A new energy vehicle in use power battery internal short circuit diagnosis method

Abstract

The present application relates to the technical field of power battery fault diagnosis, in particular to a new energy in-use vehicle power battery internal short circuit diagnosis method, comprising: S1, collecting real-time working data of the power battery; S2, constructing FFRLS combined with UKF algorithm, identifying battery working parameters, and estimating SOC of the battery in real time; S3, constructing a fault feature function of the difference of the SOC curves of each battery monomer and the reference battery; S4, substituting the identification data of each battery into the fault feature function to detect the internal short circuit fault. The present application can effectively improve the sensitivity and accuracy of the battery internal short circuit fault diagnosis, and through the online parameter identification technology, the accuracy of fault detection and the safety of the system are improved, and the present application has good adaptability and robustness, and is suitable for battery internal short circuit detection under dynamic working conditions.

Description

Technical Field

[0001] This invention relates to the field of power battery fault diagnosis technology, specifically a method for diagnosing internal short circuits in power batteries of new energy vehicles. Background Technology

[0002] An in-line short circuit fault in a power battery refers to an abnormal conduction between the positive and negative electrodes caused by various reasons (such as separator damage, metal impurities, manufacturing defects, etc.). This fault can lead to localized overheating, increased self-discharge, and may even cause thermal runaway, seriously threatening the battery's safety and lifespan.

[0003] Existing methods for diagnosing internal short-circuit faults collect battery voltage and current data or test changes in battery internal resistance, and combine this with fault mode recognition models for fault diagnosis. However, these methods have several shortcomings:

[0004] During battery operation, the voltage signal may be affected by external environmental factors, circuit interference, and other factors, resulting in noise. The signal noise acquired by the voltage sensor is relatively high, causing significant fluctuations in the voltage curve, making it difficult to directly characterize the correlation between batteries and potentially affecting the accuracy of diagnosis.

[0005] Existing fault detection methods require determining the battery's internal resistance change data. However, in practical applications, the battery internal resistance testing methods are quite complex and often require pausing the battery's operation, making it difficult to achieve online detection of battery internal resistance and unsuitable for the requirements of real-time monitoring and diagnosis.

[0006] Commonly used parameter estimation algorithms, such as least squares and recursive least squares, may be affected by historical data, leading to slower response times or decreased accuracy. This data can, to some extent, cause the algorithm to become insensitive to new data, reducing the sensitivity of real-time detection.

[0007] Dynamic changes in battery operating conditions, such as temperature and load variations, can cause changes in certain battery parameters (e.g., SOC, internal resistance). Existing models and algorithms do not adequately account for these changes, potentially leading to inaccurate diagnostic results. Summary of the Invention

[0008] This invention provides a method for diagnosing internal short circuits in the power batteries of new energy vehicles. It can effectively improve the sensitivity and accuracy of internal short circuit fault diagnosis. At the same time, through online parameter identification technology, it improves the accuracy of fault detection and the safety of the system. It also has good adaptability and robustness and is suitable for internal short circuit detection in batteries under dynamic operating conditions.

[0009] This application provides the following technical solution:

[0010] A method for diagnosing internal short circuits in the power batteries of new energy vehicles includes the following steps:

[0011] S1. Collect real-time operating data of the power battery;

[0012] S2. Construct an FFRLS algorithm combined with the UKF algorithm to identify battery operating parameters and estimate the battery's SOC in real time;

[0013] S3. Construct a fault characteristic function for the difference between the SOC curves of each battery cell and the reference battery;

[0014] S4. Substitute the identification data of each battery cell into the fault characteristic function to perform internal short circuit fault detection.

[0015] Technical principle: The FFRLS algorithm combined with the UKF algorithm is used to identify the battery's operating parameters, convert the voltage data into a SOC curve, construct a fault feature extraction function from the SOC curve, and perform internal short-circuit fault diagnosis of the power battery.

[0016] Beneficial effects: The fluctuation of the SOC curve is smaller than that of the voltage curve. Using the SOC curve for fault diagnosis can effectively reduce the probability of false alarms.

