Distributed battery health assessment and dynamic connection management system for battery swap cabinets
By using frequency domain analysis and four-dimensional vector evaluation, the problem of insufficient identification of microscopic features of lithium batteries in battery swapping cabinets is solved, enabling accurate assessment and dynamic management of battery health status, and preventing accelerated battery degradation and false protection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING LINGSHUO TECH CO LTD
- Filing Date
- 2026-04-14
- Publication Date
- 2026-06-02
Smart Images
Figure CN122131160A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of battery swapping cabinet technology, and more specifically to a distributed battery health assessment and dynamic connection management system for battery swapping cabinets. Background Technology
[0002] Currently, in the fields of urban on-demand delivery (food delivery, express delivery) and short-distance travel, the shared battery swapping model is widely used due to its high-efficiency energy replenishment advantage. As a centralized charging and storage device, a battery swapping cabinet typically includes multiple independent charging compartments, each containing a battery pack for charging.
[0003] In actual operation, with the increase of charge-discharge cycles, complex chemical changes occur inside the battery, such as electrode material structure degradation, electrolyte decomposition, and lithium dendrite growth. These microscopic changes manifest macroscopically as an increase in the battery's DC internal resistance and a decrease in capacity. More importantly, batteries at different aging stages exhibit significantly different voltage response curve characteristics during charging, thus requiring the use of a battery health assessment system.
[0004] However, defects such as early micro-short circuits, lithium plating, or abnormal thickening of the SEI film inside lithium batteries often manifest as extremely small fluctuations or distortions on the macroscopic voltage curve. Most existing battery health assessment systems cannot capture these microscopic features, causing batteries with potential problems to be misjudged as normal during routine testing and continue to be subjected to high-current fast charging, which can easily induce thermal runaway.
[0005] Meanwhile, many high-performance electric vehicles require two or more batteries connected in parallel for power. Existing battery swapping stations typically allocate batteries randomly to users based on availability or similar remaining charge levels. Due to a lack of consistent assessment of the battery's internal electrochemical characteristics (such as polarization resistance and diffusion coefficient), even batteries with the same remaining charge can exhibit significantly different dynamic response characteristics. Parallel use accelerates the degradation of the entire battery pack and can even lead to false power outages caused by the battery management system (BMS). Summary of the Invention
[0006] To address this, the present invention provides a distributed battery health assessment and dynamic connection management system for battery swapping cabinets, in order to solve the problems in the prior art that cannot identify the micro-features during the battery charging process and have low matching degree of multiple batteries in parallel connection.
[0007] To achieve the above objectives, the present invention provides the following technical solution:
[0008] The distributed battery health assessment and dynamic connection management system for battery swapping cabinets includes a voltage signal acquisition module, a feature processing module, a multi-dimensional deviation assessment module, a calculation and classification module, and a dynamic management module.
[0009] The voltage signal acquisition module continuously acquires the battery terminal voltage after the battery is inserted into the compartment and enters the constant current charging stage, and obtains the voltage-time dataset through two-stage noise reduction processing.
[0010] The feature processing module converts the time-domain voltage signal of the voltage-time dataset to the frequency domain, and performs normalization by dividing the frequency bands, calculating the energy proportion of each frequency band, and dividing by the total energy, and outputs a four-dimensional vector.
[0011] The multidimensional deviation assessment module is used to calculate the offset of the four-dimensional vector relative to its own historical benchmark and the deviation from the group average level.
[0012] The calculation and classification module calculates the battery's comprehensive risk index R based on the offset of the four-dimensional vector relative to its own historical benchmark and the deviation from the group average level, and classifies the battery into health levels.
[0013] The dynamic management module issues charging control and pairing connection management instructions in parallel according to the levels divided by the calculation and classification module, which is used to realize differentiated charging protection based on battery health status and pairing management based on the consistency of electrochemical characteristics.
[0014] Furthermore, the two-stage denoising process includes a first-stage denoising process and a second-stage denoising process.
