CN114152880A - Soft package battery sensor fault online detection method - Google Patents
Soft package battery sensor fault online detection method Download PDFInfo
- Publication number
- CN114152880A CN114152880A CN202210118609.3A CN202210118609A CN114152880A CN 114152880 A CN114152880 A CN 114152880A CN 202210118609 A CN202210118609 A CN 202210118609A CN 114152880 A CN114152880 A CN 114152880A
- Authority
- CN
- China
- Prior art keywords
- matrix
- temperature data
- preset
- value
- training
- 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]
- G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
-
- 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]
- G01R31/378—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC] specially adapted for the type of battery or accumulator
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R35/00—Testing or calibrating of apparatus covered by the other groups of this subclass
-
- 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)
- Secondary Cells (AREA)
Abstract
The invention discloses a soft package battery sensor fault on-line detection method, which comprises the steps of measuring and recording battery input current and terminal voltage data, acquiring surface temperature data of a soft package battery through a thermocouple sensor, and training and verifying a preset prediction model according to a principal component matrix, an input current vector and a terminal voltage data vector by using a rolling time domain on-line learning method on the premise of not knowing information of a system model and parameters to obtain an estimated value of the temperature data at a corresponding moment; and obtaining a fault detection residual according to the thermocouple temperature data measured value and the temperature data estimated value at the corresponding moment, and obtaining a detection result according to the fault detection residual, a preset residual evaluation function and a preset fault detection threshold. And the sensor fault detection of the soft package lithium ion battery is realized by using a small number of thermocouples and current and voltage sensors on the premise of not knowing an accurate model.
Description
Technical Field
The invention belongs to the technical field of fault detection, and particularly relates to an online fault detection method for a soft package battery sensor.
Background
Compared with square and cylindrical batteries, the soft package lithium ion battery has the characteristics of high energy density, high safety, high flexibility and low cost, is a power source of new energy automobiles, and has the operation safety which is a key concern in the scientific research community and the industry. The battery management system monitors, evaluates and manages the battery state information through various sensors (temperature, voltage, current, etc.), and is important for high performance, long service life and safe operation of the battery. However, the number of the sensors is large, and when a fault occurs, the battery management system cannot accurately obtain the state information of the battery system, so that the battery is irreversibly damaged or even has potential safety hazards. At present, a fault detection method based on a battery internal thermodynamic model needs to know information of a system model and parameters in advance, and in practical application, an accurate model is often difficult to obtain and various interferences exist, so that application and popularization of the method are restricted. Therefore, it is highly desirable to develop a data-driven online detection method for the failure of the pouch battery sensor.
Disclosure of Invention
Aiming at the technical problems, the invention provides an online detection method for the faults of a soft package battery sensor.
The technical scheme adopted by the invention for solving the technical problems is as follows:
a soft package battery sensor fault online detection method comprises the following steps:
step S100: measuring and recording battery input current and terminal voltage data, acquiring surface temperature data of the soft package battery through a thermocouple sensor, expressing the input current and terminal voltage data into a preset vector form to obtain an input current vector and a terminal voltage data vector, and expressing the surface temperature data of the soft package battery into a preset matrix form to obtain a temperature data matrix;
step S200: obtaining a singular value matrix by adopting a KL decomposition method for the temperature data matrix, selecting the largest singular value from the singular value matrix, calculating to obtain a corresponding basis function matrix, and obtaining a principal element matrix according to the basis function matrix and the temperature data matrix;
step S300: training and verifying a preset prediction model by using a rolling time domain online learning method according to a principal component matrix, an input current vector and a terminal voltage data vector to obtain an estimated value of temperature data at a corresponding moment;
step S400: the method comprises the steps of obtaining a thermocouple temperature data measured value at a moment, obtaining a fault detection residual according to the thermocouple temperature data measured value and a temperature data estimated value at a corresponding moment, and obtaining a detection result according to the fault detection residual, a preset residual evaluation function and a preset fault detection threshold.
Preferably, in step S100, the input current and terminal voltage data are expressed in a preset vector form, and the surface temperature data of the pouch battery is expressed in a preset matrix form, specifically:
wherein the content of the first and second substances,andrespectively a measured terminal voltage data vector and a current data vector,in order to be a step of time,is composed ofA matrix of temperature data measured by individual thermocouples.
Preferably, step S200 includes:
step S210: obtaining a matrix C according to the temperature data matrix, and performing singular value decomposition on the matrix C to obtain a singular value matrix;
step S220: selecting the largest from the matrix of singular valuesSingular values based on the temperature data matrix and maximumObtaining a basis function matrix by using elements corresponding to the singular values; and obtaining a principal component matrix according to the basis function matrix and the temperature data matrix.
