A classification method and system for rematching applicable scenarios for retired batteries

Abstract

The present invention relates to the technical field of secondary utilization of retired batteries, in particular to a classification method and system for re-matching applicable scenarios for retired batteries. The proposed two-layer model structure first conducts a preliminary screening through SVM to determine whether the battery is suitable for secondary utilization, and then conducts a more refined scenario classification through Softmax regression. This structure not only reflects a hierarchical decision-making process but also allows for a more detailed matching of the battery's uses. Its uniqueness lies in that it combines the advantages of two different types of models, the high efficiency of SVM for binary classification problems, and the flexibility of the Softmax regression model in multi-classification tasks, to classify the scenarios of retired batteries that meet the conditions for secondary utilization, thereby determining the application scenarios that the retired batteries can be re-applied to. This can ensure that retired batteries are effectively allocated to the application scenarios most suitable for them, achieving the maximization of resource utilization.

Description

Technical Field

[0001] The present invention relates to the technical field of secondary utilization of retired batteries, and particularly to a classification method and system for re-matching applicable scenarios for retired batteries. Background Art

[0002] The secondary utilization of retired batteries can reduce the production cost of the new energy vehicle industry and is conducive to the large-scale popularization and application of other battery usage scenarios. The secondary utilization and recycling of batteries can create economic value and improve the utilization efficiency of resources. Retired batteries can be applied to fields such as energy storage, base station backup power, two / three-wheeled electric vehicles, photovoltaic street lights, portable mobile power supplies, and UPS power supplies. Studying how to effectively classify these batteries and use them in suitable application scenarios can meet the market demand and enhance their application value.

[0003] Through the reasonable classification and treatment of retired batteries, it is possible to prevent heavy metals and chemical substances contained in retired batteries from polluting the environment. If retired batteries are not properly treated, it may cause long-term pollution to soil, water sources, etc. At the same time, re-matching the secondary utilization scenarios for retired batteries can greatly reduce the production demand for new batteries, thereby reducing the exploitation of natural resources and environmental damage. The existing classification and matching adopt a single-layer classification scheme and use the classification results of SVM models based on three different kernel functions for decision fusion, and the model is single; secondly, this scheme proposes to allocate weights through decision fusion for classification, and the results are only simply divided into three categories: excellent, good, and general, with limited categories, and no specific secondary utilization scenarios are divided. That is to say, even if the three levels of battery quality are distinguished, it is not known to which scenarios these three levels of classification can be reused; thirdly, this scheme jointly determines the classification results based on three different kernel functions through the same classification model (SVM), which means that a battery has to be detected using three functions, increasing the classification steps and time. Summary of the Invention

[0004] In view of the problems existing in the prior art, the present invention is proposed.

[0005] Therefore, the problem to be solved by the present invention is to solve the problem of re-matching the applicable scenarios for retired batteries that meet the secondary utilization and re-entering the market.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides a classification method for re-matching applicable scenarios for retired batteries, which includes constructing a new dataset that meets the secondary utilization by collecting and organizing data on the secondary utilization of retired batteries by previous scholars;

[0008] Preprocess the dataset and then perform one - layer SVM model training;

[0009] Export the trained SVM model to a file and apply it to the classification system;

[0010] Determine the applicable scenarios for retired batteries. By collecting the battery usage requirements of national standards for each scenario, establish a dataset for two - layer model training;

[0011] Preprocess the dataset and then perform two - layer Softmax regression model training;

[0012] Export the trained Softmax regression model to a file and apply it to the classification system.

[0013] As a preferred solution of the classification method for rematching applicable scenarios for retired batteries described in the present invention, wherein: the collected data feature values include the basic information, usage history, performance parameters, and safety of retired batteries. The corresponding data categories are divided into meeting secondary utilization and not meeting secondary utilization. Subsequently, perform data cleaning on the dataset to handle missing values and outliers;

[0014] The step of performing one - layer SVM model training after preprocessing the dataset. The specific implementation of the one - layer SVM model is as follows:

[0015] Data preparation: Collect including but not limited to the basic information, usage history, performance parameters, and safety of retired batteries. Subsequently, perform data cleaning on the dataset to handle missing values and outliers;

