ALO-OCSVM-based battery high-dimensional feature abnormal aging identification method and device

Abstract

A battery high-dimensional feature abnormal aging identification method and device based on ALO-OCSVM, and the method comprises the steps: A, monitoring and collecting the operation data of a battery, generating effective high-dimensional features through feature engineering, and constructing a training data set with high-dimensional feature quantities; step B, constructing an ALO-OCSVM model, solving an OCSVM algorithm on the training data set with the high-dimensional characteristic quantity constructed in the step A, and utilizing ALO to optimize parameters to obtain an abnormal aging identification model meeting error requirements; and step C, continuously monitoring and collecting battery operation data, generating a test data set with high-dimensional characteristic quantity, and carrying out online monitoring and identification on the abnormal aging battery by using the abnormal aging identification model obtained in the step B. According to the method, the classification of the battery states is realized by utilizing the OCSVM algorithm, and the global optimization of the parameters is realized by utilizing the ALO algorithm, so that the method adapts to the difference of the aging degrees of different batteries, and has relatively high calculation efficiency and reliability.

Description

Technical Field

[0001] The present invention belongs to the technical field of battery abnormality identification, and in particular relates to a battery high-dimensional feature abnormal aging identification method based on ALO-OCSVM. Background Art

[0002] Lithium batteries age slowly under normal conditions, but their performance decays faster under certain conditions (such as battery abuse). Once the battery fails abnormally and rapidly, the risk of thermal runaway and safety accidents under the same load conditions will increase significantly. There are many reasons for battery failure, including SEI thickening, loss of electrode active materials or changes in properties, and the most important manifestation is the abnormal decrease in the remaining available capacity of the battery.

[0003] The most commonly used method for identifying abnormal battery aging is big data analysis and artificial intelligence methods. Based on the capacity attenuation law of batteries under different aging paths, data features that can reflect the health status of the battery are extracted from real-time operating data, and abnormal batteries are identified by analyzing the distribution of related features.

[0004] The current research on abnormal battery aging identification based on artificial intelligence algorithms has the following shortcomings: most neural networks are supervised models, which require actual lithium battery aging data as output functions and require a large number of samples, including sensor data at different stages. Training samples are generally obtained based on experiments, which leads to the following limitations: the number of samples obtained in the experiment cannot reach a large level, and the experimental cost is high; the amount of data is insufficient to obtain the most accurate model; the battery aging status model can only be verified through experiments and cannot be verified in actual factory applications.

[0005] The prior art discloses a method and device for building an energy storage battery model and determining the state of an energy storage battery (CN116736173B), which first builds a multi-parameter health data set for each energy storage battery, then uses each multi-parameter health data set to train a preset SVDD model, and further obtains the target hypersphere center, the first target hypersphere radius, and the second target hypersphere radius of the target hypersphere by solving each preset SVDD model, and finally uses the target hypersphere center, the first target hypersphere radius, and the second target hypersphere radius to establish an energy storage battery model containing the target hypersphere. However, the shortcomings of the prior art are: (1) it relies on the collection of historical fault or abnormal data sets, which is costly and has limitations in practical applications; (2) the data set uses the original basic signal, and the feature richness needs to be enhanced; (3) the speed of model parameter search and solution is slow, and the calculation efficiency needs to be improved.

[0006] In summary, the limitations of existing methods can be summarized as follows:

[0007] First, most existing methods rely on specified operating condition data sets with limited samples or laboratory data sets with controlled test conditions. The learned battery aging laws are relatively limited and difficult to generalize and apply to engineering sites with diverse battery types and complex and changeable on-site operating conditions.

[0008] Second, traditional neural network methods require the battery's actual remaining life or state of health (SOH) as label data for learning, but this data is difficult and costly to obtain, which greatly restricts engineering applications. Summary of the invention

[0009] In order to solve the deficiencies in the prior art, the present invention provides a battery multi-feature abnormal aging identification method based on ALO-OCSVM, forming an unsupervised battery abnormal aging monitoring technology that adapts to complex data distribution, and identifies abnormal individuals under the set boundaries according to the high-dimensional features of normal battery data, effectively coping with the challenges of diverse battery aging paths, complex coupling with the internal and external environment of the battery, and incomplete failure data samples.