[0015] The first-stage denoising process uses a moving average filtering algorithm with a window size of 5 to calculate the local mean of the battery terminal voltage. ;
[0016] The second-stage noise reduction process is used to calculate the original voltage data and the local mean. The difference Obtain the residual data and calculate the standard deviation of the residual data. .
[0017] Furthermore, the specific contents of the feature processing module are as follows:
[0018] 1) Perform a Fast Fourier Transform on the voltage data in the voltage-time dataset;
[0019] 2) Based on the time-scale characteristics of the electrochemical process in lithium batteries, the spectrum is divided into four continuous frequency bands;
[0020] 3) For each frequency band, calculate the sum of squares of the amplitudes of all spectral components within that band to obtain the energy value of that band. The calculation formula is as follows:
[0021]
[0022] in, For frequency The corresponding FFT magnitude;
[0023] 4) Calculate the total energy of the four frequency bands. Then, the energy of each frequency band is divided by the total energy to obtain the normalized energy percentage. The four energy percentage values are combined into a four-dimensional vector and output.
[0024] Furthermore, the specific content of the multidimensional deviation assessment module is as follows:
[0025] 1) Retrieve the historical baseline vector of the target battery ID from the historical database;
[0026] 2) Calculate the offset of the current four-dimensional vector relative to its own historical baseline. The calculation formula is as follows:
[0027]
[0028] in, For battery history benchmark number Dimensional value, For battery historical data, number 1 Standard deviation of dimension Preset weighting coefficients;
[0029] 3) Retrieve the group mean feature vector of all batteries of the same model and production batch as this battery, and then calculate the deviation of the current feature from the group average level. .
[0030] Furthermore, the deviation The calculation formula is as follows:
[0031]
[0032] in, For all batteries of the same model and batch, the first Dimension group mean.
[0033] Furthermore, a higher comprehensive risk index R value indicates a worse battery health condition, and the calculation formula is as follows:
[0034]
[0035] in, and Preset balance coefficient, This is the offset value. This represents the deviation.
[0036] Furthermore, the specific content of the dynamic management module is as follows:
[0037] 1) In terms of charging control: Based on the battery health level generated by the calculation and grading module, a preset current setting command is sent to the charging compartment;
[0038] 2) In terms of pairing and connection management, when a power request is received from a user terminal, the controller retrieves a list of all batteries in the cabinet that are available in all states and have sufficient remaining power to meet the delivery requirements in real time.
[0039] 3) Perform pairwise combinations on all candidate batteries and calculate the similarity between the four-dimensional vectors of the two batteries in each combination. After calculating the similarity of all combinations, a decision is made and instructions are issued based on the preset similarity threshold.
[0040] Furthermore, the similarity The calculation formula is as follows:
[0041]
[0042] in, For the first The four-dimensional vector of the battery. For the first The four-dimensional vector of the battery. Let be the Euclidean distance between two four-dimensional vectors.
[0043] This invention has the following advantages: By converting the voltage value to the frequency domain and dividing it into four characteristic frequency bands, this invention calculates the normalized energy percentage of each frequency band, forming a four-dimensional vector. This four-dimensional vector quantifies the electrochemical activity distribution of the battery at different time scales, enabling early micro-defects inside the lithium battery, such as micro-short circuits, lithium plating, or abnormal thickening of the SEI film, to be accurately captured through the shift in the energy percentage of specific frequency bands. This solves the problem that existing systems cannot identify micro-features, leading to the misjudgment of potentially hazardous batteries as normal.