[0016] Data cleaning: Clean the data to remove invalid or abnormal data points;

[0017] Feature selection: Select the most relevant features from the integrated dataset for input to the SVM model;

[0018] Use the Z - score normalization method to scale the features to the same scale. For each feature, the normalized feature can be expressed as:

[0019]

[0020] where, X j ′ represents the normalized feature; X j represents the feature; μ j represents the mean of the feature X j and σ j represents the standard deviation of the feature X j ;

[0021] Subsequently, correlation analysis is carried out between features to calculate the correlation coefficient matrix of features to evaluate the relationships among numerous features. The correlation coefficient represents the degree of correlation between feature i and j:

[0022]

[0023] where r ij represents the correlation coefficient, that is, the degree of correlation between feature i and j; x ik , x jk respectively represent the values of feature i and feature j in the k-th sample; represents the sample mean of feature i and feature j; n represents the number of samples, that is, the total number of observations in the dataset; k represents the sample index, a loop variable from 1 to n;

[0024] According to the correlation coefficient matrix, features with relatively high correlations are selected for retention. A threshold of 0.5 is set, that is, only features with a correlation coefficient greater than 0.5 are retained. Through the results of the correlation analysis, the features are sorted according to the proportion of importance, and then relatively important features are selected to form a subset of the dataset, which is then used as the input for the SVM binary classification model. Selecting the remaining capacity, maximum voltage, and internal resistance as the main feature vectors, the feature vector of each battery can be expressed as:

[0025] X = (x1, x2, x3) = (remaining capacity, maximum voltage, internal resistance)

[0026] Construct an SVM model:

[0027] Perform binary classification, that is, linearly separable. At this time, the objective function of this SVM model can be expressed as:

[0028]

[0029] where W is the weight vector, b is the bias term, C is the penalty parameter, and T represents the transpose of the W weight vector; in order to allow some feature points to violate the classification boundary, ε i is the slack variable;

[0030] The constraint conditions of the above objective function can be expressed as:

[0031] y i (W T X i +b) ≥ 1 - ε i , i = 1, 2,..., n

[0032] Since the data is linearly separable, there is no need to use a complex kernel function, and only a simple linear kernel function is required. Therefore, the decision function of this binary classification model is:

[0033] f(x) = sign(WT (X + b)

[0034] Classification decision:

[0035] After the model is trained, for any new battery feature vector xj, the SVM binary classification decision model can be used to determine whether the battery meets the basic conditions for cascade utilization;

[0036] If f(x)' > 0, the battery meets the cascade utilization and can enter the subsequent scenario judgment and classification; if f(x)' < 0, the battery does not meet the cascade utilization and is thus excluded and recycled for metal reuse;

[0037] In the one-layer SVM model, retired batteries whose output results do not meet the cascade utilization conditions will be directly excluded and will not enter the next link.

[0038] As a preferred solution of the classification method for re-matching applicable scenarios for retired batteries according to the present invention, the dataset is preprocessed and then trained with a two-layer Softmax regression model; the two-layer Softmax regression model is trained, and the model accuracy is continuously optimized through parameter tuning;

[0039] The Z-score standardization method is used to standardize the features to ensure that all features have the same scale. The standardized features can be expressed as:

[0040]

[0041] where X ij ' represents the j-th feature of the i-th sample, μ j represents the average value of the j-th feature, and σ j represents the standard deviation of the j-th feature;

[0042] SoftMax regression model:

[0043] Initialize the weight matrix:

[0044] W ∈ R d×K

[0045] where d is the number of features and K is the number of classes;

[0046] b represents the bias vector:

[0047] b ∈ R K

[0048] Forward propagation:

[0049] After calculating the linear combination of the features of each input class and the model weights and adding the bias vector, the expression can be:

[0050]

[0051] Among them, W is the model weight, b represents the bias vector, i represents the category, and x represents the feature;

[0052] SoftMax function:

[0053] Use the SoftMax function to convert the result of the linear combination into a normalized probability distribution:

[0054]

[0055] Among them, p(y = i|x) represents the probability that the sample belongs to category i given the input x, and K is the total number of categories;

[0056] Loss function:

[0057] Cross-entropy loss function: For each sample, the loss function can be expressed as:

[0058] L i = -log(p(y = y i |x i ))

[0059] Total loss function:

[0060] Based on the entire training set, the total loss function is the average of all cross-entropy functions, that is, the average of the losses of all samples:

[0061]

[0062] Backpropagation:

[0063] Calculate the gradient: Calculate the gradient of the loss function with respect to each weight wij and bias:

[0064]

[0065] Among them; W ij represents each individual weight; b i represents the bias gradient; δ(y k ,j) is an indicator function and is 1 when y k = i and 0 otherwise;

[0066] Parameter update:

[0067] Gradient descent: Use the gradient descent method to update the weights and biases:

[0068]

[0069] Among them, η is the learning rate.

[0070] As a preferred embodiment of the classification method for re - matching applicable scenarios for retired batteries according to the present invention, the trained two - layer Softmax regression model can classify the scenarios of retired batteries that meet the requirements of cascade utilization, that is, re - match the applicable scenarios of retired batteries, and then enter the market again, determine the data types of the data feature sets in the dataset, which is convenient for subsequent output of classification results.

[0071] As a preferred embodiment of the classification method for re - matching applicable scenarios for retired batteries according to the present invention, the to - be - tested retired battery is sent into the system, and the appearance of the to - be - classified retired battery is detected by a high - definition camera.

[0072] Immediately afterwards, the retired battery that has passed the appearance detection is sent into the battery parameter detection link, which is mainly used to detect various parameters of the retired battery, including two aspects: battery performance and reliability.

[0073] After passing through the battery detection system link, all the detection parameters of the retired battery are collected, and these parameters are input into a one - layer binary classification model for cascade utilization judgment.

[0074] Immediately afterwards, the detection parameters of the retired battery that meets the requirements of cascade utilization are input into a two - layer multi - classification model.

[0075] According to the classification results, the retired batteries are sorted into corresponding applicable scenarios.

[0076] As a preferred embodiment of the classification method for re - matching applicable scenarios for retired batteries according to the present invention, it is detected that there are no scratches, deformations, damages or liquid leakage on the appearance, no rust on the positive and negative electrodes, and the markings are correct and clear; those that meet the conditions enter the next step, and those that do not meet the conditions are excluded and not reused.

[0077] Pass through the battery detection system, which is mainly used to detect various parameters of the retired battery, and the parameters include two aspects: battery performance and reliability.

[0078] Collect all the detection parameters of the retired battery. After inputting these parameters into a one - layer binary classification model, the retired batteries whose output results do not meet the cascade utilization conditions will be directly excluded and will not enter the next link.

[0079] Among them, the range of battery performance detection parameters includes charge - discharge voltage and capacity at different temperatures, charge - discharge energy at different rates, DC internal resistance, storage performance, standard cycle life, SOC calibration, peak power, and energy efficiency.

[0080] The range of reliability detection parameters includes electrical insulation, flame retardancy, thermal management, and high temperature and high humidity.

[0081] As a preferred solution of the classification method for re - matching applicable scenarios for retired batteries according to the present invention, wherein: the detection parameters of the retired batteries that meet the requirements of cascade utilization are input into a two - layer multi - classification model, and the classification result of the retired batteries can be output, so as to obtain the applicable scenarios for the cascade utilization of the batteries, and then enter the market again;

[0082] After passing through the battery detection system, collect the detection parameters of the retired battery and number them, and use a one - layer binary classification model to screen the batteries for cascade utilization with this series of parameters;

[0083] Input the detection parameters of the retired batteries that have passed the screening into a two - layer multi - classification model, so as to obtain the re - matching result of the numbered retired batteries, and then enter the market again.

[0084] In the second aspect, an embodiment of the present invention provides a classification system for re - matching applicable scenarios for retired batteries, which includes a construction module that constructs a new data set that meets the requirements of cascade utilization by collecting and organizing the data of previous scholars' research on the cascade utilization of retired batteries;

[0085] A training module that pre - processes the data set and then trains a one - layer SVM model;

[0086] An output module that exports the trained SVM model to a file and applies it to the classification system;

[0087] A matching module that determines the applicable scenarios for retired batteries, and establishes a data set for two - layer model training by collecting the national standards on the battery usage requirements for each scenario;

[0088] A processing module that pre - processes the data set and then trains a two - layer Softmax regression model;

[0089] A classification module that exports the trained Softmax regression model to a file and applies it to the classification system.