[0010] The present invention adopts the following technical solution.

[0011] The first aspect of the present invention provides a method for identifying abnormal aging of a battery based on high-dimensional features of ALO-OCSVM, comprising:

[0012] Step A: monitor and collect battery operation data, generate effective high-dimensional features through feature engineering, and construct a training data set with high-dimensional feature quantities;

[0013] Step B, constructing an ALO-OCSVM model, solving the OCSVM algorithm on the training data set with high-dimensional feature quantities constructed in step A, and using ALO to optimize parameters to obtain an abnormal aging recognition model that meets the error requirements;

[0014] Step C: Continue to monitor and collect battery operation data and generate a test data set with high-dimensional feature quantities, and use the abnormal aging recognition model obtained in step B to carry out online monitoring and recognition of abnormally aged batteries.

[0015] Preferably, in step A, constructing a training data set with high-dimensional feature quantities includes:

[0016] Step A1, monitoring the battery operation within a preset time period, and collecting the battery original signal at equal intervals;

[0017] Step A2, performing feature transformation on the original battery signal in step A1, including scalar and vector forms;

[0018] Step A3: Combine the original battery signals collected in each group of collection time periods in step A1 and the battery signals after feature transformation in step A2 to convert them into a training data set with high-dimensional feature quantities.

[0019] Preferably, in step A1, the original battery signal includes: voltage, current, temperature, and time.

[0020] Preferably, establishing the abnormal aging recognition model in step B includes:

[0021] Step B1, in order to map the training data set data with high-dimensional features from the original space to the feature space, establish a nonlinear function and a kernel function;

[0022] Step B2, solving the OCSVM single-class support vector machine algorithm on the training data set with high-dimensional feature quantities;

[0023] Step B3: Continue to monitor and collect battery operation data and generate a verification data set with high-dimensional feature quantities, and use the ALO ant colony algorithm to search for the optimal parameters of the OCSVM model on the verification data set with high-dimensional feature quantities to obtain an abnormal aging recognition model.

[0024] Preferably, in step B1, the nonlinear function and the kernel function are as follows:

[0025]

[0026] K(x,z)=(x T z) 2 =Φ(x) T Φ(z)

[0027] Where x and z are the eigenvectors of any two data.

[0028] Preferably, in step B2, solving the OCSVM single-class support vector machine algorithm includes:

[0029] Step B2.1, construct a hypersphere in the feature space, and construct an optimization problem to minimize the volume of the hypersphere;

[0030] Step B2.2, according to the Lagrange multiplier method, list the actual objective function to be solved;

[0031] Step B2.3: Solve the maximization problem of the objective function in step B2.2 using the conventional SVM algorithm, obtain the Lagrangian coefficients of all samples, synthesize the support vector set, and calculate the center and radius of the hypersphere.

[0032] Preferably, in step B2.1, the hypersphere constructed and the optimization problem are:

[0033]

[0034] Where o and r are the center and radius of the hypersphere, C is the penalty parameter, ξ is the relaxation factor, and i For L train / The relaxation factor of time period i in the n-time period group, X i is the training data set with high-dimensional features in time period i, L train is the preset time period, and n is the time length.

[0035] Preferably, in step B2.2, the actual objective function to be solved is:

[0036]

[0037] st0≤λ i ≤C∑λ i =1 i=1,2,…,n

[0038] In the formula, λ i For data X i The corresponding Lagrange coefficient.