Preferably, the matrix C obtained in step S210 according to the temperature data matrix is specifically:
wherein the content of the first and second substances,in the form of a matrix of temperature data,is the time step;
in step S210, singular value decomposition is performed on the matrix C, specifically:
wherein the content of the first and second substances,the matrix is a singular value matrix, and the matrixes U and V are matrixes calculated after decomposition according to the characteristic values;
s220 specifically comprises the following steps:
wherein the content of the first and second substances,in the form of a matrix of basis functions,is a main element matrix, and the main element matrix,is a matrixMiddle correspondenceThe number of the elements with the maximum singular value, the principal elements and the eigenvectors after dimensionality reduction isAnd satisfy the conditions。
Preferably, step S300 includes:
step S310: on-line learning method using rolling time domain, each principal element in principal element matrix is subjected to on-line learning according to principal element matrix, input current vector and terminal voltage data vectorTraining alone, for principal elementsOf 1 atAn input matrix of a training window is,First action ofSecond actionBy analogy, action m;
Wherein the content of the first and second substances,is the length of the training window, each training window having thereinThe number of the batches is one,is the length of data in each batch, first rowIs the training data of the first batch, and so on, the second batchThe line is the firstTraining data for individual batches;
step S320: the training algorithm adopted for training each batch of data is a lasso algorithm corresponding to the first batchIn a windowThe objective function for the individual batch parameter estimation is:
wherein the content of the first and second substances,representing the value of the variable at which the following equation is minimized,is composed ofFirst of the matrixThe rows of the image data are, in turn,andare respectively the firstEach batch of parameter vectors to be estimated and estimated,andfor the basis function terms and the regularization coefficients,andrespectively representing 2-norm and 1-norm of Euclidean space;
step S330: after trainingAfter the window's data, verify the second with the following inputsA preset prediction model trained for each window:
corresponding preset prediction model verification outputIs composed ofThe calculation formula of (c) is:
wherein the content of the first and second substances,is as followsEstimating parameter vectors of each batch;
step S340: to pairIn the first placePer window preset predictive model validation outputAnd obtaining an estimated value of corresponding temperature data by combining the basis function matrix, specifically:
wherein the content of the first and second substances,is composed ofTime of dayAn estimate of the temperature data for each thermocouple,is composed ofTime of dayAn estimate of the value of the individual pivot,is a basis function matrix;
step S350: when it comes toA first training windowAfter the data of each batch is trained, the data of the second batch is obtainedThe training window moves toA training window, continuing fromThe first batch of the training window begins to train to the second batchA first training windowAnd (5) carrying out batch processing until all the principal elements are trained.
Preferably, the step S400 of obtaining a fault detection residual according to the thermocouple temperature data measurement value and the temperature data estimation value at the corresponding time includes:
wherein the content of the first and second substances,in order to detect the residual error for the fault,is a measurement of the temperature data of the thermocouple,the temperature data estimation value of the corresponding moment is obtained;
wherein the content of the first and second substances,is composed ofFault detection residual at time.
Preferably, the step S400 of obtaining a detection result according to the fault detection residual, the preset residual evaluation function and the preset fault detection threshold includes:
step S410: obtaining a residual error evaluation value according to the fault detection residual error and a preset residual error evaluation function, specifically:
step S420: when the residual evaluation value is larger than the preset fault detection threshold value, the method indicates thatThere is a fault at any moment, otherwise,the method has no fault at any moment, and specifically comprises the following steps:
wherein the content of the first and second substances,is composed ofThe residual evaluation value at the time of the time,is a preset fault detection threshold.
Preferably, step S400 further comprises:
step S430: when the residual evaluation value is larger than the supremum of the peak norm of the Euclidean space of the fault detection residual, switching the preset training mode of the prediction model to an off-line mode, and when the preset training mode of the prediction model is in the off-line mode, estimating the parameter vectorThe real-time updating is not performed any more, and the numerical value of the moment before switching is kept; otherwise, the preset training mode of the prediction model is continuously kept in the online mode.
According to the soft package battery sensor fault on-line detection method, the input current and terminal voltage data of the battery are measured and recorded, the surface temperature data of the soft package battery are collected through the thermocouple sensor, and on the premise that the information of a system model and parameters is not known, a preset prediction model is trained and verified according to a principal component matrix, an input current vector and a terminal voltage data vector by using a rolling time domain on-line learning method to obtain an estimated value of the temperature data at the corresponding moment; and obtaining a fault detection residual error according to the thermocouple temperature data measured value and the temperature data estimated value at the corresponding moment, obtaining a detection result according to the fault detection residual error, a preset residual error evaluation function and a preset fault detection threshold value, and realizing the sensor fault detection of the soft package lithium ion battery according to the detection result.
