CN115267589A - Multi-parameter joint diagnosis method for battery faults of electric vehicle - Google Patents
Multi-parameter joint diagnosis method for battery faults of electric vehicle Download PDFInfo
- Publication number
- CN115267589A CN115267589A CN202211171127.0A CN202211171127A CN115267589A CN 115267589 A CN115267589 A CN 115267589A CN 202211171127 A CN202211171127 A CN 202211171127A CN 115267589 A CN115267589 A CN 115267589A
- Authority
- CN
- China
- Prior art keywords
- variance
- matrix
- abnormal
- temperature
- value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/392—Determining battery ageing or deterioration, e.g. state of health
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Computational Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computing Systems (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Tests Of Electric Status Of Batteries (AREA)
Abstract
The invention relates to a multi-parameter joint diagnosis method for battery faults of an electric automobile, which comprises the steps of obtaining characteristic parameter time sequence values of all monomers of a power battery of the electric automobile to be diagnosed, establishing a characteristic parameter time sequence matrix, calculating characteristic parameter variances line by line, establishing a characteristic parameter variance matrix, marking the characteristic parameter variance matrix from top to bottom according to a sliding window with a preset initial length, screening abnormal values of the variances in the sliding window to traverse the variance matrix, if abnormal variances exist in the sliding window, determining the moment corresponding to the initial abnormal variances as the battery fault occurrence moment, simultaneously intercepting the characteristic parameter time sequence matrix again within a certain period of time before the fault occurrence moment to form a submatrix, calculating the variances column by submatrix, and identifying the abnormal variances and corresponding sequence numbers by using an abnormal value identification algorithm, wherein the abnormal sequence numbers are the positions of identified fault monomers.
Description
Technical Field
The invention relates to the technical field of battery fault diagnosis, in particular to a multi-parameter joint diagnosis method for battery faults of an electric vehicle.
Background
The main factors of internal short circuit, high temperature and overcharge (voltage) are the main reasons for causing thermal runaway of the power battery. At present, faults of the power battery are mainly diagnosed through abnormity of parameters such as battery voltage and temperature. For example, some scholars propose to use a raydeta criterion (3 σ criterion), a box plot method and other single detection methods to identify abnormal values of the battery terminal voltage, but on the premise that the raydeta criterion is only limited to normal or approximately normal samples and the measurement times are sufficiently large, the box plot method does not require that the samples conform to normal distribution, but proposes to use large sample data, and the accuracy and the rationality of the judgment are higher than those of small sample data. For a few data, more than ten data, the boxplot algorithm may highlight the defects of the data, no abnormity is prompted, and the abnormal values can be diagnosed only by carrying out simultaneous judgment on the abnormity represented by the boxplot and professional abnormity. The Lauda criterion (3 sigma criterion) considers that the probability of the data concentrated in the confidence interval (mu-3 sigma, mu +3 sigma) exceeds 99.7 percent, the probability of exceeding the range only accounts for less than 0.3 percent, the probability is very small, and the data of 0.3 percent is considered to be abnormal values, but in practical situations, the abnormal parameter value when the fault of the power battery occurs is not in the 0.3 percent, and the diagnosis delay is caused when the abnormal parameter value is identified by the method.
Disclosure of Invention
The invention aims to provide a multi-parameter joint diagnosis method for battery faults of an electric vehicle aiming at the defects of the prior art, solves the problems that the representation change of early fine parameters of the battery faults cannot be identified and the like, and improves the timeliness and the accuracy of diagnosis.
The invention is realized by adopting the following technical scheme:
a multi-parameter joint diagnosis method for battery faults of electric vehicles comprises the following steps:
step 100: according to the inside module condition of arranging of battery package, found temperature probe serial number and battery cell corresponding relation matrix A:
wherein the relation matrix is a k (m/k + 1) size matrix, the number of a first column of each row of the matrix is 1, 2.,. K represents the serial number of the battery temperature probe; numbers in the second row to the (m/k + 1) th row in each row represent the serial numbers of the corresponding battery monomers detected by the serial numbers of the probes in the first row in each row, wherein 1,2,. The., m is the serial number of the battery monomer, and m is a positive integer multiple of k;
step 101: acquiring voltage data and probe temperature data of a battery pack to be diagnosed, and constructing a voltage original matrix and a temperature original matrix;
step 102: carrying out variance calculation on the voltage original matrix and the temperature original matrix line by line to obtain corresponding variance matrixes;
step 103: presetting the length of an initial sliding window, intercepting data of the variance matrix region by using the sliding window, and identifying an abnormal value of the variance in the sliding window by using an abnormal value detection method;
step 104: if the abnormal variance is not identified in the current window, the sliding window slides downwards for fixing the step length, and the abnormal value identification is carried out on the variance extracted into the new window until the voltage variance matrix and the temperature variance matrix are traversed;
step 105: if the abnormal variance is identified, stopping the sliding window from sliding downwards, taking the moment corresponding to the abnormal variance as a terminal point, namely the fault occurrence moment, and extracting partial data of the original matrix to form a sub-matrix;
step 106: calculating the variance column by column for the submatrix to form a longitudinal variance matrix, and screening abnormal values in the longitudinal variance matrix by using an abnormal value detection method to obtain a corresponding abnormal variance sequence number;
step 107: judging the parameters of the longitudinal variance matrix: if the voltage data is used for calculation, the abnormal variance serial number is the serial number of the battery fault single body; and if the abnormal variance serial number is calculated by using the temperature data and is taken as the temperature probe serial number, further establishing a corresponding relation matrix of the temperature probe serial number and the single battery to obtain the serial number of the single battery group with the fault.