[0039] Preferably, in step B2.3, the center and radius of the hypersphere are calculated as follows:

[0040]

[0041] Among them, X s For any sample in the support vector set, its Lagrangian coefficient λ s Satisfies 0<λ s <C。

[0042] Preferably, step B3 includes:

[0043] Step B3.1, set the number of iterations k for the algorithm to execute;

[0044] Step B3.2, randomly generate an initial population, and the candidate solutions include different combinations of penalty coefficients and relaxation factors;

[0045] Step B3.3, iteratively train the OCSVM model, repeat steps B2.1 to B2.3, and calculate its mean square error on the validation data set test set with high-dimensional features;

[0046] Step B3.4, by alternating between nearest neighbor and random search, take the corresponding solution with the minimum mean square error to update the candidate solution position;

[0047] Step B3.5: Replace the parameters in the OCSVM model with the corresponding solution of the minimum mean square error to obtain the abnormal aging identification model.

[0048] Preferably, step C comprises:

[0049] Step C1: During the preset time period L test Internally monitor the battery operation and collect voltage, current and temperature signals for a preset time length of n;

[0050] Step C2, performing feature transformation on the original signal in step C1 to obtain high-dimensional features;

[0051] Step C3: The original signals and high-dimensional features of the group time period are combined into a test data set A with high-dimensional features, which is used as the input of the abnormal aging recognition model;

[0052] Step C4: for each sample A in the test data set with high-dimensional features u , calculate its distance to the center of the hypersphere and determine whether it is an abnormal sample.

[0053] Preferably, in step C4, the distance calculation formula from each sample in the test data set with high-dimensional feature quantity to the center of the hypersphere is as follows:

[0054]

[0055] Where, d u is a sample A in the test dataset with high-dimensional features u The distance to the center of the hypersphere, K is the kernel function, L test is the preset time period, n is the time length, X i and X j For L test / n groups of training data sets with high-dimensional features in time periods i and j, λ i and λ j For data X i and X j The corresponding Lagrange coefficient;

[0056] If d u ≤r, indicating that the sample A in the test data set has high-dimensional features u Samples on or inside the hypersphere are identified as normal samples; otherwise, they are identified as abnormal samples.

[0057] A second aspect of the present invention provides a battery high-dimensional feature abnormal aging identification device using the above-mentioned battery high-dimensional feature abnormal aging identification method based on ALO-OCSVM, comprising:

[0058] A training data set construction module with high-dimensional feature quantities, an abnormal aging recognition model construction module, and an abnormal aging battery online monitoring and recognition module;

[0059] Among them, the training data set construction module with high-dimensional feature quantities is used to monitor and collect battery operation data, generate effective high-dimensional features through feature engineering, and construct a training data set with high-dimensional feature quantities;

[0060] The abnormal aging recognition model construction module is used to construct the ALO-OCSVM model, solve the OCSVM algorithm on the training data set with high-dimensional features constructed by the training data set construction module with high-dimensional features, and use ALO to optimize parameters to obtain the abnormal aging recognition model that meets the error requirements;

[0061] The abnormal aging battery online monitoring and identification module is used to continue to monitor and collect battery operation data and generate a test data set with high-dimensional feature quantities, and use the abnormal aging recognition model obtained in the abnormal aging recognition model construction module to carry out online monitoring and identification of abnormal aging batteries.

[0062] The beneficial effects of the present invention are that, compared with the prior art, it has the following advantages:

[0063] (A) The present invention realizes the identification of abnormally aged batteries by identifying data outside the normal category, overcoming the dependence of traditional neural network methods on abnormal data samples, and has strong adaptability and scalability in the actual application of battery energy storage health monitoring.

[0064] (B) The present invention only needs to monitor basic signals such as voltage, current, and temperature, and does not need to obtain aging data labels such as the remaining battery life and health status, thereby reducing the cost and complexity of algorithm implementation.

[0065] (C) The present invention utilizes the single-class support vector machine algorithm OCSVM to realize the classification of battery status, and utilizes the ant lion algorithm ALO to realize the global optimization of parameters to adapt to the differences in the aging degree of different batteries, and has high computational efficiency and reliability.