[0044] Meanwhile, a similarity evaluation mechanism based on four-dimensional vectors is introduced into the pairing and connection management. When responding to a user's power request, the system not only filters candidate batteries that meet the SOC requirements, but also calculates the similarity between pairs of candidate batteries. Only when the similarity is higher than a preset threshold are the two batteries assigned as a group to the user. This ensures that the batteries used in parallel after leaving the warehouse are not only consistent in capacity, but also highly matched in internal electrochemical characteristics such as high-frequency transient response, mid-frequency charge transfer process, and low-frequency ion diffusion process. This effectively avoids excessive circulating current or uneven load distribution caused by differences in polarization resistance and diffusion coefficient, prevents BMS false protection shutdowns, and slows down the accelerated degradation of the entire battery pack.
[0045] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. Attached Figure Description
[0046] To more intuitively illustrate the prior art and this application, exemplary drawings are provided below. It should be understood that the specific shapes and structures shown in the drawings should not generally be regarded as limiting conditions for implementing this application; for example, based on the technical concept disclosed in this application and the exemplary drawings, those skilled in the art are able to easily make conventional adjustments or further optimizations to the addition / reduction / classification, specific shapes, positional relationships, connection methods, size ratios, etc. of certain units (components).
[0047] Figure 1 This is a flowchart illustrating the implementation of the distributed battery health assessment and dynamic connection management system for battery swapping cabinets according to the present invention. Detailed Implementation
[0048] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. It should be understood that these embodiments are merely for further explanation of the present invention and should not be construed as limiting the scope of protection of the present invention. Technical engineers in the field can make some non-essential improvements and adjustments to the present invention based on the above-described content. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0049] Please see Figure 1 The distributed battery health assessment and dynamic connection management system for battery swapping cabinets includes a voltage signal acquisition module, a feature processing module, a multi-dimensional deviation assessment module, a calculation and classification module, and a dynamic management module.
[0050] The voltage signal acquisition module can continuously acquire the battery terminal voltage after the battery is inserted into the compartment and enters the constant current (CC) charging stage, and obtain the voltage-time dataset through two-stage noise reduction processing.
[0051] After the battery pack is inserted into the compartment and the charging state machine enters the constant current (CC) charging stage, the analog-to-digital converter (ADC) is automatically triggered. The ADC continuously reads the terminal voltage signal between the positive and negative terminals of the battery pack at a fixed sampling frequency of 50Hz to generate the original voltage time data.
[0052] Simultaneously, the real-time remaining charge (SOC) value reported by the battery management unit (BMS) is read through the communication interface. The voltage data is timestamped with the real-time remaining charge value, and data segments with real-time remaining charge in the range of 30% to 70% are filtered out. Data in the ohmic resistance-dominated region at the beginning of charging and the polarization saturation region at the end of charging are removed to ensure that subsequent analysis focuses on the active electrochemical reaction region.
[0053] The two-stage denoising process consists of a first-stage denoising process and a second-stage denoising process. The first-stage denoising process uses a moving average filtering algorithm with a window size of 5 to calculate the local mean of the battery terminal voltage. This is used to filter out high-frequency random noise; the calculation formula is as follows:
[0054]
[0055] in, To represent the first in the original voltage-time data The sampled voltage value at each moment. This represents the traversal index within the sliding window, with a value range of [ -2, +2], This indicates the index of the sampling point currently being calculated.
[0056] The second-stage noise reduction process is used to calculate the original voltage data and the local mean. The difference Obtain the residual data and calculate the standard deviation of the residual data. The calculation formula is as follows:
[0057]
[0058] in, This represents the total number of valid sampling points participating in the statistics.
[0059] If the standard deviation exceeds the preset noise threshold , ,in The reference noise standard deviation calibrated under the steady-state condition of the battery at rest; is the confidence coefficient. A high noise level indicates that a pre-defined adaptive denoising algorithm based on Empirical Mode Decomposition (EMD) is used to process the original data, obtaining the clean voltage value. The output is a voltage-time dataset including the clean voltage value and a timestamp. This effectively filters out noise while preserving, to the maximum extent possible, the subtle voltage fluctuations related to the internal processes of the battery.
