A micro-short circuit online quantitative diagnosis method and system for energy storage batteries

Abstract

The application provides a micro short circuit online quantitative diagnosis method and system for energy storage batteries, and belongs to the technical field of battery safety; the method comprises the following steps: collecting charging data of a battery pack, and adaptively matching voltage statistical step and smoothing window width according to a charging rate; based on the adaptively matched parameters, cumulative residence amounts of voltage sampling values falling in each voltage interval are counted to construct a voltage residence time spectrum; all voltage regions covered by electrochemical reactions are selected, and a relative residence gain spectrum is constructed through cumulative integral ratio calculation; based on the relative residence gain spectrum, an electrochemical capacity conservation model is used for analytical relation calculation to obtain short circuit resistance of a to-be-detected single battery; the application uses statistical integral characteristics to naturally inhibit high-frequency noise, can realize micro short circuit diagnosis and short circuit resistance estimation without complex filtering, and greatly reduces the algorithm load of an embedded system.

Description

Technical Field

[0001] This invention belongs to the field of battery safety technology, and particularly relates to a method and system for online quantitative diagnosis of micro-short circuits in energy storage batteries. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] Internal micro-short circuits within batteries are one of the main causes of thermal runaway. Therefore, timely detection and control of micro-short circuits are essential for battery safety. However, in the early stages of a micro-short circuit, the leakage current is extremely small (typically <100mA), making it difficult to detect directly by voltage sensors. Under these circumstances, existing technologies generally suffer from the following drawbacks:

[0004] (1) Existing technologies typically diagnose faults by calculating the difference in capacity increment curves. However, since the capacity increment curve is essentially a derivative of the voltage signal, sampling noise is significantly amplified. To obtain a smooth curve, it is necessary to rely on computationally intensive filtering algorithms (such as Kalman filtering and Gaussian process regression) for filtering, which is difficult to run in real time on low-cost embedded BMS chips.

[0005] (2) With the popularization of fast charging technology, the charging time has been greatly shortened, but this has also led to a sharp decrease in the number of sampling points within a unit voltage range. Based on this, the existing fixed step size diagnostic algorithm will experience calculation divergence when the data is sparse, which will lead to false alarms.

[0006] (3) Some existing methods use the difference in geometric area to infer the leakage current, but this method has a long calculation link, large cumulative error, and is difficult to decouple the effects of battery aging. Summary of the Invention

[0007] To overcome the shortcomings of the prior art, this invention provides a method and system for online quantitative diagnosis of micro short circuits in energy storage batteries. By utilizing the statistical integral characteristic to naturally suppress high-frequency noise, it can achieve micro short circuit diagnosis and short circuit resistance estimation without the need for complex filtering, which greatly reduces the computing load of embedded systems.

[0008] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0009] The first aspect of this invention provides a method for online quantitative diagnosis of micro-short circuits in energy storage batteries.

[0010] An online quantitative diagnostic method for micro-short circuits in energy storage batteries includes:

[0011] Collect voltage and current data of the individual cells under test in the battery pack, and adaptively match the voltage statistical step size and smoothing window width according to the charging rate.

[0012] Based on the parameters after adaptive matching, the voltage domain is divided into multiple discrete intervals, and the cumulative residence time of the voltage sampled values ​​falling in each discrete interval is statistically analyzed to construct a voltage residence time spectrum that reflects the voltage residence time distribution.

[0013] All voltage regions covered by the electrochemical reaction are selected, and a relative residence gain spectrum is constructed by calculating the cumulative integral ratio of the monomer under test and the reference monomer at the corresponding curves in each region of the voltage residence time spectrum.

[0014] Based on the relative residence gain spectrum, the analytical relationship is calculated using an electrochemical capacity conservation model to obtain the short-circuit resistance of the test cell.

[0015] Furthermore, the voltage statistical step size and smoothing window width are adaptively matched according to the charging rate, including: when the charging rate is lower than a preset first threshold, high-resolution parameters are used, that is, a small step size and a narrow window are selected; when the charging rate is higher than the first threshold, low-resolution parameters are used, that is, a large step size and a wide window are selected.

[0016] Furthermore, the construction of the voltage residence time spectrum includes: first, assuming that the BMS collects the battery voltage at a fixed sampling period during constant current charging; then, the voltage distribution is statistically analyzed by defining a discrete grid on the voltage axis.