[0090] In the third aspect, an embodiment of the present invention provides a computer device, including a memory and a processor, wherein: when the computer program instructions are executed by the processor, the steps of the classification method for re - matching applicable scenarios for retired batteries as described in the first aspect of the present invention are implemented.

[0091] In the fourth aspect, an embodiment of the present invention provides a computer - readable storage medium, on which a computer program is stored, wherein: when the computer program instructions are executed by the processor, the steps of the classification method for re - matching applicable scenarios for retired batteries as described in the first aspect of the present invention are implemented.

[0092] The beneficial effects of the present invention are as follows: It supplements the deficiencies in the back end of the existing technology. Currently, most scholars focus on researching and judging whether retired batteries meet the criteria for cascade utilization; or improving the accuracy of cascade utilization judgment parameters. Based on this research, the present invention uses a one-layer SVM binary classification model to judge whether retired batteries meet the requirements for cascade utilization. Subsequently, the batteries that meet the cascade utilization requirements are input into a two-layer Softmax regression multi-classification model, and then the applicable scenarios of the retired batteries are output, so that they can re-enter the market and be reused again.

[0093] The system structure is clear and the algorithm has strong robustness. The one-layer SVM binary classification model has been proven to provide high-accuracy classification results on various data sets. Moreover, the theoretical basis of the SVM algorithm is structural risk minimization, which enables it to perform excellently in avoiding overfitting and underfitting. Compared with other complex machine learning algorithms, SVM has only a few parameters that need to be adjusted (such as C, kernel function, and its parameters), which simplifies the model selection and parameter tuning process. The SVM algorithm can also perform well on small-sample data, which is very useful in cases where the data acquisition cost is high or the data is scarce, and its optimization problem has a mature solution (SMO); the two-layer Softmax regression multi-classification model has a simple structure, is easy to understand and implement. Different from binary classification algorithms (such as logistic regression), Softmax regression can be directly applied to multi-classification problems without an extension strategy. At the same time, as the battery utilization scenarios in the market are constantly updated, Softmax regression can also be efficiently extended to large-scale data sets. BRIEF DESCRIPTION OF THE DRAWINGS

[0094] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0095] Figure 1 It is a flowchart of a classification method for re-matching applicable scenarios for retired batteries;

[0096] Figure 2 It is a computer device diagram of a classification method for re-matching applicable scenarios for retired batteries;

[0097] Figure 3 It is a flowchart of the one-layer SVM binary classification model of the system of a classification method for re-matching applicable scenarios for retired batteries;

[0098] Figure 4 It is a flowchart of the two-layer Softmax regression multi-classification model of the system of a classification method for re-matching applicable scenarios for retired batteries. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0099] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will provide a detailed description of the specific embodiments of the present invention with reference to the accompanying drawings of the specification.

[0100] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0101] Secondly, as used herein, an "embodiment" or "embodiments" refers to specific features, structures, or characteristics that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it an embodiment that is separate or mutually exclusive of other embodiments.

[0102] Embodiment 1

[0103] Referring to Figures 1 to 2 , this is the first embodiment of the present invention. This embodiment provides a classification method for re-matching applicable scenarios for retired batteries, including:

[0104] S100: Construct a new dataset that meets the requirements of cascade utilization by collecting and organizing data on the cascade utilization of retired batteries by previous scholars.

[0105] S101: Collect data characteristic values including the basic information, usage history, performance parameters, and safety of retired batteries. The corresponding data categories are divided into those that meet the requirements of cascade utilization and those that do not. Subsequently, perform data cleaning on the dataset to handle missing values and outliers.

[0106] Preferably, the preliminary screening aims to quickly identify whether a retired battery is suitable for cascade utilization according to the national standard GB / T 34015-2017, ensuring the safety and effectiveness of subsequent utilization processes.

[0107] Appearance inspection: Visually inspect the battery for obvious physical damage, leakage, deformation, or other abnormal phenomena.

