A battery fault judgment method and system based on principal component analysis

Abstract

The present disclosure proposes a battery fault diagnosis method and system based on principal component analysis. The method includes obtaining historical operating data of fault-free batteries and constructing a data matrix based on the historical operating data of fault-free batteries; constructing a principal component analysis model, which obtains a covariance matrix based on the input data matrix, and then performs singular value decomposition on the covariance matrix to output a projection matrix of principal components and residual components; obtaining real-time operating data of a target battery, and calculating the Hotelling statistic, squared prediction error, and corresponding control limits of the target battery based on the real-time operating data, the principal component and residual component projection matrices, and the significance level; judging whether the target battery has a fault based on the Hotelling statistic, squared prediction error, and corresponding control limits; and determining the location of the battery fault if the target battery has a fault. According to the method disclosed in the present disclosure, battery faults and the location of the fault can be more accurately determined.

Description

Technical Field

[0001] The present disclosure relates to the field of intelligent fault diagnosis of battery energy storage systems, and in particular to a battery fault diagnosis method and system based on principal component analysis. Background Art

[0002] Global warming and environmental damage are attributed to the overuse of traditional fossil fuels such as coal. Countries have recognized the need for energy transition and are actively developing and utilizing green, low-carbon energy sources to reduce carbon emissions. To promote this energy transition, electrochemical energy storage, as a key means of improving energy efficiency, has been widely promoted in recent years. Batteries, particularly lithium-ion batteries, have become the preferred electrochemical energy storage solution due to their high energy density and lack of memory effect. Lithium-ion batteries are crucial for the safe operation of energy storage power stations. To ensure timely inspection and maintenance of these batteries, real-time prediction of potential faults and timely identification of fault types are crucial.

[0003] Existing fault diagnosis methods can be categorized into model-based and data-driven approaches based on their principles. Battery energy storage system fault diagnosis research and technology already has a substantial theoretical foundation and practical experience. With the increasing global demand for battery energy storage, particularly lithium-ion battery storage, battery energy storage systems are moving toward integration and lightweighting, and their fault diagnosis technology is also rapidly developing. However, these existing fault diagnosis technologies have low sensitivity when diagnosing minor faults. Summary of the Invention

[0004] The present disclosure aims to solve one of the technical problems in the related art at least to a certain extent.

[0005] To this end, the first purpose of the present disclosure is to propose a battery fault diagnosis method based on principal component analysis, the main purpose of which is to more accurately determine the battery fault and the location of the fault, thereby improving the sensitivity of minor fault diagnosis.

[0006] The second objective of the present disclosure is to provide a battery fault diagnosis system based on principal component analysis.

[0007] The third objective of the present disclosure is to provide a battery fault diagnosis device based on principal component analysis.

[0008] A fourth object of the present disclosure is to provide a non-transitory computer-readable storage medium.

[0009] To achieve the above objectives, a first embodiment of the present disclosure provides a battery fault diagnosis method based on principal component analysis, comprising:

[0010] Acquire historical operating data of a fault-free battery, and construct a data matrix based on the historical operating data of the fault-free battery;

[0011] Constructing a principal component analysis model, wherein the principal component analysis model obtains a covariance matrix based on an input data matrix, and then performs singular value decomposition on the covariance matrix to output a principal component projection matrix and a residual component projection matrix;

[0012] Acquire real-time operating data of a target battery, and calculate a Hotelling statistic, a squared prediction error, and corresponding control limits of the target battery based on the real-time operating data, the principal component projection matrix, the residual component projection matrix, and a significance level;

[0013] Determining whether the target battery is faulty based on the Hotelling statistic, the squared prediction error, and the corresponding control limit;

[0014] If the target battery fails, a battery failure location is determined based on the Hotelling statistic and the squared prediction error.

[0015] In one embodiment of the present disclosure, determining the battery fault location based on the Hotelling statistic and the squared prediction error includes: decomposing the Hotelling statistic to obtain the Hotelling statistic of each battery cell in the target battery; decomposing the squared prediction error to obtain the squared prediction error of each battery cell in the target battery; and determining the battery fault location based on the battery cell Hotelling statistic and the battery cell squared prediction error.

[0016] In one embodiment of the present disclosure, the control limits include a Hotelling statistic control limit and a squared prediction error control limit, and determining whether the target battery has a fault based on the Hotelling statistic, the squared prediction error and the corresponding control limits includes: if the Hotelling statistic is less than or equal to the Hotelling statistic control limit, and the squared prediction error is less than or equal to the squared prediction error control limit, then the target battery has no fault; otherwise, the target battery has a fault.

[0017] In one embodiment of the present disclosure, the Hotelling statistic, square prediction error and corresponding control limit of the target battery are calculated based on the real-time operating data, the principal component projection matrix, the residual component projection matrix and the significance level, including: calculating the Hotelling statistic of the target battery based on the real-time operating data and the principal component projection matrix; calculating the square prediction error of the target battery based on the real-time operating data and the residual component projection matrix; and calculating the Hotelling statistic control limit and the square prediction error control limit based on the significance level.