Preferably, the original matrix of voltages in step 101 is U of t × m t*m E of original matrix of temperature t x k t*k ;
Voltage original matrix U t*m Comprises the following steps:
original matrix of temperature E t*k Comprises the following steps:
wherein t is the number of sampling strips, m is the number of single batteries, k is the number of temperature probes, u tm Represents the m-th cell voltage value at the t-th sampling time, E tk Representing the kth probe temperature value at the tth sampling time.
Preferably, the voltage variance matrix in step 102 is D of t × 1 t*1 M with temperature variance matrix t x 1 t*1 ;
Voltage variance matrix D t*1 Comprises the following steps:
temperature variance matrix M t*1 Comprises the following steps:
wherein σ (t), σ * And (t) the voltage variances of all the single batteries at the tth sampling moment and the temperature variances of all the probes at the tth sampling moment are respectively obtained.
Preferably, the abnormal value detection method is at least one of a nonparametric classification algorithm, a box plot, a clustering algorithm, a Grubbs detection algorithm and a threshold method.
Preferably, in step 103, the length of the initial sliding window is N, and the voltage variance matrix and the temperature variance matrix are respectively crossed from top to bottom at the same time; screening abnormal values of the voltage variance and the temperature variance in the initial sliding window by using a Grubbs detection algorithm;
the initial sliding window length N is less than the number of sampling strips t of the voltage raw matrix and the temperature raw matrix, and the window length N is defined to be a minimum of 10.
Preferably, in step 103, according to the length N of the initial sliding window, the voltage variance matrix and the temperature variance matrix are sequentially slid to obtain a voltage variance sequence D under the corresponding window N*1 Temperature variance sequence M N*1 ;
Voltage variance sequence D N*1 Comprises the following steps:
temperature variance sequence M N*1 Comprises the following steps:
calculating the mean value of the variance and the standard deviation of the variance according to the variance in the sliding window;
subtracting the variance average value from the variance in each line in the sliding window, and then taking an absolute value to obtain a residual value to establish a residual matrix;
establishing a matrix G by using the value obtained by the ratio of each row value of the residual matrix to the standard deviation of the variance (i) ;
Wherein G is 1 ,G 2 ,G 3 ...,G i Respectively representing the ratio of residual values to standard deviations of variances at the ith sampling time, i = N;
upon determining the test level α and the sliding window length N, look up the Grubbs table for a cut-off value G (P, N), where P =1- α;
comparing the matrix G (i) The value of each row in the column and the magnitude of G (P, N), if G i G (P, N), then the matrix G is represented (i) The lower threshold value G (P, N) at the ith sampling time is an abnormal value.
Preferably, in step 104, the abnormal variance is not identified in the sliding window, the sliding window slides downwards by a fixed step length, and the abnormal value identification is performed on the variance in the new window, which includes the following specific processes:
identifying abnormal values of the variance in the initial sliding window by using a Grubbs detection algorithm, if no abnormal value exists, extracting the maximum value and the minimum value of the variance in the current window, setting the maximum value and the minimum value as the upper limit value and the lower limit value of the range of the initial dynamic safety threshold, sliding the sliding window downwards for 1 step length, and acquiring the next variance value to be detected;
if the variance to be detected is within the dynamic threshold range, judging the variance to be a non-abnormal value;
if the variance to be detected is out of the dynamic threshold range, identifying abnormal values of all variances in the new window by using a Grubbs detection algorithm; if no abnormity exists, re-determining the maximum value and the minimum value of the variance in the current window as the upper limit value and the lower limit value of the dynamic safety threshold range; and automatically sliding the sliding window downwards for 1 step length, and performing abnormal value detection by repeating the steps and traversing the residual variance values in the matrix.
Preferably, the step 105 of extracting partial data from the original matrix, and forming the sub-matrix specifically includes the following steps:
determining the sampling time t' corresponding to the variance abnormality which is firstly identified;
setting the number L of data extraction strips;
finding out a row where sampling time t 'is located from a corresponding original matrix, intercepting L rows of original matrix data from the beginning of the row where the sampling time t' is located upwards, and establishing a sub-matrix, wherein 0-type and L-type are-type and N-type;
the sub-matrix size is determined by the corresponding original matrix, the sub-matrix is matrix Z L*m And matrix Z ' L*K One of (1);
Z L*m represents a sub-matrix extracted from a voltage matrix, wherein the number of voltage data collected is L, the time of collection is from (t '-L + 1) to t' end, u t'm The voltage of the mth single battery at the t' sampling moment;
Z ' L*K represents a sub-matrix extracted from a temperature matrix, wherein the number of temperature data collected is L, the time of collection is from (t '-L + 1) to t' end, E t'K The kth probe temperature value at the t' sampling instant.