[0108] Performance test: Use professional testing equipment to perform charge and discharge tests on the battery according to the procedures specified in the standard GB / T 34015-2017, and record the performance parameters of the battery.

[0109] Passing criteria: Define the criteria for determining whether a battery is qualified, for example, the SOH of the battery needs to be higher than a certain threshold.

[0110] Treatment of unqualified batteries: Mark the batteries that do not meet the passing criteria, exclude them from subsequent cascade utilization, and directly eliminate them.

[0111] The main tests and records are as follows:

[0112] Charging test: Charge the battery according to the charging current, voltage and other conditions recommended by the standard.

[0113] Discharging test: Discharge the battery according to the discharging current, voltage and other conditions recommended by the standard.

[0114] Data recording: Record the key performance parameters of the battery such as the maximum voltage, discharging capacity, internal resistance, etc., so as to represent the SOH parameter of the battery.

[0115]

[0116] The data collection format at this time is:

[0117]

[0118] Battery number: Used to uniquely identify each battery.

[0119] New product capacity: The discharging capacity of the battery in the new product state.

[0120] Current discharging capacity: The discharging capacity of the battery in the current state.

[0121] SOC: The percentage of the state of charge of the battery calculated according to the current discharging capacity and the rated capacity of the new product.

[0122] SOH: Calculate by comparing the current performance of the battery with that of a new battery or evaluate through parameters such as the internal resistance, capacity, self-discharge rate, etc. of the battery.

[0123] In the in-depth inspection link, collect more battery information such as the maximum voltage, internal resistance, etc. through an electrical performance detector, and integrate them into a new data set for the training and testing of the SVM model to determine whether the battery meets the basic conditions for cascade utilization.

[0124] S200: After preprocessing the data set, perform one-layer SVM model training;

[0125] Preferably, many domestic retired battery detection and recycling enterprises can collect a sufficient number of data sets based on this, and divide them into a 3:7 validation set and training set, and start training the SVM binary classification model. Considering the application of this model in this system is to determine whether the battery meets the cascade utilization standard, a linear kernel is used to handle the binary classification problem.

[0126] The linear SVM binary classification process is as follows:

[0127] Data integration: Integrate the remaining capacity data collected in the preliminary screening stage with the electrical performance data collected in the in-depth inspection link to form a complete dataset in the following data format:

[0128]

[0129] S201: After preprocessing the dataset, perform one-layer SVM model training. The specific implementation of the one-layer SVM model is as follows:

[0130] Data preparation: Collect the basic information, usage history, performance parameters, and safety of retired batteries, including but not limited to these. Subsequently, clean the dataset to handle missing values and outliers;

[0131] Data cleaning: Clean the data to remove invalid or abnormal data points;

[0132] Feature selection: Select the most relevant features from the integrated dataset for input to the SVM model;

[0133] Use the Z-score normalization method to scale the features to the same scale. For each feature, the normalized feature can be expressed as:

[0134]

[0135] where, X j ′ represents the normalized feature; X j represents the feature; μ j represents the mean of feature X j ; σ j represents the standard deviation of feature X j ;

[0136] Subsequently, perform a correlation analysis between features, calculate the correlation coefficient matrix between features to evaluate the relationship between numerous features. The correlation coefficient represents the degree of correlation between feature i and j:

[0137]

[0138] where, r ij represents the correlation coefficient, that is, the degree of correlation between feature i and j; x ik and x jk represent the values of feature i and feature j in the k-th sample respectively; represents the sample means of feature i and feature j; n represents the number of samples, that is, the total number of observations in the dataset; k represents the sample index, a loop variable from 1 to n;

[0139] According to the correlation coefficient matrix, select the features with higher correlation for retention. Set a threshold of 0.5, that is, only retain the features with a correlation coefficient greater than 0.5. Through the results of correlation analysis, sort the features according to the proportion of importance, and then select the relatively important features to form a subset of the dataset, which is used as the input of the SVM binary classification model. Select the remaining capacity, maximum voltage, and internal resistance as the main feature vectors. Then, the feature vector of each battery can be expressed as:

[0140] X = (x1, x2, x3) = (remaining capacity, maximum voltage, internal resistance)