[0018] In one embodiment of the present disclosure, the singular value decomposition of the covariance matrix to output a principal component projection matrix and a residual component projection matrix includes: performing singular value decomposition on the covariance matrix to obtain a target matrix; obtaining the principal component projection matrix based on the eigenvectors of a preset number of columns of the target matrix; and obtaining the residual component projection matrix based on the eigenvectors of the remaining number of columns in the target matrix excluding the preset number of columns.

[0019] In one embodiment of the present disclosure, it also includes: using the operating data of the target battery when it is fault-free to update the historical operating data of the fault-free battery, and then updating the data matrix; inputting the updated data matrix into the principal component analysis model to output a new principal component projection matrix and a residual component projection matrix.

[0020] To achieve the above objectives, a second embodiment of the present disclosure further provides a battery fault diagnosis system based on principal component analysis, comprising:

[0021] An acquisition module is used to obtain historical operating data of fault-free batteries and real-time operating data of target batteries;

[0022] a modeling module, configured to construct a data matrix based on the historical operating data of the fault-free battery, and to construct a principal component analysis model, wherein the principal component analysis model obtains a covariance matrix based on the input data matrix, and then performs singular value decomposition on the covariance matrix to output a principal component projection matrix and a residual component projection matrix;

[0023] A calculation module, configured to calculate the Hotelling statistic, the squared prediction error, and the corresponding control limit of the target battery based on the real-time operating data, the principal component projection matrix, the residual component projection matrix, and the significance level;

[0024] A first judgment module is configured to judge whether a fault occurs in the target battery based on the Hotelling statistic, the squared prediction error, and the corresponding control limit;

[0025] The second judgment module is configured to determine a battery fault location based on the Hotelling statistic and the squared prediction error if the target battery fails.

[0026] In one embodiment of the present disclosure, the battery fault judgment system also includes an update module, which is used to update the historical operating data of the fault-free battery using the operating data of the target battery when it is fault-free; the modeling module updates the data matrix based on the updated historical operating data of the fault-free battery, and inputs the updated data matrix into the principal component analysis model to output a new principal component projection matrix and a residual component projection matrix.

[0027] To achieve the above-mentioned purpose, the third aspect embodiment of the present disclosure also provides a battery fault judgment device based on the principal component analysis method, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the battery fault judgment method based on the principal component analysis method proposed in the first aspect embodiment of the present disclosure.

[0028] In order to achieve the above-mentioned objectives, the fourth embodiment of the present disclosure proposes a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable the computer to execute the battery fault judgment method based on principal component analysis proposed in the first embodiment of the present disclosure.

[0029] In one or more embodiments of the present disclosure, historical operating data of a fault-free battery is obtained, and a data matrix is ​​constructed based on the historical operating data of the fault-free battery. A principal component analysis model is constructed, and the principal component analysis model obtains a covariance matrix based on the input data matrix, and then performs singular value decomposition on the covariance matrix to output a principal component projection matrix and a residual component projection matrix. Real-time operating data of a target battery is obtained, and the Hotelling statistic, squared prediction error, and corresponding control limits of the target battery are calculated based on the real-time operating data, the principal component projection matrix, the residual component projection matrix, and the significance level. Whether the target battery has a fault is determined based on the Hotelling statistic, the squared prediction error, and the corresponding control limits. If the target battery has a fault, the battery fault location is determined based on the Hotelling statistic and the squared prediction error. In this case, the principal component analysis model is used to obtain the principal component projection matrix and the residual component projection matrix based on the historical operating data of the fault-free battery. Then, the Hotelling statistic, squared prediction error, and corresponding control limits of the target battery are calculated based on the real-time operating data, the principal component projection matrix, the residual component projection matrix, and the significance level, thereby determining whether the target battery has a fault and the location of the fault. This makes it possible to more accurately determine battery failures and their locations, improving the sensitivity of minor fault diagnosis.

[0030] Additional aspects and advantages of the present disclosure will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The above and / or additional aspects and advantages of the present disclosure will become apparent and readily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0032] Figure 1 A schematic diagram illustrating a flow chart of a battery fault diagnosis method based on principal component analysis provided by an embodiment of the present disclosure is shown;

[0033] Figure 2 A schematic diagram showing a flow chart of another battery fault diagnosis method based on principal component analysis provided by an embodiment of the present disclosure;

[0034] Figure 3 A block diagram of a battery fault diagnosis system based on principal component analysis according to an embodiment of the present disclosure is shown;

[0035] Figure 4 The block diagram is a battery fault judgment device based on principal component analysis method for implementing the battery fault judgment method based on principal component analysis method according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0036] Exemplary embodiments are described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all possible implementations consistent with the present disclosure. Rather, they are merely examples of apparatuses and methods consistent with certain aspects of the present disclosure, as detailed in the appended claims.

[0037] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine different embodiments or examples described in this specification and features of different embodiments or examples, unless they are mutually inconsistent.

