A method for detecting and locating an initial short circuit fault of a vehicle power battery pack

Abstract

The present invention discloses a method for detecting and locating an initial short-circuit fault of a vehicle-mounted power battery pack, which belongs to the field of power battery pack voltage initial short-circuit fault diagnosis, and the steps are: establishing a battery voltage signal measurement scheme; detecting an initial short-circuit fault signal evaluation and establishing a detection threshold; calculating a real-time initial short-circuit fault detection index; and determining the location of the initial short-circuit fault. The present invention takes into account the impact of the inconsistency between single cells in the battery pack under the operating condition fluctuation of the electric vehicle on the initial short-circuit fault detection performance. The virtual voltage measurement scheme adopted adjusts the measurement channel in the virtual space so that the sampled voltage vector can be rearranged, thereby reducing the impact of voltage fluctuations under normal operating conditions on the fault detection and positioning process; the designed fault detection index establishes a quantitative correlation between the index threshold and the fault amplitude, thereby being able to provide an effective estimate of the severity of the short-circuit fault, and ultimately achieving adaptive adjustment of the detection threshold for a variety of scenarios.

Description

Technical Field

[0001] The present invention relates to the field of early stage short circuit fault diagnosis of power battery packs, and in particular to an early stage short circuit fault detection and positioning method for a vehicle-mounted power battery pack. Background Art

[0002] As a means of transportation, the safety of electric vehicles has received widespread attention. Vehicle fires, explosions and other severe incidents caused by short-circuit failures in on-board battery packs have occurred one after another. Timely detection and control of electrical short-circuit failures in the early stages can avoid the occurrence of the above accidents to a great extent. Therefore, early short-circuit fault diagnosis of power battery packs is crucial to the safe and reliable operation of electric vehicles.

[0003] Existing power battery pack short-circuit fault diagnosis methods are mainly divided into model-driven methods and data-driven methods. Model-based fault detection methods convert fault detection problems into model parameter or state estimation. By comparing the estimated value with the measured value, the residual of the parameter or state is obtained, and the residual is used to diagnose the battery short-circuit fault. However, the estimation effectiveness of this type of method depends on the accuracy of the battery model. The initial short-circuit fault has the characteristics of weak signal characteristics. The single battery model has limited ability to estimate the battery dynamics and is insufficient to support effective initial short-circuit fault diagnosis. The multi-model coupling used to improve the state estimation capability requires a high computing hardware cost, which will hinder the actual online application of the method; the data-driven method no longer relies on the battery model, but performs statistical analysis on the sampled data, extracts abnormal features, and then performs fault diagnosis. Its diagnostic effectiveness depends on the significance of the abnormal features. At the same time, the original physical meaning of the data will be lost during the data processing and analysis process, resulting in a lack of sufficient fault information in the diagnosis results, and it is impossible to effectively distinguish the evolution stage of the fault through the diagnosis results. Therefore, the current data-driven method still has deficiencies in the diagnostic performance of initial short-circuit faults that need to be made up. Summary of the invention

[0004] In view of the above-mentioned technical problems existing in the prior art, the present invention provides a method for detecting and locating an initial short-circuit fault of a vehicle-mounted power battery pack, which has a reasonable design, solves the shortcomings of the prior art, and has good effects.

[0005] In order to achieve the above object, the present invention adopts the following technical scheme:

[0006] A method for detecting and locating an initial short circuit fault of a vehicle power battery pack voltage comprises the following steps:

[0007] S1: Establish a battery voltage signal measurement plan;

[0008] Preset the near-cut-off voltage to align the charging procedure, and adaptively determine the corresponding near-cut-off voltage according to the battery type; construct a virtual transformation matrix, and rearrange the order of sampling channels in the sampled voltage column vector through a virtual transformation voltage measurement sequence;

[0009] S2: It can detect the initial short circuit fault signal amplitude evaluation and establish the detection threshold;

[0010] Obtain the reference differential voltage mean vector and the reference differential voltage standard deviation diagonal matrix of the reference differential sampling voltage timing matrix after virtual transformation under the reference working condition of the power battery pack; calculate the covariance matrix of the reference differential sampling voltage timing matrix after virtual transformation after standardization, and obtain its maximum reference eigenvalue and maximum reference eigenvalue standard deviation; set the detectable initial short circuit fault signal amplitude, and obtain the detection index threshold;

[0011] S3: Calculate real-time fault detection indicators;

[0012] According to the voltage data acquired in real time, a normalized and virtually transformed real-time differential sampling voltage time series matrix of the power battery pack is constructed, and the corresponding real-time initial short-circuit fault detection index is calculated;

[0013] S4: Fault location;

[0014] When the initial short-circuit fault detection index exceeds the detection index threshold, the abnormal characteristic value contribution of the differential voltage channel of the fault is calculated to determine the occurrence location of the initial short-circuit fault.