[0141] Construct an SVM model:

[0142] Perform binary classification, that is, linearly separable. At this time, the objective function of this SVM model can be expressed as:

[0143]

[0144] Among them, W is the weight vector, b is the bias term, C is the penalty parameter, and T represents the transpose of the W weight vector; in order to allow some feature points to violate the classification boundary, ε is added i as the slack variable;

[0145] The constraint conditions of the above objective function can be expressed as:

[0146] y i (W T X i +b) ≥ 1 - ε i , i = 1, 2,..., n

[0147] Since the data is linearly separable, there is no need to use a complex kernel function, and only a simple linear kernel function is required. Therefore, the decision function of this binary classification model is:

[0148] f(x) = sign(W T X + b)

[0149] Classification decision:

[0150] After the model is trained, for any new battery feature vector xj, it can be judged whether the battery meets the basic conditions for cascade utilization through this SVM binary classification decision model;

[0151] If f(x)' > 0, the battery meets the cascade utilization conditions, that is, it can enter the subsequent scenario judgment classification; if f(x)' < 0, the battery does not meet the cascade utilization conditions, that is, it is excluded and recycled for metal reuse;

[0152] In a single-layer SVM model, the retired batteries whose output results do not meet the cascade utilization conditions will be directly excluded and will not enter the next link.

[0153] Through this step, it is possible to deeply determine whether the batteries that meet the complementary energy detection requirements also meet the requirements for cascade utilization, avoiding subsequent waste of resources caused by unstable prediction, and at the same time improving the effectiveness of cascade utilization of retired batteries. So far, all the battery information that can be used for cascade utilization has been obtained, and then the process will enter the final step of re-matching the usage scenarios.

[0154] S300: Export the trained SVM model to a file and apply it to the classification system;

[0155] S400: Determine the applicable scenarios for retired batteries. By collecting the national standards on battery usage requirements for each scenario, establish a dataset for training the two-layer model;

[0156] S500: After preprocessing the dataset, train the two-layer Softmax regression model;

[0157] S501: After preprocessing the dataset, train the two-layer Softmax regression model; Train the two-layer Softmax regression model and continuously optimize the model accuracy through parameter tuning;

[0158] The two-layer multi-classification model is implemented using the Softmax regression model as follows:

[0159] Use the Z-score normalization method to normalize the features to ensure that all features have the same scale. The normalized features can be expressed as:

[0160]

[0161] where, X ij ′ represents the j-th feature of the i-th sample, μ j represents the average value of the j-th feature, and σ j represents the standard deviation of the j-th feature;

[0162] SoftMax regression model:

[0163] Initialize the weight matrix:

[0164] W ∈ R d×K

[0165] where, d is the number of features and K is the number of classes;

[0166] b represents the bias vector:

[0167] b ∈ R K

[0168] Forward propagation:

[0169] After calculating the linear combination of the features and model weights for each category in the input and adding the bias vector, the expression can be:

[0170] z i =W i T x + b i

[0171] Where W is the model weight, b represents the bias vector, i represents the category, and x represents the feature;

[0172] SoftMax function:

[0173] Use the SoftMax function to convert the result of the linear combination into a normalized probability distribution:

[0174]

[0175] Where p(y = i|x) represents the probability that a sample belongs to category i given the input x, and K is the total number of categories;

[0176] Loss function:

[0177] Cross-entropy loss function: For each sample, the loss function can be expressed as:

[0178] L i =-log(p(y = y i |x i ))

[0179] Total loss function:

[0180] Based on the entire training set, the total loss function is the average of all cross-entropy functions, that is, the average of the losses of all samples:

[0181]

[0182] Backpropagation:

[0183] Calculate the gradient: Calculate the gradient of the loss function with respect to each weight wij and bias:

[0184]

[0185] Where; W ij represents each individual weight; b i represents the bias gradient; δ(y k , j) is an indicator function and is 1 when y k =i and 0 otherwise;

[0186] Parameter update:

[0187] Gradient descent: Use the gradient descent method to update the weights and biases:

[0188]

[0189] where η is the learning rate.

[0190] S600: Export the trained Softmax regression model to a file and apply it to the classification system.