[0038] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of the present disclosure, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined. It should also be understood that the term "and / or" used in the present disclosure refers to and includes any or all possible combinations of one or more associated listed items.

[0039] The following describes in detail embodiments of the present disclosure, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present disclosure, and should not be construed as limiting the present disclosure.

[0040] The present disclosure provides a battery fault diagnosis method and system based on principal component analysis, the main purpose of which is to more accurately determine battery faults and fault locations, thereby improving the sensitivity of minor fault diagnosis.

[0041] In a first embodiment, Figure 1 A flow chart of a battery fault diagnosis method based on principal component analysis provided by an embodiment of the present disclosure is shown.

[0042] like Figure 1 Specifically, the battery fault judgment method based on principal component analysis includes:

[0043] Step S11 : acquiring historical operating data of the battery without faults, and constructing a data matrix based on the historical operating data of the battery without faults.

[0044] In step S11 , the battery targeted by the battery fault judgment method may be, but is not limited to, a lithium-ion battery.

[0045] In step S11 , the historical operating data of the battery without faults may be various types of operating data such as current and voltage stored when the battery is operating without faults.

[0046] In step S11 , the battery includes a plurality of battery cells, so the historical operating data of the battery without faults includes various types of operating data such as the current and voltage of each battery cell during the fault-free operation.

[0047] In step S11, a corresponding data matrix is ​​constructed using any type of operating data. For example, the current of each battery cell during trouble-free operation can be selected to construct a data matrix corresponding to the battery current, or the voltage of each battery cell during trouble-free operation can be selected to construct a data matrix corresponding to the battery voltage. The data matrices corresponding to various operating data can be fed into subsequent steps for battery fault diagnosis.

[0048] In step S11, the data matrix corresponding to the battery voltage is constructed based on the voltage of each battery cell during trouble-free operation as an example. The data matrix constructed based on the trouble-free battery historical operation data can be represented by the symbol Z. The data matrix Z satisfies Z∈R N×mWhere R represents a real matrix. Each element in the real matrix represents the battery voltage. m represents the number of battery cells (i.e., the number of variables), and N represents the total number of voltages detected by each battery cell (i.e., the number of samples for each variable). When detecting the voltage, the voltage can be collected at a preset time interval. The preset time interval can be determined based on different battery types and different actual operating conditions.

[0049] Step S12: constructing a principal component analysis model. The principal component analysis model obtains a covariance matrix based on the input data matrix, and then performs singular value decomposition on the covariance matrix to output a principal component projection matrix and a residual component projection matrix.

[0050] In step S12, the principal component analysis model constructed is a principal component analysis (PCA) method, which obtains the covariance matrix of the input data matrix and then obtains the principal component projection matrix and the residual component projection matrix of the covariance matrix.

[0051] Specifically, in step S12, the covariance matrix is ​​obtained based on the input data matrix using formula (1). Formula (1) is:

[0052]

[0053] Where C Z represents the covariance matrix, N represents the number of samples for each variable, Z represents the data matrix, and Z T represents the transpose of the data matrix.

[0054] In step S12, in order to eliminate the order of magnitude differences between various variable data and avoid a major impact on the subsequent principal component analysis, before obtaining the covariance matrix based on the input data matrix, the data matrix is ​​first normalized so that the sample mean of each variable is 0, and then the covariance matrix of the normalized data matrix is ​​calculated.

[0055] In step S12, singular value decomposition is performed on the covariance matrix to output a principal component projection matrix and a residual component projection matrix, including: performing singular value decomposition on the covariance matrix to obtain a target matrix; obtaining a principal component projection matrix based on the eigenvectors of a preset number of columns of the target matrix; and obtaining a residual component projection matrix based on the eigenvectors of the remaining number of columns in the target matrix excluding the preset number of columns.

[0056] Specifically, the covariance matrix obtained by formula (1) is subjected to singular value decomposition (SVD) by formula (2), which is:

[0057]

[0058] Where U∈R m×m ,∑∈R m×m , V∈R m×m . Matrix U is a unitary matrix, V * is the transposed conjugate matrix of the unitary matrix V. Since the covariance matrix C Z Symmetric and positive definite, so the target matrix P is obtained based on the unitary matrix U, and the target matrix P satisfies P=U=V * , when the diagonal matrix ∧ satisfies ∧=∑, the covariance matrix C Z It can be converted into formula (3):

[0059] C Z =P∧P T (3)

[0060] In the formula, each column of the matrix P is the eigenvector of the covariance matrix, and the diagonal matrix ∧ is expressed as formula (4):

[0061] ∧=diag(λ1,...,λ m ) (4)

[0062] Where diag represents the construction of a diagonal matrix, λ1,...,λ m Denotes the diagonal values ​​of the diagonal matrix ∧. Where the diagonal values ​​λ1, ..., λ m are the eigenvalues ​​of the covariance matrix sorted in descending order.

[0063] By selecting the eigenvectors of a preset number of columns from the target matrix P, the projection matrix of the principal component (PC) of the covariance matrix can be obtained. The preset number of columns can be represented by the symbol γ. The preset number of columns γ can be obtained by the Cumulative Variance Percentage (CVP) method. The projection matrix of the principal component of the covariance matrix is ​​the principal component projection matrix P pc , P pc ∈R m×γ .