Preferably, the specific steps of forming the vertical variance matrix in step 106 are as follows:
the vertical variance matrix may be a voltage vertical variance matrix S1 1×m And temperature longitudinal variance matrix S2 1×k The matrix size mainly depends on the sub-matrix size;
voltage longitudinal variance matrix S1 1×m Comprises the following steps:
wherein S m Indicates that the m-th cell is at S1 1×m All voltage variances in the m-th column;
temperature longitudinal variance matrix S2 1×k Comprises the following steps:
Preferably, if for the vertical variance matrix S1 1×m Identification using outlier detection algorithmsDistinguishing the abnormal variance and the corresponding serial number, wherein the serial number corresponding to the abnormal variance is the serial number of the suspected fault battery monomer;
if for the vertical variance matrix S2 1×k And identifying the abnormal variance and the corresponding serial number by using an abnormal value detection algorithm, wherein the corresponding serial number of the abnormal variance is the serial number of the temperature probe, and is further matched with the serial numbers of the temperature probes in the first column in the relation matrix A, and the serial number of the detected battery monomer corresponding to the serial number of the temperature probe is the serial number of the suspected faulty battery monomer.
Compared with the prior art, the invention has the following beneficial technical effects:
the invention can identify whether the battery has faults or not by detecting abnormal values of two parameters of the voltage of the single battery and the temperature value of the probe, and the diagnosis method adopting multi-parameter fusion and simultaneous is superior to the single-parameter fault diagnosis;
the method is different from the existing method for diagnosing the faults by data needing a full life cycle, can be transplanted to a large data monitoring platform, and realizes real-time monitoring of the faults of the new energy batteries.
Drawings
The invention is further described with reference to the accompanying drawings in which:
FIG. 1 is a flow chart of a multi-parameter joint diagnosis method for battery faults of an electric vehicle according to the present invention.
Detailed Description
In order that the above objects, features and advantages of the present invention can be more clearly understood, a detailed description of the present invention will be given below with reference to the accompanying drawings and specific embodiments. It should be noted that the embodiments and features of the embodiments of the present application may be combined with each other without conflict. In the description of the present invention, it should also be noted that, unless otherwise explicitly specified or limited, the terms "disposed," "mounted," and "connected" are to be construed broadly, e.g., as meaning fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood in specific cases to those skilled in the art.
The first embodiment is as follows:
a multi-parameter joint diagnosis method for battery faults of an electric vehicle is characterized in that a non-parameter classification algorithm (for example, a KNN: K adjacent algorithm) is adopted for abnormal value identification; a boxplot (used as a statistical graph for displaying a group of data dispersion situation data), a clustering algorithm (for example, a density-based clustering algorithm: DBSCAN is selected), a Grubbs detection algorithm, and a threshold method (for example, a threshold method for setting upper and lower limits);
a multi-parameter joint diagnosis method for battery faults of electric vehicles adopts a Grubbs detection algorithm and comprises the following steps:
s100: constructing a corresponding relation matrix A of the serial numbers of the temperature probes and the single batteries according to the arrangement condition of modules in the battery pack;
s101: acquiring voltage data of a battery pack to be diagnosed and probe temperature data, and constructing an original voltage matrix and an original temperature matrix;
s102: carrying out variance calculation on the voltage original matrix and the temperature original matrix line by line to obtain corresponding variance matrixes;
s103: presetting the length of a sliding window, intercepting data of the variance matrix region by using the sliding window, and identifying abnormal values of the variance in the sliding window by using an abnormal value detection method;
s104: if the abnormal variance is not identified in the current window, the sliding window moves downwards, and abnormal value identification is carried out on the variance extracted into the new window until the voltage variance matrix and the temperature variance matrix are traversed;
s105: if the abnormal variance is identified, stopping the sliding window from sliding downwards, and taking the moment corresponding to the abnormal variance as an end point (namely the moment when the fault occurs), extracting partial data of the original matrix to form a sub-matrix;
s106: calculating the variance column by column for the submatrix to form a longitudinal variance matrix, and screening abnormal values in the longitudinal variance matrix by using an abnormal value detection method to obtain a corresponding abnormal variance sequence number;
s107: judging the parameters of the longitudinal variance matrix: if the voltage data is used for calculation, the abnormal variance serial number is the serial number of the battery fault single body; and if the abnormal variance serial number is calculated by using the temperature data and is taken as the serial number of the temperature probe, further establishing a corresponding relation matrix of the serial number of the temperature probe and the single battery to obtain the serial number of the single battery group with the fault.
The steps are described in further detail below.
Specifically, in step 100, a corresponding relationship between the serial number of the temperature probe and the detected battery cell is determined according to the battery module in the battery box, the arrangement of the battery cells in the module, and a temperature sensor arrangement diagram, and a corresponding relationship matrix table is established. For example, 156 single batteries and 52 temperature probes are arranged in a certain battery box to be detected, and according to the actual arrangement situation, the corresponding relation matrix a of the serial numbers of the temperature probes and the serial numbers of the single batteries to be detected is as follows:
the size of the matrix is 52 x 4, and the number of the matrix 1 st column 1, 2.. 52 represents the serial number of the battery temperature probe; the numbers in the 2 nd to 4 th columns represent that the probe serial numbers of the rows correspond to the serial numbers of the detected battery monomers, and every 1 probe temperature is responsible for detecting the temperature value of 3 battery monomers. For example, probe No. 1 is responsible for detecting the temperature of cell No. 1,2,3, and probe No. 52 is responsible for detecting the temperature of cell No. 154, 155, 156.
In step 101, obtaining a voltage value of each battery cell and a temperature value of a battery probe in a time period to be diagnosed from a big data platform of a new energy vehicle, and constructing a voltage original matrix and a temperature original matrix, for example, obtaining 1000 battery voltages and probe temperature time sequence data, and constructing a corresponding parameter matrix.