Drawings
Fig. 1 is a flowchart of a method for online detecting a fault of a pouch battery sensor according to an embodiment of the present invention;
fig. 2 is a schematic diagram of an online learning method of a rolling time domain according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a distribution of 20 thermocouples provided in accordance with an embodiment of the present invention;
FIG. 4 shows an input current under no fault condition according to an embodiment of the present inventionA curve;
FIG. 5 shows a schematic view of the present inventionEmbodiments provide for real terminal voltage without failureA curve;
fig. 6 is a failure detection result in a mode switching manner for a first failure according to an embodiment of the present invention;
FIG. 7 is a diagram illustrating a failure detection result in a continuous online update mode for a first failure according to an embodiment of the present invention;
fig. 8 is a failure detection result in a mode switching manner for a second failure according to an embodiment of the present invention;
fig. 9 is a failure detection result in the continuous online update mode for the second failure in an embodiment of the present invention.
Detailed Description
In order to make the technical solutions of the present invention better understood, the present invention is further described in detail below with reference to the accompanying drawings.
In one embodiment, as shown in fig. 1, an online detection method for a pouch cell sensor fault includes the following steps:
step S100: measuring and recording battery input current and terminal voltage data, acquiring surface temperature data of the soft package battery through a thermocouple sensor, expressing the input current and terminal voltage data into a preset vector form to obtain an input current vector and a terminal voltage data vector, and expressing the surface temperature data of the soft package battery into a preset matrix form to obtain a temperature data matrix;
step S200: obtaining a singular value matrix by adopting a KL decomposition method on the temperature data matrix, and selecting the largest singular value matrixCalculating the singular values to obtain corresponding basis function matrixes, and obtaining a principal element matrix according to the basis function matrixes and the temperature data matrix;
step S300: training and verifying a preset prediction model by using a rolling time domain online learning method according to a principal component matrix, an input current vector and a terminal voltage data vector to obtain an estimated value of temperature data at a corresponding moment;
step S400: the method comprises the steps of obtaining a thermocouple temperature data measured value at a moment, obtaining a fault detection residual according to the thermocouple temperature data measured value and a temperature data estimated value at a corresponding moment, and obtaining a detection result according to the fault detection residual, a preset residual evaluation function and a preset fault detection threshold.
Compared with other fault detection methods, the fault detection method provided by the invention only needs sensor data, and does not need to know the accurate information of the system model. And the sensor fault detection of the soft package lithium ion battery is realized by using a small number of thermocouples and current and voltage sensors on the premise of not knowing an accurate model.
In one embodiment, in step S100, the input current and terminal voltage data are expressed in a preset vector form, and the surface temperature data of the pouch battery is expressed in a preset matrix form, specifically:
wherein the content of the first and second substances,andrespectively a measured terminal voltage data vector and a current data vector,in order to be a step of time,is composed ofA matrix of temperature data measured by individual thermocouples.
Specifically, the current and voltage data are input into the battery thermal distribution model, and the temperature measurement data are output.
In one embodiment, step S200 includes:
step S210: obtaining a matrix C according to the temperature data matrix, and performing singular value decomposition on the matrix C to obtain a singular value matrix;
step S220: selecting the largest from the matrix of singular valuesSingular values based on the temperature data matrix and maximumObtaining a basis function matrix by using elements corresponding to the singular values; and obtaining a principal component matrix according to the basis function matrix and the temperature data matrix.
In one embodiment, the matrix C obtained in step S210 according to the temperature data matrix is specifically:
wherein the content of the first and second substances,in the form of a matrix of temperature data,is the time step;
in step S210, singular value decomposition is performed on the matrix C, specifically:
wherein the content of the first and second substances,the matrix is a singular value matrix, and the matrixes U and V are matrixes calculated after decomposition according to the characteristic values;
s220 specifically comprises the following steps:
wherein the content of the first and second substances,in the form of a matrix of basis functions,is a main element matrix, and the main element matrix,is a matrixMiddle correspondenceThe number of the elements with the maximum singular value, the principal elements and the eigenvectors after dimensionality reduction isAnd satisfy the conditions。
Specifically, forThe matrix adopts a Karhunen-Loe've (KL) decomposition method, and the specific steps are that the matrix is subjected to(the superscript T represents the transpose of the matrix) and Singular Value Decomposition (SVD) is performed to obtainWherein, in the step (A),as singular valuesThe elements on the diagonal line of the matrix are arranged from large to small in singular value, and the maximum element is selectedThe singular values and the corresponding basis function matrix and principal component matrix are calculated as:
wherein the content of the first and second substances,in the form of a matrix of basis functions,is a main element matrix, and the main element matrix,is a matrixMiddle correspondenceThe number of the elements with the maximum singular value, the principal elements and the eigenvectors after dimensionality reduction isAnd satisfy the conditions。
In one embodiment, step S300 includes:
step S310: on-line learning method using rolling time domain, each principal element in principal element matrix is subjected to on-line learning according to principal element matrix, input current vector and terminal voltage data vectorTraining alone, for principal elementsOf 1 atAn input matrix of a training window is,First action ofSecond actionBy analogy, action m;
Wherein the content of the first and second substances,is the length of the training window, each training window having thereinThe number of the batches is one,is the length of data in each batch, first rowIs the training data of the first batch, and so on, the second batchThe line is the firstTraining data for individual batches;
step S320: the training algorithm adopted for training each batch of data is a lasso algorithm corresponding to the first batchIn a windowThe objective function for the individual batch parameter estimation is:
wherein the content of the first and second substances,representing the value of the variable at which the following equation is minimized,is composed ofFirst of the matrixThe rows of the image data are, in turn,andare respectively the firstEach batch of parameter vectors to be estimated and estimated,andfor basis function terms and regularization systemsThe number of the first and second groups is,andrespectively representing 2-norm and 1-norm of Euclidean space;
step S330: after trainingAfter the window's data, verify the second with the following inputsA preset prediction model trained for each window:
corresponding preset prediction model verification outputIs composed ofThe calculation formula of (c) is:
wherein the content of the first and second substances,is as followsEstimating parameter vectors of each batch;
step S340: for the firstPresetting of individual windowsIs output by the predictive model verificationAnd obtaining an estimated value of corresponding temperature data by combining the basis function matrix, specifically:
wherein the content of the first and second substances,is composed ofTime of dayAn estimate of the temperature data for each thermocouple,is composed ofTime of dayAn estimate of the value of the individual pivot,is a basis function matrix;
step S350: when it comes toA first training windowAfter the data of each batch is trained, the data of the second batch is obtainedThe training window moves toA training window, continuing fromThe first batch of the training window begins to train to the second batchA first training windowAnd (5) carrying out batch processing until all the principal elements are trained.
Specifically, for the principal elements after the dimension reduction, a new online learning method of a rolling time domain is proposed for model training and prediction, and a schematic diagram of the method is shown in fig. 2.
In one embodiment, the step S400 of obtaining the fault detection residual according to the thermocouple temperature data measurement value and the temperature data estimation value at the corresponding time includes:
wherein the content of the first and second substances,in order to detect the residual error for the fault,is a measurement of the temperature data of the thermocouple,the temperature data estimation value of the corresponding moment is obtained;
wherein the content of the first and second substances,is composed ofFault detection residual at time.
In one embodiment, the obtaining the detection result according to the fault detection residual, the preset residual evaluation function and the preset fault detection threshold in step S400 includes:
step S410: obtaining a residual error evaluation value according to the fault detection residual error and a preset residual error evaluation function, specifically:
step S420: when the residual evaluation value is larger than the preset fault detection threshold value, the method indicates thatThere is a fault at any moment, otherwise,the method has no fault at any moment, and specifically comprises the following steps:
wherein the content of the first and second substances,is composed ofThe residual evaluation value at the time of the time,is a preset fault detection threshold.
In one embodiment, step S400 further comprises:
step S430: when the residual evaluation value is larger than the supremum of the peak norm of the Euclidean space of the fault detection residual, switching the preset training mode of the prediction model to an off-line mode, and when the preset training mode of the prediction model is in the off-line mode, estimating the parameter vectorThe real-time updating is not performed any more, and the numerical value of the moment before switching is kept; otherwise, the preset training mode of the prediction model is continuously kept in the online mode.
Specifically, the fault detection threshold is obtained by a Kernel Density Estimation (KDE)The specific steps are to adopt Gaussian to check an evaluation function under the condition of no faultIs estimated, defining a significance levelTime of day correspondingDetecting thresholds for faults。
In a detailed embodiment, an experimental platform comprising a battery test cabinet, a thermostat and a battery management system is set up to collect battery surface temperature data, an experimental object is a LiFePO 4/graphite soft package lithium ion battery, the length, the width and the thickness of the battery are respectively 0.24 meter, 0.18 meter and 0.00783 meter (the thickness is negligible), and the total thickness is 0.24 meter, 0.18 meter and 0.00783 meter20 thermocouples were located on the surface, as schematically shown in FIG. 3. FIG. 4 and FIG. 5 show the input current in the fault-free caseAnd true terminal voltageCurve line. The following two types of sensor faults are injected respectively:
failure 1: constant measurement bias of 1K was injected from 2500 seconds for thermocouples No. 10 to No. 15 in fig. 3;
and (3) failure 2: the end-to-end voltage measurement sensor injects a constant measurement bias of 3V starting at 2500 seconds.
Fig. 6 and 7 show the results of failure detection in the case where the mode switching manner in step S430 is adopted for failure 1) and the continuous online update mode is adopted, respectively, and fig. 8 and 9 show the results of failure detection in the case where the mode switching manner in step S430 is adopted for failure 2) and the continuous online update mode is adopted, respectively. Comparing fig. 6 and 7, and fig. 8 and 9, it can be seen that the Fault Detection Rate (FDR) can be effectively improved by using the mode switching manner proposed in this patent, and table 1 is a specific value of the Fault detection rate under different situations.