[0017] The UKF algorithm can more accurately capture the nonlinear characteristics of the battery system through unscented transformation. By recursively using the UKF algorithm, the optimal estimate of the battery's state of charge (SOC) can be achieved, effectively addressing the nonlinearity and uncertainty issues of the battery model.

[0018] The online parameter identification technology improves the accuracy of fault detection and the safety of the system, and has good adaptability and robustness, making it suitable for detecting internal short circuits in batteries under dynamic operating conditions.

[0019] Furthermore, S2 includes:

[0020] S21. Using the voltage sequence data collected in real time, select the cell with the smallest standard deviation of the voltage sequence as the reference cell and calculate the voltage curve of the reference cell.

[0021] S22. Establish the equivalent circuit model and the discretized equations corresponding to the equivalent circuit model;

[0022] S23. Based on the discretized equations of the equivalent circuit model, the online parameters of the equivalent circuit model are identified using the FFRLS method with a variable forgetting factor.

[0023] S24. Online estimation of battery SOC based on the unscented Kalman filter algorithm UKF.

[0024] Furthermore, S23 includes:

[0025] S231. Obtain the transfer function of the equivalent circuit model based on the dynamic equation of the equivalent circuit model;

[0026] S232, Calculate the variable forgetting factor and calculate the gain matrix K0;

[0027] S233. Update the covariance matrix P0(k);

[0028] S234. Update the parameter θ(k) to be estimated;

[0029] S235. Return the parameter identification value of the next time step to S232, and iterate through S232 to S235. When the change in the covariance matrix is ​​less than the set threshold, the FFRLS algorithm ends.

[0030] Furthermore, step S232 uses a simulated annealing algorithm to solve for the optimal forgetting factor.

[0031] Beneficial effects: The forgetting factor is dynamically updated based on changes in system state to adapt to different operating conditions, while avoiding performance degradation caused by a fixed forgetting factor, making the RLS algorithm more robust. Simulated annealing can avoid getting trapped in local optima and find the global optimum for the forgetting factor.

[0032] Furthermore, S24 includes:

[0033] S241. Assign initial values ​​to the model parameters;

[0034] S242. Determine the state equation vector and observation vector equation of the battery system;

[0035] S243. After obtaining the state equation and observation equation of the battery system, the optimal estimate of the battery SOC is achieved according to the recursive steps of the UKF algorithm.

[0036] Furthermore, in step S3, t0-t e The data point at time is defined as

[0037]

[0038] Where n is the number of voltage data points collected by the sensor;

[0039] Calculate the mean at each time step using the data points. for:

[0040]

[0041] The standard deviation of the data points at each time step is:

[0042]

[0043] Sort the data points at each time step from largest to smallest, and define them as follows: Then, the median of the voltage data at each time point is calculated:

[0044]

[0045] The reference data for each time step is defined as follows:

[0046] and

[0047] Where, n sel The number of voltage data points required to satisfy the inequality;

[0048] The fault characteristic function is then:

[0049] Beneficial effects: The fault characteristic function based on the exponential function can amplify the small differences in the SOC curves between the target battery and the reference battery, which is beneficial to improving the accuracy of fault detection.

[0050] Furthermore, S4 employs a functional data outlier detection algorithm based on deformation anomaly degree to determine the internal short-circuit fault of the power battery.

[0051] Beneficial effects: The use of a functional outlier detection algorithm based on deformation anomaly degree to identify anomalous cell characteristic curves effectively improves the sensitivity and accuracy of battery internal short-circuit fault diagnosis, enhancing the system's robustness.

[0052] Furthermore, S4 includes:

[0053] S41. Calculate the discrete Fréchet distance of the SOC curve between each cell and the reference cell, and generate a single cell curve plot after time changes.

[0054] S42. Determine the benchmark curve for deformation anomaly detection based on the individual cell curves of all batteries;

[0055] S43. Calculate the deformation anomaly degree of the SOC characteristic curve of each battery cell, and then combine the deformation anomaly degree with the depth value MBD. Change the vertical distance from the original depth point (MEL,MBD) to the parabola, and compare it with the reference curve to finally identify the anomaly value.