Wherein U is t*m 、E t*k Respectively a voltage original matrix and a temperature original matrix;
where t is the number of sampled data pieces, m is the number of cells, and k is the number of temperature probes, in this example t =1000, m =156, k =52. u. of tm Representing the voltage value of the 156 th cell under the 1000 th sampling data, E tk Represents the temperature value measured by probe number 52 under sample data item 1000.
After the step 101 and before the step 102, preprocessing data in the voltage original matrix and the temperature original matrix, wherein the data preprocessing method comprises the following steps: and (4) data cleaning and data normalization processing.
Data cleaning strategy: setting reasonable effective value ranges of all characteristic parameters, such as a voltage normal range of 0-5v and a temperature normal range of 0-254 degrees, and deleting all corresponding data at a certain sampling moment if the numerical value of the sampling moment exceeds the corresponding effective value range or is empty.
The data normalization processing method comprises the following steps: and carrying out normalization processing on each matrix subjected to data cleaning line by line, wherein the normalization processing method selects one of minimum-maximum normalization and zero-mean normalization.
102, carrying out variance calculation on the preprocessed voltage matrix and temperature matrix line by line to obtain a corresponding variance matrix; e.g. pretreated voltage matrix, temperature matrix, D t*1 ,M t*1 For a corresponding voltage variance matrix, temperature variance matrix, where σ (t), σ * (t) represents all cell voltage variances at the t-th sampling time and all probe temperature variances at the t-th sampling time, respectively.
103, in step 104, setting the size of the sliding window, sliding the voltage variance matrix and the temperature variance matrix from top to bottom, and using the Grubbs detection algorithm joint threshold method to jointly identify the abnormal value of the variance sequence value in each sliding window in the voltage variance matrix and the temperature variance matrix, for example, setting the length of the initial sliding window N =50 in the example, and respectively identifying the abnormal value of the size of the voltage variance matrix and the temperature variance matrix as the initial sliding window length N =501000 x 1D t*1 Voltage variance matrix sum M t*1 The temperature variance matrix is sequentially crossed from top to bottom by a sliding window with the length of 50, abnormal value identification is carried out on variance sequence values in each sliding window by using a Grubbs detection algorithm and a threshold value method, and if the sliding window traverses the whole variance matrix, sliding is carried out 951 times.
The specific method of the Grubbs detection algorithm is as follows:
calculating the corresponding G of each difference value in the sliding window according to a Grubbs test method formula j The formula is as follows:
whereinIs the mean of the variance sequence values within the window, S is the standard deviation of the variance sequence values within the window, G j The variance value of the jth sampling time in the sliding window is shown.
While selecting a suitable confidence interval p. WhereinN isLength of sliding window,Looking up Grubbs value table to obtain corresponding G (P, N), and comparing G j And G (P, N) are of greater size, if G j If < G (P, N), it is a non-abnormal value, if G j G (P, N) or more, the variance value in the jth sampling time is abnormal. In this embodiment, a 95% confidence interval is selected, the number of data pieces is the window length 50, a Grubbs table is looked up to obtain G (95, 50) =2.956, and the difference value G between each party in the sliding window is compared j And G (95, 50), in the embodiment, when the Grubbs detection algorithm is used for identifying abnormal single bodies, in order to improve the operation efficiency, only the statistic G of the minimum variance value and the maximum variance value in each window is calculated at the beginning min ,G max And compared to the G (95, 50) value. If the difference does not exceed G (95, 50), the window has no abnormal variance, and the abnormal single battery can be judged. If G in a certain sliding window is calculated min Or G max If the variance exceeds G (95, 50), stopping the calculation of the downward sliding window, and calculating the corresponding statistic value G of all variances in the window j And determining the first exceedingG (95, 50) corresponds to the sampling instant. The method judges whether an abnormal value exists by presetting upper and lower limit statistic thresholds in the sliding window and comparing the upper and lower limit statistic thresholds with G (95, 50), and does not need to compare all the statistic G in the sliding window j The values are compared with G (95, 50), thereby improving the overall abnormal value determination time efficiency. The method is characterized in that the abnormal variance value of the voltage variance matrix and the temperature variance matrix is detected by using a sliding window principle, and a sampling time point when the abnormal variance occurs earlier is taken as the earliest time point when the fault occurs.
In this embodiment, abnormal values of the voltage variance and the temperature variance in the initial sliding window are identified by Grubbs, if there is no abnormal value, the maximum variance and the minimum variance of the voltage variance and the temperature variance in the sliding window are respectively set as the upper limit value and the lower limit value of the initial dynamic critical threshold, and the sliding window moves downward by one step at the same time to obtain the voltage variance value and the temperature variance value to be detected at the next sampling time. And judging whether the voltage variance and the temperature variance to be detected are within the set initial dynamic threshold range, and if so, judging that the variance is a non-abnormal value.
And if the variance to be detected next is out of the dynamic threshold range, identifying abnormal values of all variances in the new window by using a Grubbs detection algorithm. If the abnormal value exists, the variance to be detected is judged to be the abnormal value, and the corresponding sampling time is the time corresponding to the single body abnormity. If no abnormity exists, the maximum value and the minimum value of the variance in the current window are re-determined to be the upper limit value and the lower limit value of the latest dynamic threshold range. And by analogy, traversing the residual variance values in the matrix by the method, and carrying out abnormal variance detection. Further, the method for jointly diagnosing the battery faults by using the Grubbs detection algorithm and the dynamic threshold method can greatly reduce a large amount of calculation amount by independently using the Grubbs detection algorithm in each window in terms of time efficiency, and can effectively improve the calculation and identification efficiency.