[0015] Furthermore, step S1 specifically includes the following steps:

[0016] S101: determining the near cut-off voltage for alignment according to the operating voltage history data of the on-board power battery pack of the electric vehicle, and adaptively selecting the near cut-off voltage to be aligned for different types of batteries. If the voltage of the on-board power battery pack is aligned near the near cut-off voltage, the battery balancing in the battery management system will be disabled in the remaining discharge and subsequent charge and discharge cycles;

[0017] S102: Obtaining a reference sampling voltage timing matrix of a series power battery pack based on a reference operating condition during a complete discharge process Its expression is shown in formula (1):

[0018]

[0019] Where n is the number of voltage sensors, and m is the time series length of the reference sampling voltage timing matrix; The reference sampling voltage timing matrix The reference sampling voltage column vector at time t in , for The reference sampling voltage value of the i-th sampling channel in; wherein, i=1,2,…,n, t=1,2,…,m; the reference operating condition is the light vehicle test cycle operating condition, which is a discharge current operating condition, that is, the state of charge of the power battery pack at the end of the operating condition is lower than the state of charge of the power battery pack at the beginning of the operating condition; therefore, Corresponding to the reference sampling voltage column vector when the battery is fully charged, A reference sampling voltage vector corresponding to the case where the battery is completely exhausted;

[0020] In formula (1), the reference sampling voltage column vector The expression of is shown in formula (2):

[0021]

[0022] in, is the voltage of the ith single cell at time t, R i,i+1 is the resistance of the connector between single cells, R b,1 and R n,b are the connecting resistances of the batteries at both ends of the series battery pack and the busbar, R i,i+1 , R b,1 , R n,b The resistance value is the same;

[0023] S103: Sampling the reference voltage timing matrix Perform differential operation to obtain the reference differential sampling voltage timing matrix The expression is as follows:

[0024]

[0025] in, The reference differential sampling voltage timing matrix The reference differential sampling voltage column vector at the tth moment in , for The differential sampling voltage value between the i-th sampling channel and the j-th sampling channel in , and In the above example, the reference differential sampling voltage column vector is The specific difference calculation is shown in the following formula:

[0026]

[0027] Among them, A d As a difference operation matrix, the specific elements are as follows:

[0028]

[0029] S104: Rearrange the order of sampling channels in the sampled voltage column vector by using a virtual conversion voltage measurement sequence, as shown in the following formula:

[0030]

[0031] Among them, A v is the virtual transformation matrix; is the virtual reference sampling voltage sequence rearranged after virtual transformation; where A v Each element Satisfy the following formula:

[0032]

[0033] in, express The jth element in is the jth sampling channel; express The i-th element in , that is, the i-th sampling channel rearranged after virtual transformation;

[0034] Then the final reference differential sampling voltage timing matrix after virtual transformation can be expressed by the following formula:

[0035]

[0036] Furthermore, step S2 specifically includes the following steps:

[0037] S201: Obtaining a reference differential voltage mean vector and a reference differential voltage standard deviation diagonal matrix of a reference differential sampling voltage timing matrix after virtual transformation under a reference working condition of a power battery pack;

[0038] Reference differential voltage mean vector The expression is shown in formula (8):

[0039]

[0040] in, is the reference differential sampling voltage timing matrix after virtual transformation The average value of the reference differential voltage between the i-th sampling channel and the j-th sampling channel is calculated as follows:

[0041]

[0042] Diagonal matrix of reference differential voltage standard deviation As shown in formula (11):

[0043]

[0044] Where diag{·} represents a diagonal matrix with the elements in {·} as diagonal elements in sequence; Represents the reference differential sampling voltage timing matrix after virtual transformation The standard deviation of the reference differential voltage between the i-th sampling channel and the j-th sampling channel in is calculated as shown in formula (12):

[0045]

[0046] S202: The virtual transformed reference differential sampling voltage timing matrix Perform standardization operations as shown below:

[0047]

[0048] in, is the virtually transformed reference differential sampling voltage timing matrix The standardized matrix of

[0049] The standardized and virtually transformed reference differential sampling voltage timing matrix Divided into two groups of standardized and virtually transformed reference differential sampling voltage timing sub-matrices and The two sub-matrices are the reference differential sampling voltage timing matrix after normalization and virtual transformation The row element options are: choose The {1,2,…,n-1}th row of choose No. OK;

[0050] Standardized and virtually transformed reference differential sampling voltage timing sub-matrix and The maximum benchmark eigenvalue mean and The expression is shown in formula (14):

[0051]

[0052] Among them, λ max,k,base,1 and λ max,k,base,2 They are respectively the reference differential sampling voltage timing sub-matrices after normalization and virtual transformation and The kth sliding time sequence submatrix with ω as matrix width in and The covariance matrix of and The maximum eigenvalue of ; where the standardized sliding time series submatrix The timing sub-matrix of the reference differential sampling voltage after normalization and virtual transformation The k-ω+1 to k columns of the standard sliding time series submatrix are composed in sequence The timing sub-matrix of the reference differential sampling voltage after normalization and virtual transformation The k-ω+1 column to the k column are sequentially composed; k=ω,ω+1,…,m;

[0053] Sliding time submatrix and The covariance matrix of and The expression of is shown in formula (15):

[0054]

[0055] And there are:

[0056]

[0057] Among them, λ max,base,1 and λ max,base,2 They are respectively the reference differential sampling voltage timing sub-matrices after normalization and virtual transformation and The maximum base eigenvalue vector of ;