[0056] Beneficial effects: Based on the discrete Fréchet distance of the SOC curve of a single cell estimated by the UKF algorithm, it can sensitively capture the small differences between the SOC characteristic curves of different battery cells and detect battery faults in the early stage; it can better adapt to the complex behavior of battery systems, effectively filter noise and interference, and is suitable for real-time online detection application scenarios. Attached Figure Description

[0057] Figure 1 A flowchart of a short-circuit diagnosis method for the power battery of a new energy vehicle in use;

[0058] Figure 2 A schematic diagram illustrating the process of constructing a battery SOC estimation method using FFRLS combined with the UKF algorithm;

[0059] Figure 3 The voltage curve of the 5Ω internal short-circuit resistance under DST conditions;

[0060] Figure 4 This is a schematic diagram of the second-order RC equivalent circuit model;

[0061] Figure 5 This is a schematic diagram of the SOC estimation results under dynamic conditions.

[0062] Figure 6 The result is the short-circuit detection result within a 5Ω short-circuit resistor. Detailed Implementation

[0063] The following detailed description illustrates the specific implementation method:

[0064] Example 1

[0065] An algorithm for diagnosing short circuits in the power batteries of new energy vehicles, such as... Figure 1 As shown, it includes the following steps:

[0066] S1. Collect real-time operating data of the power battery.

[0067] When the battery is charging or discharging, the voltage (V), total current (I) flowing into the battery, and sampling time (t) of each battery cell are collected in real time through vehicle-side monitoring or monitoring data uploaded to the cloud data platform in real time, providing a data basis for subsequent analysis and fault diagnosis.

[0068] Based on the different models of the power batteries, the battery operating data is classified and several sets are established. Each set includes the operating data of several power batteries of the same model but different serial numbers. Subsequent modeling analysis and fault diagnosis algorithms are all targeted at a specific set.

[0069] S2. Construct an FFRLS algorithm combined with a UKF algorithm to identify battery operating parameters and estimate the battery's SOC in real time. The algorithm flow is as follows: Figure 2 As shown, it includes the following steps:

[0070] S21. Using the voltage sequence data collected in real time, select the cell with the smallest standard deviation of the voltage sequence as the reference cell and calculate the voltage curve of the reference cell.

[0071] Figure 3 The figure shows the voltage curve of the 5Ω internal short-circuit resistor under DST conditions.

[0072] S22. Establish the equivalent circuit model and the discretization equations corresponding to the equivalent circuit model.

[0073] The equivalent circuit model uses some basic circuit components to construct a circuit model that simulates the operating characteristics of a battery, and also establishes the relationship between the external characteristics exhibited by the battery during operation and its internal state. This method has the advantages of low computational cost, easy parameter identification, and high accuracy, making model verification easier to achieve. Therefore, it is widely used in practical engineering applications.

[0074] In this embodiment, the following is adopted: Figure 4 The second-order RC model is shown. In this model, U t U represents the terminal voltage of a lithium-ion battery. ocv R0 represents the open-circuit voltage of the lithium-ion battery, R0 represents the ohmic resistance, which is used to describe the internal ohmic resistance of the battery, and the RC network, i.e., the resistor-capacitor network, is used to describe the resistance generated inside the battery due to polarization reactions.

[0075] The dynamic equations of the second-order RC equivalent circuit model are summarized as follows:

[0076]

[0077] In the formula: I represents current, which is negative during discharge and positive during charging, and the unit is A;

[0078] Ut represents the terminal voltage, in volts (V).

[0079] U ocv This represents the open-circuit voltage, in volts (V).

[0080] R0 represents the ohmic internal resistance, simulating the relationship between the terminal voltage and the open-circuit voltage during charging and discharging, in Ω;

[0081] R1 and R2 represent the polarization resistances of the first and second RC networks, respectively, in Ω.

[0082] C1 and C2 represent the polarization capacitances of the first and second RC networks, respectively, in F.

[0083] These represent the currents flowing through resistors R1 and R2, respectively, in amperes (A).