TABLE 1 Fault detection results for different training modes
Compared with other fault detection methods, the fault detection method of the soft package battery sensor only needs sensor data, and does not need to know accurate information of a system model. Compared with the method adopting continuous online updating, the mode switching method provided by the invention can effectively improve the fault detection rate. And the sensor fault detection of the soft package lithium ion battery is realized by using a small number of thermocouples and current and voltage sensors on the premise of not knowing an accurate model.
The online fault detection method for the soft package battery sensor provided by the invention is described in detail above. The principles and embodiments of the present invention are explained herein using specific examples, which are presented only to assist in understanding the core concepts of the present invention. It should be noted that, for those skilled in the art, it is possible to make various improvements and modifications to the present invention without departing from the principle of the present invention, and those improvements and modifications also fall within the scope of the claims of the present invention.
Claims (8)
1. The online detection method for the fault of the soft package battery sensor is characterized by comprising the following steps:
step S100: measuring and recording battery input current and terminal voltage data, acquiring surface temperature data of the soft package battery through a thermocouple sensor, expressing the input current and the terminal voltage data into a preset vector form to obtain an input current vector and a terminal voltage data vector, and expressing the surface temperature data of the soft package battery into a preset matrix form to obtain a temperature data matrix;
step S200: obtaining a singular value matrix by adopting a KL decomposition method for the temperature data matrix, selecting the largest singular value from the singular value matrix, calculating to obtain a corresponding basis function matrix, and obtaining a principal element matrix according to the basis function matrix and the temperature data matrix;
step S300: training and verifying a preset prediction model according to the principal component matrix, the input current vector and the terminal voltage data vector by using a rolling time domain online learning method to obtain an estimated value of temperature data at a corresponding moment;
step S400: the method comprises the steps of obtaining a thermocouple temperature data measured value at a moment, obtaining a fault detection residual according to the thermocouple temperature data measured value and the temperature data estimated value at the corresponding moment, and obtaining a detection result according to the fault detection residual, a preset residual evaluation function and a preset fault detection threshold.
2. The method according to claim 1, wherein in step S100, the input current and the terminal voltage data are expressed in a preset vector form, and the surface temperature data of the pouch battery is expressed in a preset matrix form, specifically:
3. The method according to claim 2, wherein step S200 comprises:
step S210: obtaining a matrix C according to the temperature data matrix, and performing singular value decomposition on the matrix C to obtain a singular value matrix;
step S220: selecting the largest of the singular value matricesA singular value according to the temperatureData matrix and maximumObtaining a basis function matrix by using elements corresponding to the singular values; and obtaining a principal component matrix according to the basis function matrix and the temperature data matrix.
4. The method according to claim 3, wherein the obtaining of the matrix C according to the temperature data matrix in step S210 specifically includes:
wherein the content of the first and second substances,in the form of a matrix of temperature data,is the time step;
in step S210, singular value decomposition is performed on the matrix C, specifically:
wherein the content of the first and second substances,the matrix is a singular value matrix, and the matrixes U and V are matrixes calculated after decomposition according to the characteristic values;
s220 specifically comprises the following steps:
wherein the content of the first and second substances,in the form of a matrix of basis functions,is a main element matrix, and the main element matrix,is a matrixMiddle correspondenceThe number of the elements with the maximum singular value, the principal elements and the eigenvectors after dimensionality reduction isAnd satisfy the conditions。
5. The method of claim 4, wherein step S300 comprises:
step S310: on-line learning method using rolling time domain, according to the principal component matrix, the input current vector and the terminal voltage data vector, each principal component in the principal component matrixTraining alone, for principal elementsOf 1 atAn input matrix of a training window is,First action ofSecond actionBy analogy, action m;
Wherein the content of the first and second substances,is the length of the training window, each training window having thereinThe number of the batches is one,is the length of data in each batch, first rowIs the training data of the first batch, and so on, the second batchThe line is the firstTraining data for individual batches;
step S320: the training algorithm adopted for training each batch of data is a lasso algorithm corresponding to the first batchIn a windowThe objective function for the individual batch parameter estimation is:
wherein the content of the first and second substances,representing the value of the variable at which the following equation is minimized,is composed ofFirst of the matrixThe rows of the image data are, in turn,andare respectively the firstEach batch of parameter vectors to be estimated and estimated,andfor the basis function terms and the regularization coefficients,andrespectively representing 2-norm and 1-norm of Euclidean space;
step S330: after trainingAfter the window's data, the following inputs are used to verify the second windowA preset prediction model trained for each window:
corresponding preset prediction model verification outputIs composed ofThe calculation formula of (c) is:
wherein the content of the first and second substances,is as followsEstimating parameter vectors of each batch;
step S340: for the said firstPer window preset predictive model validation outputAnd obtaining an estimated value of corresponding temperature data by combining the basis function matrix, specifically:
wherein the content of the first and second substances,is composed ofTime of dayAn estimate of the temperature data for each thermocouple,is composed ofTime of dayAn estimate of the value of the individual pivot,is a basis function matrix;
step S350: when it comes toA first training windowAfter the data of each batch is trained, the data of the second batch is obtainedThe training window moves toA training window, continuing fromThe first batch of the training window begins to train to the second batchA first training windowAnd (5) carrying out batch processing until all the principal elements are trained.