In step 105, the earliest fault occurrence time is identified in step 104, and partial data extraction is performed on the original matrix by taking the time as the ending time, so as to establish a sub-matrix. The selection principle of the original matrix corresponding to the abnormal variance is as follows:
identifying abnormal values of the voltage variance matrix and the temperature variance matrix by a sliding window method;
if the voltage variance is recognized to have an abnormal value, stopping all the sliding windows to move downwards for calculation. The time corresponding to the abnormal variance value is the fault occurrence time of the single battery.
If the abnormal value of the temperature variance is identified, the time corresponding to the abnormal value of the temperature variance is the time when the battery fault occurs, and the calculation of downward movement of all the sliding windows is stopped. The time corresponding to the variance abnormal value is the occurrence time of the battery fault.
Selecting an original matrix corresponding to the variance matrix to extract partial data based on the variance matrix where the abnormal variance value is firstly identified;
that is, if the sampling time corresponding to the abnormal value in the voltage variance matrix is identified to be earlier than the sampling time corresponding to the abnormal value in the temperature variance matrix, only partial data extraction is performed on the voltage original matrix, and vice versa.
In this embodiment, it is determined that the variance anomaly is in the temperature variance sequence first, and then only partial data extraction is performed on the temperature matrix, where the extraction method is as follows: and determining a row corresponding to the temperature matrix from the earliest fault occurrence moment in the temperature matrix, and intercepting 20 rows of data from the row to the temperature matrix upwards to establish a temperature sub-matrix.
In step 106, the submatrix constructed in the step 105 is subjected to variance calculation column by column to form a longitudinal variance matrix, and an abnormal value detection method is used for screening abnormal values of all values in the longitudinal variance matrix. In this embodiment, a longitudinal temperature variance matrix with a size of 1 × 52 is obtained by calculating a temperature submatrix, an abnormal value is identified for the longitudinal temperature variance matrix, preferably, a Grubbs detection method is used for identifying an abnormal value, 95% confidence intervals are selected, a data length in the longitudinal variance matrix is 52, a Grubbs table is checked to obtain G (95, 52) =2.971, and a statistic G corresponding to each value in the longitudinal variance matrix is calculated 1*52 =(G 1 ,G 2 ,G 3 ...,G 52 ),G 1 ,G 2 ,G 3 ...,G 52 Respectively represent the temperature of No. 1 and No. 2The statistic obtained by variance calculation is determined G i The temperature probe serial number i which is greater than or equal to G (95, 52) is a suspected fault position, and meanwhile, in combination with the corresponding relation matrix of the probe serial number and the detected single battery serial number obtained in the step 100, in this embodiment, it is determined that 3 single battery serial numbers corresponding to the probe temperature serial number i are diagnostic fault single positions.
If the voltage sub-matrix is established by the voltage matrix, and the longitudinal variance matrix established by the voltage sub-matrix is used for calculating statistics corresponding to each variance, the Grubbs detection method is preferably adopted for abnormal value identification, and the serial number corresponding to the determined abnormal statistics of the voltage variance is the diagnosed and determined fault single body position.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential attributes thereof. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned.
The above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; although the present invention has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A multi-parameter joint diagnosis method for battery faults of electric vehicles is characterized by comprising the following steps:
step 100: according to the inside module condition of arranging of battery package, found temperature probe serial number and battery cell corresponding relation matrix A:
wherein the relation matrix is a k (m/k + 1) size matrix, the number of a first column of each row of the matrix is 1, 2.,. K represents the serial number of the battery temperature probe; numbers in the second row to the (m/k + 1) th row in each row represent the serial numbers of the corresponding battery monomers detected by the serial numbers of the probes in the first row in each row, wherein 1,2,. The., m is the serial number of the battery monomer, and m is a positive integer multiple of k;
step 101: acquiring voltage data and probe temperature data of a battery pack to be diagnosed, and constructing a voltage original matrix and a temperature original matrix;
step 102: carrying out variance calculation on the voltage original matrix and the temperature original matrix line by line to obtain corresponding variance matrixes;
step 103: presetting the length of an initial sliding window, intercepting data of the variance matrix region by using the sliding window, and identifying an abnormal value of the variance in the sliding window by using an abnormal value detection method;
step 104: if the abnormal variance is not identified in the current window, the sliding window slides downwards for fixing the step length, and the abnormal value identification is carried out on the variance extracted into the new window until the voltage variance matrix and the temperature variance matrix are traversed;
step 105: if the abnormal variance is identified, stopping the sliding window from sliding downwards, taking the moment corresponding to the abnormal variance as an end point, namely the fault occurrence moment, and extracting partial data of the original matrix to form a sub-matrix;
step 106: calculating the variance column by column for the submatrix to form a longitudinal variance matrix, and screening abnormal values in the longitudinal variance matrix by using an abnormal value detection method to obtain a corresponding abnormal variance sequence number;
step 107: judging the parameters of the longitudinal variance matrix: if the voltage data is used for calculation, the abnormal variance serial number is the serial number of the battery fault single body; and if the abnormal variance serial number is calculated by using the temperature data and is taken as the serial number of the temperature probe, further establishing a corresponding relation matrix of the serial number of the temperature probe and the single battery to obtain the serial number of the single battery group with the fault.