[0084] U1(t) and U2(t) represent the voltage values ​​across the two RC networks, in units of V;

[0085] η(t) represents the Coulomb efficiency, which is generally considered to be equal to 1 during discharge and not greater than 1 during charging.

[0086] Based on the kinetic equations, the discretization equations for the second-order RC equivalent battery model are established as follows:

[0087]

[0088] S23. Based on the discretized equations of the equivalent circuit model, the online parameters of the equivalent circuit model are identified using the FFRLS method.

[0089] A crucial foundation for ensuring the accuracy of SOC estimation is the accurate identification of relevant parameters in the second-order RC equivalent circuit model. This allows for the real-time acquisition of key system parameters, reflecting the current operating status of the equipment or system, and promptly identifying whether parameter changes deviate from the normal range, thus providing a basis for early fault warning.

[0090] This embodiment introduces recursive least squares with a variable forgetting factor (FFRLS) for dynamic parameter identification, which is particularly suitable for online estimation of parameters of equivalent circuit models. This method can adapt to changes in system parameters over time, while balancing the importance of historical and current data by adjusting the forgetting factor. The algorithm includes the following steps:

[0091] S231. Obtain the transfer function of the equivalent circuit model based on the dynamic equations of the equivalent circuit model:

[0092] When performing parameter identification on the established equivalent circuit model based on FFRLS, it is necessary to obtain the transfer function of the established circuit model. The least-squares form of the system input and output is obtained by transforming the expression of the second-order RC equivalent circuit model:

[0093]

[0094] Where y(k)=E(k) represents the system output (such as voltage, current, etc.);

[0095] θ(k) = [θ1θ2θ3θ4θ5] T , where is the parameter vector to be identified;

[0096]

[0097] v k This indicates measurement noise.

[0098] S232. Calculate the variable forgetting factor λ and the gain matrix K0:

[0099] Recursive least squares (RLS) algorithms can minimize computation while maintaining accuracy. The basic principle is to use the estimated value from the previous time step and the measured value from the current time step to recursively obtain the estimated value for the current time step. However, as the recursion process progresses, new data can be influenced by historical data, leading to slow algorithm response and decreased computational accuracy. Therefore, a variable forgetting factor is introduced into the RLS algorithm to reduce the impact of historical data on the recursive algorithm and improve the effectiveness of new data and computational accuracy.

[0100] The objective function of FFRLS can be defined as the output voltage function E obtained after each parameter update, and the calculation formula is as follows:

[0101]

[0102] In the formula, E(θ) is the actual output value, and E(θ,λ) is the model prediction value based on the current forgetting factor.

[0103] In the formula, λ is the forgetting factor, which is used to assign weights to new and old data. Typically, the forgetting factor λ in FFRLS is assigned a value between 0.9 and 0.999. A smaller value results in stronger tracking ability, but may lead to parameter jumps; a larger value slows down the algorithm; a value of 1 degenerates FFRLS into recursive least squares. In the above formula, P0(k) is the covariance matrix at time k, which is usually assigned a value based on practical experience.

[0104] To adaptively set the forgetting factor λ during the gain matrix calculation, a simulated annealing algorithm is introduced to find the globally stochastic optimal solution for the forgetting factor.

[0105] The steps of the simulated annealing algorithm include:

[0106] First, set the initial forgetting factor λ0, initial temperature T0, and cooling rate α related to the calculation.

[0107] To meet the boundary requirements of FFRLS, the range of the randomly generated forgetting factor is limited to the interval [0.9, 0.999]. In each iteration, the objective function and initial value are substituted into the acceptance probability formula, and the decision to accept the new solution is made based on the current objective function value f(λ) and the objective function value f(λ′) of the new forgetting factor: if f(λ) ≥ f(λ′), the new solution is accepted; otherwise, a random number r in the range [0,1] is generated. If r < λ... accept If the new solution is not found, the new solution will be accepted; otherwise, the current solution λ will be retained. The probability formula for acceptance is as follows:

[0108]

[0109] Next step: Update temperature value.