6. The method of claim 5, wherein the step S400 of obtaining a fault detection residual according to the thermocouple temperature data measured value and the temperature data estimated value at the corresponding time comprises:
wherein the content of the first and second substances,in order to detect the residual error for the fault,is a measurement of the temperature data of the thermocouple,the temperature data estimation value of the corresponding moment is obtained;
7. The method according to claim 6, wherein obtaining the detection result according to the fault detection residual, the preset residual evaluation function and the preset fault detection threshold in step S400 includes:
step S410: obtaining a residual error evaluation value according to the fault detection residual error and a preset residual error evaluation function, specifically:
step S420: when the residual evaluation value is larger than a preset fault detection threshold value, indicating that the residual evaluation value is larger than the preset fault detection threshold valueThere is a fault at any moment, otherwise,the method has no fault at any moment, and specifically comprises the following steps:
8. The method of claim 7, wherein step S400 further comprises:
step S430: when the residual error evaluation value is larger than the supremum of the peak norm of the Euclidean space of the fault detection residual error, switching the preset training mode of the prediction model to an off-line mode, and when the preset training mode of the prediction model is in the off-line mode, estimating the parameter vectorThe real-time updating is not performed any more, and the numerical value of the moment before switching is kept; otherwise, the preset training mode of the prediction model is continuously kept in the online mode.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210118609.3A CN114152880B (en) | 2022-02-08 | 2022-02-08 | Soft package battery sensor fault online detection method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210118609.3A CN114152880B (en) | 2022-02-08 | 2022-02-08 | Soft package battery sensor fault online detection method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114152880A true CN114152880A (en) | 2022-03-08 |
CN114152880B CN114152880B (en) | 2022-04-12 |
Family
ID=80450254
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210118609.3A Active CN114152880B (en) | 2022-02-08 | 2022-02-08 | Soft package battery sensor fault online detection method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114152880B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114859246A (en) * | 2022-07-07 | 2022-08-05 | 江苏中兴派能电池有限公司 | Soft package battery detection method and device, computer equipment and storage medium |
CN116879763A (en) * | 2023-09-07 | 2023-10-13 | 上海融和元储能源有限公司 | Battery fault early warning method based on Kalman filtering algorithm |
CN117007984A (en) * | 2023-09-27 | 2023-11-07 | 南通国轩新能源科技有限公司 | Dynamic monitoring method and system for operation faults of battery pack |
Citations (22)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20080254347A1 (en) * | 2007-04-12 | 2008-10-16 | International Truck Intellectual Property Company, Llc | Vehicle battery state of charge indicator |
CN102340811A (en) * | 2011-11-02 | 2012-02-01 | 中国农业大学 | Method for carrying out fault diagnosis on wireless sensor networks |
CN104102773A (en) * | 2014-07-05 | 2014-10-15 | 山东鲁能软件技术有限公司 | Equipment fault warning and state monitoring method |
CN106526488A (en) * | 2016-09-27 | 2017-03-22 | 北京理工大学 | Fault diagnosis method of sensors in tandem type power battery pack |
CN107035490A (en) * | 2017-03-29 | 2017-08-11 | 北京航空航天大学 | A kind of SCR system of diesel engine nitrogen oxides input pickup method for diagnosing faults |
JP2017211183A (en) * | 2016-05-23 | 2017-11-30 | シャープ株式会社 | Charge rate detection device and charge rate detection method for secondary batteries |
CN109061537A (en) * | 2018-08-23 | 2018-12-21 | 重庆大学 | Electric vehicle lithium ion battery sensor fault diagnosis method based on observer |
CN109752113A (en) * | 2019-01-22 | 2019-05-14 | 南京市计量监督检测院 | Location determining method and circuit design method in web temperature sensor and its application |
CN111190066A (en) * | 2020-01-14 | 2020-05-22 | 中南大学 | Fault diagnosis method for matrix converter motor driving system |
US20200217897A1 (en) * | 2017-07-26 | 2020-07-09 | Invenox Gmbh | Method and Device for Detecting Battery Cell States and Battery Cell Parameters |
CN111578931A (en) * | 2020-05-21 | 2020-08-25 | 中国人民解放军海军航空大学 | High-dynamic aircraft autonomous attitude estimation method based on online rolling time domain estimation |