[0017] Furthermore, the definition of the discrete grid on the voltage axis includes: defining each voltage interval by setting the starting point and sampling step size of the voltage statistics; and then determining the corresponding original dwell count value falling in each voltage interval by introducing an indicator function.

[0018] Furthermore, the construction of the voltage residence time spectrum also includes: statistically analyzing the original residence count values ​​of the voltage sampling sequence falling within each discrete voltage interval; and performing convolution smoothing on the original residence count values ​​using a Gaussian kernel function to generate continuous voltage residence density spectral lines.

[0019] Furthermore, the selection of the reference unit includes:

[0020] When the BMS stores historical health status data of the battery pack and the data is valid, the cell with the highest SOH value is selected as the benchmark reference cell.

[0021] When historical SOH data is lacking in the BMS or the SOH data dispersion is below the effective threshold, the median curve of the VRTS curves of all cells in the battery pack is used as the benchmark.

[0022] Furthermore, the relative residence gain spectrum is obtained by comparing the voltage residence time spectrum of the cell under test with the voltage residence time spectrum of the reference cell at the corresponding voltage point, and is used to characterize the charging time expansion effect caused by micro-short circuit.

[0023] Furthermore, the analytical relationship of the short-circuit resistance is as follows: the short-circuit resistance is directly proportional to the average voltage of the selected characteristic voltage range and inversely proportional to the product of the charging current and a coefficient determined based on the comprehensive dwell gain ratio.

[0024] Furthermore, when applied to the static voltage drop phase, the relative residence gain spectrum reflects the shortened voltage residence time of the tested cell due to self-discharge, and the calculation relationship of the short-circuit resistance is adjusted accordingly.

[0025] A second aspect of the present invention provides a micro short-circuit online quantitative diagnostic system for energy storage batteries.

[0026] A micro-short circuit online quantitative diagnostic system for energy storage batteries includes:

[0027] The data acquisition and parameter matching module is configured to: acquire voltage and current data of the individual cells under test in the battery pack, and adaptively match the voltage statistical step size and smoothing window width according to the charging rate;

[0028] The residence spectrum construction module is configured to: divide the voltage domain into multiple discrete intervals based on the parameters after adaptive matching, and count the cumulative residence of voltage sample values ​​falling in each discrete interval in order to construct a voltage residence time spectrum that reflects the voltage residence time distribution.

[0029] The feature extraction and gain spectrum calculation module is configured to: select all voltage regions covered by the electrochemical reaction, and construct a relative residence gain spectrum by calculating the cumulative integral ratio of the monomer under test and the reference monomer at the corresponding curves in each region of the voltage residence time spectrum;

[0030] The short-circuit resistance calculation module is configured to perform analytical relationship calculations based on the relative residence gain spectrum using an electrochemical capacity conservation model to obtain the short-circuit resistance of the cell under test.

[0031] The above one or more technical solutions have the following beneficial effects:

[0032] (1) This invention first collects charging data from the battery pack and adaptively matches the voltage statistical step size and smoothing window width according to the charging rate; then, based on the adaptively matched parameters, it statistically calculates the cumulative residence time of the voltage sampled values ​​within each voltage range to construct a voltage residence time spectrum. Compared with the prior art, this invention abandons differential operations and utilizes the integral smoothing characteristics of statistical counting to naturally suppress high-frequency noise. It can obtain a smooth fault characteristic curve without the need for additional complex filter design. Therefore, it can realize micro-short circuit diagnosis and short circuit resistance estimation without complex filtering, which greatly reduces the computing load of the embedded system.

[0033] (2) In the process of constructing the voltage residence time spectrum, the present invention preserves the time dilation difference between the battery under test and the reference battery in the characteristic voltage range by statistically analyzing the absolute residence amount. Thus, battery fault diagnosis can be performed based on the characteristic value. Compared with the prior art, it effectively avoids the false alarm problem caused by calculation divergence.

[0034] (3) Based on the established relative residence gain spectrum, this invention uses an electrochemical capacity conservation model to perform analytical relationship calculations to obtain the short-circuit resistance of the cell under test. The entire process does not require iterative solutions. This enables this invention to run on various low-cost BMS embedded chips with extremely low computing power costs, and has extremely high engineering practical value.

[0035] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0036] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0037] Figure 1 This is a flowchart of an online quantitative diagnostic method for micro-short circuits in energy storage batteries according to Embodiment 1 of the present invention.

