

[0050] The balancing efficiency analysis submodule in the performance evaluation module obtains a balancing efficiency coefficient by calculating the ratio of the number of successful balancing to the number of failed balancing within a preset period, and obtains a balancing quality score by performing principal component analysis on parameters such as energy loss rate, balancing duration and voltage convergence speed. When the balancing efficiency coefficient or the balancing quality score is lower than the corresponding threshold, a balancing strategy optimization instruction is generated and fed back to the active balancing control module, realizing continuous optimization of the balancing strategy and further improving the overall performance and stability of the battery pack. BRIEF DESCRIPTION OF DRAWINGS

[0051] Fig. 1 The working principle diagram of the battery management system with active monitoring function described in the present application;

[0052] Fig. 2 The architecture diagram for battery aging prediction;

[0053] Fig. 3 The working principle diagram of thermal runaway prediction. DETAILED DESCRIPTION

[0054] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0055] Referring to Figs. 1-3 The present application provides a technical solution: based on a battery management system with active monitoring function, the system includes:

[0056] Battery state monitoring module: responsible for real-time monitoring of the working state of the battery pack in the charging, discharging and standing three stages. In the charging stage, the single cell voltage, charging current and ambient temperature of the battery are collected; in the discharging stage, the single cell voltage, discharging current and internal temperature are collected; in the standing stage, the open circuit voltage and self-discharge rate are collected.

[0057] Abnormal signal generation and transmission: the battery state monitoring module analyzes and processes the collected data through a dynamic threshold algorithm to generate voltage abnormal signal, current abnormal signal or temperature abnormal signal. These abnormal signals are forwarded to the active balancing control module and fault warning module through the processor.

[0058] Fault warning module: after receiving the abnormal signal, it displays it in stages and triggers the corresponding warning level according to different abnormal conditions, so that the user can learn about the abnormal state of the battery in a timely manner.

[0059] Active balancing control module: when receiving an abnormal signal, based on fuzzy logic algorithm, dynamically adjusts the balancing strategy of the battery pack to ensure that the state of each single battery in the battery pack is as consistent as possible, improving the overall performance and service life of the battery pack.

[0060] Performance evaluation module: tracks and evaluates the balancing process of the active balancing control module, records the duration from the time the abnormal signal is triggered to the time the balancing is completed. Compare the duration with the preset balancing time threshold, if it exceeds the threshold, generate a balancing timeout flag and update the number of balancing failures; if it does not exceed, update the number of balancing successes.

[0061] The present application will be further described in conjunction with Examples 1 to 5:

[0062] Example 1:

[0063] This embodiment details the specific implementation of the dynamic threshold algorithm in different working stages of the battery. Its unique role is to more accurately determine whether the battery state is abnormal. Compared with traditional fixed threshold monitoring, it can greatly improve the accuracy and reliability of monitoring. The specific method includes:

[0064] Charging stage: during the charging process, the battery state monitoring module calculates the instantaneous change rate of the single cell voltage in real time, i.e. the ratio of the change amount of the single cell voltage at a certain time to the time interval, which can be expressed by the formula: where ΔV is the change of the cell voltage in Δt time. At the same time, the fluctuation amplitude of the charging current is calculated, which is the difference between the maximum and minimum values of the charging current in a certain time period. In addition, the gradient value of the ambient temperature is calculated, that is, the change of the ambient temperature per unit distance or per unit time.

[0065] The calculated instantaneous change rate of the cell voltage is subtracted from the average of the historical charging data to obtain the voltage dynamic deviation value. Similarly, current dynamic deviation values and temperature dynamic deviation values are obtained. If any dynamic deviation value exceeds its corresponding adaptive threshold range, a corresponding abnormal signal is generated. For example, when the voltage dynamic deviation value is greater than the preset upper voltage deviation threshold value or less than the preset lower voltage deviation threshold value, a voltage abnormal signal is generated. These adaptive thresholds are dynamically adjusted according to the historical data and actual use of the battery, which can better adapt to the changes of the battery under different use conditions.

[0066] Discharge phase: In the discharge phase, the battery state monitoring module counts the number of consecutive decreases in the cell voltage through a sliding window. The sliding window is a data window with a fixed size, which slides on the data sequence over time, and only the data within the window is focused on each time. When the number of consecutive decreases in the cell voltage within the window exceeds a preset safety number threshold, a voltage abnormal signal is generated.