[0110] T new =α·T

[0111] At high temperatures, the computation process is more likely to accept a poor solution, allowing the algorithm to explore a wider solution space. As the temperature decreases, the probability of accepting a poor solution decreases, and the algorithm tends to converge to a better solution.

[0112] When the set maximum number of iterations is reached or the temperature drops to the threshold, the algorithm stops and outputs the current optimal forgetting factor.

[0113] The formula for calculating the corresponding gain matrix K0 is as follows:

[0114]

[0115] S233. Update the covariance matrix P0(k) using the following formula:

[0116]

[0117] S234. Update the parameter θ(k) to be estimated, as shown in the following formula:

[0118]

[0119] In the formula, the initial value θ(0) is a sufficiently small real matrix.

[0120] S235. Return the parameter identification value of the next time step to S232, and iterate through S232 to S235 until the change in the covariance matrix is ​​less than the set threshold. The FFRLS algorithm ends.

[0121] It should be noted that in the least squares recursive algorithm containing genetic factors, the open-circuit voltage at time k is obtained by fitting the corresponding SOC-OCV curve. Based on the open-circuit voltage identification experiment, the fitting polynomials for SOC and OCV under charging and discharging states can be obtained. Furthermore, the FFRLS algorithm can be implemented through MATLAB programming.

[0122] S24. Online estimation of battery SOC based on the unscented Kalman filter algorithm UKF.

[0123] In practical engineering applications, most systems or models are nonlinear. To achieve filtering in nonlinear systems, the unscented Kalman filter algorithm approximates the probability density distribution of the nonlinear function, using a series of deterministic samples to approximate the posterior probability density of the state. It handles the nonlinear propagation problem of the mean and covariance through the UKF unscented transform, avoiding the process of linearizing the nonlinear function. It eliminates the need for approximation of the nonlinear function and differentiation of the Jacobian matrix, thus performing Kalman filtering more accurately. The steps include:

[0124] S241. Assign initial values ​​to the model parameters.

[0125] Before starting the algorithm, x0, P0, and Q should be evaluated. k and R k Assign initial values, where x0 is the initial value of the polarization capacitor voltage, P0 is the initial covariance matrix, and Q... k Let R be the process noise covariance matrix. k To observe the noise covariance matrix. The state variables in the battery model are determined by the state of charge (SOC).k It is composed of polarization capacitor voltages U1(k) and U2(k). In MATLAB programming, the battery open-circuit voltage U is used. ocv Determine SOC k The value of x0 is such that the polarization capacitor voltage is generally zero, therefore x0 can be determined, and x0 has high fault tolerance; P0 is the identity matrix, Q k and R k Values ​​are determined based on experience.

[0126] S242. Determine the state equation vector and observation vector equation of the battery system.

[0127] To successfully apply the UKF algorithm to lithium-ion battery SOC estimation, it is necessary to determine the state vector, observation equation, and output vector of the battery as the controlled system.

[0128] The battery SOC and the voltages U1 and U2 of the two RC circuits are selected as the state vectors of the system model. The change equation of the battery model terminal voltage is used as the observation equation of the system. In addition, since these variables are affected by the current, the current is used as the input of the entire system.

[0129] Based on the established equivalent circuit model and related expressions, the state equations of the discrete battery system are obtained as follows:

[0130]

[0131] The observation equations for the discretized battery system are as follows:

[0132] U t (k)=U OCV (SOC(k))-R0I(k)-U1(k)-U2(k)

[0133] The state vector equation and observation vector equation can be obtained from the system's state equation and observation equation.

[0134] The state vector equations are as follows:

[0135]

[0136] make:

[0137]

[0138] u(k-1)=I(k-1)

[0139] The state equation of the battery is then obtained as follows:

[0140] x k =Ax k-1 +Bu k-1 +wk-1

[0141] The observation vector equation is as follows:

[0142] U t (k)=U OCV (SOC(k))-R0I(k)-U1(k)-U2(k)

[0143] Similarly, let:

[0144]

[0145] D = [-r0]

[0146] The state equation of the battery is then obtained as follows:

[0147] Y k =cX k +dU k +V k

[0148] S243. After obtaining the state equation and observation equation of the battery system, the optimal estimate of the battery SOC is achieved according to the recursive steps of the UKF algorithm.