[0064] Based on the above, we know that the number of columns of matrix U is m, so the number of columns of target matrix P is m, and the number of columns of eigenvectors of the remaining columns excluding the preset number of columns in the target matrix is ​​m-γ. Select the eigenvectors of the remaining columns m-γ excluding the preset number of columns γ from the target matrix P to obtain the projection matrix of the remaining components of the covariance matrix. The projection matrix of the remaining components of the covariance matrix is ​​the remaining component projection matrix P res , P res ∈R m×(m-γ) .

[0065] Based on the preset number of columns γ, the first γ eigenvalues ​​can be selected from the diagonal matrix ∧ to obtain the principal component projection matrix P pc The main diagonal matrix ∧ in the corresponding diagonal matrix pc , ∧ pc ∈R γ×γ Based on the remaining number of columns m-γ, the remaining m-γ eigenvalues ​​can be selected from the diagonal matrix ∧ to obtain the remaining component projection matrix P res The residual focus matrix ∧ in the corresponding diagonal matrix res ∧ res ∈R (m-γ)×(m-γ) .

[0066] In summary, the principal component analysis model in step S12 outputs the principal component projection matrix P based on the input data matrix pc And the remaining component projection matrix P res , and also outputs the main diagonal matrix ∧ pc and the remaining focus matrix ∧ res .

[0067] In step S12, based on the principal component projection matrix P pc , the main diagonal matrix ∧ pc , the remaining component projection matrix P res , the remaining focus matrix ∧ res The covariance matrix C can be Z Convert to formula (5):

[0068]

[0069] In addition, in step S12, the data matrix Z can be projected into two orthogonal subspaces using the PCA method to obtain the principal subspace and the residual component subspace. In other words, the principal component projection matrix P output by the principal component analysis model can be used. pc And the remaining component projection matrix P res Obtain the principal component subspace and the residual component subspace. That is, the principal component subspace S pc Indicated as S pc =span{P pc}, the remaining component subspace S res Indicated as S res =span{P res Considering the orthogonality of the two subspaces, in the principal component subspace S pc and the remaining component subspace S res The projection can be expressed as formula (6) and formula (7) respectively:

[0070]

[0071]

[0072] Where, P T pc P pc The transpose of P T res P res The transpose of . The principal component subspace S pc The projection of (i.e. the principal component matrix of the data matrix Z), is the remaining component subspace S res The projection of (i.e., the residual component matrix of the data matrix Z).

[0073] Based on equations (6) and (7), the expression of the data matrix Z shown in equation (8) can be obtained:

[0074]

[0075] The data matrix Z is further decomposed into formula (9):

[0076]

[0077] Where, T pc =ZP pc , T res =ZP res , is the score matrix, due to the singular value decomposition (SVD), so The columns of are orthogonal to each other. The score matrix is the projected data matrix Z. Where T pc Contains the main measurement information, T res Contains the remaining information of the data matrix Z. Therefore, through the principal component projection matrix P pc And the remaining component projection matrix P res The space of the data matrix Z can be converted into a dimensionality-reduced space by selecting the principal component projection matrix P pc The preset number of columns γ (also called degrees of freedom) can reduce the dimension of the data that needs to be processed.

[0078] Step S13: Acquire the real-time operating data of the target battery, and calculate the Hotelling statistic, squared prediction error, and corresponding control limit of the target battery based on the real-time operating data, the principal component projection matrix, the residual component projection matrix, and the significance level.

[0079] In step S13, the real-time operating data of the target battery can be represented by the symbol z A Indicates. A ∈R m m represents the number of battery cells. Based on real-time operating data z AThe output of the principal component analysis model in step S12 is Hotelling's T 2 (Hotelling's T statistic) and SPE (squared prediction error) are used as monitoring statistics to perform fault analysis on the target battery. 2 represents systematic variation in battery cell data, and SPE represents random or unknown variation in measurements.

[0080] In step S13 , the control limits include Hotelling statistic control limits and squared prediction error control limits.

[0081] In step S13, the Hotelling statistic, square prediction error and corresponding control limits of the target battery are calculated based on the real-time operation data, the principal component projection matrix, the residual component projection matrix and the significance level, including: calculating the Hotelling statistic of the target battery based on the real-time operation data and the principal component projection matrix; calculating the square prediction error of the target battery based on the real-time operation data and the residual component projection matrix; and calculating the Hotelling statistic control limit and the square prediction error control limit based on the significance level.

[0082] In step S13, when calculating Hotelling's T 2 Before using Hotelling statistic and SPE (squared prediction error), we first run the real-time data z A Normalize the data and make its mean equal to 0.

[0083] In step S13, the Hotelling statistic of the target battery is calculated by formula (10):

[0084]

[0085] Where, T 2 is the Hotelling statistic of the target battery, z A T z A The device, P T pc P pc The transpose of is the principal diagonal matrix ∧ pc The inverse matrix of .