[0038] Figure 2 This is a schematic diagram of the charging voltage curves of the two batteries in Embodiment 1 of the present invention.

[0039] Figure 3 The image shows the spectral diagrams of two batteries in Embodiment 1 of the present invention.

[0040] Figure 4 This is a ratio diagram of the relative residence gain spectrum in Embodiment 1 of the present invention.

[0041] Figure 5 This is a comparison chart of the estimation results of the short-circuit resistance in Embodiment 1 of the present invention. Detailed Implementation

[0042] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0043] It should be noted that the terminology used herein is for the purpose of describing particular implementations only and is not intended to limit the exemplary implementations of the present invention.

[0044] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0045] The overall approach of this invention is as follows: This invention provides an online quantitative diagnostic method for micro-short circuits in energy storage batteries. First, battery charging data is acquired in real time, and the voltage statistical step size and smoothing window width are adaptively adjusted according to the charging rate to overcome the data sparsity problem under high-rate fast charging. Then, by statistically analyzing the cumulative residence time of voltage samples in each discrete interval, a voltage residence time spectrum (VRTS) is constructed to fully preserve the charging time expansion characteristics caused by micro-short circuit leakage. Next, by selecting characteristic voltage regions with significant electrochemical phase transitions, the relative residence gain spectrum of the tested cell and the reference cell in these regions is calculated. Finally, based on the principle of electrochemical capacity conservation, a comprehensive residence gain ratio for the selected voltage range is constructed, and the short-circuit resistance is directly calculated using a closed-loop analytical formula.

[0046] Example 1

[0047] This embodiment discloses an online quantitative diagnostic method for micro short circuits in energy storage batteries.

[0048] like Figure 1 As shown, an online quantitative diagnostic method for micro-short circuits in energy storage batteries includes:

[0049] Step S1: Collect the voltage and current data of the cell under test in the battery pack, and adaptively match the voltage statistical step size and smoothing window width according to the charging rate.

[0050] Step S2: Based on the parameters after adaptive matching, the voltage domain is divided into multiple discrete intervals, and the cumulative residence time of the voltage sampled value falling in each discrete interval is counted to construct a voltage residence time spectrum that reflects the voltage residence time distribution.

[0051] Step S3: Select all voltage regions covered by the electrochemical reaction, and construct a relative residence gain spectrum by calculating the cumulative integral ratio of the monomer to be tested and the reference monomer at the corresponding curves in each region of the voltage residence time spectrum.

[0052] Step S4: Based on the relative residence gain spectrum, the analytical relationship is calculated using the electrochemical capacity conservation model to obtain the short-circuit resistance of the cell under test.

[0053] Based on the above process, this invention utilizes the statistical integral characteristic to naturally suppress high-frequency noise, enabling micro-short-circuit diagnosis and short-circuit resistance estimation without the need for complex filtering, thus significantly reducing the computational load on embedded systems. To facilitate understanding of the technical solution of this invention, the specific implementation methods are further explained and described below.

[0054] In step S1, data acquisition and sampling mode matching are performed, that is, by acquiring the voltage and current data of the cell under test in the battery pack, and adaptively matching the voltage statistical step size and smoothing window width according to the charging rate.

[0055] First, the voltage and current data of the battery pack are acquired to identify the current operating condition (constant current charging or resting and falling back), and statistical parameters are adaptively matched based on the current amplitude. As an optional implementation, a high-precision voltage sensor (such as an AD sampling chip) built into the BMS can be connected in parallel across the individual cells to collect the voltage, and a Hall sensor or shunt can be connected in series in the main circuit to collect the current. When the current direction is detected as the charging direction, and the variance of the current amplitude fluctuation over a period of time is less than a preset threshold (such as 0.1A) and the current amplitude is close to the preset charging current, it indicates that the current state is constant current charging; when the current amplitude is continuously less than the current sensor zero drift threshold (such as 0.05A) and the duration exceeds a preset duration (such as 10 minutes), it indicates that the current state is resting and falling back.

[0056] Subsequently, the charging rate (C-rate) and voltage statistical step size (Bin Width) were established. ) and smooth window width ( The mapping relationship. Specifically:

[0057] 1) When the charging rate At this time, high-resolution parameters (small step size, narrow window) are used. Specifically, the specific values ​​of the high-resolution parameters can be determined according to the sensor accuracy and noise level. In this embodiment, the step size in the high-resolution parameters ranges from 2mV, and the smoothing window value is 5. This indicates the threshold for determining the charging rate.