[0067] At the same time, the harmonic distortion rate of the discharge current, which reflects the size of the harmonic content in the discharge current, and the time domain variance of the internal temperature, which measures the dispersion degree of the internal temperature over time, are calculated. When the harmonic distortion rate of the discharge current exceeds a preset current harmonic distortion rate threshold, or the time domain variance of the internal temperature exceeds a preset temperature variance threshold, a current abnormal signal and a temperature abnormal signal are generated, respectively.

[0068] Rest phase: When the battery is at rest, the battery state monitoring module calculates the decay slope of the open circuit voltage, that is, the rate of change of the open circuit voltage over time, which is represented by the formula: where ΔV oc is the change of the open circuit voltage in Δt time. Then the correlation coefficient of the decay slope of the open circuit voltage and the self-discharge rate is calculated. If the correlation coefficient is lower than a preset correlation threshold, a self-discharge abnormal signal is generated. This process can effectively monitor the self-discharge of the battery in the resting state and timely discover potential problems.

[0069] Example 2:

[0070] This embodiment focuses on battery aging prediction related content, and its unique role is to accurately predict the remaining life of the battery pack, providing a scientific basis for users to reasonably arrange battery maintenance, replacement, etc. The specific method includes:

[0071] The aging feature extraction module is in communication connection with the processor, and the cycle number, the maximum capacity attenuation rate and the internal resistance growth rate of the battery pack are collected in real time. The cycle number is obtained by recording the number of battery charge and discharge cycles through the battery management system; the maximum capacity attenuation rate is the ratio of the difference between the current maximum capacity and the initial maximum capacity to the initial maximum capacity, i.e. C0 is the initial maximum capacity, and C n is the current maximum capacity; the internal resistance growth rate is calculated by monitoring the change of the battery internal resistance with time.

[0072] The cycle number is normalized to map the cycle number to the interval [0, 1] to obtain the cycle aging coefficient. The specific normalization formula can be: N is the current cycle number, N min is the minimum cycle number, and N max is the maximum cycle number. At the same time, the product of the maximum capacity attenuation rate and the internal resistance growth rate is marked as the composite aging index.

[0073] A historical battery data set is collected, which includes the cycle number, the capacity attenuation curve, the internal resistance change curve and the corresponding actual remaining life value. The second derivative of the capacity attenuation curve and the first derivative of the internal resistance change curve are convolved to extract the aging acceleration feature. Convolution operation can integrate the change characteristics of the two curves, and more comprehensively reflect the acceleration change in the battery aging process.

[0074] The cycle number, the aging acceleration feature and the actual remaining life value are used as training samples, and the particle swarm optimization algorithm is used to adjust the weight parameters of the radial basis function neural network. The particle swarm optimization algorithm is a kind of optimization algorithm based on swarm intelligence, which simulates the foraging behavior of bird flock to find the optimal solution. In the training process, the weight parameters are adjusted constantly until the prediction error converges within the preset accuracy range.

[0075] The trained aging prediction model takes the cycle aging coefficient and the composite aging index as input and outputs the remaining life prediction value of the battery pack. When the remaining life prediction value is lower than the preset life threshold, a life warning signal is generated, which is sent to the fault warning module by the processor to remind the user to pay attention to the battery life in time and take measures in advance.

[0076] Embodiment 3:

[0077] This embodiment details the fuzzy logic algorithm and fuzzy rule base updating mechanism of the active balancing control module, which has the unique function of intelligently and efficiently adjusting the balancing strategy according to the real-time state of the battery, improving the balancing effect and overall performance of the battery pack. The specific method includes:

[0078] The active balancing control module defines input variables as voltage deviation level, current deviation level, and temperature deviation level. Voltage deviation level is determined based on the difference between the individual cell voltage and the average voltage of the battery pack. For example, a difference greater than a certain set value indicates a high deviation level, within a certain range it indicates a medium deviation level, and less than another set value it indicates a low deviation level. Current and temperature deviation levels are determined similarly, based on the differences between the actual current and temperature and the set reference values.

[0079] The output variables are balancing priority and balancing power. A fuzzy rule base is established, with rules in the form of: if the voltage deviation level is high and the temperature deviation level is medium, then the balancing priority is extremely high and the balancing power is full. Through these rules, taking into account the voltage, current, and temperature deviations, appropriate balancing priority and balancing power are determined.