[0086] In step S13, the square prediction error of the target battery is calculated by formula (11):

[0087]

[0088] Where SPE is the squared prediction error of the target battery, P T res P resThe transpose of , I represents the identity matrix.

[0089] In formula (10) and formula (11), Hotelling's T 2 (Hotelling statistic) can be obtained by running the data z in real time in the PCA model space A The Mahalanobis distance is given by , and the SPE (squared prediction error) is given by the quadratic orthogonal distance to the PCA model space.

[0090] In step S13, the Hotelling's T control limit is calculated based on the significance level. 2 control limits) and squared prediction error control limits (SPE control limits).

[0091] Specifically, for a significance level α, Hotelling's T 2 The control limit of is expressed as formula (12):

[0092]

[0093] Where, is the control limit of Hotelling statistic, N is the number of battery cells (that is, the number of samples of each variable), γ is the preset number of columns (that is, the degrees of freedom), F α (γ,N-γ) is the critical value of the F distribution. This critical value satisfies formula (13):

[0094]

[0095] Where, and are χ with degrees of freedom γ and N-γ respectively 2 The significance level α is usually set between 90% and 95%.

[0096] The control limits of SPE are expressed as follows:

[0097]

[0098] Where, is the control limit of SPE, where θ1, θ2, and h0 are calculated using formula (15)

[0099]

[0100] Where, Represents λ j k-th power, k=1,2,3, when k=1, θ1 can be calculated, when k=2, θ2 can be calculated, when k=3, θ3 can be calculated, λ jrepresents the j-th eigenvalue of the diagonal matrix ∧, where j is [γ+1,m].

[0101] c under normal distribution corresponding to significance level α α Calculated by formula (16):

[0102]

[0103] Where erfc represents the complementary error function.

[0104] Step S14 , determining whether a target battery fails based on the Hotelling statistic, the squared prediction error, and the corresponding control limit.

[0105] In step S14, whether the target battery is faulty is determined based on the Hotelling statistic, the squared prediction error, and the corresponding control limit, including: if the Hotelling statistic is less than or equal to the Hotelling statistic control limit, and the squared prediction error is less than or equal to the squared prediction error control limit, then the target battery is not faulty; otherwise, the target battery is faulty.

[0106] That is, determine whether formula (17) satisfies:

[0107] and

[0108] If equation (17) is satisfied, the target battery is fault-free. If at least one of the Hotelling statistic and the squared prediction error is not satisfied (i.e., at least one exceeds the corresponding control limit), the target battery is faulty.

[0109] Step S15: If the target battery fails, the battery failure location is determined based on the Hotelling statistic and the squared prediction error.

[0110] In step S15, the battery fault location is determined based on the Hotelling statistic and the squared prediction error, including: decomposing the Hotelling statistic to obtain the Hotelling statistic of each battery cell in the target battery; decomposing the squared prediction error to obtain the squared prediction error of each battery cell in the target battery; and determining the battery fault location based on the battery cell Hotelling statistic and the battery cell squared prediction error.

[0111] In step S15, the fault location can be located by analyzing the contribution of each variable (i.e., each battery cell) to the SPE. The variable with the largest contribution is considered to be the variable most affected by the fault. The SPE statistic is decomposed into the sum of the contributions of each variable to it using equation (18):

[0112]

[0113] Among them, SPE i Indicates the target battery z AThe squared prediction error of the ith battery cell. i ranges from 1 to m.

[0114] In step S15, the fault location can be located by analyzing the contribution of each variable to the Hotelling statistic. The formula for decomposing and obtaining the Hotelling statistic of each battery cell in the target battery can be obtained by analogy with formula (18).

[0115] In step S15, the battery fault location is determined based on the Hotelling statistics of the battery cells and the squared prediction error of the battery cells. Specifically, the maximum value of the squared prediction error of each battery cell is screened out, SPE i The battery cell corresponding to the maximum value fails, and the battery cell corresponding to the maximum value of the Hotelling statistic of the battery cell also fails.

[0116] Figure 2 FIG. 1 is a flow chart showing another method for determining battery faults based on principal component analysis according to an embodiment of the present disclosure. Figure 2 As shown, the battery fault diagnosis method based on principal component analysis includes: calculating a PCA model using the operating data of a fault-free battery; calculating the statistics and control limits of the target battery; and judging whether the statistics meet the conditions. If so, the target battery is fault-free; if not, the target battery is faulty and fault location is performed. The operating data of the fault-free battery is the historical operating data of the fault-free battery in step S11 above. The calculation of the PCA model using the operating data of the fault-free battery can be specifically referred to the relevant description of steps S11 and S12 above. The statistics of the target battery include the Hotelling statistic and the squared prediction error. The calculation of the statistics and control limits of the target battery can be specifically referred to the relevant description of step S13 above. Judging whether the statistics meet the conditions can be specifically referred to the relevant description of step S14 above. If not, the target battery is faulty and fault location is performed, which can be specifically referred to the relevant description of step S15 above.