[0149] For a nonlinear discrete system, its state equations and observation equations are as follows:

[0150]

[0151] Among them, the nonlinear function F(X) k U k ) and g(x k ,u k It is differentiable at all sampling points. k Let u be the system state variable at time k. k Z represents the system input at time k. k w represents the system output at time k. k For system process noise, v k For system observation noise, w k v k The noise is composed of independent Gaussian white noise, with process noise and observation noise covariance matrices Q and R, respectively. UKF is used to obtain the mean of the current system state. The covariance P is used to generate sigma points. The mean and covariance of the predicted state are obtained by weighted averaging of these sigma points, and the state estimate and covariance matrix are updated using Kalman gain. The specific steps are as follows:

[0152] (1) Obtain a set of sampling points (called the Sigma point set) using the unscented transformation formula:

[0153]

[0154] The weight of each sampling point is:

[0155]

[0156] (2) Calculate the prediction for the 2n+1 Sigma point set:

[0157] X (i) (k+1|k)=f[k,x (i) [(K|k)],i=1~2n+1

[0158] (3) Calculate the prediction and covariance matrix of the system state variables:

[0159]

[0160] (4) Generate a new Sigma point set using unscented transformation based on the prediction results:

[0161]

[0162] (5) Substitute the predicted Sigma point set into the observation equation to obtain the predicted observations:

[0163] Z (i) (k+1|k)=h[X (i) [(k+1|k)], i = 1 to 2n+1

[0164] (6) Obtain the mean and covariance of the system predictions by weighted summation of the observed predictions:

[0165]

[0166] (7) Calculate the cross covariance and Kalman gain:

[0167]

[0168] (8) Calculate the state update and covariance update of the system:

[0169]

[0170] P(k+1|k+1)=P(k+1|k)-K(k+1)P xz K T (k+1) The SOC estimation results under dynamic conditions based on the UKF algorithm are as follows: Figure 5 As shown.

[0171] S3. Construct a fault characteristic function for the difference in SOC curve between each battery cell and the reference battery.

[0172] By utilizing the explosive growth characteristic of the exponential function to construct a fault characteristic function, the subtle differences in the SOC curves between individual battery cells and a reference cell are amplified, which helps to better identify potential internal short-circuit faults. The SOC value does not follow sudden changes in current and fluctuates less compared to voltage or current parameters, effectively reducing the probability of false alarms. Specifically:

[0173] t0-t e The data point at time is defined as

[0174]

[0175] Where n is the number of voltage data points collected by the sensor. The mean value at each time step is calculated using these data points. for:

[0176]

[0177] The standard deviation of the data points at each time step is:

[0178]

[0179] Sort the data points at each time step from largest to smallest, and define them as follows: Then, the median of the voltage data at each time point is calculated:

[0180]

[0181] The reference data for each time step is defined as follows:

[0182] and

[0183] Where, n sel The number of voltage data points to satisfy the inequality.

[0184] The fault characteristic function is then:

[0185]

[0186] S4. Substitute the identification data of each battery into the fault characteristic function to perform internal short-circuit fault detection. This includes:

[0187] S41. Calculate the discrete Fréchet distance of the SOC curve between each cell and the reference cell, and generate a single cell curve graph that changes over time.

[0188] The Fraser distance is a method for quantifying path space similarity. It considers the time series space between curves and quantifies curve similarity by calculating the distance between two curves. Therefore, it is used for fault feature extraction.

[0189] Assuming curve The curve consists of p data points. Composed of q data points, passing through curve X p And curve Y q We can obtain a sequence pair dataset L, which is represented as follows:

[0190]

[0191] In the formula, a1=1, b1=1, a m =p,b m =q, for any i = 1, 2, ..., m, we have a i+1 =a i Or a i+1 =a i +1, and b i+1 =b i The curve X is defined using the sequence pair dataset L. p And curve Y q The distance between them is

[0192]

[0193] In the formula Let L be the Euclidean distance between sequence pairs, and ||L(X) p ,Y q )|| represents the maximum value of the Euclidean distance between the sequence pairs.