[0058] Standardized and virtually transformed reference differential sampling voltage timing sub-matrix and Maximum benchmark eigenvalue standard deviation and The calculation formula is shown in formula (17):

[0059]

[0060] S203: Setting a fault characteristic amplitude representing the severity of the fault. The minimum detectable fault characteristic amplitude for an initial short circuit fault can be f sc,min It is represented by the minimum detectable fault characteristic amplitude f of the initial short circuit fault. sc,min The fault detection threshold is determined by deriving the quantitative relationship between the fault detection threshold:

[0061]

[0062] Among them, D th Indicates the initial short-circuit fault detection threshold; t f represents the fault duration in the sliding window, represents the standard deviation of the (n-1) eigenvalues ​​of the virtual transformed reference differential sampling voltage timing submatrix after normalization, ξ l,baseIt represents the standard deviation of the lth voltage differential channel under the reference condition, that is, the diagonal matrix of the reference differential voltage standard deviation The lth element.

[0063] Furthermore, step S3 specifically includes the following steps:

[0064] S301: During the operation of the electric vehicle, as the voltage sensor measures, a standardized and virtually transformed real-time differential sampling voltage timing matrix of the power battery pack is constructed. The specific process is as follows:

[0065] Get the real-time voltage vector V of the power battery pack based on the standard measurement topology after virtual transformation at time t t sv , as shown in formula (19):

[0066]

[0067] in, is the real-time voltage value of the i-th sampling channel obtained based on the standard measurement topology after virtual transformation at time t;

[0068] Calculate the real-time differential voltage vector ΔV of the power battery pack based on the virtual transformation of the standard measurement topology at time t t sv As shown in formula (20):

[0069]

[0070] When t≥ω, construct the real-time differential voltage measurement timing matrix of the power battery pack based on the standard measurement topology after virtual transformation at time t As shown in formula (21):

[0071]

[0072] The timing matrix of the real-time differential voltage measurement of the power battery pack after standardization and virtual transformation based on the standard measurement topology at time t As shown in formula (22):

[0073]

[0074] The real-time differential voltage measurement timing matrix of the power battery pack after standardization and virtual transformation Divided into two groups of normalized and virtually transformed power battery pack real-time differential voltage measurement timing sub-matrices and Among them, the real-time differential voltage measurement timing sub-matrix of the power battery pack after virtual transformation Real-time differential voltage measurement timing matrix of power battery pack after virtual transformation The 1st to n-1th rows are sequentially composed of the real-time differential voltage measurement timing sub-matrix of the power battery pack after virtual transformation Real-time differential voltage measurement timing matrix of the power battery pack after virtual transformation No. The rows are composed in order;

[0075] S302: The calculation steps of the real-time initial short-circuit fault detection index are specifically as follows:

[0076] Get the standardized real-time differential voltage measurement timing sub-matrix at time t and The covariance matrix of is shown in formula (23):

[0077]

[0078] The standardized real-time differential voltage measurement timing submatrix at time t is obtained by the eigenvalue solution method. and The covariance matrix of and The maximum eigenvalue λ max,t,1 and λ max,t,2 And its corresponding eigenvector is v max,t,1 and v max,t,2 ;

[0079] For the eigenvalue λ max,t,1 and λ max,t,2 They are standardized respectively, as shown in formula (24):

[0080]

[0081] in, and They are the standardized real-time differential voltage measurement timing sub-matrices at time t and The covariance matrix of and The standardized maximum eigenvalue of ;

[0082] Get the real-time initial short circuit fault detection index D of the power battery pack at time t t , its expression is shown in formula (25):

[0083]

[0084] Furthermore, step S4 specifically includes the following steps:

[0085] When the real-time initial short-circuit fault detection index D at time t t Satisfy D t >D thWhen the battery pack is short-circuited, it indicates that the battery pack has an initial short-circuit fault;

[0086] Then the fault location is carried out. The specific steps for locating the location of the initial short circuit fault of the battery are as follows:

[0087] S401: Formula (26) can be used to calculate the contribution of the eigenvector corresponding to the maximum eigenvalue after normalization:

[0088]

[0089] in, Represents the maximum eigenvalue λ after normalization max,t Eigenvector v λmax,t The kth item in ;

[0090] Grouped standardized real-time differential voltage measurement timing sub-matrix and The covariance matrix of and Finally, the real-time voltage sampling channel contribution vector η of the initial short-circuit fault at time t is t ={η k',t}Calculate as in formula (27), where k'=1,…,n;

[0091]

[0092] in, and Respectively represent the standardized real-time differential voltage measurement timing sub-matrices and The covariance matrix of and The contribution generated by the k′th item in the eigenvector corresponding to the abnormal maximum eigenvalue;

[0093] S402: When the real-time differential voltage channel contribution vector η of the initial short-circuit fault at time t t The largest two elements η p,t and η q,t The subscripts p and q satisfy equation (28):

[0094]

[0095] And there are:

[0096]

[0097] Then the initial short-circuit fault occurs at the pth single cell at time t. In formula (29), η th =0.8.