[0080] In actual operation, based on the collected real-time battery data, the voltage deviation level, current deviation level, and temperature deviation level are determined. Then, according to the fuzzy rule base, the corresponding balancing priority and balancing power are derived as fuzzy outputs. Defuzzification calculation then converts the fuzzy outputs into precise balancing commands and dynamically allocates balancing energy to the target battery cells. Defuzzification calculation can employ common methods such as the centroid method to transform the fuzzy set into specific numerical values ​​for practical control.

[0081] During the equalization process, the actual equalization efficiency is collected in real time. The actual equalization efficiency can be determined by calculating the ratio of the battery pack's performance improvement indicators after equalization (such as the degree of improvement in voltage consistency and the percentage increase in capacity utilization) to the theoretically expected performance improvement indicators. The ratio of the theoretical efficiency to the actual efficiency of the equalization command is calculated.

[0082] If this ratio falls below the preset efficiency threshold three times consecutively, it indicates that the current fuzzy rule base may not be well adapted to the actual battery conditions and needs adjustment. In this case, an adaptive adjustment of the rule base is triggered, using a genetic algorithm to re-optimize the membership functions of the fuzzy rules. The genetic algorithm simulates the inheritance, mutation, and selection mechanisms in biological evolution. Through continuous optimization of the membership functions, the fuzzy rule base can better adapt to the real-time state of the battery, improving the balancing effect.

[0083] Example 4:

[0084] This embodiment illustrates the specific working process of the thermal runaway prediction module and the calculation method of the dynamic critical threshold. Its unique function is to promptly detect the risk of battery thermal runaway, trigger the emergency chain disconnection mechanism, and ensure the safe operation of the battery system. The specific methods include:

[0085] Thermal runaway risk index calculation: During the battery charging or discharging phase, the thermal runaway prediction module collects the surface temperature distribution, gas release rate, and internal pressure change of the battery in real time. The surface temperature distribution is obtained through a temperature sensor array, which can reflect the temperature at different positions on the battery surface.

[0086] The surface temperature distribution is subjected to Fourier transform to extract the high-frequency component energy. Fourier transform is a mathematical transform that can convert time-domain signals into frequency-domain signals. The high-frequency component energy reflects the rapidly changing part of the temperature distribution, which is often related to abnormal reactions inside the battery.

[0087] At the same time, the gas release rate and the derivative of the internal pressure change are coupled to calculate the thermal runaway risk index. The derivative of the internal pressure change represents the speed of pressure change, and the coupling of the gas release rate and the pressure change rate can more comprehensively reflect the severity of the reaction inside the battery, thus more accurately assessing the risk of thermal runaway.

[0088] Dynamic critical threshold calculation and early warning: According to the historical working temperature of the battery pack and the current environmental humidity, the temperature-humidity coupling coefficient is calculated. The historical working temperature can be obtained from the historical data record of the battery management system, and the current environmental humidity is measured by a humidity sensor. The temperature-humidity coupling coefficient can be calculated using an empirical formula or a formula fitted from experimental data, for example: Temperature-humidity coupling coefficient = k1T h +k2H, T h is the average value of the historical working temperature, H is the current environmental humidity, and k1, k2 are coefficients determined according to experiments.

[0089] The basic critical threshold and the temperature-humidity coupling coefficient are weighted and summed to obtain the real-time updated dynamic critical threshold. When the thermal runaway risk index exceeds the dynamically adjusted critical threshold, a thermal runaway warning signal is generated, and an emergency disconnection mechanism is triggered. The emergency disconnection mechanism cuts off the circuit connection of the battery pack to prevent the further development of thermal runaway and protect the safety of the battery system and surrounding equipment.

[0090] Example 5:

[0091] This embodiment details the working process of the equalization performance analysis submodule and the specific steps of principal component analysis. Its unique role is to evaluate the effectiveness of the equalization strategy, provide data support for optimizing the equalization strategy, and further improve the equalization performance of the battery pack. The specific method includes:

[0092] The equalization performance analysis submodule calculates the ratio of the number of successful equalizations to the number of failed equalizations within the preset period, marked as the equalization performance coefficient. For example, in a one-month preset period, the equalization is successful 50 times and fails 10 times, then The balancing performance coefficient can intuitively reflect the overall effect of the balancing strategy in a certain time, and the higher the coefficient, the higher the success rate of the balancing strategy.