2. The multi-parameter joint diagnosis method for battery faults of electric vehicles according to claim 1, wherein the voltage original matrix in the step 101 is U with t x m t*m E with original matrix of temperature t x k t*k ;
Voltage original matrix U t*m Comprises the following steps:
original matrix of temperature E t*k Comprises the following steps:
wherein t is the number of sampling strips, m is the number of single batteries, k is the number of temperature probes, and u tm Represents the m-th cell voltage value at the t-th sampling time, E tk Representing the kth probe temperature value at the t-th sampling instant.
3. The multi-parameter joint diagnosis method for battery faults of electric vehicles according to claim 1, wherein the voltage variance matrix in step 102 is D of t x 1 t*1 M with temperature variance matrix of t x 1 t*1 ;
Voltage variance matrix D t*1 Comprises the following steps:
temperature variance matrix M t*1 Comprises the following steps:
wherein σ (t), σ * (t) are respectively the t-th sampling timeThe voltage variance of all the lower single batteries and the temperature variance of all the probes at the tth sampling moment.
4. The multi-parameter joint diagnosis method for battery faults of electric vehicles according to claim 1, characterized in that the abnormal value detection method adopts at least any one of a nonparametric classification algorithm, a box line graph, a clustering algorithm, a Grubbs detection algorithm and a threshold method.
5. The multi-parameter joint diagnosis method for battery faults of electric vehicles according to claim 4, wherein in step 103, the length of an initial sliding window is N, and the voltage variance matrix and the temperature variance matrix are respectively crossed from top to bottom at the same time; screening abnormal values of the voltage variance and the temperature variance in the initial sliding window by using a Grubbs detection algorithm;
the initial sliding window length N is less than the number of samples t of the voltage original matrix and the temperature original matrix, and the window length N is defined to be a minimum of 10.
6. The multi-parameter joint diagnosis method for battery faults of electric vehicles according to claim 5, wherein in step 103, a voltage variance matrix and a temperature variance matrix are sequentially slipped according to the length N of the set initial sliding window to obtain a voltage variance sequence D under the corresponding window N*1 Temperature variance sequence M N*1 ;
Voltage variance sequence D N*1 Comprises the following steps:
temperature variance sequence M N*1 Comprises the following steps:
calculating the mean value of the variance and the standard deviation of the variance according to the variance in the sliding window;
subtracting the variance of each row in the sliding window from the variance average value, and then taking an absolute value to obtain a residual value, and establishing a residual matrix;
establishing a matrix G by using the value obtained by the ratio of each row value of the residual matrix to the standard deviation of the variance (i) ;
Wherein G is 1 ,G 2 ,G 3 ...,G i Respectively, 1 st, 2 nd, 3 rd, residual value to variance standard deviation ratio at the ith sampling time, i = N;
upon determining the test level α and the sliding window length N, look up the Grubbs table for a cut-off value G (P, N), where P =1- α;
comparing the matrix G (i) The value of each row in the column and the magnitude of G (P, N), if G i G (P, N), then the matrix G is represented (i) The lower critical value G (P, N) at the ith sampling time is an abnormal value.
7. The multi-parameter joint diagnosis method for battery faults of electric vehicles according to claim 6, wherein in step 104, no abnormal variance is identified in the sliding window, the sliding window slides downwards for a fixed step length, and abnormal value identification is performed on the variance in the new window, which comprises the following specific processes:
identifying abnormal values of the variance in the initial sliding window by using a Grubbs detection algorithm, if the abnormal values do not exist, extracting the maximum value and the minimum value of the variance in the current window, setting the maximum value and the minimum value as the upper limit value and the lower limit value of the range of the initial dynamic safety threshold, sliding the sliding window downwards for 1 step length, and acquiring the next variance value to be detected;
if the variance to be detected is within the dynamic threshold range, judging the variance to be a non-abnormal value;
if the variance to be detected is out of the dynamic threshold range, identifying abnormal values of all variances in the new window by using a Grubbs detection algorithm; if no abnormity exists, re-determining the maximum value and the minimum value of the variance in the current window as the upper limit value and the lower limit value of the dynamic safety threshold range; and automatically sliding the sliding window downwards for 1 step length, and performing abnormal value detection by repeating the steps and traversing the residual variance values in the matrix.
8. The multi-parameter joint diagnosis method for battery faults of electric vehicles according to claim 7, wherein in step 105, partial data extraction is performed on the original matrix, and the forming of the submatrix specifically comprises the following steps:
determining the sampling time t' corresponding to the variance abnormality identified at first;
setting the number L of data extraction strips;
finding out a row where sampling time t' is located from a corresponding original matrix, intercepting original matrix data of L rows from the beginning of the row, and establishing a sub-matrix, wherein 0-L-N are constructed;
the sub-matrix size is determined by the corresponding original matrix, the sub-matrix is matrix Z L*m Sum matrix Z ' L*K One of (1);
Z L*m represents a sub-matrix extracted from a voltage matrix, wherein the number of voltage data collected is L, the time of collection is from (t '-L + 1) to t' end, u t'm The voltage of the mth single battery at the t' sampling moment;
Z ' L*K represents a sub-matrix extracted from a temperature matrix, wherein the number of temperature data collected is L, the time of collection is from (t '-L + 1) to t' end, E t'K Is the kth probe temperature value at the tth sampling time.