[0191] S601: The trained two-layer Softmax regression model can perform scenario classification on retired batteries that meet the requirements of cascade utilization, that is, re-match the applicable scenarios of retired batteries, and then enter the market again, determine the data type of the data feature set in the data set, and facilitate the subsequent output of classification results.

[0192] Preferably, the Softmax multi-classification regression model includes 1. Data preparation: Dataset construction: According to the national standards for battery usage requirements in each scenario, construct a dataset containing batteries that meet the conditions of cascade utilization and requirements for different application scenarios.

[0193] Assume the requirements for each scenario are as follows:

[0194] Energy storage: Remaining capacity > 2400 mAh, internal resistance < 0.06 Ω

[0195] Base station backup power: Remaining capacity > 2200 mAh, internal resistance < 0.07 Ω

[0196] Portable power supply: Remaining capacity > 2000 mAh, internal resistance < 0.08 Ω

[0197] Electric vehicle: Remaining capacity > 1800 mAh, internal resistance < 0.1 Ω

[0198] Dataset example:

[0199] Battery ID Remaining capacity (mAh) Maximum voltage (V) Internal resistance (Ω) Scenario B001 2500 3.7 0.05 Energy storage B002 2200 3.6 0.07 Base station backup power B003 2000 3.5 0.06 Portable power supply B004 1800 3.4 0.08 Electric vehicle

[0200] Feature selection: Select key performance parameters related to the application scenario, such as battery capacity, maximum voltage, internal resistance, etc.

[0201] Data preprocessing: Clean and standardize the data to ensure data quality.

[0202] 2. Observation and analysis of feature correlation

[0203] Feature selection: Refer to the feature engineering in the SVM model for feature selection, and select the most relevant features for model training according to the results of correlation analysis.

[0204] 3. Softmax regression model training

[0205] Data splitting: Divide the dataset into a training set and a test set.

[0206] Standardization processing: The Z-score standardization method is used to standardize the features.

[0207] Model initialization: Initialize the weight matrix W and the bias vector b of the model.

[0208] Model training: Use the Softmax regression model for training. The training process includes forward propagation, application of the Softmax function, calculation of the loss function, backpropagation, and parameter update.

[0209] Loss function: Use the cross-entropy loss function to measure the difference between the probability distribution predicted by the model and the true labels.

[0210] Parameter update: Use optimization algorithms such as gradient descent or the Adam optimizer to update the weights and biases to reduce the value of the loss function.

[0211] 4. Model evaluation

[0212] Test set evaluation: Use the trained model to make predictions on the test set and use appropriate evaluation metrics (such as accuracy, precision, recall, etc.) to evaluate the model performance.

[0213] 5. Model application

[0214] Model deployment: Incorporate the trained Softmax regression model into the three-layer multi-classification process of the system.

[0215] Scene classification: Input the detection parameters of the retired batteries that meet the conditions for cascade utilization screened by the SVM model into the Softmax regression model, and output the classification results of the retired batteries to determine the applicable application scenarios for the retired batteries.

[0216] S602: Send the retired battery to be tested into the system, and perform appearance detection on the retired battery to be classified through a high-definition camera;

[0217] Immediately send the retired battery that has passed the appearance detection into the battery parameter detection process, which is mainly used to detect various parameters of the retired battery, including battery performance and reliability;

[0218] After passing through the battery detection system process, collect all the detection parameters of the retired battery, input the parameters into a one-layer binary classification model, and perform cascade utilization judgment;

[0219] Immediately input the detection parameters of the retired battery that meets the cascade utilization into a two-layer multi-classification model;

[0220] According to the classification results, organize the retired batteries into the corresponding applicable scenarios.

[0221] S603: Inspect that the appearance has no scratches, deformations, damages, or liquid leakage, the positive and negative electrodes have no rust, and the markings are correct and clear; if the conditions are met, proceed to the next step; if not, reject it and do not reuse it.

[0222] Pass through the battery detection system, which is mainly used to detect various parameters of retired batteries, including battery performance and reliability.

[0223] Collect all the detection parameters of the retired battery. After inputting these parameters into a one-layer binary classification model, if the output result shows that the retired battery does not meet the conditions for cascade utilization, it will be directly rejected and will not enter the next link.