[0058] 2) When the charging rate When using low-resolution parameters (large step size, wide window), in this embodiment, the step size in the low-resolution parameters is 10mV and the smoothing window is 20.

[0059] By reducing the voltage axis resolution, the sample size within a single interval is increased to suppress discretization noise at high magnification. At high magnification, the data is extremely sparse, making the signal-to-noise ratio (SNR) more important than resolution. If the step size is too small, each... There might only be 0 or 1 points, rendering the statistical properties ineffective. Although the resolution is reduced to some extent, the confidence level of the statistics is guaranteed.

[0060] In step S2, based on the parameters after adaptive matching, the voltage domain is divided into multiple discrete intervals, and the cumulative residence time of the voltage sampled values ​​falling in each discrete interval is counted to construct a voltage residence time spectrum (VRTS) to reflect the voltage residence time distribution.

[0061] The voltage domain is divided into several discrete intervals. The cumulative dwell time of voltage sampled values ​​falling within each interval is statistically analyzed. After smoothing, a VRTS curve is generated without normalizing the total number of sampling points. The smoothing is achieved through discrete convolution operations. Specifically, using the original dwell count values ​​(Craw()) as the input sequence and a Gaussian kernel function as the convolution kernel, a weighted sum of the count values ​​of the surrounding grid points is calculated for each voltage grid point. The weights are determined by the value of the Gaussian function. This transforms the discrete histogram into a continuous, smooth density curve.

[0062] The core objective of this step is to transform the voltage sampling sequence in the time domain. The mapping is represented as a voltage residence time spectrum. To overcome the quantization noise introduced by discretization and retain the time dilation feature, this invention employs a non-normalized density estimation method based on Gaussian kernel convolution. The specific derivation and implementation process are as follows:

[0063] Assume that the BMS operates at a fixed sampling period during constant current charging. The battery voltage is collected; and the collected voltage time series is defined as a set. :

[0064] ;

[0065] in, Sampling time, This represents the total number of sampling points for this charging segment. It directly reflects the total charging time. For micro-short-circuit batteries, charging is slower due to leakage current shunting. The value will definitely be greater than that of a normal battery.

[0066] To statistically analyze the voltage distribution, a discrete grid is first defined on the voltage axis. Let the starting point for voltage statistics be... The sampling step size is Then the first Each voltage range is defined as ,in Building upon this, we introduce indicator functions:

[0067] ;

[0068] in, The indicator function is used to determine whether a condition is true or false; it returns 1 if true and 0 if false. The input logic conditions for the characteristic function are represented.

[0069] Then it falls on the 1st The original dwell count values ​​for each voltage range for:

[0070] ;

[0071] To obtain smooth and physically continuous spectral lines, this invention introduces a Gaussian kernel function. Convolution processing is performed. Therefore, the relative voltage residence time spectrum can be defined as:

[0072] ;

[0073] in, For bandwidth The Gaussian kernel function for (smoothing window width) can be specifically expressed as:

[0074] ;

[0075] in, The independent variable represents the Gaussian kernel function.

[0076] In practical engineering discretization calculations, for the first... voltage grid points Its smoothed spectral value can be expressed as:

[0077] ;

[0078] The above formula represents the discrete convolution operation, which smooths the original counting histogram using a Gaussian kernel function. Wherein, This represents the traversal index of all original voltage discrete grids. This represents the voltage residence density value. This indicates the number of original sampling points counted.

[0079] In step S3, all voltage regions covered by the electrochemical reaction are selected, and a relative residence gain spectrum is constructed by calculating the cumulative integral ratio of the corresponding curves of the monomer to be tested and the reference monomer in each region of the voltage residence time spectrum.

[0080] All voltage regions covered by the electrochemical reaction are selected, and the cumulative integral ratio of the VRTS curves of the cell under test and the reference cell within that region is calculated. The reference cell is selected in the following ways: when the BMS stores historical state of health data of the battery pack and the data is valid, the cell with the highest SOH value is selected as the reference cell; when the BMS lacks historical SOH data (e.g., after a new battery pack or BMS reset) or the SOH data dispersion is below the effective threshold (i.e., extremely small dispersion), the median curve of the VRTS curves of all cells in the battery pack is used as the reference; the effective threshold can be set according to actual needs, and this embodiment does not impose specific limitations on it.