[0117] In some embodiments, as Figure 2 As shown, the battery fault judgment method based on principal component analysis also includes adaptively updating model data, wherein the model data is the output data of the PCA model.

[0118] Specifically, considering that battery systems are not always in a steady state, the PCA model will change due to changes in battery parameters (e.g., changes caused by temperature, state of charge (SoC), or aging). Therefore, the battery fault diagnosis method based on principal component analysis also includes: using a recursive method to adaptively update the principal component analysis model data. The recursive adaptive update process includes: using the target battery's fault-free operating data to update the historical operating data of the fault-free battery, thereby updating the data matrix; inputting the updated data matrix into the principal component analysis model to output a new principal component projection matrix and a residual component projection matrix. In this case, because the adaptive method is highly robust to parameter changes, the adaptive method uses new fault-free samples to update the PCA model input data to obtain new principal component projection matrices and residual component projection matrices, thereby improving the accuracy of fault diagnosis. A forgetting factor can be used during each update to limit the influence of the old model on the update.

[0119] In a battery fault diagnosis method based on principal component analysis in an embodiment of the present disclosure, historical operating data of a fault-free battery is obtained, and a data matrix is ​​constructed based on the historical operating data of the fault-free battery; a principal component analysis model is constructed, and the principal component analysis model obtains a covariance matrix based on the input data matrix, and then performs singular value decomposition on the covariance matrix to output a principal component projection matrix and a residual component projection matrix; real-time operating data of a target battery is obtained, and the Hotelling statistic, squared prediction error, and corresponding control limits of the target battery are calculated based on the real-time operating data, the principal component projection matrix, the residual component projection matrix, and the significance level; whether the target battery has a fault is determined based on the Hotelling statistic, the squared prediction error, and the corresponding control limits; if the target battery has a fault, the battery fault location is determined based on the Hotelling statistic and the squared prediction error. In this case, using the principal component analysis model, the principal component projection matrix and the residual component projection matrix are obtained based on the historical operating data of the fault-free battery, and then the Hotelling statistic, squared prediction error, and corresponding control limits of the target battery are calculated based on the real-time operating data, the principal component projection matrix, the residual component projection matrix, and the significance level, thereby determining whether the target battery has a fault and the location of the fault. As a result, the battery fault and the location of the fault can be determined more accurately, and the sensitivity of small fault diagnosis is improved. The method disclosed in the present invention detects and locates faults through statistical evaluation based on principal component analysis. First, the operating data of the fault-free battery system is subjected to principal component analysis using the PCA model. Then, the operating data of the real-time running battery is obtained to calculate two statistics (i.e., Hotelling's T and squared prediction error) for monitoring, and the individual battery cell signal is analyzed to determine the Hotelling's T 2The method disclosed herein also utilizes a recursive method to adaptively update the principal component analysis model. In summary, the method disclosed herein is capable of diagnosing battery faults and their locations. The method disclosed herein improves the detectability of minor faults, has high sensitivity, and the adaptive method is highly robust to parameter changes.

[0120] The following are system embodiments of the present disclosure, which can be used to implement the method embodiments of the present disclosure. For details not disclosed in the system embodiments of the present disclosure, please refer to the method embodiments of the present disclosure.

[0121] Figure 3 A block diagram of a battery fault diagnosis system based on principal component analysis provided by an embodiment of the present disclosure is shown. Figure 3 The battery fault diagnosis system 10 based on principal component analysis includes an acquisition module 11, a modeling module 12, a calculation module 13, a first judgment module 14 and a second judgment module 15, wherein:

[0122] An acquisition module 11 is used to acquire historical operating data of a fault-free battery and real-time operating data of a target battery;

[0123] A modeling module 12 is used to construct a data matrix based on the historical operating data of the fault-free battery, and to construct a principal component analysis model. The principal component analysis model obtains a covariance matrix based on the input data matrix, and then performs singular value decomposition on the covariance matrix to output a principal component projection matrix and a residual component projection matrix;

[0124] A calculation module 13 is used to calculate the Hotelling statistic, squared prediction error and corresponding control limit of the target battery based on real-time operation data, the principal component projection matrix, the residual component projection matrix and the significance level;

[0125] A first judgment module 14 is configured to judge whether a target battery has a fault based on the Hotelling statistic, the squared prediction error, and the corresponding control limit;

[0126] The second judgment module 15 is configured to determine the battery fault location based on the Hotelling statistic and the squared prediction error if a target battery fault occurs.

[0127] Optionally, the second judgment module 15 is specifically used to: decompose the Hotelling statistic to obtain the Hotelling statistic of each battery cell in the target battery; decompose the square prediction error to obtain the square prediction error of each battery cell in the target battery; and determine the battery fault location based on the battery cell Hotelling statistic and the battery cell square prediction error.

[0128] Optionally, the control limits include Hotelling statistic control limits and squared prediction error control limits.