[0098] Beneficial technical effects brought by the present invention:

[0099] The present invention fully considers the influence of inconsistency between single cells in a battery pack under fluctuations in operating conditions of electric vehicles on the performance of initial short-circuit fault detection. The virtual voltage measurement scheme adopted can rearrange the sampled voltage vector by adjusting the measurement channel in the virtual space, thereby reducing the influence of voltage fluctuations under normal operating conditions on the fault detection and positioning process; and the fault detection index designed by the present invention based on full-dimensional statistical analysis establishes a quantitative correlation between the index threshold and the fault amplitude, thereby being able to provide an effective estimation of the severity of the short-circuit fault, and ultimately achieving adaptive adjustment of the detection threshold for a variety of scenarios, providing theoretical support for initial short-circuit fault detection of power battery packs. BRIEF DESCRIPTION OF THE DRAWINGS

[0100] Figure 1 This is a flow chart of the initial short-circuit fault detection and positioning of the vehicle-mounted power battery pack in the present invention;

[0101] Figure 2 It is a schematic diagram of the standard voltage measurement topology in the present invention; DETAILED DESCRIPTION

[0102] The specific implementation of the present invention is further described below in conjunction with the accompanying drawings and specific embodiments:

[0103] A method for detecting and locating an initial short-circuit fault of a vehicle power battery pack, such as Figure 1 As shown, the following steps are included:

[0104] S1: Establish a battery voltage signal measurement plan;

[0105] Preset the near-cut-off voltage to align the charging procedure, and adaptively determine the corresponding near-cut-off voltage according to the battery type; construct a virtual transformation matrix, and rearrange the order of sampling channels in the sampled voltage column vector through a virtual transformation voltage measurement sequence;

[0106] The specific steps include:

[0107] S101: According to Figure 2 The standard measurement topology shown in the figure determines the near cut-off voltage for alignment based on extensive historical data of operating voltage of on-board power battery packs of electric vehicles. Different types of batteries adaptively select the near cut-off voltage to be aligned. Once the on-board power battery pack voltage is aligned near the near cut-off voltage, the battery balancing in the battery management system will be disabled in the remaining discharge and subsequent charge and discharge cycles.

[0108] S102: Obtaining a reference sampling voltage timing matrix of a series power battery pack based on a reference operating condition during a complete discharge process Its expression is shown in formula (1):

[0109]

[0110] Where n is the number of voltage sensors, and m is the time series length of the reference sampling voltage timing matrix; The reference sampling voltage timing matrix The reference sampling voltage column vector at time t in , for The reference sampling voltage value of the i-th sampling channel in; wherein, i=1,2,…,n, t=1,2,…,m; the reference operating condition is the light vehicle test cycle operating condition, which is a discharge current operating condition, that is, the state of charge of the power battery pack at the end of the operating condition is lower than the state of charge of the power battery pack at the beginning of the operating condition; therefore, Corresponding to the reference sampling voltage column vector when the battery is fully charged, The reference sampling voltage vector corresponds to the case where the battery power is completely exhausted; and the battery state of charge can be calculated in real time according to the operating current, thereby obtaining the corresponding relationship between the battery voltage and the state of charge.

[0111] In formula (1), the reference sampling voltage column vector The expression of is shown in formula (2):

[0112]

[0113] in, is the voltage of the ith single cell at time t, R i,i+1 is the resistance of the connector between single cells, R b,1 and R n,b are the connecting resistances of the batteries at both ends of the series battery pack and the busbar, R i,i+1 , R b,1 , R n,b The resistance value is the same;

[0114] S103: In order to reduce the voltage fluctuation related to the working condition under normal working conditions, the reference sampling voltage timing matrix Perform differential operation to obtain the reference differential sampling voltage timing matrix The expression is as follows:

[0115]

[0116] in, The reference differential sampling voltage timing matrix The reference differential sampling voltage column vector at the tth moment in , for The differential sampling voltage value between the i-th sampling channel and the j-th sampling channel in , and In the above example, the reference differential sampling voltage column vector is The specific difference calculation is shown in the following formula:

[0117]

[0118] Among them, A d As a difference operation matrix, the specific elements are as follows:

[0119]

[0120] S104: Considering that the correlation between the open circuit voltage and the internal resistance between adjacent single cells in the battery pack is not optimal, which will cause additional differential voltage offset during differentiation, the order of the sampling channels in the sampled voltage column vector is rearranged through a virtual conversion voltage measurement sequence, as shown in the following formula:

[0121]

[0122] Among them, A v is the virtual transformation matrix; is the virtual reference sampling voltage sequence rearranged after virtual transformation; where A v Each element Satisfy the following formula:

[0123]

[0124] in, express The jth element in is the jth sampling channel; express The i-th element in , that is, the i-th sampling channel rearranged after virtual transformation;

[0125] Then the final reference differential sampling voltage timing matrix after virtual transformation can be expressed by the following formula:

[0126]

[0127] S2: It can detect the initial short circuit fault signal amplitude evaluation and establish the detection threshold;