[0081] Define the voltage residence time spectrum of the single cell under test as follows: The reference unit is Based on this, the present invention constructs a relative residency gain spectrum. At voltage point The mathematical definition of a place is:

[0082] ;

[0083] in, It is a very small positive number, which is used to increase the robustness of the algorithm in low voltage or data missing segments, and has almost no impact on the final calculation result.

[0084] Therefore, if the cell under test has no short circuit, then ,at this time It fluctuates around the value 1 across the entire voltage domain. If a short circuit occurs in the cell under test, the effective charging current of the cell decreases due to leakage current, resulting in a longer residence time within the unit voltage. The value will be greater than 1 across the entire voltage domain. Furthermore, as the short circuit becomes more severe, the leakage current increases, and the charging deceleration effect intensifies. The curve shows an overall trend of slight upward movement as the severity of the short circuit increases.

[0085] In step S4, based on the relative residence gain spectrum, the electrochemical capacity conservation model is used to perform analytical relationship calculations to obtain the short-circuit resistance of the monomer under test.

[0086] Under constant current charging conditions, let the charging current be... Select the characteristic voltage range In order to complete the voltage range The corresponding voltage increase (i.e., to complete the same electrochemical phase transition), the internal electrochemical capacity required for the battery's active materials to intercalate lithium ions. It should be constant.

[0087] For a reference cell, i.e., a fault-free cell, its external charging current will be entirely converted into internal chemical current. At this time:

[0088] ;

[0089] in, Reference battery through voltage range The required physical absolute duration can be specifically expressed as:

[0090] ;

[0091] in, The cumulative statistical count of the reference single entity in the characteristic interval.

[0092] For the monomer under test, if a short-circuit fault exists, according to Kirchhoff's current law, the effective current actually participating in the chemical reaction is... equals charging current minus leakage current ,Right now:

[0093] ;

[0094] At this time, fill with the same The time required is :

[0095] ;

[0096] in, The voltage range through which the battery under test passes The required physical absolute duration can be specifically expressed as:

[0097] ;

[0098] in, This represents the cumulative statistical count of the individual being tested within the characteristic interval.

[0099] Therefore, in summary, based on The calculation formula yields the following expression:

[0100] ;

[0101] Therefore, and Substituting into the calculation formula, we get:

[0102] ;

[0103] The sampling period can be eliminated from both sides of the equation. This proves that the resistance calculation method is independent of the sampling period.

[0104] By rearranging and transforming the above equation, the leakage current can be solved:

[0105] ;

[0106] ;

[0107] According to Ohm's law, the relationship between leakage current and short-circuit resistance is as follows:

[0108] ;

[0109] in, The average voltage over the characteristic interval, This is the short-circuit resistance. It is determined by the leakage current. The calculation formula, after simplification, yields:

[0110] ;

[0111] Based on this, the overall residence gain ratio within the characteristic interval is defined. The ratio of the cumulative values ​​over the characteristic voltage range: Therefore, the short-circuit resistance can be obtained:

[0112] .

[0113] To further illustrate the technical solution of the present invention and its practical application effects, this embodiment describes in detail the entire process of quantitative diagnosis of micro-short circuits in a ternary lithium-ion battery using the method proposed in this invention, which can be achieved through the following process:

[0114] 1) Experiment setup.

[0115] This embodiment uses a cylindrical lithium battery with a nominal capacity of 3.5Ah as an example. One battery was selected as the faulty cell, and a high-precision power resistor was connected in parallel across its positive and negative terminals. The measured resistance was 50Ω, used to simulate an early high-impedance micro-short circuit fault inside the battery. Simultaneously, a normal battery from the same batch and in the same healthy state was selected as a reference cell. Constant current charging tests were conducted under the same ambient temperature. Voltage, current, and time data were recorded using a high-precision battery testing system at a sampling frequency of 1Hz.

[0116] 2) Data preprocessing and adaptive parameter matching.

[0117] After reading the battery's current, voltage, and other BMS sampling data, the system can obtain the current during the constant current charging phase. The charging rate is calculated. The system determines that this charging rate falls under low-rate charging conditions. To capture minute short-circuit characteristics, the system adaptively matches high-resolution calculation parameters, including the voltage statistical step size. Smooth window The charging voltage curves of the two batteries are as follows: Figure 2 As shown, the two curves represent the charging voltage changes over time for the reference cell (dashed line) and the faulty cell under test (solid line), respectively. It can be seen that under the same constant current charging conditions, the faulty cell, due to the presence of a small short-circuit leakage current, experiences a slightly slower voltage rise rate than the reference cell, resulting in a slightly longer time required to reach the cutoff voltage.