[0129] Optionally, the first judgment module 14 is specifically configured to: if the Hotelling statistic is less than or equal to the Hotelling statistic control limit, and the square prediction error is less than or equal to the square prediction error control limit, then the target battery has no fault; otherwise, the target battery has a fault.

[0130] Optionally, the calculation module 13 is specifically used to: calculate the Hotelling statistic of the target battery based on the real-time operation data and the principal component projection matrix; calculate the square prediction error of the target battery based on the real-time operation data and the residual component projection matrix; and calculate the Hotelling statistic control limit and the square prediction error control limit based on the significance level.

[0131] Optionally, the modeling module 12 is specifically used to: perform singular value decomposition on the covariance matrix to output a principal component projection matrix and a residual component projection matrix, including: performing singular value decomposition on the covariance matrix to obtain a target matrix; obtaining a principal component projection matrix based on the eigenvectors of a preset number of columns of the target matrix; obtaining a residual component projection matrix based on the eigenvectors of the remaining number of columns in the target matrix excluding the preset number of columns.

[0132] Optionally, the battery fault judgment system 10 based on principal component analysis method also includes an update module, which is used to update the historical operating data of the fault-free battery using the operating data of the target battery when it is fault-free; the modeling module updates the data matrix based on the updated historical operating data of the fault-free battery, and inputs the updated data matrix into the principal component analysis model to output a new principal component projection matrix and a residual component projection matrix.

[0133] It should be noted that the above explanation of the embodiment of the battery fault diagnosis method based on the principal component analysis method is also applicable to the battery fault diagnosis system based on the principal component analysis method of this embodiment, and will not be repeated here.

[0134] In the battery fault judgment system based on the principal component analysis method of the embodiment of the present disclosure, the acquisition module is used to obtain the historical operating data of the fault-free battery and the real-time operating data of the target battery; the modeling module is used to construct a data matrix based on the historical operating data of the fault-free battery, and to construct a principal component analysis model. The principal component analysis model obtains a covariance matrix based on the input data matrix, and then performs singular value decomposition on the covariance matrix to output a principal component projection matrix and a residual component projection matrix; the calculation module is used to calculate the Hotelling statistic, square prediction error and corresponding control limit of the target battery based on the real-time operating data, the principal component projection matrix, the residual component projection matrix and the significance level; the first judgment module is used to judge whether the target battery has a fault based on the Hotelling statistic, the square prediction error and the corresponding control limit; the second judgment module is used to determine the battery fault location based on the Hotelling statistic and the square prediction error if the target battery has a fault. In this case, the principal component analysis model is used to obtain the principal component projection matrix and the residual component projection matrix based on the historical operating data of the fault-free battery. Then, based on the real-time operating data, the principal component projection matrix, the residual component projection matrix and the significance level, the Hotelling statistic, the square prediction error and the corresponding control limit of the target battery are calculated to determine whether the target battery has a fault and the location of the fault. In this way, the battery fault and the location of the fault can be judged more accurately, and the sensitivity of small fault diagnosis is improved. The system disclosed in the present invention detects and locates faults through statistical evaluation based on principal component analysis. First, the operating data of the fault-free battery system is subjected to principal component analysis by the PCA model. Then, the operating data of the real-time operating battery is obtained to calculate two statistics (i.e., the Hotelling statistic and the square prediction error) for monitoring, and the single battery cell signal is analyzed to determine whether the fault has occurred and the location of the fault has occurred. 2 The system disclosed herein also utilizes a recursive method to adaptively update the principal component analysis model. In summary, the system disclosed herein is capable of diagnosing battery faults and their locations. The system disclosed herein improves the detectability of minor faults, exhibits high sensitivity, and its adaptive method is highly robust to parameter changes.

[0135] According to an embodiment of the present disclosure, the present disclosure also provides a battery fault diagnosis device based on principal component analysis, a readable storage medium, and a computer program product.

[0136] Figure 4It is a block diagram of a battery fault judgment device based on principal component analysis method for implementing the battery fault judgment method based on principal component analysis method of an embodiment of the present disclosure. The battery fault judgment device based on principal component analysis method is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The battery fault judgment device based on principal component analysis method can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable electronic devices and other similar computing devices. The components shown in this disclosure, the connections and relationships of the components, and the functions of the components are merely examples and are not intended to limit the implementation of the present disclosure described and / or required in this disclosure.

[0137] like Figure 4 As shown, the battery failure diagnosis device 20 based on principal component analysis includes a computing unit 21, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 22 or a computer program loaded from a storage unit 28 into a random access memory (RAM) 23. RAM 23 may also store various programs and data required for the operation of the battery failure diagnosis device 20 based on principal component analysis. The computing unit 21, ROM 22, and RAM 23 are connected to each other via a bus 24. An input / output (I / O) interface 25 is also connected to the bus 24.

[0138] Multiple components of the battery failure diagnosis device 20 based on principal component analysis are connected to an I / O interface 25, including an input unit 26, such as a keyboard and mouse; an output unit 27, such as various types of displays and speakers; a storage unit 28, such as a magnetic disk and optical disk, which is communicatively connected to the computing unit 21; and a communication unit 29, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 29 allows the battery failure diagnosis device 20 based on principal component analysis to exchange information / data with other electronic devices via a computer network such as the Internet and / or various telecommunication networks.