[0128] Obtain the reference differential voltage mean vector and the reference differential voltage standard deviation diagonal matrix of the reference differential sampling voltage timing matrix after virtual transformation under the reference working condition of the power battery pack; calculate the covariance matrix of the reference differential sampling voltage timing matrix after virtual transformation after standardization, and obtain its maximum reference eigenvalue and maximum reference eigenvalue standard deviation; set the detectable initial short circuit fault signal amplitude, and obtain the detection index threshold;

[0129] The specific steps include:

[0130] S201: Obtaining a reference differential voltage mean vector and a reference differential voltage standard deviation diagonal matrix of a reference differential sampling voltage timing matrix after virtual transformation under a reference working condition of a power battery pack;

[0131] Reference differential voltage mean vector The expression is shown in formula (8):

[0132]

[0133] in, is the reference differential sampling voltage timing matrix after virtual transformation The average value of the reference differential voltage between the i-th sampling channel and the j-th sampling channel is calculated as follows:

[0134]

[0135] Diagonal matrix of reference differential voltage standard deviation As shown in formula (11):

[0136]

[0137] Where diag{·} represents a diagonal matrix with the elements in {·} as diagonal elements in sequence; Represents the reference differential sampling voltage timing matrix after virtual transformation The standard deviation of the reference differential voltage between the i-th sampling channel and the j-th sampling channel in is calculated as shown in formula (12):

[0138]

[0139] S202: The virtual transformed reference differential sampling voltage timing matrix Perform standardization operations as shown below:

[0140]

[0141] in, is the virtually transformed reference differential sampling voltage timing matrix The standardized matrix of

[0142] The standardized and virtually transformed reference differential sampling voltage timing matrix Divided into two groups of standardized and virtually transformed reference differential sampling voltage timing sub-matrices and The two sub-matrices are the reference differential sampling voltage timing matrix after normalization and virtual transformation The row element options are: choose The {1,2,…,n-1}th row of choose No. OK;

[0143] Standardized and virtually transformed reference differential sampling voltage timing sub-matrix and The maximum benchmark eigenvalue mean and The expression is shown in formula (14):

[0144]

[0145] Among them, λ max,k,base,1 and λ max,k,base,2 They are respectively the reference differential sampling voltage timing sub-matrices after normalization and virtual transformation and The kth sliding time sequence submatrix with ω as matrix width in and The covariance matrix of and The maximum eigenvalue of ; where the standardized sliding time series submatrix The timing sub-matrix of the reference differential sampling voltage after normalization and virtual transformation The k-ω+1 to k columns of the standard sliding time series submatrix are composed in sequence The timing sub-matrix of the reference differential sampling voltage after normalization and virtual transformation The k-ω+1 column to the k column are sequentially composed; k=ω,ω+1,…,m;

[0146] Sliding time submatrix and The covariance matrix of and The expression of is shown in formula (15):

[0147]

[0148] And there are:

[0149]

[0150] Among them, λ max,base,1 and λ max,base,2 They are respectively the reference differential sampling voltage timing sub-matrices after normalization and virtual transformation and The maximum base eigenvalue vector of ;

[0151] Standardized and virtually transformed reference differential sampling voltage timing sub-matrix and Maximum benchmark eigenvalue standard deviation and The calculation formula is shown in formula (17):

[0152]

[0153] S203: Setting a fault characteristic amplitude representing the severity of the fault. The minimum detectable fault characteristic amplitude for an initial short circuit fault can be f sc,min It is represented by the minimum detectable fault characteristic amplitude f of the initial short circuit fault. sc,min The fault detection threshold is determined by deriving the quantitative relationship between the fault detection threshold:

[0154]

[0155] Among them, D th Indicates the initial short-circuit fault detection threshold; t f represents the fault duration in the sliding window, represents the standard deviation of the (n-1) eigenvalues ​​of the virtual transformed reference differential sampling voltage timing submatrix after normalization, ξ l,base It represents the standard deviation of the lth voltage differential channel under the reference condition, that is, the diagonal matrix of the reference differential voltage standard deviation The lth element.

[0156] S3: Calculate real-time fault detection indicators;

[0157] According to the voltage data acquired in real time, a normalized and virtually transformed real-time differential sampling voltage time series matrix of the power battery pack is constructed, and the corresponding real-time initial short-circuit fault detection index is calculated;

[0158] The specific steps include:

[0159] S301: During the operation of the electric vehicle, as the voltage sensor measures, a standardized and virtually transformed real-time differential sampling voltage timing matrix of the power battery pack is constructed. The specific process is as follows:

[0160] Get the real-time voltage vector V of the power battery pack based on the standard measurement topology after virtual transformation at time t t sv , as shown in formula (19):

[0161]

[0162] in, is the real-time voltage value of the i-th sampling channel obtained based on the standard measurement topology after virtual transformation at time t;

[0163] Calculate the real-time differential voltage vector ΔV of the power battery pack based on the virtual transformation of the standard measurement topology at time t t sv As shown in formula (20):

[0164]

[0165] When t≥ω, construct the real-time differential voltage measurement timing matrix of the power battery pack based on the standard measurement topology after virtual transformation at time t As shown in formula (21):

[0166]

[0167] The timing matrix of the real-time differential voltage measurement of the power battery pack after standardization and virtual transformation based on the standard measurement topology at time t As shown in formula (22):