9. The multi-parameter joint diagnosis method for battery faults of electric vehicles according to claim 8, wherein the specific steps of forming the longitudinal variance matrix in step 106 are as follows:
the vertical variance matrix may be a voltage vertical variance matrix S1 1×m And temperature longitudinal variance matrix S2 1×k The matrix size mainly depends on the sub-matrix size;
voltage longitudinal variance matrix S1 1×m Comprises the following steps:
wherein S m Indicates that the m-th cell is at S1 1×m All voltage variances in the m-th column;
temperature longitudinal variance matrix S2 1×k Comprises the following steps:
10. The multi-parameter joint diagnosis method for battery faults of electric vehicles according to claim 9, characterized in that if the longitudinal variance matrix S1 is matched, the method is used for the diagnosis 1×m Identifying abnormal variance and corresponding serial number by using an abnormal value detection algorithm, wherein the serial number corresponding to the abnormal variance is the serial number of the suspected faulty battery monomer;
if for the vertical variance matrix S2 1×k And identifying the abnormal variance and the corresponding serial number by using an abnormal value detection algorithm, wherein the corresponding serial number of the abnormal variance is the serial number of the temperature probe, the abnormal variance is further matched with the serial number of the temperature probe in the first column in the relation matrix A, and the serial number of the detected battery monomer corresponding to the serial number of the temperature probe is the serial number of the suspected fault battery monomer.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211171127.0A CN115267589B (en) | 2022-09-26 | 2022-09-26 | Multi-parameter joint diagnosis method for battery faults of electric vehicle |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211171127.0A CN115267589B (en) | 2022-09-26 | 2022-09-26 | Multi-parameter joint diagnosis method for battery faults of electric vehicle |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115267589A true CN115267589A (en) | 2022-11-01 |
CN115267589B CN115267589B (en) | 2023-01-06 |
Family
ID=83757957
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211171127.0A Active CN115267589B (en) | 2022-09-26 | 2022-09-26 | Multi-parameter joint diagnosis method for battery faults of electric vehicle |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115267589B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116400231A (en) * | 2023-06-09 | 2023-07-07 | 中国华能集团清洁能源技术研究院有限公司 | Battery multi-fault detection method and device of energy storage system and electronic equipment |
CN117150269A (en) * | 2023-09-04 | 2023-12-01 | 淮阴师范学院 | Electrical equipment operation abnormality diagnosis system based on data analysis |
CN117922301A (en) * | 2024-03-25 | 2024-04-26 | 北京厚方科技有限公司 | Battery power supply safety supervision system suitable for new energy automobile |
CN117150269B (en) * | 2023-09-04 | 2024-06-07 | 淮阴师范学院 | Electrical equipment operation abnormality diagnosis system based on data analysis |
Citations (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2015074065A1 (en) * | 2013-11-18 | 2015-05-21 | California Institute Of Technology | Electrochemical separators with inserted conductive layers |
CN110161414A (en) * | 2019-06-13 | 2019-08-23 | 北京理工大学 | A kind of power battery thermal runaway on-line prediction method and system |
CN110907833A (en) * | 2019-10-16 | 2020-03-24 | 北京理工大学 | Power battery fault diagnosis method based on sliding window |
CN111653840A (en) * | 2020-06-08 | 2020-09-11 | 中国第一汽车股份有限公司 | Early warning method, device, equipment and storage medium |
CN111913114A (en) * | 2020-07-23 | 2020-11-10 | 蜂巢能源科技有限公司 | Battery thermal runaway detection method and device, storage medium and electronic equipment |
CN111929588A (en) * | 2020-09-01 | 2020-11-13 | 杭州颉码能源科技有限公司 | Charging safety monitoring method based on extreme learning machine |
CN113537525A (en) * | 2021-07-23 | 2021-10-22 | 四川大学 | Self-adaptive early warning method for fault state of battery energy storage system |
KR20210141211A (en) * | 2020-05-15 | 2021-11-23 | 주식회사 엘지에너지솔루션 | Apparatus and method for diagnosing battery |
CN114325417A (en) * | 2021-12-22 | 2022-04-12 | 北京国家新能源汽车技术创新中心有限公司 | Power battery internal short circuit detection method |
CN114509685A (en) * | 2022-02-21 | 2022-05-17 | 重庆工商大学 | Method and system for constructing thermal runaway prediction model of lithium ion battery |
CN114559819A (en) * | 2022-01-25 | 2022-05-31 | 重庆标能瑞源储能技术研究院有限公司 | Electric vehicle battery safety early warning method based on signal processing |
CN114966427A (en) * | 2021-08-03 | 2022-08-30 | 长城汽车股份有限公司 | Performance detection method, device and system of battery thermal management system |
CN115032542A (en) * | 2022-06-07 | 2022-09-09 | 中国人民解放军陆军工程大学 | Hybrid model-based battery thermal runaway pre-judgment method for energy storage system |
-
2022
- 2022-09-26 CN CN202211171127.0A patent/CN115267589B/en active Active
Patent Citations (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2015074065A1 (en) * | 2013-11-18 | 2015-05-21 | California Institute Of Technology | Electrochemical separators with inserted conductive layers |
CN110161414A (en) * | 2019-06-13 | 2019-08-23 | 北京理工大学 | A kind of power battery thermal runaway on-line prediction method and system |
CN110907833A (en) * | 2019-10-16 | 2020-03-24 | 北京理工大学 | Power battery fault diagnosis method based on sliding window |