[0139] The computing unit 21 can be any general-purpose and / or specialized processing component with processing and computing capabilities. Some examples of the computing unit 21 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 21 performs the various methods and processes described above, such as executing the battery fault diagnosis method based on principal component analysis. For example, in some embodiments, the battery fault diagnosis method based on principal component analysis can be implemented as a computer software program tangibly embodied in a machine-readable medium, such as the storage unit 28. In some embodiments, part or all of the computer program can be loaded and / or installed into the battery fault diagnosis device 20 based on principal component analysis via the ROM 22 and / or the communication unit 29. When the computer program is loaded into the RAM 23 and executed by the computing unit 21, one or more steps of the battery fault diagnosis method based on principal component analysis described above can be performed. Alternatively, in other embodiments, the computing unit 21 may be configured to execute the battery fault determination method based on principal component analysis in any other appropriate manner (for example, by means of firmware).

[0140] Various embodiments of the systems and techniques described above in the present disclosure can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0141] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0142] In the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by an instruction execution system, an apparatus, or a battery fault diagnosis device based on principal component analysis, or for use in conjunction with an instruction execution system, an apparatus, or a battery fault diagnosis device based on principal component analysis. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or electronic device, or any suitable combination of the foregoing. More specific examples of machine-readable storage media may include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage electronic device, a magnetic storage electronic device, or any suitable combination of the foregoing.

[0143] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0144] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), the Internet, and a blockchain network.

[0145] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact via a communication network. This client-server relationship is established by computer programs running on the respective computers, establishing a client-server relationship. The server may be a cloud server, also known as a cloud computing server or cloud host, a host product within the cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosts and VPS services ("Virtual Private Servers" or simply "VPS"). The server may also be a server in a distributed system or a server integrated with blockchain.

[0146] It should be understood that the various forms of processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved. This disclosure is not limited here.

[0147] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A battery fault diagnosis method based on principal component analysis, characterized in that: include: Acquire historical operating data of a fault-free battery, and construct a data matrix based on the historical operating data of the fault-free battery; Constructing a principal component analysis model, wherein the principal component analysis model obtains a covariance matrix based on an input data matrix, and then performs singular value decomposition on the covariance matrix to output a principal component projection matrix and a residual component projection matrix; Acquire real-time operating data of a target battery, and calculate a Hotelling statistic, a squared prediction error, and corresponding control limits of the target battery based on the real-time operating data, the principal component projection matrix, the residual component projection matrix, and a significance level; Determining whether the target battery is faulty based on the Hotelling statistic, the squared prediction error, and the corresponding control limit; If a fault occurs in the target battery, decomposing the Hotelling statistic to obtain a Hotelling statistic of each battery cell in the target battery, decomposing the square prediction error to obtain a square prediction error of each battery cell in the target battery, and determining a battery fault location based on the battery cell Hotelling statistic and the battery cell square prediction error; The control limits include a Hotelling statistic control limit and a squared prediction error control limit, and determining whether the target battery has a fault based on the Hotelling statistic, the squared prediction error, and the corresponding control limits includes: If the Hotelling statistic is less than or equal to the Hotelling statistic control limit, and the squared prediction error is less than or equal to the squared prediction error control limit, then the target battery has no fault; otherwise, the target battery has a fault.

2. The battery fault diagnosis method based on principal component analysis according to claim 1, characterized in that: The Hotelling statistic, the squared prediction error, and the corresponding control limit of the target battery are calculated based on the real-time operating data, the principal component projection matrix, the residual component projection matrix, and the significance level, including: Calculate the Hotelling statistics of the target battery based on the real-time operating data and the principal component projection matrix; Calculating a square prediction error of the target battery based on the real-time operating data and the residual component projection matrix; The control limits of the Hotelling statistic and the squared prediction error are calculated based on the significance level.

3. The battery fault diagnosis method based on principal component analysis according to claim 1, characterized in that: The step of performing singular value decomposition on the covariance matrix to output a principal component projection matrix and a residual component projection matrix includes: Performing singular value decomposition on the covariance matrix to obtain a target matrix; Obtaining the principal component projection matrix based on the eigenvectors of a preset number of columns of the target matrix; The remaining component projection matrix is ​​obtained based on the eigenvectors of the remaining number of columns in the target matrix except the preset number of columns.

4. The battery fault diagnosis method based on principal component analysis according to claim 1, wherein: Also includes: Using the operating data of the target battery when it is fault-free to update the historical operating data of the fault-free battery, and then updating the data matrix; The updated data matrix is ​​input into the principal component analysis model to output a new principal component projection matrix and a residual component projection matrix.

5. A battery fault diagnosis device based on principal component analysis, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the battery fault judgment method based on principal component analysis according to any one of claims 1 to 4.

6. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to enable the computer to execute the battery fault judgment method based on principal component analysis according to any one of claims 1 to 4.

Images