[0168]

[0169] The real-time differential voltage measurement timing matrix of the power battery pack after standardization and virtual transformation Divided into two groups of normalized and virtually transformed power battery pack real-time differential voltage measurement timing sub-matrices and Among them, the real-time differential voltage measurement timing sub-matrix of the power battery pack after virtual transformation Real-time differential voltage measurement timing matrix of power battery pack after virtual transformation The 1st to n-1th rows are sequentially composed of the real-time differential voltage measurement timing sub-matrix of the power battery pack after virtual transformation Real-time differential voltage measurement timing matrix of the power battery pack after virtual transformation No. The rows are composed in order;

[0170] S302: The calculation steps of the real-time initial short-circuit fault detection index are specifically as follows:

[0171] Get the standardized real-time differential voltage measurement timing sub-matrix at time t and The covariance matrix of is shown in formula (23):

[0172]

[0173] The standardized real-time differential voltage measurement timing submatrix at time t is obtained by the eigenvalue solution method. and The covariance matrix of and The maximum eigenvalue λ max,t,1 and λmax,t,2 And its corresponding eigenvector is v max,t,1 and v max,t,2 ;

[0174] For the eigenvalue λ max,t,1 and λ max,t,2 They are standardized respectively, as shown in formula (24):

[0175]

[0176] in, and They are the standardized real-time differential voltage measurement timing sub-matrices at time t and The covariance matrix of and The standardized maximum eigenvalue of ;

[0177] Get the real-time initial short circuit fault detection index D of the power battery pack at time t t , its expression is shown in formula (25):

[0178]

[0179] S4: Fault location

[0180] When the initial short-circuit fault detection index exceeds the detection index threshold, the abnormal characteristic value contribution of the differential voltage channel of the fault is calculated to determine the occurrence location of the initial short-circuit fault.

[0181] The specific steps include:

[0182] When the real-time initial short-circuit fault detection index D at time t t Satisfy D t <D th , it indicates that there is no fault, the system continues to run, and returns to step S3 to perform fault diagnosis at the next moment;

[0183] When the real-time initial short-circuit fault detection index D at time t t Satisfy D t >D th When the battery pack is short-circuited, it indicates that the battery pack has an initial short-circuit fault;

[0184] Then the fault location is carried out. The specific steps for locating the location of the initial short circuit fault of the battery are as follows:

[0185] S401: Calculate the eigenvector contribution corresponding to the normalized maximum abnormal eigenvalue using formula (26):

[0186]

[0187] in, Represents the maximum eigenvalue λ after normalization max,t Eigenvector The kth item in ;

[0188] Standardized real-time differential voltage measurement timing sub-matrix and The covariance matrix of and Finally, the real-time voltage sampling channel contribution vector η of the initial short-circuit fault at time t is t ={η k',t}Calculate as in formula (27), where k'=1,…,n;

[0189]

[0190] in, and Respectively represent the standardized real-time differential voltage measurement timing sub-matrices and The covariance matrix of and The contribution generated by the k′th item in the eigenvector corresponding to the abnormal maximum eigenvalue;

[0191] S402: When the real-time differential voltage channel contribution vector η of the initial short-circuit fault at time t t The largest two elements η p,t and η q,t The subscripts i and j of satisfy equation (28):

[0192]

[0193] And there are:

[0194]

[0195] At time t, the initial short-circuit fault occurs in the pth single cell, and the battery management system is shut down. In formula (29), η th =0.8.

[0196] If equation (28) is not satisfied, the battery management system does not shut down and returns to step S3.

[0197] The above is the complete implementation process of this embodiment.

[0198] Of course, the above description is not a limitation of the present invention, and the present invention is not limited to the above examples. Changes, modifications, additions or substitutions made by technicians in this technical field within the essential scope of the present invention should also fall within the protection scope of the present invention.

Claims

1. A method for detecting and locating an initial short-circuit fault of a vehicle power battery pack, characterized in that: The following steps are involved: S1: Establish a battery voltage signal measurement plan; Preset the near cut-off voltage to align the charging procedure, and adaptively determine the corresponding near cut-off voltage according to the battery type; Constructing a virtual transformation matrix, and rearranging the order of sampling channels in the sampled voltage column vector through a virtual transformation voltage measurement sequence; S2: It can detect the initial short circuit fault signal amplitude evaluation and establish the detection threshold; Obtaining a reference differential voltage mean vector and a reference differential voltage standard deviation diagonal matrix of a reference differential sampling voltage timing matrix after virtual transformation under a reference working condition of a power battery pack; Calculate the covariance matrix of the normalized and virtually transformed reference differential sampling voltage time series matrix, obtain its maximum reference eigenvalue and maximum reference eigenvalue standard deviation; set the detectable initial short circuit fault signal amplitude, and obtain the detection index threshold; S3: Calculate real-time fault detection indicators; According to the voltage data acquired in real time, a normalized and virtually transformed real-time differential sampling voltage time series matrix of the power battery pack is constructed, and the corresponding real-time initial short-circuit fault detection index is calculated; S4: Fault location; When the initial short-circuit fault detection index exceeds the detection index threshold, the abnormal characteristic value contribution of the differential voltage channel of the fault is calculated to determine the occurrence location of the initial short-circuit fault.