[0060] Preset noise threshold Dynamic adaptive setting is adopted, and the calculation formula is as follows: ,in The reference noise standard deviation calibrated under the steady-state condition of the battery at rest; is the confidence coefficient, with a value range of [3, 5].
[0061] The adaptive denoising algorithm based on Empirical Mode Decomposition (EMD) is a data-driven time-frequency analysis method that can adaptively decompose the signal according to its own time-scale characteristics, effectively filtering out noise while preserving local fluctuation details. Details are as follows:
[0062] The algorithm performs empirical mode decomposition on the raw voltage data. By identifying local extrema, fitting upper and lower envelopes, and iteratively filtering, it decomposes the signal into a series of intrinsic mode functions (IMFs) arranged from high frequency to low frequency, and a residual term. The high-frequency IMF components mainly include noise components, while the low-frequency IMF components and the residual term reflect the electrochemical process characteristics of the battery.
[0063] Then, the noise-dominant component is automatically identified by calculating the correlation coefficient between each IMF component and the original signal. First, the Pearson correlation coefficient between each IMF component and the original residual signal is calculated. Then preset the correlation coefficient threshold. If the Pearson correlation coefficient is less than the correlation coefficient threshold If the IMF component is determined to be a noise-dominant component, soft thresholding denoising is performed on that component. Otherwise, the decision is made... This is the dominant component of the signal. We retain this component directly without any modification; Pearson correlation coefficient. The calculation formula is as follows:
[0064]
[0065] in, For the first The value of each IMF component for The standard deviation (calculated using the same formula as above).
[0066] Soft thresholding is applied to the identified noise-dominant IMF components. Comparison processing, ,in This is an estimate of the noise standard deviation of the IMF component. This is the signal length.
[0067] Coefficients below a threshold are set to zero, while coefficients above the threshold are shrunk towards zero, thus suppressing noise while preserving the main shape of the components. Finally, all thresholded high-frequency IMF components are added to the unprocessed low-frequency IMF components to obtain the pure voltage value, and the output is a voltage-time dataset including the pure voltage value and a timestamp.
[0068] The feature processing module converts the time-domain voltage signal of the voltage-time dataset to the frequency domain using the Fast Fourier Transform (FFT) algorithm. It then normalizes the signal by dividing it into frequency bands, calculating the energy percentage of each band, and dividing by the total energy, outputting a four-dimensional vector. This four-dimensional vector quantifies the energy distribution of the battery at different time scales (corresponding to different electrochemical processes). When aging phenomena such as micro-short circuits or lithium plating occur inside the battery, the energy percentage of each frequency band will shift measurably, thus reflecting changes in the battery's health status through vector changes.
[0069] The specific contents of the feature processing module are as follows:
[0070] 1) Perform a Fast Fourier Transform on the voltage data in the voltage-time dataset. Assume the sequence length is N and the sampling frequency is... Then, the FFT yields N complex spectral components, corresponding to a frequency range of 0 to... Frequency resolution Since the input signal is a real sequence with a conjugate symmetric spectrum, only the positive frequency portion (0~25Hz) is used for subsequent calculations.
[0071] 2) Based on the time-scale characteristics of the electrochemical process in lithium batteries, the spectrum is divided into four continuous frequency bands, each corresponding to a different physical process: Band 1 (high-frequency transient reaction) 12.5Hz-25Hz; Band 2 (mid-frequency charge transfer process) 6.25Hz-12.5Hz; Band 3 (low-frequency ion diffusion process) 3.125Hz-6.25Hz; Band 4 (overall trend) 0Hz-3.125Hz;
[0072] The boundaries of each frequency band are determined by the frequency resolution. Determine the corresponding discrete frequency index. For example, the starting index of frequency band 1 is (12.5 / The ending index is (25 / ).
[0073] 3) For each frequency band, calculate the sum of squares of the amplitudes of all spectral components within that band to obtain the energy value of that band. The calculation formula is as follows:
[0074]
[0075] in, For frequency The corresponding FFT magnitude.