[0194] The discrete Friesian distance is defined as:

[0195] DFD(X p ,Y q )=min||L(X p ,Y q )||

[0196] Then data points With average voltage The discrete Frescher distance between them is

[0197]

[0198] To diagnose faults in real time, a sliding window-based forgetting mechanism is used to calculate the discrete Fréchet distance at each sampling time.

[0199] S42. Determine the benchmark curve for deformation anomaly detection based on the individual cell curves of all batteries.

[0200] At the data level, anomalous individual characteristic curves occur frequently. To detect abnormal fluctuations in these characteristic curves, a function-based outlier detection algorithm based on deformation anomaly degree is used. The deformation anomaly degree detection curve reflects the degree to which its shape deviates from the normal curve; the greater the deviation, the more obvious the anomaly of the curve.

[0201] When selecting the baseline normal curve, to reduce the impact of outliers, the curves are sorted according to their depth values, and only the portions with larger depth values ​​are averaged. This results in an average curve that is less affected by outliers and more representative of the fluctuations in normal data. The specific steps are as follows:

[0202] (1) Calculate the MBD, MEI, and parabolic value P of the i-th curve. i :

[0203] MBD primarily measures the frequency with which curve i forms a banded region with other curves. The main formula is:

[0204]

[0205] Where n is the total number of curves and m is the number of time points, usually taken as m = [n / 2] + 1. Considering all the bands formed by pairs of curves, calculate the proportion of the band area contained in the minimum and maximum values ​​of the target curve at each time point. I(·) is an indicator function; if x i (t) satisfies min(t)≤x on all t. i (t k If t ≤ max(t), then increment the count by 1. After calculating all proportions, take the average of all curve pairs.

[0206] MEI primarily combines the positional relationship between the upper and lower envelopes of the curve to reflect the degree to which the curve lies in the upper part of the dataset. The main formula is:

[0207]

[0208] Calculate curve x i Exceeding other curves x at each time point j The frequency of each curve and time point is then averaged. For ease of subsequent calculations, MBD and MEI calculations are described as follows:

[0209]

[0210] The formula for the value of a parabola is:

[0211] (2) Determine the baseline curve:

[0212] For x1,...,x nFind the derivative curve x′(t), and if the data is discrete, perform a fitting operation.

[0213] Sort x′(t) from largest to smallest based on the statistical depth values ​​to obtain x′ (1) ,...,x′ (n) .

[0214] After removing the curves with the smallest depth value [an] (the largest integer not greater than an), calculate the mean to obtain the baseline curve x′. mean (t):

[0215]

[0216] S43. Calculate the deformation anomaly degree of the SOC characteristic curve of each battery cell, then combine the deformation anomaly degree with the depth value MBD to establish a depth point, change the vertical distance from the original depth point to the parabola, and compare it with the baseline curve to finally identify the outlier. The specific steps are as follows:

[0217] (1) Calculate the shape fluctuation deviation s i :

[0218] s i =∫ I (x′ i =x′ mean )ω(t)dt;

[0219] (2) Normalized deviation s i Calculate the deformation anomaly S i :

[0220]

[0221] (3) Determine the deformation anomaly coefficient β, and add the deformation anomaly to mb. i Get Amb i :

[0222]

[0223] Amb i =mb i -βs i .

[0224] (4) Calculate the perpendicular distance d from the point (MEL, MBD) to the parabola. i Calculate the cutoff:

[0225] D i =p i -amb i i = 1, ..., n

[0226] Where MEL represents the horizontal position of the measurement point; MBD represents the depth value of the measurement point.

[0227] (5) Calculate d i upper quartile q d3 and interquartile range (IQR) d This serves as a parameter for subsequent calculation of the anomaly boundary cutoff:

[0228] cutoff = q d3 +1.5×IQR d

[0229] (6) Identifying outliers based on outlier boundary cutoff:

[0230] SO={i|d i ≥cutoff}

[0231] When the deformation anomaly of a battery relative to the SOC curve of a reference battery exceeds the anomaly cutoff, it is determined that the battery has experienced an internal short-circuit fault. Figure 6 As shown, an alarm message is generated for each battery cell identified as having an internal short circuit fault.