[0066] (D) The present invention designs high-dimensional features, maps the nonlinear problems of the original space to the high-dimensional space, and performs linear division, thereby effectively solving complex nonlinear problems and helping to improve the accuracy of anomaly recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 It is a flow chart of the method for identifying abnormal aging of a battery based on ALO-OCSVM in the present invention;

[0068] Figure 2 This is a schematic diagram of constructing the characteristic space hypersphere of the present invention. DETAILED DESCRIPTION

[0069] In order to make the purpose, technical scheme and advantages of the present invention clearer, the technical scheme of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. The embodiments described in this application are only embodiments of a part of the present invention, rather than all embodiments. Based on the spirit of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the protection scope of the present invention.

[0070] The purpose of the present invention is to provide a battery high-dimensional feature abnormal aging identification method based on ALO-OCSVM, to form an unsupervised battery abnormal aging monitoring technology that adapts to complex data distribution, to identify abnormal individuals under the set boundaries according to the high-dimensional features of normal battery data, and to effectively cope with the challenges of diverse battery aging paths, complex coupling with the internal and external environment of the battery, and incomplete failure data samples.

[0071] The present invention is implemented by adopting the following technical solutions:

[0072] like Figure 1 As shown, an embodiment of the present invention provides a method for identifying abnormal aging of a battery based on high-dimensional features of ALO-OCSVM, comprising the following steps:

[0073] Step A: monitor and collect battery operation data, generate effective high-dimensional features through feature engineering, and construct a training data set X with high-dimensional feature quantities as input of the OCSVM (One-Class Support Vector Machine) algorithm;

[0074] Step A specifically includes the following steps:

[0075] Step A1: In the preset time period L train The battery operation is monitored internally, and the original battery signal S is collected at equal intervals, including time t, voltage U, current I, and temperature T, with a length of n.

[0076] Step A2: Perform feature transformation on the original battery signal in step A1, and obtain features in two forms: scalar and vector, as shown in the following table:

[0077] Table 1 Description of battery scalar and vector features

[0078]

[0079] Step A3: The original battery signal collected in step A1 of the group time period and the battery signal after feature transformation in step A2 are combined into a training data set X with high-dimensional feature quantities, with a dimension of Where d is the number of signals, that is, the number of feature quantities;

[0080] Step B, constructing an ALO-OCSVM model, solving the OCSVM algorithm on the training data set X with high-dimensional feature quantities constructed in step A, and optimizing parameters using ALO (Ant Lion Optimizer) to obtain an abnormal aging recognition model that meets the error requirements;

[0081] Step B specifically includes the following steps:

[0082] Step B1: To map the data of the training data set with high-dimensional feature quantities from the original space to the feature space, establish a non-linear function Φ and a kernel function K:

[0083]

[0084] K(x,z) = (x T z) 2 = Φ(x) T Φ(z)

[0085] In the formula, x and z are the feature vectors of any two data.

[0086] Step B2: Solve the one-class support vector machine algorithm of OCSVM on the training data set X with high-dimensional feature quantities, specifically including:

[0087] Step B2.1: Construct a hypersphere in the feature space. As Figure 2 shown, construct the following optimization problem to minimize the volume of this hypersphere:

[0088]

[0089] In the formula, o and r are the center and radius of the hypersphere respectively, C is the penalty parameter, ξ is the relaxation factor, used to balance the classification accuracy and the model complexity determined by the volume of the hypersphere, ξ i is the relaxation factor in the i-th time period of L train / n groups of time periods, and X i is the data of the training data set with high-dimensional feature quantities in the i-th time period.

[0090] Step B2.2: According to the Lagrange multiplier method, the actual objective function to be solved is:

[0091]

[0092] s.t. 0 ≤ λ i ≤ C ∑λ i = 1 i = 1,2,…,n

[0093] In the formula, λ i is the Lagrange coefficient corresponding to the data X i .