[0093] The energy loss rate, balancing duration and voltage convergence speed of each balancing process are extracted. The energy loss rate refers to the ratio of the energy consumed in the balancing process to the total energy of the battery pack, reflecting the energy utilization efficiency of the balancing process; the balancing duration is the time spent from the start of balancing to the completion of balancing; and the voltage convergence speed represents the speed at which the battery pack voltage reaches consistency during the balancing process.

[0094] The energy loss rate, balancing duration and voltage convergence speed are standardized to eliminate the influence of different dimensions. Standardization can use the Z-score standardization method, and the formula is: x is the original data, is the mean of the data, and s is the standard deviation of the data.

[0095] The eigenvalues and eigenvectors of the covariance matrix are calculated, and the first two principal components with a cumulative contribution rate of more than 85% are selected. These two principal components can retain most of the information of the original data and represent the main features in the balancing process.

[0096] The principal component scores are mapped to a two-dimensional space, and the K-means clustering algorithm is used to divide the balancing quality levels. The K-means clustering algorithm is a commonly used clustering algorithm, which divides the data points into K clusters through continuous iteration, and each cluster represents a balancing quality level. According to the division levels, the corresponding balancing quality scores are output.

[0097] If the balancing performance coefficient is lower than the first threshold or the balancing quality score is lower than the second threshold, it means that the current balancing strategy may have deficiencies, and balancing strategy optimization instructions are generated and fed back to the active balancing control module. The active balancing control module adjusts the balancing strategy according to the optimization instructions, such as adjusting the balancing priority, balancing power and other parameters, to improve the balancing effect and enhance the performance of the battery pack.

[0098] It should be noted that in this article, relationship terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the terms "include", "contain" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or equipment including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or equipment.

[0099] While embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary of the principles and application of the present application. Numerous modifications and adaptions can be effected without departing from the spirit and scope of the present application, which is not limited to the exact construction and arrangement described. It is intended, therefore, to cover all modifications and adaptions that fall within the scope of the claims and their equivalents.

Claims

1. A battery management system with active monitoring function, characterized in that, It includes a processor, a battery status monitoring module, an active balancing control module, a performance evaluation module, and a fault early warning module; the battery status monitoring module monitors the working status of the battery pack in real time, which includes the charging stage, the discharging stage, and the resting stage. During the charging stage, it collects the individual cell voltage, charging current, and ambient temperature of the battery; during the discharging stage, it collects the individual cell voltage, discharging current, and internal temperature of the battery; and during the resting stage, it collects the open circuit voltage and self-discharge rate of the battery. The battery status monitoring module generates voltage abnormality signals, current abnormality signals, or temperature abnormality signals through a dynamic threshold algorithm, and sends the abnormal signals to the active balancing control module and the fault early warning module via a processor. The fault early warning module displays the abnormal signals in a graded manner and triggers the early warning level. When the active balancing control module receives an abnormal signal, it dynamically adjusts the balancing strategy of the battery pack based on a fuzzy logic algorithm. The performance evaluation module tracks and evaluates the equalization process of the active equalization control module, and records the duration from the time of the abnormal signal trigger to the time of equalization completion. If the duration exceeds the preset equalization duration threshold, an equalization timeout flag is generated and the number of equalization failures is updated; otherwise, the number of equalization successes is updated. The specific operation process of the dynamic threshold algorithm is as follows: During the charging phase, the instantaneous rate of change of the individual cell voltage, the fluctuation amplitude of the charging current, and the gradient value of the ambient temperature are calculated. The voltage dynamic deviation value is obtained by calculating the difference between the instantaneous rate of change and the mean of historical charging data. Similarly, the current dynamic deviation value and the temperature dynamic deviation value are obtained. If any dynamic deviation value exceeds its corresponding adaptive threshold range, a corresponding abnormal signal is generated. During the discharge phase, the number of consecutive drops in the cell voltage, the harmonic distortion rate of the discharge current, and the time-domain variance of the internal temperature are statistically analyzed using a sliding window. The number of consecutive drops is compared with a preset safe number threshold. If the threshold is exceeded, a voltage abnormality signal is generated. Similarly, a current or temperature abnormality signal is generated based on the harmonic distortion rate and the time-domain variance. During the resting phase, the correlation coefficient between the attenuation slope of the open-circuit voltage and the self-discharge rate is calculated. If the correlation coefficient is lower than the preset correlation threshold, an abnormal self-discharge signal is generated. The fuzzy logic algorithm of the active equalization control module specifically includes: Define the input variables as voltage deviation level, current deviation level and temperature deviation level, and the output variables as equalization priority and equalization power; Establish a fuzzy rule base with the following rule form: if the voltage deviation level is high and the temperature deviation level is medium, then the balancing priority is extremely high and the balancing power is full. The fuzzy output is converted into a precise equalization command through defuzzification calculation, and the equalization energy is dynamically allocated to the target battery cell.