CN111902731A (en) * | 2018-03-21 | 2020-11-06 | 祖克斯有限公司 | Automatic detection of sensor calibration errors |
CN111953009A (en) * | 2019-05-17 | 2020-11-17 | 天津科技大学 | Fault diagnosis method for island multi-inverter parallel sensor |
CN111965547A (en) * | 2020-09-27 | 2020-11-20 | 哈尔滨工业大学(威海) | Battery system sensor fault diagnosis method based on parameter identification method |
CN112098851A (en) * | 2020-11-06 | 2020-12-18 | 北京理工大学 | Intelligent battery and online state of charge estimation method and application thereof |
CN113343633A (en) * | 2021-06-10 | 2021-09-03 | 上海交通大学 | Thermal runaway fault classification and risk prediction method and system for power lithium battery |
CN113534000A (en) * | 2021-07-05 | 2021-10-22 | 合肥工业大学 | New energy automobile driving system inverter and current sensor fault diagnosis method |
CN113571742A (en) * | 2021-06-08 | 2021-10-29 | 北京格睿能源科技有限公司 | Fault diagnosis method and device for fuel cell thermal management system |
CN113608127A (en) * | 2021-10-08 | 2021-11-05 | 湖南大学 | Thermal anomaly detection and positioning method for cylindrical lithium ion battery |
CN113625182A (en) * | 2021-07-23 | 2021-11-09 | 北京理工大学 | Method for on-line estimating battery state |
CN113671380A (en) * | 2021-08-23 | 2021-11-19 | 哈尔滨工业大学(威海) | Deep learning-based multi-fault diagnosis method for power battery system |
CN113884884A (en) * | 2021-10-21 | 2022-01-04 | 山东大学 | Power battery pack fault diagnosis method and system based on correlation |
-
2022
- 2022-02-08 CN CN202210118609.3A patent/CN114152880B/en active Active
Patent Citations (22)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20080254347A1 (en) * | 2007-04-12 | 2008-10-16 | International Truck Intellectual Property Company, Llc | Vehicle battery state of charge indicator |
CN102340811A (en) * | 2011-11-02 | 2012-02-01 | 中国农业大学 | Method for carrying out fault diagnosis on wireless sensor networks |
CN104102773A (en) * | 2014-07-05 | 2014-10-15 | 山东鲁能软件技术有限公司 | Equipment fault warning and state monitoring method |
JP2017211183A (en) * | 2016-05-23 | 2017-11-30 | シャープ株式会社 | Charge rate detection device and charge rate detection method for secondary batteries |
CN106526488A (en) * | 2016-09-27 | 2017-03-22 | 北京理工大学 | Fault diagnosis method of sensors in tandem type power battery pack |
CN107035490A (en) * | 2017-03-29 | 2017-08-11 | 北京航空航天大学 | A kind of SCR system of diesel engine nitrogen oxides input pickup method for diagnosing faults |
US20200217897A1 (en) * | 2017-07-26 | 2020-07-09 | Invenox Gmbh | Method and Device for Detecting Battery Cell States and Battery Cell Parameters |
CN111902731A (en) * | 2018-03-21 | 2020-11-06 | 祖克斯有限公司 | Automatic detection of sensor calibration errors |
CN109061537A (en) * | 2018-08-23 | 2018-12-21 | 重庆大学 | Electric vehicle lithium ion battery sensor fault diagnosis method based on observer |
CN109752113A (en) * | 2019-01-22 | 2019-05-14 | 南京市计量监督检测院 | Location determining method and circuit design method in web temperature sensor and its application |
CN111953009A (en) * | 2019-05-17 | 2020-11-17 | 天津科技大学 | Fault diagnosis method for island multi-inverter parallel sensor |
CN111190066A (en) * | 2020-01-14 | 2020-05-22 | 中南大学 | Fault diagnosis method for matrix converter motor driving system |
CN111578931A (en) * | 2020-05-21 | 2020-08-25 | 中国人民解放军海军航空大学 | High-dynamic aircraft autonomous attitude estimation method based on online rolling time domain estimation |
CN111965547A (en) * | 2020-09-27 | 2020-11-20 | 哈尔滨工业大学(威海) | Battery system sensor fault diagnosis method based on parameter identification method |
CN112098851A (en) * | 2020-11-06 | 2020-12-18 | 北京理工大学 | Intelligent battery and online state of charge estimation method and application thereof |
CN113571742A (en) * | 2021-06-08 | 2021-10-29 | 北京格睿能源科技有限公司 | Fault diagnosis method and device for fuel cell thermal management system |
CN113343633A (en) * | 2021-06-10 | 2021-09-03 | 上海交通大学 | Thermal runaway fault classification and risk prediction method and system for power lithium battery |
CN113534000A (en) * | 2021-07-05 | 2021-10-22 | 合肥工业大学 | New energy automobile driving system inverter and current sensor fault diagnosis method |
CN113625182A (en) * | 2021-07-23 | 2021-11-09 | 北京理工大学 | Method for on-line estimating battery state |
CN113671380A (en) * | 2021-08-23 | 2021-11-19 | 哈尔滨工业大学(威海) | Deep learning-based multi-fault diagnosis method for power battery system |
CN113608127A (en) * | 2021-10-08 | 2021-11-05 | 湖南大学 | Thermal anomaly detection and positioning method for cylindrical lithium ion battery |