[0076] 4) Calculate the total energy of the four frequency bands. , Then, the energy of each frequency band is divided by the total energy to obtain the normalized energy percentage. The four energy percentage values are combined into a four-dimensional vector and output. The calculation formula is as follows:
[0077]
[0078] The multidimensional deviation assessment module is used to calculate the offset of the four-dimensional vector relative to its own historical benchmark and the deviation from the group average level, so as to achieve individual longitudinal assessment and group lateral assessment of the battery using the four-dimensional vector.
[0079] The specific contents of the multidimensional deviation assessment module are as follows:
[0080] 1) Retrieve the historical baseline vector of the target battery ID from the historical database. If the battery is being charged for the first time in this system and there is no historical record in the database, the current four-dimensional vector is temporarily stored in the cache, and the average value is taken as the initial value after three charging cycles are completed. For batteries with existing records, the most recently updated health baseline data is read directly. At the same time, the standard deviation of the battery's historical characteristic data is read to characterize the normal fluctuation range of this individual.
[0081] 2) Calculate the offset of the current four-dimensional vector relative to its own historical baseline. This offset is represented by a weighted Mahalanobis distance. The calculation formula is as follows:
[0082]
[0083] in, This is the first historical benchmark for this battery. Dimensional value (obtained from the average of the fingerprints from the previous three charging cycles or from historical records). This is the first historical data point for this battery. Standard deviation of dimension This is a preset weighting coefficient. The larger the value, the more significant the deviation of the current feature from its own historical benchmark, and the more obvious the individual's aging trend.
[0084] 3) Retrieve the group mean feature vector of all batteries of the same model and production batch as this battery. The group mean is calculated from the latest characteristics of all batteries of the same model and batch in the database according to the steps described above. It represents the typical health status of this batch of batteries. Then, the deviation of the current characteristic from the group average is calculated. This reflects the consistency of battery performance within a batch, and the degree of deviation. The calculation formula is as follows:
[0085]
[0086] in, For all batteries of the same model and batch, the first Dimension group mean.
[0087] The grading module calculates the battery's comprehensive risk index R based on the offset of the four-dimensional vector relative to its own historical baseline and the deviation from the group average level, and then classifies the battery into health levels. A higher comprehensive risk index R value indicates a worse battery health condition. The calculation formula is as follows:
[0088]
[0089] in, and Preset balance coefficient, + =1.
[0090] Batteries are classified into three levels based on a preset first and second threshold (the specific values are determined by experimental calibration or based on historical data statistics):
[0091] L1 (Healthy) The threshold indicates that the aging trend of individual batteries is normal and consistent with the group, with no significant abnormalities.
[0092] L2 (Note) First threshold ≤ The second threshold indicates that the battery shows slight signs of aging or deviates from the group, requiring monitoring of its subsequent development and the adoption of protective charging strategies.
[0093] L3 (Risk) A reading above the second threshold indicates that the battery has obvious microscopic defects or abnormal aging, such as internal micro-short circuits or severe lithium plating, and its use should be restricted immediately and repairs should be arranged.
[0094] The dynamic management module issues charging control and pairing connection management commands in parallel according to the levels divided by the calculation and classification module. It can realize differentiated charging protection based on battery health status and pairing management based on the consistency of electrochemical characteristics. This can block the risk of fast charging of potentially dangerous batteries, while avoiding excessive circulating current, accelerated degradation and BMS false protection problems caused by internal characteristic differences of parallel batteries.
[0095] The specific contents of the dynamic management module are as follows:
[0096] 1) In terms of charging control: Based on the battery health level generated by the calculation and grading module, a preset current setting command is sent to the DC-DC power module of the charging compartment.
[0097] For example: for L1 level batteries, a standard fast charging command is issued, allowing charging at a rate of 1.0C; for L2 level batteries, the controller activates a protection strategy, locking the upper limit of the charging current to 0.7C and extending the cutoff current threshold of the constant voltage stage to reduce polarization stress; for L3 level batteries, the controller immediately reduces the current to 0.3C trickle mode, disconnects the main circuit relay after charging is complete, marks the compartment as "fault locked" and prohibits it from being reassigned to the user until manual intervention.
[0098] 2) In terms of pairing and connection management, when the receiver receives a power request (including the required number of batteries N) from a user terminal, the controller searches for a list of all batteries in the cabinet that are available (non-L3 level) and whose SOC meets the delivery requirements.
[0099] If N=1, prioritize the L1-level battery with the lowest comprehensive risk index R when releasing it from the cabinet;
[0100] When a user initiates a power request to rent two batteries (N=2), the system first filters out all candidate batteries from the battery swapping station's inventory that are available in a certain state (i.e., not L3 level) and whose SOC meets the user's needs.
[0101] 3) Perform pairwise combinations on all candidate batteries and calculate the similarity between the four-dimensional vectors of the two batteries in each combination. The value ranges from (0, 1). The smaller the distance, the closer the similarity is to 1; the larger the distance, the closer the similarity is to 0. The calculation formula is as follows:
[0102]
[0103] in, For the first The four-dimensional vector of the battery. For the first A four-dimensional vector of a battery cell. This represents the Euclidean distance between two four-dimensional vectors. The calculation formula is as follows:
[0104]
[0105] After calculating the similarity of all combinations, a decision is made and an instruction is issued based on a preset similarity threshold. If a pair of batteries has a similarity greater than or equal to the similarity threshold, these two batteries will be prioritized for release to the user. If the similarity of all possible combinations in the inventory is below the threshold, it is determined that there is currently no suitable pair.
[0106] At this point, if the user urgently needs to use the device, only one optimal battery will be allocated (i.e., downgraded to a single rental of N=1). At the same time, a prompt message will be sent to the user's APP terminal, informing them that no best matching battery has been found and suggesting single use or retrying later.
[0107] This invention converts voltage values to the frequency domain and divides them into four characteristic frequency bands, calculating the normalized energy percentage of each band to form a four-dimensional vector. This four-dimensional vector quantifies the electrochemical activity distribution of the battery at different time scales, enabling early micro-defects inside the lithium battery, such as micro-short circuits, lithium plating, or abnormal thickening of the SEI film, to be accurately captured through the shift in the energy percentage of specific frequency bands. This solves the problem that existing systems cannot identify micro-features, leading to the misjudgment of potentially hazardous batteries as normal.
[0108] Meanwhile, a similarity evaluation mechanism based on four-dimensional vectors is introduced into the pairing and connection management. When responding to a user's power request, the system not only filters candidate batteries that meet the SOC requirements, but also calculates the similarity between pairs of candidate batteries. Only when the similarity is higher than a preset threshold are the two batteries assigned as a group to the user. This ensures that the batteries used in parallel after leaving the warehouse are not only consistent in capacity, but also highly matched in internal electrochemical characteristics such as high-frequency transient response, mid-frequency charge transfer process, and low-frequency ion diffusion process. This effectively avoids excessive circulating current or uneven load distribution caused by differences in polarization resistance and diffusion coefficient, prevents BMS false protection shutdowns, and slows down the accelerated degradation of the entire battery pack.
[0109] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A distributed battery health assessment and dynamic connection management system for battery swapping cabinets, characterized in that, It includes a voltage signal acquisition module, a feature processing module, a multi-dimensional deviation evaluation module, a calculation and classification module, and a dynamic management module. The voltage signal acquisition module continuously acquires the battery terminal voltage after the battery is inserted into the compartment and enters the constant current charging stage, and obtains the voltage-time dataset through two-stage noise reduction processing. The feature processing module converts the time-domain voltage signal of the voltage-time dataset to the frequency domain, and performs normalization by dividing the frequency bands, calculating the energy proportion of each frequency band, and dividing by the total energy, and outputs a four-dimensional vector. The multidimensional deviation assessment module is used to calculate the offset of the four-dimensional vector relative to its own historical benchmark and the deviation from the group average level. The calculation and classification module calculates the battery's comprehensive risk index R based on the offset of the four-dimensional vector relative to its own historical benchmark and the deviation from the group average level, and classifies the battery into health levels. The dynamic management module issues charging control and pairing connection management instructions in parallel according to the levels divided by the calculation and classification module, which is used to realize differentiated charging protection based on battery health status and pairing management based on the consistency of electrochemical characteristics.
2. The distributed battery health assessment and dynamic connection management system for battery swapping cabinets according to claim 1, characterized in that, The two-stage noise reduction process includes a first-stage noise reduction process and a second-stage noise reduction process. The first-stage denoising process uses a moving average filtering algorithm with a window size of 5 to calculate the local mean of the battery terminal voltage. ; The second-stage noise reduction process is used to calculate the original voltage data and the local mean. The difference Obtain the residual data and calculate the standard deviation of the residual data. .
3. The distributed battery health assessment and dynamic connection management system for battery swapping cabinets according to claim 1, characterized in that, The specific contents of the feature processing module are as follows: 1) Perform a Fast Fourier Transform on the voltage data in the voltage-time dataset; 2) Based on the time-scale characteristics of the electrochemical process of lithium batteries, the spectrum is divided into four continuous frequency bands; 3) For each frequency band, calculate the sum of squares of the amplitudes of all spectral components within that band to obtain the energy value of that band. The calculation formula is as follows: in, For frequency The corresponding FFT magnitude; 4) Calculate the total energy of the four frequency bands. Then, the energy of each frequency band is divided by the total energy to obtain the normalized energy percentage. The four energy percentage values are combined into a four-dimensional vector and output.
4. The distributed battery health assessment and dynamic connection management system for battery swapping cabinets according to claim 1, characterized in that, The specific contents of the multidimensional deviation assessment module are as follows: 1) Retrieve the historical baseline vector of the target battery ID from the historical database; 2) Calculate the offset of the current four-dimensional vector relative to its own historical baseline. The calculation formula is as follows: in, For battery history benchmark number Dimensional value, For battery historical data, number 1 Standard deviation of dimension Preset weighting coefficients; 3) Retrieve the group mean feature vector of all batteries of the same model and production batch as this battery, and then calculate the deviation of the current feature from the group average level. .
5. The distributed battery health assessment and dynamic connection management system for battery swapping cabinets according to claim 4, characterized in that, The deviation The calculation formula is as follows: in, For all batteries of the same model and batch, the first Dimension group mean.
6. The distributed battery health assessment and dynamic connection management system for battery swapping cabinets according to claim 1, characterized in that, The higher the comprehensive risk index R value, the worse the battery health condition. The calculation formula is as follows: in, and Preset balance coefficient, This is the offset value. This represents the deviation degree.
7. The distributed battery health assessment and dynamic connection management system for battery swapping cabinets according to claim 1, characterized in that, The specific contents of the dynamic management module are as follows: 1) In terms of charging control: Based on the battery health level generated by the calculation and grading module, a preset current setting command is sent to the charging compartment; 2) In terms of pairing and connection management, when a power request is received from a user terminal, the controller retrieves a list of all batteries in the cabinet that are available in all states and have sufficient remaining power to meet the delivery requirements in real time. 3) Perform pairwise combinations on all candidate batteries and calculate the similarity between the four-dimensional vectors of the two batteries in each combination. After calculating the similarity of all combinations, a decision is made and instructions are issued based on the preset similarity threshold.
8. The distributed battery health assessment and dynamic connection management system for battery swapping cabinets according to claim 7, characterized in that, The similarity The calculation formula is as follows: in, For the first The four-dimensional vector of the battery. For the first The four-dimensional vector of the battery. Let be the Euclidean distance between two four-dimensional vectors.