2. The method for detecting and locating an initial short-circuit fault of a vehicle power battery pack according to claim 1, characterized in that: Step S1 specifically includes the following steps: S101: determining the near cut-off voltage for alignment according to the operating voltage history data of the on-board power battery pack of the electric vehicle, and adaptively selecting the near cut-off voltage to be aligned for different types of batteries. If the voltage of the on-board power battery pack is aligned near the near cut-off voltage, the battery balancing in the battery management system will be disabled in the remaining discharge and subsequent charge and discharge cycles; S102: Obtaining a reference sampling voltage timing matrix of a series power battery pack based on a reference operating condition during a complete discharge process Its expression is shown in formula (1): Where n is the number of voltage sensors, and m is the time series length of the reference sampling voltage timing matrix; The reference sampling voltage timing matrix The reference sampling voltage column vector at time t in , for The reference sampling voltage value of the i-th sampling channel in; wherein, i=1,2,…,n, t=1,2,…,m; the reference operating condition is the light vehicle test cycle operating condition, which is a discharge current operating condition, that is, the state of charge of the power battery pack at the end of the operating condition is lower than the state of charge of the power battery pack at the beginning of the operating condition; therefore, The column vector of the reference sampled voltage corresponding to the fully charged battery, A reference sampling voltage vector corresponding to the case where the battery is completely exhausted; In formula (1), the reference sampling voltage column vector The expression of is shown in formula (2): in, is the voltage of the ith single cell at time t, R i,i+1 is the resistance of the connector between single cells, R b,1 and R n,b are the resistance of the connection between the batteries at both ends of the series battery pack and the busbar, R i,i+1 , R b,1 , R n,b The resistance value is the same; S103: Sampling the reference voltage timing matrix Perform differential operation to obtain the reference differential sampling voltage timing matrix ΔV b s ase , the expression is as follows: in, The reference differential sampling voltage timing matrix The reference differential sampling voltage column vector at the tth moment in , for The differential sampling voltage value between the i-th sampling channel and the j-th sampling channel in , and In formula (3), the reference differential sampling voltage column vector The specific difference calculation is shown in the following formula: Among them, A d As a difference operation matrix, the specific elements are as follows: S104: Rearrange the order of sampling channels in the sampled voltage column vector by using a virtual conversion voltage measurement sequence, as shown in the following formula: Among them, A v is the virtual transformation matrix; is the virtual reference sampling voltage sequence rearranged after virtual transformation; where A v Each element Satisfy the following formula: in, express The jth element in is the jth sampling channel; express The i-th element in , that is, the i-th sampling channel rearranged after virtual transformation; Then the final reference differential sampling voltage timing matrix after virtual transformation can be expressed by the following formula:

3. The method for detecting and locating an initial short-circuit fault of a vehicle power battery pack according to claim 2, characterized in that: Step S2 specifically includes the following steps: S201: Obtaining a reference differential voltage mean vector and a reference differential voltage standard deviation diagonal matrix of a reference differential sampling voltage timing matrix after virtual transformation under a reference working condition of a power battery pack; Reference differential voltage mean vector The expression is shown in formula (8): in, is the reference differential sampling voltage timing matrix after virtual transformation The average value of the reference differential voltage between the i-th sampling channel and the j-th sampling channel is calculated as follows: Diagonal matrix of reference differential voltage standard deviation As shown in formula (11): Where diag{·} represents a diagonal matrix with the elements in {·} as diagonal elements in sequence; Represents the reference differential sampling voltage timing matrix after virtual transformation The standard deviation of the reference differential voltage between the i-th sampling channel and the j-th sampling channel in is calculated as shown in formula (12): S202: The virtual transformed reference differential sampling voltage timing matrix Perform standardization operations as shown below: in, is the virtually transformed reference differential sampling voltage timing matrix The standardized matrix of The standardized and virtually transformed reference differential sampling voltage timing matrix Divided into two groups of standardized and virtually transformed reference differential sampling voltage timing sub-matrices and The two sub-matrices are the reference differential sampling voltage timing matrix after normalization and virtual transformation The row element options are: choose The {1,2,…,n-1}th row of choose of {1,2,…,, …,n} rows; Standardized and virtually transformed reference differential sampling voltage timing sub-matrix and The maximum baseline eigenvalue mean and The expression is shown in formula (14): Among them, λ max,k,base,1 and λ max,k,base,2 They are respectively the reference differential sampling voltage timing sub-matrices after normalization and virtual transformation and The kth sliding time sequence submatrix with ω as matrix width in and The covariance matrix of and The maximum eigenvalue of ; where the standardized sliding time series submatrix The timing sub-matrix of the reference differential sampling voltage after normalization and virtual transformation The k-ω+1 to k columns of the standard sliding time series submatrix are composed in sequence The timing sub-matrix of the reference differential sampling voltage after normalization and virtual transformation The k-ω+1 column to the k column are sequentially composed; k=ω,ω+1,…,m; Sliding time submatrix and The covariance matrix of and The expression of is shown in formula (15): And there are: Among them, λ max,base,1 and λ max,base,2 They are respectively the reference differential sampling voltage timing sub-matrices after normalization and virtual transformation and The maximum base eigenvalue vector of ; Standardized and virtually transformed reference differential sampling voltage timing sub-matrix and Maximum benchmark eigenvalue standard deviation and The calculation formula is shown in formula (17): S203: Setting a fault characteristic amplitude representing the severity of the fault. The minimum detectable fault characteristic amplitude for an initial short circuit fault can be f sc,min It is represented by the minimum detectable fault characteristic amplitude f of the initial short circuit fault. sc,min The fault detection threshold is determined by deriving the quantitative relationship between the fault detection threshold: Among them, D th Indicates the initial short-circuit fault detection threshold; t f represents the fault duration in the sliding window, represents the standard deviation of the (n-1) eigenvalues ​​of the virtual transformed reference differential sampling voltage timing submatrix after normalization, ξ l,base It represents the standard deviation of the lth voltage differential channel under the reference working condition, that is, the diagonal matrix of the reference differential voltage standard deviation The lth element.