CN113884884A (en) * | 2021-10-21 | 2022-01-04 | 山东大学 | Power battery pack fault diagnosis method and system based on correlation |
Non-Patent Citations (4)
Title |
---|
OLAOLUWA OJO等: "A Neural Network Based Method for Thermal Fault Detection in Lithium-Ion Batteries", 《 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS》 * |
PINGYU ZHU等: "Reducing Residual Strain in Fiber Bragg Grating Temperature Sensors Embedded in Carbon Fiber Reinforced Polymers", 《 JOURNAL OF LIGHTWAVE TECHNOLOGY》 * |
宋绍民等: "传感器数据的精确重构方法及其性能研究", 《传感技术学报》 * |
朱浩等: "锂电池异常状态监测的FEKF方法研究", 《湖南大学学报(自然科学版)》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114859246A (en) * | 2022-07-07 | 2022-08-05 | 江苏中兴派能电池有限公司 | Soft package battery detection method and device, computer equipment and storage medium |
CN114859246B (en) * | 2022-07-07 | 2022-09-09 | 江苏中兴派能电池有限公司 | Soft package battery detection method and device, computer equipment and storage medium |
CN116879763A (en) * | 2023-09-07 | 2023-10-13 | 上海融和元储能源有限公司 | Battery fault early warning method based on Kalman filtering algorithm |
CN117007984A (en) * | 2023-09-27 | 2023-11-07 | 南通国轩新能源科技有限公司 | Dynamic monitoring method and system for operation faults of battery pack |
CN117007984B (en) * | 2023-09-27 | 2023-12-15 | 南通国轩新能源科技有限公司 | Dynamic monitoring method and system for operation faults of battery pack |
Also Published As
Publication number | Publication date |
---|---|
CN114152880B (en) | 2022-04-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN114152880B (en) | Soft package battery sensor fault online detection method | |
Chang et al. | Lithium-ion battery state of health estimation using the incremental capacity and wavelet neural networks with genetic algorithm | |
Wang et al. | A transferable lithium-ion battery remaining useful life prediction method from cycle-consistency of degradation trend | |
CN112147512B (en) | Diagnosis and separation method for short-circuit and abuse faults of lithium ion battery | |
Wei et al. | Lyapunov-based thermal fault diagnosis of cylindrical lithium-ion batteries | |
Sun et al. | State of charge estimation for lithium‐ion battery based on an intelligent adaptive unscented Kalman filter | |
CN113484770B (en) | Method and system for measuring and calculating core temperature inside battery on line based on charge and discharge data | |
Yu et al. | State of health estimation method for lithium-ion batteries based on multiple dynamic operating conditions | |
CN113447828B (en) | Lithium battery temperature estimation method and system based on Bayesian neural network | |
CN114167284B (en) | Lithium battery RUL prediction method and equipment based on BMS big data and integrated learning | |
CN113255764A (en) | Method, system and device for detecting electrochemical energy storage system fault by using machine learning | |
CN113962154A (en) | Lithium battery nuclear temperature evaluation method and system based on transfer learning | |
Wu et al. | A fault detection method of electric vehicle battery through Hausdorff distance and modified Z-score for real-world data | |
Ding et al. | Diagnosis of connection fault for parallel-connected lithium-ion batteries based on long short-term memory networks | |
Shen et al. | Source-free cross-domain state of charge estimation of lithium-ion batteries at different ambient temperatures | |
CN117154263A (en) | Lithium battery cascade utilization charging and discharging system and control method | |
CN113884884B (en) | Power battery pack fault diagnosis method and system based on correlation | |
Lin et al. | Algorithm of BPNN‐UKF based on a fusion model for SOC estimation in lithium‐ion batteries | |
Zhang et al. | A deep learning approach to estimate the state of health of lithium-ion batteries under varied and incomplete working conditions | |
CN114184972A (en) | Method and equipment for automatically estimating SOH (state of health) of battery by combining data driving with electrochemical mechanism | |
Yu | Data-driven prognostics and remaining useful life estimation for lithium-ion battery: A review | |
Gu et al. | A precise minor-fault diagnosis method for lithium-ion batteries based on phase plane sample entropy | |
Sun et al. | State-of-health estimation for lithium-ion battery using model-based feature optimization and deep extreme learning machine | |
Yang et al. | Multi-fault diagnosis for battery pack based on adaptive correlation sequence and sparse classification model | |
CN114720901A (en) | Method for estimating health state of lithium ion battery |
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 |