[0232] The above are merely embodiments of the present invention, and the invention is not limited to the fields covered by these embodiments. Commonly known structures and characteristics in the solutions are not described in detail here. It should be noted that those skilled in the art can make various modifications and improvements without departing from the structure of the present invention, and these should also be considered within the scope of protection of the present invention. These modifications and improvements will not affect the effectiveness of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. A method for diagnosing internal short circuits in the power batteries of new energy vehicles, characterized in that: Includes the following steps: S1. Collect real-time operating data of the power battery; S2. Construct an FFRLS algorithm combined with the UKF algorithm to identify battery operating parameters and estimate the battery's SOC in real time; S3. Construct a fault characteristic function to quantify the difference between the SOC curves of each battery and the reference battery based on the following method. ,Will The data point at time is defined as: Where n is the number of voltage data points collected by the sensor; Calculate the mean at each time step using the data points. for: The standard deviation of the data points at each time step is: Sort the data points at each time step from largest to smallest, and define them as follows: Then, the median of the voltage data at each time point is calculated: The reference data for each time step is defined as follows: in, The number of voltage data points required to satisfy the inequality; The fault characteristic function is then: ; S4. Substitute the identification data of each battery into the fault characteristic function to perform internal short circuit fault detection.

2. The method for diagnosing internal short circuits in the power battery of a new energy vehicle according to claim 1, characterized in that: S2 includes: S21. Using the voltage sequence data collected in real time, select the cell with the smallest standard deviation of the voltage sequence as the reference cell and calculate the voltage curve of the reference cell. S22. Establish the equivalent circuit model and the discretized equations corresponding to the equivalent circuit model; S23. Based on the discretized equations of the equivalent circuit model, the online parameters of the equivalent circuit model are identified using the FFRLS method with a variable forgetting factor. S24. Online estimation of battery SOC based on the unscented Kalman filter algorithm UKF.

3. The method for diagnosing internal short circuits in the power battery of a new energy vehicle according to claim 2, characterized in that: S23 includes: S231. Obtain the transfer function of the equivalent circuit model based on the dynamic equation of the equivalent circuit model; S232. Calculate the variable forgetting factor λ and the gain matrix. ; S233, regarding the covariance matrix Update; S234, Parameters to be estimated Update; S235. Return the parameter identification value of the next time step to S232, and iterate through S232 to S235. When the change in the covariance matrix is ​​less than the set threshold, the FFRLS algorithm ends.

4. The method for diagnosing internal short circuits in the power battery of a new energy vehicle according to claim 3, characterized in that: S232 uses the simulated annealing algorithm to solve for the optimal forgetting factor.

5. The method for diagnosing internal short circuits in the power battery of a new energy vehicle according to claim 4, characterized in that: S24 includes: S241. Assign initial values ​​to the model parameters; S242. Determine the state equation vector and observation vector equation of the battery system; S243. After obtaining the state equation and observation equation of the battery system, the optimal estimate of the battery SOC is achieved according to the recursive steps of the UKF algorithm.

6. The method for diagnosing internal short circuits in the power battery of a new energy vehicle according to claim 1, characterized in that: The S4 uses a functional data outlier detection algorithm based on deformation anomaly degree to determine the internal short circuit fault of the power battery.

7. A method for diagnosing internal short circuits in the power battery of a new energy vehicle according to claim 6, characterized in that: S4 includes: S41. Calculate the discrete Fréchet distance of the SOC curve between each cell and the reference cell, and generate a single cell curve graph that changes over time. S42. Determine the benchmark curve for deformation anomaly detection based on the individual cell curves of all batteries; S43. Calculate the deformation anomaly degree of the SOC characteristic curve of each battery cell, then combine the deformation anomaly degree with the depth value MBD, change the vertical distance from the original depth point to the parabola, and finally identify the anomaly value by comparing it with the reference curve.

Images