[0094] Step B2.3: Solve the maximization problem of the objective function in B2.2 according to the conventional SVM algorithm, obtain the Lagrange coefficients λ of all samples, and according to the support vector set X synthesized by the samples satisfying 0 < λ s < C, X s = {X * = {Xs}, calculate the center and radius of the hypersphere:

[0095]

[0096] Step B3: Continue to monitor and collect battery operation data and generate a verification data set with high-dimensional features. Use the ALO ant colony algorithm to search for the optimal parameters of the OCSVM model on the verification data set with high-dimensional features to obtain an abnormal aging recognition model, which specifically includes:

[0097] Step B3.1, set the number of iterations k for the algorithm to execute;

[0098] Step B3.2, randomly generate an initial population, and the candidate solutions include different combinations of penalty coefficient C and relaxation factor ξ;

[0099] Step B3.3, iteratively train the OCSVM model, i.e., repeat steps B2.1 to B2.3, and calculate its mean square error (MSE) on the validation data set test set with high-dimensional feature quantities;

[0100] Step B3.4, by alternating between nearest neighbor and random search, take the corresponding solution with the minimum mean square error (MSE) to update the candidate solution position;

[0101] Step B3.5: Replace the parameters in the OCSVM model with the corresponding solutions of the minimum mean square error (MSE) to obtain the abnormal aging identification model.

[0102] Step C: Continue to monitor and collect battery operation data and generate a test data set with high-dimensional feature quantities, and use the abnormal aging recognition model obtained in step B to carry out online monitoring and recognition of abnormally aged batteries.

[0103] Step C specifically includes the following steps:

[0104] Step C1: During the preset time period L test Internally monitor the battery operation and collect voltage, current and temperature signals for a preset time length of n;

[0105] Step C2, performing feature transformation on the original signal in step C1 to obtain high-dimensional features;

[0106] Step C3: The original signals and high-dimensional features of the group time period are combined into a test data set A with high-dimensional features, which is used as the input of the abnormal aging recognition model;

[0107] Step C4: for each sample A in the test data set with high-dimensional features u , calculate its distance to the center of the hypersphere and determine whether it is an abnormal sample:

[0108]

[0109] Where, d u is a sample A in the test dataset with high-dimensional features u The distance to the center of the hypersphere, K is the kernel function, L test is the preset time period, n is the time length, X i and X j For L test / n groups of training data sets with high-dimensional features in time periods i and j, λ i and λ j For data X i and X j The corresponding Lagrange coefficient.

[0110] If d u ≤r, indicating that the sample A in the test data set has high-dimensional features u Samples on or inside the hypersphere are identified as normal samples; otherwise, they are identified as abnormal samples.

[0111] The embodiment of the present invention further provides a battery high-dimensional feature abnormal aging identification device based on the above-mentioned battery high-dimensional feature abnormal aging identification method based on ALO-OCSVM, comprising:

[0112] A training data set construction module with high-dimensional feature quantities, an abnormal aging recognition model construction module, and an abnormal aging battery online monitoring and recognition module;

[0113] Among them, the training data set construction module with high-dimensional feature quantities is used to monitor and collect battery operation data, generate effective high-dimensional features through feature engineering, and construct a training data set with high-dimensional feature quantities;

[0114] The abnormal aging recognition model construction module is used to construct the ALO-OCSVM model, solve the OCSVM algorithm on the training data set with high-dimensional features constructed by the training data set construction module with high-dimensional features, and use ALO to optimize parameters to obtain the abnormal aging recognition model that meets the error requirements;

[0115] The abnormal aging battery online monitoring and identification module is used to continue to monitor and collect battery operation data and generate a test data set with high-dimensional feature quantities, and use the abnormal aging recognition model obtained in the abnormal aging recognition model construction module to carry out online monitoring and identification of abnormal aging batteries.

[0116] The present invention can identify abnormal aging behavior without requiring abnormal battery aging data, has simple calculation and strong applicability, and has strong generalizability in practical applications of battery energy storage health monitoring.

[0117] The present disclosure may be a system, a method and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for causing a processor to implement various aspects of the present disclosure.

[0118] A computer-readable storage medium may be a tangible device that can hold and store instructions used by an instruction execution device. A computer-readable storage medium may be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples of computer-readable storage media (a non-exhaustive list) include: 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), a static random access memory (SRAM), a portable compact disk read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punch card or a raised structure in a groove on which instructions are stored, and any suitable combination of the foregoing. As used herein, a computer-readable storage medium is not to be interpreted as a transient signal per se, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., a light pulse through a fiber optic cable), or an electrical signal transmitted through a wire.

[0119] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, optical fiber transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing / processing device.

[0120] The computer program instructions for performing the operation of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages, such as Smalltalk, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. Computer-readable program instructions may be executed completely on a user's computer, partially on a user's computer, as an independent software package, partially on a user's computer, partially on a remote computer, or completely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., using an Internet service provider to connect via the Internet). In some embodiments, an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), may be customized by utilizing the state information of the computer-readable program instructions, and the electronic circuit may execute the computer-readable program instructions, thereby realizing various aspects of the present disclosure.

[0121] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for identifying abnormal aging of batteries based on ALO-OCSVM high-dimensional features, characterized in that: include: Step A: monitor and collect battery operation data, generate effective high-dimensional features through feature engineering, and construct a training data set with high-dimensional feature quantities; Step B, constructing an ALO-OCSVM model, solving the OCSVM algorithm on the training data set with high-dimensional feature quantities constructed in step A, and using ALO to optimize parameters to obtain an abnormal aging recognition model that meets the error requirements; Step C: Continue to monitor and collect battery operation data and generate a test data set with high-dimensional feature quantities, and use the abnormal aging recognition model obtained in step B to carry out online monitoring and recognition of abnormally aged batteries.

2. The method for identifying abnormal aging of batteries based on ALO-OCSVM according to claim 1, characterized in that: In step A, constructing a training data set with high-dimensional features includes: Step A1, monitoring the battery operation within a preset time period, and collecting the battery original signal at equal intervals; Step A2, performing feature transformation on the original battery signal in step A1, including scalar and vector forms; Step A3: Combine the original battery signals collected in each group of collection time periods in step A1 and the battery signals after feature transformation in step A2 to convert them into a training data set with high-dimensional feature quantities.

3. The method for identifying abnormal aging of batteries based on ALO-OCSVM according to claim 2, characterized in that: In step A1, the original battery signal includes: voltage, current, temperature, and time.

4. The method for identifying abnormal aging of batteries based on ALO-OCSVM according to claim 1, characterized in that: The establishment of the abnormal aging recognition model in step B includes: Step B1, in order to map the training data set data with high-dimensional features from the original space to the feature space, establish a nonlinear function and a kernel function; Step B2, solving the OCSVM single-class support vector machine algorithm on the training data set with high-dimensional feature quantities; Step B3: Continue to monitor and collect battery operation data and generate a verification data set with high-dimensional feature quantities, and use the ALO ant colony algorithm to search for the optimal parameters of the OCSVM model on the verification data set with high-dimensional feature quantities to obtain an abnormal aging recognition model.

5. The method for identifying abnormal aging of batteries based on ALO-OCSVM according to claim 4 is characterized in that: In step B1, the nonlinear function and kernel function are as follows: K(x,z)=(x T z) 2 =Φ(x) T Φ(z) Where x and z are the eigenvectors of any two data.

6. The method for identifying abnormal aging of batteries based on ALO-OCSVM according to claim 5, characterized in that: In step B2, solving the OCSVM single-class support vector machine algorithm includes: Step B2.1, construct a hypersphere in the feature space, and construct an optimization problem to minimize the volume of the hypersphere; Step B2.2, according to the Lagrange multiplier method, list the actual objective function to be solved; Step B2.3: Solve the maximization problem of the objective function in step B2.2 using the conventional SVM algorithm, obtain the Lagrangian coefficients of all samples, synthesize the support vector set, and calculate the center and radius of the hypersphere.

7. The method for identifying abnormal aging of batteries based on ALO-OCSVM high-dimensional features according to claim 6, characterized in that: In step B2.1, the hypersphere constructed and the optimization problem are: Where o and r are the center and radius of the hypersphere, C is the penalty parameter, ξ is the relaxation factor, and i For L train / The relaxation factor of time period i in the n-time period group, X i is the high-dimensional feature of time period i The amount of training data set data, L train is the preset time period, and n is the time length.

8. The method for identifying abnormal aging of batteries based on ALO-OCSVM high-dimensional features according to claim 7, characterized in that: In step B2.2, the actual objective function to be solved is: s.t.0≤λ i ≤C∑λ i =1 i=1,2,...,n In the formula, λ i For data X i The corresponding Lagrange coefficient.

9. The method for identifying abnormal aging of batteries based on ALO-OCSVM high-dimensional features according to claim 8, characterized in that: In step B2.3, the center and radius of the hypersphere are calculated as follows: Among them, X s For any sample in the support vector set, its Lagrangian coefficient λ s Satisfies 0<λ s <C。 10. The method for identifying abnormal aging of batteries based on ALO-OCSVM high-dimensional features according to claim 4, characterized in that: Step B3 includes: Step B3.1, set the number of iterations k for the algorithm to execute; Step B3.2, randomly generate an initial population, and the candidate solutions include different combinations of penalty coefficients and relaxation factors; Step B3.3, iteratively train the OCSVM model, repeat steps B2.1 to B2.3, and calculate Its mean square error on the validation dataset test set with high-dimensional features; Step B3.4: Use the nearest neighbor and random search method to find the solution with the minimum mean square error. Update the candidate solution position; Step B3.5: Replace the parameters in the OCSVM model with the corresponding solution of the minimum mean square error, and obtain to abnormal aging identification models.

11. The method for identifying abnormal aging of a battery based on ALO-OCSVM high-dimensional features according to claim 9, characterized in that: Step C includes: Step C1: During the preset time period L test Internally monitor the battery operation and collect voltage, current and temperature signals for a preset time length of n; Step C2, performing feature transformation on the original signal in step C1 to obtain high-dimensional features; Step C3: The original signals and high-dimensional features of the group time period are combined into a test data set A with high-dimensional features, which is used as the input of the abnormal aging recognition model; Step C4: for each sample A in the test data set with high-dimensional features u , calculate its distance to the center of the hypersphere and determine whether it is an abnormal sample.

12. The method for identifying abnormal aging of batteries based on ALO-OCSVM high-dimensional features according to claim 11, characterized in that: In step C4, the distance calculation formula from each sample in the test data set with high-dimensional features to the center of the hypersphere is as follows: Where, d u is a sample A in the test dataset with high-dimensional features u The distance to the center of the hypersphere, K is the kernel function, L test is the preset time period, n is the time length, X i and X j For L test / n groups of training data sets with high-dimensional features in time periods i and j, λ i and λ j For data X i and X j The corresponding Lagrange coefficient; If d u ≤r, indicating that the sample A in the test data set has high-dimensional features u Samples on or inside the hypersphere are identified as normal samples; otherwise, they are identified as abnormal samples.

13. A battery high-dimensional feature abnormal aging identification device using the battery high-dimensional feature abnormal aging identification method based on ALO-OCSVM according to any one of claims 1 to 12, characterized in that: include: A training data set construction module with high-dimensional feature quantities, an abnormal aging recognition model construction module, and an abnormal aging battery online monitoring and recognition module; Among them, the training data set construction module with high-dimensional feature quantities is used to monitor and collect battery operation data, generate effective high-dimensional features through feature engineering, and construct a training data set with high-dimensional feature quantities; The abnormal aging recognition model construction module is used to construct the ALO-OCSVM model, solve the OCSVM algorithm on the training data set with high-dimensional features constructed by the training data set construction module with high-dimensional features, and use ALO to optimize parameters to obtain the abnormal aging recognition model that meets the error requirements; The abnormal aging battery online monitoring and identification module is used to continue to monitor and collect battery operation data and generate a test data set with high-dimensional feature quantities, and use the abnormal aging recognition model obtained in the abnormal aging recognition model construction module to carry out online monitoring and identification of abnormal aging batteries.

Images