Retired battery health state evaluation method and system based on auto-encoder

Abstract

The invention belongs to the technical field of battery health state evaluation, and particularly relates to a retired battery health state evaluation method and system based on an auto-encoder, and the method comprises the steps: forming a to-be-processed data set through collecting the charging and discharging data and charging curve characteristics of a retired single battery, and obtaining a sequence sample comprising temperature, voltage and capacity changes; and performing feature learning and dimension reduction processing on the sequence sample by using an auto-encoder to obtain a low-dimensional feature vector, and inputting the low-dimensional feature vector into a multi-layer perceptron model for iterative training to obtain an evaluation model. The system comprises a data acquisition module, a sequence acquisition module, a feature extraction module, an evaluation model training module and a health state evaluation module. According to the method, the maximum available capacity of the battery can be estimated only through the charging test data, the test time and energy consumption are reduced, the evaluation precision and the calculation efficiency are improved, a scientific basis is provided for secondary utilization of the retired battery, and resource waste and safety risks are avoided.

Description

Technical Field

[0001] The present invention belongs to the technical field of battery health status assessment, and in particular relates to a retired battery health status assessment method and system based on an autoencoder. Background Art

[0002] As the core energy storage component of new energy vehicles, power batteries need to be retired after long-term use. However, retired batteries have significant differences in maximum available capacity, ohmic internal resistance and polarization internal resistance. This inconsistency weakens the overall performance of the battery pack and poses a challenge to the reuse of retired batteries. Therefore, before retired batteries are put back into use, their health status needs to be assessed.

[0003] Existing retired battery screening technologies mainly include measuring battery parameters based on professional tools, testing battery capacity based on charge and discharge cycles, and evaluating battery performance using specially designed charge and discharge conditions. However, these methods have many disadvantages, such as being highly dependent on the battery's remaining state of charge (SOC), time-consuming and energy-consuming, and having high technical barriers.

[0004] Specifically, the existing technology has the following difficulties: It requires a lot of tedious and meticulous feature engineering work, requires a high level of expertise in the field of electrochemistry, has a high technical threshold, and its scope of application is not wide enough.

[0005] Methods based on charge and discharge cycles require complex timing regression models, such as long short-term memory (LSTM), gated recurrent unit (GRU), and temporal convolutional network (TCN) to achieve good results, but they have high application costs and low efficiency.

[0006] The current feature screening work mainly uses statistical methods such as correlation analysis, which makes it difficult to explore the nonlinear relationship between features and battery health status, resulting in insufficient feature mining and low model accuracy. Summary of the invention

[0007] The purpose of the present invention is to provide a retired battery health status assessment method and system based on an autoencoder to solve the technical problem of how to efficiently and accurately assess the health status of retired batteries for secondary utilization.

[0008] The present invention achieves the above-mentioned purpose through the following technical solutions: In a first aspect, the present invention proposes a retired battery health status assessment method based on an autoencoder, the method comprising: Collect the charge and discharge data of retired single cells and the charge curve characteristics of preset charge and discharge conditions to form a data set to be processed; Acquire a sequence sample of the preset charge and discharge conditions including temperature, voltage and capacity changes based on the data set to be processed; Use the pre-built autoencoder to perform feature learning and dimensionality reduction on the pre-processed sequence samples to obtain a low-dimensional feature vector; Inputting the low-dimensional feature vector into a pre-built multi-layer perceptron model for iterative training to obtain an evaluation model; The health status of the target retired battery is evaluated according to the evaluation model.

[0009] Furthermore, before collecting the charging and discharging data of the retired single battery and the charging curve characteristics under the preset charging and discharging conditions, the method further includes: Generate identification code for each retired single battery and build battery information database; Controlling the charging and discharging of single cells based on preset charging and discharging strategies and preset charging and discharging conditions; The charging and discharging strategy includes setting a time period with the same time interval and time starting point as the charging and discharging time interval; and the preset charging and discharging operating condition includes performing constant current charging and discharging control on the single battery based on the time interval.

[0010] Further, the obtaining of sequence samples including temperature, voltage and capacity changes under the preset charge and discharge conditions based on the data set to be processed includes: Record the temperature changes under preset charging and discharging conditions and obtain the temperature change time series ; ; Record the voltage changes under preset charging and discharging conditions and obtain the voltage change time series ; ; Calculate and record the cumulative charging capacity change time series ; ; in, Indicates time, from arrive ; Temperature change time series , voltage change time series and capacity change time series One, get the normalized sequence , and ; Get the sequence samples under the preset charge and discharge conditions : .

[0011] Furthermore, the pre-constructed autoencoder is used to perform feature learning and dimensionality reduction processing on the pre-processed sequence samples to obtain a low-dimensional feature vector, including: Build an autoencoder based on the encoder and decoder; The sequence sample The pre-partitioned training set is input to the encoder and compressed into a low-dimensional latent representation h; The latent representation h is input into the decoder to obtain the reconstructed input ; The autoencoder is iteratively trained by the stochastic gradient descent algorithm to minimize the sequence samples. and reconstruct the input The difference between The sequence sample The optimized autoencoder is input to obtain a low-dimensional feature vector, wherein the low-dimensional feature vector includes low-dimensional features corresponding to each battery.

[0012] Furthermore, the low-dimensional feature vector is input into a pre-built multi-layer perceptron model for iterative training to obtain an evaluation model, including: Inputting the low-dimensional features extracted from the training set into a pre-built multi-layer perceptron model to obtain the corresponding actual maximum available capacity of the battery; The multi-layer perceptron model is iteratively trained using the low-dimensional features in the training set and the actual maximum available capacity of the battery corresponding thereto to obtain an evaluation model.

[0013] Furthermore, after the pre-processed sequence samples are subjected to feature learning and dimensionality reduction processing using the pre-built autoencoder to obtain a low-dimensional feature vector, the method further includes: storing the low-dimensional feature vector in the battery information database corresponding to the identification code to update the battery information database.

[0014] Furthermore, the health status of the target retired battery is evaluated according to the evaluation model, including: Retrieving the identification code of the target retired battery from the battery information database; Inputting the low-dimensional feature vector corresponding to the identification code into the evaluation model for processing to obtain the maximum available capacity of the battery; A battery health status assessment result is determined based on the maximum available capacity of the battery.

[0015] In a second aspect, the present invention proposes a retired battery health status assessment system based on an autoencoder, the system comprising: A data acquisition module is used to collect the charge and discharge data of retired single cells and the charging curve characteristics under set charge and discharge conditions to form a data set to be processed; A sequence acquisition module, used to acquire sequence samples including temperature, voltage and capacity changes under the preset charge and discharge conditions based on the data set to be processed; The feature extraction module is used to perform feature learning and dimensionality reduction processing on the preprocessed sequence samples using a pre-built autoencoder to obtain a low-dimensional feature vector; An evaluation model training module, used for inputting the low-dimensional feature vector into a pre-built multi-layer perceptron model for iterative training to obtain an evaluation model; A health status assessment module, used to assess the health status of the target retired battery according to the assessment model; Wherein, the data acquisition module includes: A code scanning unit, used to generate an identification code for each single battery; A charge and discharge control unit, used to execute a preset charge and discharge strategy and preset charge and discharge conditions to perform charge and discharge tests on single cells; The data acquisition unit is used to collect the battery charging and discharging information, including current, voltage and time, as well as external physical parameters, including temperature changes and capacity changes.

[0016] Furthermore, the health status assessment module includes: A database for storing battery identification codes and corresponding low-dimensional feature vectors; A retrieval unit, used to retrieve a corresponding low-dimensional feature vector from a database according to an identification code of a target retired battery; The evaluation unit is used to input the retrieved low-dimensional feature vector into the evaluation model for processing, and determine the health state of the battery according to the evaluation result.

[0017] The beneficial effects of the present invention are: The present invention can estimate the maximum available capacity of the battery only through the battery charging test data. Compared with the method of determining the maximum available capacity of the battery using a complete charge and discharge cycle, it can reduce the test time and energy consumption; and combined with the autoencoder to screen features, the complexity of the back-end multi-layer perceptron model is reduced, thereby reducing the computational cost and improving the generalization ability of the model. The proposed health status assessment method can achieve high-precision battery health status estimation, thereby improving the use efficiency of retired batteries and avoiding resource waste and safety risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A schematic diagram of a flow chart of a retired battery health status assessment method based on an autoencoder provided in an embodiment of the present application; Figure 2 A system structure diagram of a retired battery health status assessment system based on an autoencoder provided in an embodiment of the present application; Figure 3 A schematic diagram of a health status assessment process of a retired battery health status assessment system based on an autoencoder provided in an embodiment of the present application; Figure 4 A schematic diagram of another process of a retired battery health status assessment method based on an autoencoder provided in an embodiment of the present application; Figure 5 Another flowchart of a retired battery health status assessment method based on an autoencoder is provided in an embodiment of the present application. DETAILED DESCRIPTION

[0019] The present application is further described in detail below in conjunction with the accompanying drawings. It is necessary to point out here that the following specific implementation methods are only used to further illustrate the present application and cannot be understood as limiting the scope of protection of the present application. Technical personnel in this field can make some non-essential improvements and adjustments to the present application based on the above application content.

[0020] Example 1

[0021] like Figure 1-5 As shown, this embodiment proposes a retired battery health status assessment method based on an autoencoder, and the method includes the following steps: S1. Generate an identification code for each retired single cell and build a battery information database; control the charge and discharge of the single cell based on a preset charge and discharge strategy and preset charge and discharge conditions; wherein the charge and discharge strategy includes setting a time period with the same time interval and time starting point as a charge and discharge time interval; the preset charge and discharge conditions include constant current charge and discharge control of the single cell based on the time interval.

[0022] S2. Collect the charge and discharge data of retired single cells and the charge curve characteristics of preset charge and discharge conditions to form a data set to be processed.

[0023] Exemplarily, in this step, a barcode scanner is first used to generate a unique identification code for each retired single cell, such as a QR code, barcode or electronic tag, and a battery information database is established to store these identification codes and related information. Subsequently, a charge and discharge instrument is used to perform charge and discharge tests on the single cell according to a preset charge and discharge strategy and preset charge and discharge conditions. The preset charge and discharge conditions may include constant current charge and discharge control of the single cell based on a set time interval. The data acquisition instrument is used to collect the charge and discharge information of the battery, including current, voltage and time, as well as external physical parameters such as temperature changes and capacity changes. This information is integrated to form a data set to be processed.

[0024] In specific implementation, this solution can estimate the maximum available capacity of the battery only through battery charging test data, which can reduce the time required to determine the maximum available capacity of the battery compared to the method of using a complete charge and discharge cycle to determine the maximum available capacity of the battery. The above test time and above energy consumption.

[0025] S3. Based on the data set to be processed, a sequence sample including temperature, voltage and capacity changes under a preset charge and discharge condition is obtained.

[0026] In a further preferred embodiment of the present invention, obtaining a sequence sample including temperature, voltage and capacity changes under a preset charge and discharge condition based on the data set to be processed includes: Record the temperature changes under preset charging and discharging conditions and obtain the temperature change time series ; ; Record the voltage changes under preset charging and discharging conditions and obtain the voltage change time series ; ; Calculate and record the cumulative charging capacity change time series ; ; in, Indicates time, from arrive ; Temperature change time series , voltage change time series and capacity change time series One, get the normalized sequence , and ; Get the sequence samples under the preset charge and discharge conditions : .

[0027] It is understandable that after obtaining the data set to be processed, this step further processes these data to extract key sequence samples. Specifically, the temperature change under the preset charge and discharge conditions is recorded to obtain the temperature change time series; the voltage change under the preset charge and discharge conditions is recorded to obtain the voltage change time series; and the cumulative charge capacity change time series is calculated and recorded at the same time. In order to improve the efficiency and accuracy of model training, these time series data are normalized to obtain normalized sequence samples.

[0028] S4. Use the pre-built autoencoder to perform feature learning and dimensionality reduction on the preprocessed sequence samples to obtain a low-dimensional feature vector.

[0029] In a further preferred embodiment of the present invention, a pre-constructed autoencoder is used to perform feature learning and dimensionality reduction processing on the pre-processed sequence samples to obtain a low-dimensional feature vector, including: Build an autoencoder based on the encoder and decoder; The sequence sample The pre-partitioned training set is input to the encoder and compressed into a low-dimensional latent representation h; Input the latent representation h into the decoder to obtain the reconstructed input ; The autoencoder is iteratively trained using the stochastic gradient descent algorithm to minimize the number of sequence samples. and reconstruct the input The difference between The sequence sample The optimized autoencoder is input to obtain a low-dimensional feature vector, which includes low-dimensional features corresponding to each battery.

[0030] In this step, a pre-built autoencoder is used to perform feature learning and dimensionality reduction on the normalized sequence samples. The autoencoder consists of two parts: an encoder and a decoder. The encoder compresses the input data into a potential representation (a low-dimensional feature vector), and the decoder attempts to reconstruct the original input based on this low-dimensional representation. The autoencoder is iteratively trained using a stochastic gradient descent algorithm to minimize the difference between the sequence samples and the reconstructed input. After training, the sequence samples are input into the optimized autoencoder to obtain low-dimensional feature vectors. These low-dimensional feature vectors contain key information corresponding to each battery and are used for subsequent evaluation model training.

[0031] This embodiment uses an autoencoder to screen features and reduces the complexity of the back-end multi-layer perceptron model from Reduce to .in, is the sampling number of charging test data, is the number of features of the hidden layer of the autoencoder, satisfying relationship.

[0032] In a further preferred embodiment of the present invention, after using a pre-built autoencoder to perform feature learning and dimensionality reduction processing on the preprocessed sequence samples to obtain a low-dimensional feature vector, it also includes: storing the low-dimensional feature vector in a battery information database corresponding to the identification code to update the battery information database.

[0033] S5. Input the low-dimensional feature vector into a pre-built multi-layer perceptron model for iterative training to obtain an evaluation model; and evaluate the health status of the target retired battery according to the evaluation model.

[0034] In a further preferred embodiment of the present invention, the low-dimensional feature vector is input into a pre-built multi-layer perceptron model for iterative training to obtain an evaluation model, including: The low-dimensional features extracted from the training set are input into the pre-built multi-layer perceptron model to obtain the corresponding actual maximum available capacity of the battery; The multi-layer perceptron model is iteratively trained using the low-dimensional features in the training set and the corresponding actual maximum available capacity of the battery to obtain an evaluation model.

[0035] In a further preferred embodiment of the present invention, the health status of the target retired battery is evaluated according to the evaluation model, including: Retrieving an identification code of a target retired battery from a battery information database; The low-dimensional feature vector corresponding to the identification code is input into the evaluation model for processing to obtain the maximum available capacity of the battery; The battery health status assessment result is determined based on the maximum available capacity of the battery.

[0036] In this step, the low-dimensional feature vector extracted from the training set is input into the pre-built multi-layer perceptron (MLP) model for iterative training. The multi-layer perceptron is a feedforward neural network that can fit complex nonlinear relationships by adjusting internal parameters. During the training process, the low-dimensional feature vector and its corresponding actual maximum available capacity of the battery are used as input and output to train the MLP model to minimize the gap between the actual maximum available capacity and the predicted result. After the training is completed, an evaluation model for evaluating the battery health status is obtained.

[0037] Furthermore, in the evaluation phase, the identification code of the target retired battery is first retrieved from the battery information database, and the low-dimensional feature vector corresponding to the identification code is found. Then, the low-dimensional feature vector is input into the evaluation model for processing to obtain an estimated value of the maximum available capacity of the battery. Finally, the battery health status evaluation result is determined based on the maximum available capacity of the battery, such as good, fair or poor.

[0038] In addition, after obtaining the low-dimensional feature vector, the low-dimensional feature vector can also be stored in the battery information database corresponding to the identification code to update the battery information database to facilitate subsequent health status assessment.

[0039] Based on the same inventive concept, this embodiment also proposes a retired battery health status assessment system based on an autoencoder, the system comprising: A data acquisition module is used to collect the charge and discharge data of retired single cells and the charging curve characteristics under set charge and discharge conditions to form a data set to be processed; A sequence acquisition module, used to acquire sequence samples including temperature, voltage and capacity changes under preset charge and discharge conditions based on the data set to be processed; The feature extraction module is used to perform feature learning and dimensionality reduction processing on the preprocessed sequence samples using a pre-built autoencoder to obtain a low-dimensional feature vector; An evaluation model training module is used to input the low-dimensional feature vector into a pre-built multi-layer perceptron model for iterative training to obtain an evaluation model; A health status assessment module, used to assess the health status of the target retired battery according to the assessment model; The data acquisition module includes: A code scanning unit, used to generate an identification code for each single battery; A charge and discharge control unit, used to execute a preset charge and discharge strategy and preset charge and discharge conditions to perform charge and discharge tests on single cells; The data acquisition unit is used to collect the battery charging and discharging information, including current, voltage and time, as well as external physical parameters, including temperature changes and capacity changes.

[0040] In a further preferred embodiment of the present invention, the health status assessment module includes: A database for storing battery identification codes and corresponding low-dimensional feature vectors; A retrieval unit, used to retrieve a corresponding low-dimensional feature vector from a database according to an identification code of a target retired battery; The evaluation unit is used to input the retrieved low-dimensional feature vector into the evaluation model for processing, and determine the health state of the battery according to the evaluation result.

[0041] According to the above evaluation system, the specific implementation includes the following steps: (1) Generate an identification code (such as a QR code, barcode, electronic tag, etc.) for each retired single battery through a barcode scanner, and establish a corresponding database in the data storage device corresponding to the identification code; (2) The host computer controls the charging and discharging instrument to execute the charging and discharging strategy of the agreed working conditions; (3) The data acquisition instrument collects the charge and discharge data of the retired single battery and the required external physical parameter data of the battery, and stores the collected data in the database established for the corresponding retired single battery after dimension reduction through the autoencoder; (4) The host computing device obtains the retired battery data from the database, processes and analyzes it according to the designed battery health status assessment algorithm, and finally obtains the status assessment result.

[0042] According to the above embodiment of the present invention, the system uses autoencoder technology to evaluate the health status of retired batteries. In principle, the system first collects the charge and discharge data and charging curve characteristics of retired single cells through the data acquisition module to form a data set to be processed. Subsequently, the autoencoder is used to perform feature learning and dimensionality reduction on the preprocessed sequence samples to extract low-dimensional feature vectors. These feature vectors are input into the multi-layer perceptron model for iterative training to construct an evaluation model. In specific implementation, the system generates an identification code for each battery through a barcode scanner and establishes a database. The charge and discharge instrument executes the charge and discharge strategy of the preset working conditions, and stores the collected data after dimensionality reduction through the autoencoder. Finally, the system evaluates the health status of the target retired battery according to the evaluation model to determine whether it is suitable for secondary use. The system can significantly improve the evaluation accuracy, reduce test time and energy consumption, and provide a scientific basis for the reuse of retired batteries.

[0043] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0044] In addition, each functional module in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0045] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments can still be modified, or some of the technical features can be replaced by equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A retired battery health status assessment method based on autoencoder, characterized in that: The method comprises: Collect the charge and discharge data of retired single cells and the charge curve characteristics under preset charge and discharge conditions to form a data set to be processed; Acquire a sequence sample including temperature, voltage and capacity changes under the preset charge and discharge conditions based on the data set to be processed; Use the pre-built autoencoder to perform feature learning and dimensionality reduction on the pre-processed sequence samples to obtain a low-dimensional feature vector; Inputting the low-dimensional feature vector into a pre-built multi-layer perceptron model for iterative training to obtain an evaluation model; The health status of the target retired battery is evaluated according to the evaluation model.

2. According to claim 1, a retired battery health status assessment method based on autoencoder is characterized in that: Before collecting the charging and discharging data of the retired single battery and the charging curve characteristics under the preset charging and discharging conditions, the method further includes: Generate identification code for each retired single battery and build battery information database; Controlling the charging and discharging of single cells based on preset charging and discharging strategies and preset charging and discharging conditions; The charging and discharging strategy includes setting a time period with the same time interval and time starting point as the charging and discharging time interval; and the preset charging and discharging operating condition includes performing constant current charging and discharging control on the single battery based on the time interval.

3. The method for assessing the health status of retired batteries based on an autoencoder according to claim 2, characterized in that: The acquiring, based on the data set to be processed, sequence samples including temperature, voltage and capacity changes under the preset charge and discharge conditions comprises: Record the temperature changes under preset charging and discharging conditions and obtain the temperature change time series ; ； Record the voltage changes under preset charging and discharging conditions and obtain the voltage change time series ; ； Calculate and record the cumulative charging capacity change time series ; ； in, Indicates time, from arrive ; Temperature change time series , voltage change time series and capacity change time series One, get the normalized sequence , and ; Get the sequence samples under the preset charge and discharge conditions : .

4. The method for assessing the health status of retired batteries based on an autoencoder according to claim 3, characterized in that: The pre-constructed autoencoder is used to perform feature learning and dimensionality reduction processing on the pre-processed sequence samples to obtain a low-dimensional feature vector, including: Build an autoencoder based on the encoder and decoder; The sequence sample The pre-partitioned training set is input to the encoder and compressed into a low-dimensional latent representation h; The latent representation h is input into the decoder to obtain the reconstructed input ; The autoencoder is iteratively trained by the stochastic gradient descent algorithm to minimize the sequence samples. and reconstruct the input The difference between The sequence sample The optimized autoencoder is input to obtain a low-dimensional feature vector, wherein the low-dimensional feature vector includes low-dimensional features corresponding to each battery.

5. The method for assessing the health status of retired batteries based on an autoencoder according to claim 4, characterized in that: The low-dimensional feature vector is input into a pre-built multi-layer perceptron model for iterative training to obtain an evaluation model, including: Inputting the low-dimensional features extracted from the training set into a pre-built multi-layer perceptron model to obtain the corresponding actual maximum available capacity of the battery; The multi-layer perceptron model is iteratively trained using the low-dimensional features in the training set and the actual maximum available capacity of the battery corresponding thereto to obtain an evaluation model.

6. The method for assessing the health status of retired batteries based on an autoencoder according to claim 5, characterized in that: After the pre-processed sequence samples are subjected to feature learning and dimensionality reduction processing using the pre-built autoencoder to obtain a low-dimensional feature vector, the method further includes: storing the low-dimensional feature vector in the battery information database corresponding to the identification code to update the battery information database.

7. The method for assessing the health status of retired batteries based on an autoencoder according to claim 6, characterized in that: The evaluating the health status of the target retired battery according to the evaluation model includes: Retrieving the identification code of the target retired battery from the battery information database; Inputting the low-dimensional feature vector corresponding to the identification code into the evaluation model for processing to obtain the maximum available capacity of the battery; A battery health status assessment result is determined based on the maximum available capacity of the battery.

8. A retired battery health status assessment system based on autoencoder, characterized in that: The system comprises: A data acquisition module is used to collect the charge and discharge data of retired single cells and the charging curve characteristics under preset charge and discharge conditions to form a data set to be processed; A sequence acquisition module, used to acquire sequence samples including temperature, voltage and capacity changes under the preset charge and discharge conditions based on the data set to be processed; The feature extraction module is used to perform feature learning and dimensionality reduction processing on the preprocessed sequence samples using a pre-built autoencoder to obtain a low-dimensional feature vector; An evaluation model training module, used for inputting the low-dimensional feature vector into a pre-built multi-layer perceptron model for iterative training to obtain an evaluation model; A health status assessment module, used to assess the health status of the target retired battery according to the assessment model; Wherein, the data acquisition module includes: A code scanning unit, used to generate an identification code for each single battery; A charge and discharge control unit, used to execute a preset charge and discharge strategy and preset charge and discharge conditions to perform charge and discharge tests on single cells; The data acquisition unit is used to collect the battery charging and discharging information, including current, voltage and time, as well as external physical parameters, including temperature changes and capacity changes.

9. The retired battery health status assessment system based on autoencoder according to claim 8, characterized in that: The health status assessment module includes: A database for storing battery identification codes and corresponding low-dimensional feature vectors; A retrieval unit, used to retrieve a corresponding low-dimensional feature vector from a database according to an identification code of a target retired battery; The evaluation unit is used to input the retrieved low-dimensional feature vector into the evaluation model for processing, and determine the health state of the battery according to the evaluation result.

Images