4. A method for detecting and locating an initial short-circuit fault of a vehicle-mounted power battery pack according to claim 3, characterized in that: Step S3 specifically includes the following steps: S301: During the operation of the electric vehicle, as the voltage sensor measures, a standardized and virtually transformed real-time differential sampling voltage timing matrix of the power battery pack is constructed. The specific process is as follows: Obtain the real-time voltage vector of the power battery pack at time t after virtual transformation based on the standard measurement topology As shown in formula (19): in, is the real-time voltage value of the i-th sampling channel obtained based on the standard measurement topology after virtual transformation at time t; Calculate the real-time differential voltage vector of the power battery pack based on the virtual transformation of the standard measurement topology at time t As shown in formula (20): When t≥ω, construct the real-time differential voltage measurement timing matrix of the power battery pack based on the standard measurement topology after virtual transformation at time t As shown in formula (21): The timing matrix of the real-time differential voltage measurement of the power battery pack after standardization and virtual transformation based on the standard measurement topology at time t As shown in formula (22): The real-time differential voltage measurement timing matrix of the power battery pack after standardization and virtual transformation Divided into two groups of normalized and virtually transformed power battery pack real-time differential voltage measurement timing sub-matrices and Among them, the real-time differential voltage measurement timing sub-matrix of the power battery pack after virtual transformation Real-time differential voltage measurement timing matrix of power battery pack after virtual transformation The 1st to n-1th rows are sequentially composed of the real-time differential voltage measurement timing sub-matrix of the power battery pack after virtual transformation Real-time differential voltage measurement timing matrix of the power battery pack after virtual transformation of {1,2,…, …,n} lines are composed in sequence; S302: The calculation steps of the real-time initial short-circuit fault detection index are specifically as follows: Get the standardized real-time differential voltage measurement timing sub-matrix at time t and The covariance matrix of is shown in formula (23): The standardized real-time differential voltage measurement timing submatrix at time t is obtained by the eigenvalue solution method. and The covariance matrix of and The maximum eigenvalue λ max,t,1 and λ max,t,2 And its corresponding eigenvector is v max,t,1 and v max,t,2 ; For the eigenvalue λ max,t,1 and λ max,t,2 They are standardized respectively, as shown in formula (24): Among them, λ max,t,1 and λ max,t,2 They are the standardized real-time differential voltage measurement timing sub-matrices at time t and The covariance matrix of and The standardized maximum eigenvalue of ; Get the real-time initial short circuit fault detection index D of the power battery pack at time t t , its expression is shown in formula (25): D t =max{λ max,t,1 ,l max,t,2 } (25).

5. A method for detecting and locating an initial short-circuit fault of a vehicle-mounted power battery pack according to claim 4, characterized in that: Step S4 specifically includes the following steps: When the real-time initial short-circuit fault detection index D at time t t Satisfy D t >D th When the battery pack is short-circuited, it indicates that the battery pack has an initial short-circuit fault; Then the fault location is carried out. The specific steps for locating the location of the initial short circuit fault of the battery are as follows: S401: Formula (26) can be used to calculate the contribution of the eigenvector corresponding to the maximum eigenvalue after normalization: in, Represents the maximum eigenvalue λ after normalization max,t Eigenvector v λmax,t The kth item in ; Grouped standardized real-time differential voltage measurement timing sub-matrix and The covariance matrix of and Finally, the real-time voltage sampling channel contribution vector η of the initial short-circuit fault at time t is t ={η k',t }Calculate as in formula (27), where k'=1,...,n...; in, and Respectively represent the standardized real-time differential voltage measurement timing sub-matrices and The covariance matrix of and The contribution generated by the k′th item in the eigenvector corresponding to the abnormal maximum eigenvalue; S402: When the real-time differential voltage channel contribution vector η of the initial short-circuit fault at time t t The largest two elements η p,t and η q,t The subscripts p and q satisfy equation (28): And there are: Then the initial short-circuit fault occurs at the pth single cell at time t. In formula (29), η th =0.8.

Images