2. The battery management system according to claim 1, characterized in that, The processor is communicatively connected to the aging feature extraction module. The aging feature extraction module collects the number of cycles, maximum capacity decay rate and internal resistance growth rate of the battery pack. The number of cycles is normalized to obtain the cycle aging coefficient. The product of the maximum capacity decay rate and the internal resistance growth rate is marked as the composite aging index. The cycle aging coefficient and composite aging index are input into the aging prediction model based on radial basis function neural network, and the remaining life prediction value of the battery pack is output. If the remaining life prediction value is lower than the preset life threshold, a life warning signal is generated and sent to the fault warning module through the processor.

3. The battery management system according to claim 2, characterized in that, The training process of the aging prediction model includes: Collect historical battery datasets including cycle count, capacity decay curve, internal resistance change curve, and corresponding actual remaining lifetime value; The second derivative of the capacity decay curve is convolved with the first derivative of the internal resistance change curve to extract the aging acceleration features. The number of cycles, aging acceleration features, and actual remaining lifespan are used as training samples. The weight parameters of the radial basis function neural network are adjusted using the particle swarm optimization algorithm until the prediction error converges within the preset accuracy range.

4. The battery management system according to claim 3, characterized in that, The update mechanism of the fuzzy rule base is as follows: The actual balancing efficiency during the balancing process is collected, and the ratio of the theoretical efficiency to the actual efficiency of the balancing command is calculated. If the ratio falls below the preset efficiency threshold three times consecutively, the rule base is triggered to make adaptive adjustments, and a genetic algorithm is used to re-optimize the membership function of the fuzzy rules.

5. The battery management system according to claim 1, characterized in that, The processor is communicatively connected to the thermal runaway prediction module, and the specific operation process of the thermal runaway prediction module includes: During the charging or discharging phase, the surface temperature distribution, gas release rate, and internal pressure change of the battery cells are collected in real time. The high-frequency component energy is extracted by performing Fourier transform on the surface temperature distribution, and the thermal runaway risk index is obtained by coupling the derivative of the gas release rate and the internal pressure change. If the thermal runaway risk index exceeds the dynamically adjusted critical threshold, a thermal runaway early warning signal will be generated and an emergency chain disconnection mechanism will be triggered.

6. The battery management system according to claim 5, characterized in that, The method for calculating the dynamically adjusted critical threshold is as follows: Calculate the temperature-humidity coupling coefficient based on the battery pack's historical operating temperature and current ambient humidity; The dynamic critical threshold is obtained by weighting and summing the basic critical threshold with the temperature-humidity coupling coefficient.

7. The battery management system according to claim 1, characterized in that, The performance evaluation module also includes a balanced performance analysis submodule, the specific analysis process of which is as follows: The ratio of the number of successful balancing attempts to the number of balancing failures within a preset period is denoted as the balancing efficiency coefficient. The energy loss rate, equilibrium duration, and voltage convergence speed of each equilibrium process are extracted, and principal component analysis is performed on the three to obtain the equilibrium quality score. If the equilibrium efficiency coefficient is lower than the first threshold or the equilibrium quality score is lower than the second threshold, an equilibrium strategy optimization instruction is generated and fed back to the active equilibrium control module.

8. The battery management system according to claim 7, characterized in that, The specific steps of the principal component analysis include: The energy loss rate, equalization time, and voltage convergence speed are standardized. Calculate the eigenvalues ​​and eigenvectors of the covariance matrix, and select the first two principal components with a cumulative contribution rate exceeding 85%. The principal component scores are mapped to a two-dimensional space, and the equilibrium quality levels are divided using the K-means clustering algorithm, and the corresponding equilibrium quality scores are output.

Images