[0118] 3) Construct a relative voltage residence time spectrum.

[0119] The system establishes a discrete voltage grid. The number of sampling points where the voltage of the reference cell and the cell under test falls within each interval is counted respectively. .

[0120] Subsequently, the Gaussian kernel function was used to process the data to obtain the following results: Figure 3 The reference cell and the cell under test are shown in the spectrum. and . Figure 3 In the middle, the dashed line represents the baseline spectrum. The solid line represents the spectrum to be measured. Because a 50Ω resistor was connected in parallel, the tested cell exhibited leakage current. This results in a decrease in effective charging current and a slower rate of voltage rise in the battery. Statistically, this manifests as a longer "dwell time" of the tested cell within the same voltage range, hence the amplitude of the solid line is higher than that of the dashed line across the entire range.

[0121] 4) Feature extraction and interval aggregation.

[0122] The system automatically selects the electrochemical phase transition plateau region with the highest signal-to-noise ratio as the feature interval. In this implementation case, the interval is... Within this interval, calculate the cumulative residence of both (i.e., the area under the spectral line, corresponding to...). Figure 3 (Shadow-filled area in the image). The cumulative value of the reference unit is calculated. Cumulative count of the faulty individual units under test .

[0123] 5) Calculation of overall residency gain ratio.

[0124] ;

[0125] like Figure 4 As shown, this ratio This quantifies the degree of time dilation caused by micro-short circuits. Specifically, this value means that the time taken for the tested battery to pass through this voltage range is approximately longer than that of a normal battery. .

[0126] 6) Analytical calculation of short-circuit resistance.

[0127] System calculates characteristic interval Average voltage within Therefore, the short-circuit resistance can be calculated using the following formula:

[0128] .

[0129] 7) Results analysis.

[0130] like Figure 5 As shown, the short-circuit resistance estimated by the method of the present invention is: The actual analog resistance is The relative error was only 3.19%, which indicates that the method proposed in this invention has higher estimation accuracy and can accurately quantify the micro-short-circuit resistance value inside the battery, verifying the effectiveness and accuracy of the method.

[0131] Furthermore, it should be noted that the diagnostic quantitative method based on relative voltage residence time spectrum proposed in this invention has broad applicability, not limited to the constant current charging stage in the above embodiments, but also applicable to the resting and drop-off stage after charging. In the resting stage, the external current is zero. If a micro-short circuit exists in the battery, the internal short-circuit resistance will form a self-discharge loop, causing the voltage drop rate of the tested battery to be significantly faster than that of a normal battery. At this time, applying the residence time spectrum constructed by this invention, the cumulative residence time of the tested cell within the characteristic voltage range will be observed to be significantly less than that of the reference cell. This inverse characteristic complements the charging stage. When applied to the resting or discharging stage, due to the leakage current causing accelerated voltage drop and shortened residence time, the ratio relationship in the above formula needs to be adjusted accordingly to adapt to the change in leakage direction.

[0132] Example 2

[0133] This embodiment discloses an online quantitative diagnostic system for micro short circuits in energy storage batteries.

[0134] A micro-short circuit online quantitative diagnostic system for energy storage batteries includes:

[0135] The data acquisition and parameter matching module is configured to: acquire voltage and current data of the individual cells under test in the battery pack, and adaptively match the voltage statistical step size and smoothing window width according to the charging rate;

[0136] The residence spectrum construction module is configured to: divide the voltage domain into multiple discrete intervals based on the parameters after adaptive matching, and count the cumulative residence of voltage sample values ​​falling in each discrete interval in order to construct a voltage residence time spectrum that reflects the voltage residence time distribution.

[0137] The feature extraction and gain spectrum calculation module is configured to: select all voltage regions covered by the electrochemical reaction, and construct a relative residence gain spectrum by calculating the cumulative integral ratio of the monomer under test and the reference monomer at the corresponding curves in each region of the voltage residence time spectrum;

[0138] The short-circuit resistance calculation module is configured to perform analytical relationship calculations based on the relative residence gain spectrum using an electrochemical capacity conservation model to obtain the short-circuit resistance of the cell under test.