KR20210141211A (en) * | 2020-05-15 | 2021-11-23 | 주식회사 엘지에너지솔루션 | Apparatus and method for diagnosing battery |
CN111653840A (en) * | 2020-06-08 | 2020-09-11 | 中国第一汽车股份有限公司 | Early warning method, device, equipment and storage medium |
CN111913114A (en) * | 2020-07-23 | 2020-11-10 | 蜂巢能源科技有限公司 | Battery thermal runaway detection method and device, storage medium and electronic equipment |
CN111929588A (en) * | 2020-09-01 | 2020-11-13 | 杭州颉码能源科技有限公司 | Charging safety monitoring method based on extreme learning machine |
CN113537525A (en) * | 2021-07-23 | 2021-10-22 | 四川大学 | Self-adaptive early warning method for fault state of battery energy storage system |
CN114966427A (en) * | 2021-08-03 | 2022-08-30 | 长城汽车股份有限公司 | Performance detection method, device and system of battery thermal management system |
CN114325417A (en) * | 2021-12-22 | 2022-04-12 | 北京国家新能源汽车技术创新中心有限公司 | Power battery internal short circuit detection method |
CN114559819A (en) * | 2022-01-25 | 2022-05-31 | 重庆标能瑞源储能技术研究院有限公司 | Electric vehicle battery safety early warning method based on signal processing |
CN114509685A (en) * | 2022-02-21 | 2022-05-17 | 重庆工商大学 | Method and system for constructing thermal runaway prediction model of lithium ion battery |
CN115032542A (en) * | 2022-06-07 | 2022-09-09 | 中国人民解放军陆军工程大学 | Hybrid model-based battery thermal runaway pre-judgment method for energy storage system |
Non-Patent Citations (2)
Title |
---|
ZHENYU SUN 等: "Modified Relative Entropy-Based Lithium-Ion Battery Pack Online Short-Circuit Detection for Electric Vehicle", 《 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION》 * |
徐可 等: "基于Grubbs准则和EKF的锂电池组故障诊断策略", 《第19届中国系统仿真技术及其应用学术年会论文集(19TH CCSSTA 2018)》 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116400231A (en) * | 2023-06-09 | 2023-07-07 | 中国华能集团清洁能源技术研究院有限公司 | Battery multi-fault detection method and device of energy storage system and electronic equipment |
CN116400231B (en) * | 2023-06-09 | 2023-10-03 | 中国华能集团清洁能源技术研究院有限公司 | Battery multi-fault detection method and device of energy storage system and electronic equipment |
CN117150269A (en) * | 2023-09-04 | 2023-12-01 | 淮阴师范学院 | Electrical equipment operation abnormality diagnosis system based on data analysis |
CN117150269B (en) * | 2023-09-04 | 2024-06-07 | 淮阴师范学院 | Electrical equipment operation abnormality diagnosis system based on data analysis |
CN117922301A (en) * | 2024-03-25 | 2024-04-26 | 北京厚方科技有限公司 | Battery power supply safety supervision system suitable for new energy automobile |
CN117922301B (en) * | 2024-03-25 | 2024-05-24 | 北京厚方科技有限公司 | Battery power supply safety supervision system suitable for new energy automobile |
Also Published As
Publication number | Publication date |
---|---|
CN115267589B (en) | 2023-01-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN115267589B (en) | Multi-parameter joint diagnosis method for battery faults of electric vehicle | |
CN108254696B (en) | Battery health state evaluation method and system | |
CN111551860B (en) | Battery internal short circuit diagnosis method based on relaxation voltage characteristics | |
CN114430080B (en) | Power battery cell abnormal self-discharge identification method based on operation data | |
CN110794305A (en) | Power battery fault diagnosis method and system | |
CN111641384B (en) | Photovoltaic power station string fault diagnosis method, device, equipment and readable storage medium | |
CN114559819B (en) | Electric automobile battery safety early warning method based on signal processing | |
CN109061537A (en) | Electric vehicle lithium ion battery sensor fault diagnosis method based on observer | |
CN111579121B (en) | Method for diagnosing faults of temperature sensor in new energy automobile battery pack on line | |
CN113791350B (en) | Battery fault prediction method | |
CN114035086B (en) | Multi-fault diagnosis method for battery pack based on signal processing | |
CN113765179A (en) | Cloud battery pack consistency analysis method based on hierarchical clustering and class center distance | |
CN114545234A (en) | Method for monitoring battery state based on battery temperature gradient change | |
CN113406502B (en) | Defective battery screening method, apparatus and medium based on machine learning | |
CN113900035B (en) | Battery detection method, device, equipment and storage medium | |
CN112009252B (en) | Fault diagnosis and fault-tolerant control method for power battery system | |
CN113311344A (en) | Thermal fault diagnosis method and system suitable for lithium ion battery with gradient utilization | |
CN116865672A (en) | Photovoltaic module abnormality recognition method and device and computer readable storage medium | |
CN115469226A (en) | Real-time safety early warning method for electric vehicle power battery based on operation big data | |
CN115166555A (en) | Battery module internal resistance abnormity identification method and system for battery big data platform | |
CN114879043A (en) | Lithium ion battery lithium analysis diagnosis method, device, equipment and medium | |
Seo et al. | Detection of internal short circuit for lithium-ion battery using convolutional neural networks with data pre-processing | |
CN114035082B (en) | Rapid diagnosis method for abnormal battery cells of new energy vehicle battery system | |
CN114814593B (en) | Multi-fault rapid diagnosis method for battery pack with two-step detection strategy | |
CN117148175A (en) | Method and system for detecting quality of battery cell |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |