Retired battery rapid sorting method based on support vector machine algorithm

Abstract

The invention discloses a fast retired battery sorting method based on a support vector machine algorithm, and the method comprises the steps: building a dichotomy sorting feature vector according to the external characteristic parameters of retired batteries, and carrying out the first-level sorting based on the support vector machine algorithm, thereby achieving the preliminary classification of the batteries capable of being used in an echelon manner and the directly recycled batteries. After the binary classification result is obtained, high-rate charging is conducted on the battery, an incremental capacity (IC) curve is obtained based on capacity-voltage data in the charging process, the characteristic peak value of the IC curve and the corresponding voltage are extracted to serve as secondary sorting indexes, the first-level sorting result is combined to form a multi-classification characteristic vector set of the battery, the multi-classification characteristic vector set is input into a multi-classification model, and the multi-classification characteristic vector set of the battery is obtained. And finally, multi-class classification of the retired lithium ion batteries is realized, and the sorting consistency of the retired batteries is improved. According to the method, the sorting problem of the retired batteries in different health states can be effectively solved, the sorting efficiency of the retired batteries is improved, and the sorting precision is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of battery recycling, and in particular to a method for quickly sorting retired batteries based on a support vector machine algorithm. Background Art

[0002] With the continued exponential growth in the number of electric vehicles, the upcoming large-scale retirement of electric vehicle batteries in the future has brought unprecedented challenges to the cascade utilization and resource recycling of batteries. Faced with this severe situation, how to efficiently recycle and utilize retired batteries has become a key issue that needs to be solved urgently. The cascade utilization of retired batteries has shown great potential in many fields such as power storage, smart grid, renewable energy grid connection, and household energy storage. It can not only maximize the remaining use value of batteries and reduce the cost of the battery throughout its life cycle, but also effectively reduce the potential pollution of waste batteries to the environment and alleviate the problem of resource shortages. Through a scientific and reasonable cascade utilization strategy, retired batteries can extend their lifespan while promoting the development of the green economy and providing strong technical support for achieving the goal of carbon neutrality.

[0003] At present, the sorting of retired batteries is mainly focused on batteries with a battery health state (SOH) of more than 80%, and is usually evaluated by a combination of multiple sorting characteristic parameters, such as capacity, internal resistance, voltage, temperature, impedance, etc. Although these indicators can evaluate the consistency of batteries to a certain extent, they are often difficult to fully reflect the actual remaining capacity of retired automotive power batteries. In addition, the combination of characteristic parameters is complex, which leads to a lot of time and labor costs in the sorting process, reducing the sorting efficiency. Therefore, the existing sorting methods are difficult to meet the needs of large-scale retired battery processing, and fail to achieve efficient and full coverage of retired battery sorting.

[0004] In summary, the existing technology has obvious limitations in the process of sorting retired batteries, and it is difficult to achieve comprehensive and accurate sorting of batteries. Therefore, there is an urgent need for an innovative method that can simplify the sorting process and improve the sorting accuracy, so as to reduce time and labor costs and achieve rapid and large-scale sorting of retired batteries. This improvement can not only significantly improve the efficiency of battery recycling and maximize the residual value of batteries, but also have a positive role in promoting the sustainable development of the electric vehicle industry chain, and help to cope with the resource and environmental challenges brought about by large-scale battery retirement in the future. Summary of the invention

[0005] In view of the shortcomings of existing retired battery sorting technology, the present invention proposes a fast sorting method for retired batteries based on a support vector machine algorithm. This method not only broadens the sorting range of retired batteries, but also effectively reduces the consistency differences between batteries by optimizing the consistency of batteries. Compared with traditional methods, this technology effectively reduces the time and labor costs in the sorting process, improves the sorting efficiency, extends the service life of retired batteries, and maximizes their cascade utilization value.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: A method for quickly sorting retired batteries based on a support vector machine algorithm comprises the following steps: Step 1: Perform an appearance inspection on the retired battery to see if the surface of the retired battery is damaged, leaking, or deformed, and check if the voltage of the retired battery is within the normal range. If both the appearance and voltage are qualified, proceed to step 2, otherwise, remove the battery; Step 2, performing a discharge test on the retired battery, and stopping the discharge when the retired battery voltage reaches a predetermined cut-off voltage; Step 3: Conduct external characteristic tests on retired batteries, and collect and record data; Step 4: construct an SVM binary classifier based on the collected data to roughly classify the retired batteries; Step 5: Charge the retired battery at a high rate, collect charging data in real time, and stop charging when the voltage reaches the charging cut-off voltage; Step 6, calculating the incremental capacity IC curve of the retired battery based on the charging data; Step 7, extracting the sorting characteristic data of the IC curve: IC curve peak value and corresponding voltage; Step 8: preprocess the extracted sorting feature data, remove abnormal data, and smooth the IC curve; Step nine, construct a SVM multi-classifier, and further subdivide the recyclable batteries into multiple categories based on the preprocessed sorting feature data.

[0007] Compared with the prior art, the present invention has the following beneficial effects: The present invention simplifies the selection process of key battery features and optimizes the evaluation and sorting criteria for retired batteries. By using easily accessible feature parameters, it is ensured that the batteries in the multi-classification results all meet the capacity requirements for cascade utilization, thereby improving the sorting efficiency of retired batteries and effectively reducing manpower and time costs. The present invention further expands the scope of application of sorting retired batteries, solves the problem of inseparability of batteries in different SOH states, and makes the sorting results more in line with actual needs. Through this method, batteries with consistent performance can be quickly and effectively screened out from a large number of retired batteries to ensure their consistency in practical applications. In addition, through the improved method of the present invention, the service life of retired batteries can be extended and the residual value of retired batteries can be maximized. The technical improvements of the present invention will be described in detail in the following description, and its advantages will be reflected in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 This is a flow chart of a method for rapidly sorting retired batteries based on a support vector machine algorithm according to the present invention; Figure 2 A flowchart of obtaining characteristic data for sorting retired batteries according to the present invention; Figure 3 shows an IC curve diagram according to an embodiment of the present invention; Figure 4 A flow chart of the improved support vector machine sorting method of the present invention; Figure 5 The binary distribution diagram of retired batteries is shown. DETAILED DESCRIPTION

[0009] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0010] Embodiment 1 like Figure 1 As shown, this embodiment provides a method for quickly sorting retired batteries based on a support vector machine algorithm, comprising the following steps: Step 1: Check the appearance of the battery to see if it is damaged, leaking, or deformed, and check if the voltage is within the normal range. If the appearance and voltage tests are qualified, proceed to step 2, otherwise, discard it.

[0011] In some embodiments, a camera or other device may be used to collect images of the battery appearance, and an image processing model may be used to automatically determine whether the appearance is qualified. The image processing model is supported by an existing intelligent algorithm and is used to identify whether the battery appearance is damaged, deformed, or leaking, and to remove unqualified batteries.

[0012] In some embodiments, manual inspection is also feasible.

[0013] In addition, batteries with voltages in abnormal ranges should also be classified as unqualified.

[0014] Step 2: Perform a discharge test on the battery and stop when the cut-off voltage is reached.

[0015] The battery discharge test should be performed at a discharge rate of 0.5C, and the cut-off voltage needs to be set according to the instructions provided by the battery manufacturer. The cut-off voltage is set to the voltage when the battery state of charge is 0%.

[0016] Step three: perform external characteristic test on the battery and collect data.

[0017] The external characteristic test of the battery includes: battery internal resistance test and open circuit voltage test. It should be noted that this test is a test after the battery is fully discharged. Among them, a battery internal resistance tester is used to test the internal resistance and open circuit voltage, and the data is recorded by a computer.

[0018] Step 4: Construct a support vector machine (SVM) binary classifier based on the collected data to roughly classify the retired batteries.

[0019] The coarse classification of batteries is mainly aimed at batteries with a battery state of health (SOH) below 70%. The SVM binary classifier is used to divide batteries with SOH above 70% and below 70% into two categories. Batteries with SOH above 70% meet the requirements for cascade utilization and will be further refined in the future.

[0020] Step 5: Charge the battery at a high rate, collect charging data, and stop when the cut-off voltage is reached.

[0021] In one embodiment, the high rate charging operation of the battery is placed under a charging rate of 1.5C, and the cut-off voltage is still set according to the instructions provided by the battery manufacturer.

[0022] During the battery charging process, charging data needs to be collected. Charging data refers to the battery capacity (Q) and voltage (V) change data during the charging process.

[0023] Step six, calculating the incremental capacity (IC) curve of the retired battery based on the charging data; the IC curve of the retired battery is obtained by numerically deriving the capacity-voltage (QV) curve.

[0024] Step 7, extracting the sorting feature data of the IC curve, the sorting feature data includes: the open circuit voltage, ohmic internal resistance, IC curve peak value and corresponding voltage of the battery; Step 8: preprocess the sorting feature data; remove abnormal data and smooth the IC curve; the preprocessing may include: setting a reasonable sampling interval, performing reasonable noise reduction on the original data, and standardizing the data to eliminate the influence between different dimensions; Step nine, construct a SVM multi-classifier, and further subdivide the recyclable batteries into multiple categories based on the preprocessed sorting feature data. In one embodiment, each category is divided into category 1, category 2, and category 3 at intervals of 10% of the battery health status.

[0025] The characteristic peak of the IC curve is the ordinate value of the highest peak of the IC curve, and the characteristic voltage is the abscissa value of the corresponding highest peak.

[0026] Figure 2 This is a flow chart for obtaining characteristic data for sorting retired batteries according to the present invention.

[0027] like Figure 2 As shown, the retired battery is discharged to the lower limit of the cut-off voltage using a 0.5C constant current, and the ohmic internal resistance and open circuit voltage of the battery in the fully discharged (0% SOC) state are tested and recorded using a battery internal resistance tester, and then the recorded data is detected and eliminated for abnormal values. Subsequently, the battery is charged to the upper limit of the cut-off voltage using a 1.5C constant current and constant voltage to ensure that the battery is fully charged (100% SOC), and the battery power-terminal voltage (QV) curve is obtained, and the QV curve is used as the basic data for calculating the IC curve in the next step. In the next step, the IC curve of the battery is obtained by deriving the QV curve.

[0028] Specifically, the calculation formula of the IC curve is as follows: , Where: Represents the battery level, represents the terminal voltage, and is its corresponding discrete form, Represents time, represents the index of a discrete time point, Indicates the sampling interval.

[0029] It is worth mentioning that Figure 3 As shown, the IC curve is a denoised IC curve after data denoising, so as to avoid the problem of large noise in the original IC curve obtained by derivation of the QV curve.

[0030] like Figure 4 As shown, the sorting feature data It is the characteristic vector combination of recyclable batteries, which is composed of the following formula: , Where: Represent the corresponding sorting characteristics, which are: internal resistance, open circuit voltage, characteristic peak value of IC curve, and characteristic peak voltage of IC curve.

[0031] It should be noted that there are two combinations of sorting features. In the SVM binary classification model, the sorting features are: internal resistance and open circuit voltage; in the SVM multi-classification model, the sorting features are the above-mentioned sorting feature vectors.

[0032] in: Figure 4 The flowchart of the improved support vector machine (SVM) sorting method of the present invention is as follows. Figure 4 Describe step 4: Based on the collected data, construct an SVM binary classifier to roughly classify the retired batteries. The SVM binary classifier is used to classify retired batteries into batteries that can be used in cascades and batteries that can be directly recycled, where i represents the sorting round, where i = 1 represents the rough sorting stage, and i = 2 represents the fine sorting stage.

[0033] The construction of the SVM binary classifier includes the following steps: S01, input the sorting feature data and the number of battery types; S02, construct the SVM binary classifier, select the appropriate kernel function according to the distribution and characteristics of the sorting feature data, and map the sorting feature data to the high-dimensional space for nonlinear transformation; S03, use the grid search method to tune the parameters of the SVM binary classifier; S04, set the core parameters of the SVM binary classifier, including: penalty parameter C and kernel function parameters; S05, use the selected kernel function and core parameter configuration to train the SVM binary classifier; S06, use the test data set to test the performance of the SVM binary classifier; S07, calculate the accuracy and precision of the SVM binary classifier, and use the battery consistency parameter index to evaluate whether it effectively divides the sample points into two categories with a health state higher than 70% and lower than 70% through the spatial distribution of the health state of the sample points. The battery consistency parameter index refers to the state of health (SOH) of retired batteries.

[0034] The algorithm of SVM binary classifier includes the following steps: Step S10, projecting sample data into a high-dimensional space: using a kernel function (such as a linear kernel, a polynomial kernel, or a Gaussian kernel, etc.) to map low-dimensional samples (such as two-dimensional data of internal resistance and open-circuit voltage) into a higher-dimensional feature space; Step S11, determine a hyperplane: The goal of SVM is to find a hyperplane that can maximize the interval between two types of data. The equation of the hyperplane can be expressed as: , Where: is the support vector corresponding to the hyperplane, is the normal vector of the hyperplane, is the bias term. The goal is to find the optimal and , so as to maximize the distance from the support vector to the hyperplane.

[0035] Step S12, define support vectors and calculate the margin: support vectors are those sample points that are closest to the hyperplane and are located on the boundary between the two types of data. Assume is a support vector, and the support vector corresponding to the hyperplane satisfies: , Margin is defined as the distance from the support vector to the hyperplane, which can be expressed as: , Where: Represents the hyperplane normal vector The norm of .

[0036] Step S13, maximizing the support vector interval and determining the hyperplane parameters: maximizing the interval from the support vector to the hyperplane by optimizing the following problem: , Where: is the class label of the sample, which takes a value of +1 or -1. By solving this optimization problem, the optimal hyperplane parameters are determined and .

[0037] Step S14: Introduce slack variables to allow for classification errors: In practical problems, the data may not be completely linearly separable, so slack variables are introduced To allow some sample points to be on the wrong side of classification. The new optimization problem is: , Where: is a regularization parameter that is used to balance the interval size and the penalty for misclassified samples.

[0038] Step S15: Introduce kernel function to solve the linear inseparable problem: When the sample points are linearly inseparable in the original space, SVM introduces kernel function Map the sample points to a high-dimensional space. In a high-dimensional space, the sample points may become linearly separable. Commonly used kernel functions include: Linear kernel: , Polynomial kernel: , Gaussian Kernel: .

[0039] Where: is the feature vector of the input data; is a constant term; is the order of the polynomial; is a hyperparameter that controls the width of the kernel function.

[0040] In this embodiment, 264 retired ternary lithium-ion batteries are selected as sample sets for training and testing. The sample ratio of the training set to the test set is about 6:4, and the specific sample distribution is shown in Table 1.

[0041] Table 1 Distribution of binary classification training and validation samples of retired ternary lithium-ion batteries

[0042]

[0043] In order to verify the classification effect of the proposed classification method on retired batteries, retired batteries are divided into recycled batteries and second-life batteries according to whether the battery capacity is higher than 70% SOH. Some samples of each type of battery are selected for training, and the classification effect of this method is verified by comparing the predicted and actual labels. The SVM binary classification model uses the linear kernel with the highest classification accuracy. The binary classification results of retired batteries are shown in the figure below. Figure 5 As shown, the classification of the second-life batteries was all correct, and only two cells in the recycled batteries were classified incorrectly, with an overall sorting accuracy of up to 98.1%.

[0044] In addition, since there are mature testing methods for battery internal resistance and open circuit voltage, the test time is less than 1 second, which significantly shortens the test time. In summary, this method can preliminarily screen out batteries that meet the second-life utilization standards and accurately classify batteries into second-life batteries and recycled batteries, which not only saves a lot of manpower and time costs, but also effectively reduces the calculation pressure of subsequent multi-classification models, speeds up the convergence speed of multi-classification models, further improves the safety of second-life batteries, and prolongs their service life.

[0045] Step nine, constructing an SVM multi-classifier to further classify the recyclable batteries into multiple categories, wherein the SVM multi-classification algorithm of the SVM multi-classifier includes the following steps: Step S20, determine multiple hyperplanes: adopt a one-vs-one (OvO) strategy and train an SVM binary classification model for each pair of categories. categories, training is required Models; Step S21, repeating step: for each hyperplane, re-perform all steps except S11 in the algorithm steps of the SVM binary classifier; Step S22: Determine the multi-classification result by voting through multiple SVM binary classification models: Each SVM binary classification model predicts that the sample belongs to one of the two categories. The final category is determined by the voting results of all SVM binary classification models, and the category with the most votes is selected as the final classification result.

[0046] In this embodiment, 202 batteries of the SVM binary classification model sorting results were selected to form a sample set of the multi-classification model for training and testing. The sample ratio of the training set and the test set is consistent with the above method, and the specific sample number is shown in Table 2.

[0047] Table 2 Distribution of multi-classification training and validation samples of retired ternary lithium-ion batteries

[0048]

[0049] In order to demonstrate the classification effect of the multi-classification method, the retired batteries are divided into three categories according to the battery capacity, and the same training and testing methods as the two-classification method are used. The calculation results of the mean and standard deviation of the comprehensive evaluation indicators of various retired battery sorting are shown in Table 3.

[0050] Among them, the maximum standard deviation of the capacity of retired batteries is 0.085Ah, which is about 2.8% of the nominal capacity; the maximum standard deviation of the open circuit voltage is 0.087V, which is about 2.5% of the standard voltage; the maximum standard deviation of the internal resistance is 1.582mΩ, which is about 5% of the standard internal resistance; and the maximum standard deviation of the peak voltage is 0.056V. It can be seen that the data distribution of the four evaluation indicators is relatively concentrated, which verifies the rationality of this classification method.

[0051] In addition, the mean values ​​of categories 1, 2, and 3 are significantly different, and the data distribution between categories is relatively discrete, indicating that the sample points of the sorted retired batteries in the same category are compactly distributed, that is, the consistency of the same type of batteries is high. This proves that the sorting method proposed in the present invention can effectively improve the consistency of retired batteries, reduce safety risks in subsequent use, enhance the reliability of retired batteries, and extend their service life.

[0052] Table 3 Evaluation indicators of retired batteries

[0053]

[0054] Embodiment 2 This embodiment provides a system for rapid comprehensive sorting of retired batteries based on a support vector machine algorithm, including the following modules: A data acquisition module is used to capture test data of retired batteries.

[0055] The feature vector data construction module is used to preprocess the test data of retired batteries, including data verification, cleaning and noise reduction, and generate binary and multi-classification feature vectors for each retired battery, and finally construct a comprehensive sorting feature vector matrix for retired batteries.

[0056] The binary classification module, based on the binary classification feature vector of retired batteries and the support vector machine algorithm, classifies the batteries by finding the hyperplane of the binary classification feature vector data, and evaluates the classification results based on the battery external characteristic parameters. This module can dynamically adjust the model parameters, find the optimal hyperplane, output the binary classification label of each retired battery, and classify retired batteries with the same classification label into the same category.

[0057] The multi-classification module, based on the classification results of the binary classification module, further uses the support vector machine algorithm to map the data from the low-dimensional space to the high-dimensional space, and realizes the multi-classification of linearly inseparable data. This module evaluates the multi-classification results according to the consistency parameters of the battery, outputs the multi-classification labels of each retired battery, and classifies retired batteries with the same classification labels into the same category.

[0058] Embodiment 3 This embodiment provides an incremental capacity curve acquisition strategy suitable for high rate conditions, and retired batteries can generate corresponding incremental capacity curves at different voltage intervals. Although the incremental capacity curves will change, the proposed retired battery rapid sorting method based on the support vector machine algorithm is still applicable.

[0059] Embodiment 4 This embodiment provides a computer-readable storage medium, on which a computer program is stored. When a processor executes the program, the steps of the above-mentioned method for rapid sorting of retired batteries based on a support vector machine algorithm can be implemented.

[0060] Those skilled in the art should understand that the embodiments of the present invention can be provided as methods, systems or computer program products. Therefore, the present invention can be implemented by hardware implementation, software implementation, or a combination of hardware and software. In addition, the present invention can also take the form of a computer program product, which can be implemented by storing computer-usable program codes through one or more computer-readable storage media (including but not limited to disk storage and optical storage, etc.).

[0061] The present invention is described by referring to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products in the embodiments of the present invention. It should be understood that each process and / or box in these flowcharts and / or block diagrams, as well as their combinations, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device, thereby generating a machine that executes these instructions through the processor to implement the functions specified by one or more processes in the flowchart and / or one or more boxes in the block diagram.

[0062] These computer program instructions can also be loaded onto a computer or other programmable data processing device to enable it to perform a series of operation steps, thereby generating a computer-implemented process. By executing these instructions on a computer or other programmable device, the functions specified by one or more processes in the flowchart and / or one or more blocks in the block diagram can be implemented.

[0063] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention can be subjected to various modifications and variations. Any modification, equivalent replacement or improvement made within the spirit and principle of the present invention shall be deemed to be included in the protection scope of the present invention.

[0064] Although the above contents are combined with the accompanying drawings to illustrate the specific implementation of the present invention, these descriptions do not constitute a limitation on the protection scope of the present invention. Those skilled in the art should understand that any modifications or variations that can be made without creative work based on the technical solution of the present invention still fall within the protection scope of the present invention.

Claims

1. A method for rapid sorting of retired batteries based on support vector machine algorithm, characterized in that: The following steps are involved: Step 1: Perform an appearance inspection on the retired battery to see if the surface of the retired battery is damaged, leaking, or deformed, and check if the voltage of the retired battery is within the normal range. If both the appearance and voltage are qualified, proceed to step 2, otherwise, remove the battery; Step 2, performing a discharge test on the retired battery, and stopping the discharge when the retired battery voltage reaches a predetermined cut-off voltage; Step 3: Conduct external characteristic tests on retired batteries, and collect and record data; Step 4: construct an SVM binary classifier based on the collected data to roughly classify the retired batteries; Step 5: Charge the retired battery at a high rate, collect charging data in real time, and stop charging when the voltage reaches the charging cut-off voltage; Step 6, calculating the incremental capacity IC curve of the retired battery based on the charging data; Step 7, extracting the sorting characteristic data of the IC curve: IC curve peak value and corresponding voltage; Step 8: preprocess the extracted sorting feature data, remove abnormal data, and smooth the IC curve; Step nine, construct a SVM multi-classifier to further subdivide the recyclable batteries into multiple categories based on the preprocessed sorting feature data.

2. A method for rapid sorting of retired batteries based on a support vector machine algorithm as claimed in claim 1, characterized in that: The external characteristic test of the retired battery includes: measuring the open circuit voltage and ohmic internal resistance of the battery.

3. A method for rapid sorting of retired batteries based on support vector machine algorithm as claimed in claim 2, characterized in that: The SVM binary classifier is used to classify retired batteries into batteries that can be reused and batteries that can be directly recycled, and the construction steps include: S01, input sorting feature data and the number of battery types; S02, construct an SVM binary classifier, select a suitable kernel function according to the distribution and characteristics of the sorting feature data, and map the sorting feature data to a high-dimensional space for nonlinear transformation; S03, use the grid search method to tune the parameters of the SVM binary classifier; S04, set the core parameters of the SVM binary classifier, including: penalty parameter C and kernel function parameters; S05, use the selected kernel function and core parameter configuration to train the SVM binary classifier; S06, use the test data set to test the performance of the SVM binary classifier; S07, calculate the classification accuracy and precision of the SVM binary classifier, use the battery consistency parameter index, and evaluate whether it effectively divides the sample points into two categories with a health status above 70% and below 70% through the spatial distribution of the health status of the sample points.

4. A method for rapid sorting of retired batteries based on a support vector machine algorithm as claimed in claim 1, characterized in that: The charging data refers to the battery capacity and voltage change data during the charging process.

5. A method for rapid sorting of retired batteries based on support vector machine algorithm as claimed in claim 1, characterized in that: The IC curve of the retired battery is obtained by numerically deriving the capacity-voltage curve.

6. A method for rapid sorting of retired batteries based on support vector machine algorithm as claimed in claim 1, characterized in that: The sorting characteristic data include: the open circuit voltage of the battery, the ohmic internal resistance, the characteristic peak value of the IC curve and its corresponding voltage.

7. A method for rapid sorting of retired batteries based on support vector machine algorithm as claimed in claim 1, characterized in that: The preprocessing of the sorting characteristic data includes: setting a reasonable sampling interval, performing reasonable noise reduction processing on the original data, and standardizing the data to eliminate the influence between different dimensions.

8. A method for rapid sorting of retired batteries based on support vector machine algorithm as claimed in claim 1, characterized in that: The SVM multi-classifier is constructed by SVM binary classifier, where is the number of battery categories.

Images