[0139] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for online quantitative diagnosis of micro-short circuits in energy storage batteries, characterized in that, include: Collect voltage and current data of the individual cells under test in the battery pack, and adaptively match the voltage statistical step size and smoothing window width according to the charging rate. Based on the parameters after adaptive matching, the voltage domain is divided into multiple discrete intervals, and the cumulative residence time of the voltage sampled values ​​falling in each discrete interval is statistically analyzed to construct a voltage residence time spectrum that reflects the voltage residence time distribution. The construction of the voltage residence time spectrum includes: first, assuming that the BMS collects battery voltage at a fixed sampling period during constant current charging; then, by defining a discrete grid on the voltage axis, the voltage distribution is statistically analyzed; the original residence count values ​​of the voltage sampling sequence falling within each discrete voltage interval are statistically analyzed; the original residence count values ​​are convolved and smoothed using a Gaussian kernel function to generate a continuous voltage residence density spectrum; wherein, the definition of the discrete grid on the voltage axis includes: defining each voltage interval by setting the starting point and sampling step size of the voltage statistics; then, by introducing an indicator function, the corresponding original residence count values ​​falling within each voltage interval are determined; All voltage regions covered by the electrochemical reaction are selected, and a relative residence gain spectrum is constructed by calculating the cumulative integral ratio of the test monomer and the reference monomer at the corresponding curves in each region of the voltage residence time spectrum. The relative residence gain spectrum is obtained by comparing the voltage residence time spectrum of the test monomer with that of the reference monomer at the corresponding voltage points, and is used to characterize the charging time expansion effect caused by micro-short circuit. Based on the relative residence gain spectrum, the analytical relationship is calculated using an electrochemical capacity conservation model to obtain the short-circuit resistance of the test cell.

2. The method for online quantitative diagnosis of micro-short circuits in energy storage batteries as described in claim 1, characterized in that, The adaptive matching voltage statistical step size and smoothing window width based on the charging rate include: when the charging rate is lower than a preset first threshold, high-resolution parameters are used, i.e., small step size and narrow window are selected; when the charging rate is higher than the first threshold, low-resolution parameters are used, i.e., large step size and wide window are selected.

3. The method for online quantitative diagnosis of micro-short circuits in energy storage batteries as described in claim 1, characterized in that, The selection of the reference unit includes: When the BMS stores historical health status data of the battery pack and the data is valid, the cell with the highest SOH value is selected as the benchmark reference cell. When historical SOH data is lacking in the BMS or the SOH data dispersion is below the effective threshold, the median curve of the VRTS curves of all cells in the battery pack is used as the benchmark.

4. The online quantitative diagnostic method for micro-short circuits in energy storage batteries as described in claim 1, characterized in that, The analytical relationship of the short-circuit resistance is as follows: the short-circuit resistance is directly proportional to the average voltage of the selected characteristic voltage range and inversely proportional to the product of the charging current and the coefficient determined based on the comprehensive dwell gain ratio.

5. The online quantitative diagnostic method for micro-short circuits in energy storage batteries as described in claim 1, characterized in that, When applied to the static voltage drop phase, the relative residence gain spectrum reflects the shortened voltage residence time of the tested cell due to self-discharge, and the calculation relationship of the short-circuit resistance is adjusted accordingly.

6. A micro-short circuit online quantitative diagnostic system for energy storage batteries, employing the micro-short circuit online quantitative diagnostic method as described in any one of claims 1-5, characterized in that, include: The data acquisition and parameter matching module is configured to: acquire voltage and current data of the individual cells under test in the battery pack, and adaptively match the voltage statistical step size and smoothing window width according to the charging rate; The residence spectrum construction module is configured to: divide the voltage domain into multiple discrete intervals based on the parameters after adaptive matching, and count the cumulative residence of voltage sample values ​​falling in each discrete interval in order to construct a voltage residence time spectrum that reflects the voltage residence time distribution. The feature extraction and gain spectrum calculation module is configured to: select all voltage regions covered by the electrochemical reaction, and construct a relative residence gain spectrum by calculating the cumulative integral ratio of the monomer under test and the reference monomer at the corresponding curves in each region of the voltage residence time spectrum; The short-circuit resistance calculation module is configured to perform analytical relationship calculations based on the relative residence gain spectrum using an electrochemical capacity conservation model to obtain the short-circuit resistance of the cell under test.

Images