[0224] Among them, the battery performance detection parameter range includes charge and discharge voltages and capacities at different temperatures, charge and discharge energies at different rates, direct current internal resistance, storage performance, standard cycle life, SOC calibration, peak power, and energy efficiency.

[0225] The reliability detection parameter range includes electrical insulation, flame retardancy, thermal management, and high temperature and high humidity.

[0226] S604: Input the detection parameters of the retired battery that meet the cascade utilization conditions into a two-layer multi-classification model, and the classification result of the retired battery can be output, and then the applicable scenarios for the cascade utilization of the battery can be obtained, and then it enters the market again.

[0227] After passing through the battery detection system, collect the detection parameters of the retired battery and number them. Pass this series of parameters through a one-layer binary classification model to screen the batteries for cascade utilization.

[0228] Input the detection parameters of the retired battery that pass the screening into a two-layer multi-classification model, and then obtain the re-matching result of the numbered retired battery, and then enter the market again.

[0229] Preferably, after training the double-layer model, the classification system in the present invention can start to be built and run, and the specific implementation is as follows:

[0230] Send the retired battery to be tested into the system, and use a high-definition camera to perform an appearance inspection on the retired battery to be classified.

[0231] Inspect that the appearance has no scratches, deformations, damages, or liquid leakage, the positive and negative electrodes have no rust, and the markings are correct and clear; if the conditions are met, proceed to the next step; if not, reject it and do not reuse it.

[0232] Immediately send the retired battery that has passed the appearance inspection into the battery parameter detection link, which is mainly used to detect various parameters of the retired battery, including battery performance and reliability.

[0233] Among them, the battery performance detection parameter range includes but is not limited to the data features used to train the double-layer model.

[0234] After passing through the battery detection system, all the detection parameters of the retired battery are collected. These parameters are input into a one-layer binary classification model for secondary utilization judgment to determine whether the retired battery meets the requirements for secondary utilization. If it meets the requirements, it proceeds to the next step; otherwise, it is directly excluded.

[0235] Subsequently, the detection parameters of the retired battery that meet the requirements for secondary utilization are input into a two-layer multi-classification model, and the classification result of the retired battery can be output, indicating that the retired battery can be reapplied to this classification scenario and then enter the market again.

[0236] Furthermore, this embodiment also provides a classification system for re-matching applicable scenarios for retired batteries, including:

[0237] A construction module that constructs a new dataset that meets the requirements for secondary utilization by collecting and organizing data on the secondary utilization of retired batteries by previous scholars.

[0238] A training module that preprocesses the dataset and then trains a one-layer SVM model.

[0239] An output module that exports the trained SVM model to a file and deploys it in the classification system.

[0240] A matching module that determines the applicable scenarios for retired batteries and establishes a dataset for two-layer model training by collecting the national standards for battery usage requirements in each scenario.

[0241] A processing module that preprocesses the dataset and then trains a two-layer Softmax regression model.

[0242] A classification module that exports the trained Softmax regression model to a file and deploys it in the classification system.

[0243] This embodiment also provides a computer device applicable to a classification method for re-matching applicable scenarios for retired batteries, including a memory and a processor. The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement a classification method for re-matching applicable scenarios for retired batteries as proposed in the above embodiment.

[0244] The computer device may be a terminal, which includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner, and the wireless manner can be achieved through WIFI, carrier network, NFC (Near Field Communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, a trackball or a touchpad set on the shell of the computer device, or an external keyboard, touchpad or mouse, etc.

[0245] This embodiment also provides a storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the classification method for re-matching an applicable scenario for a retired battery as proposed in the above embodiment.

[0246] In summary, the purpose of the present invention is to solve the problem of re-matching the applicable scenario of retired batteries that meet the requirements of cascade utilization and re-entering the market. A classification method and system for re-matching the applicable scenario of retired batteries are provided. First, a preliminary screening of retired batteries is carried out with reference to the GB / T34015-2017 standard, mainly completing the detection of the appearance and remaining energy of retired batteries; at the same time, the appearance and remaining energy of the batteries passing the preliminary screening are recorded; then, they are sent into a trained one-layer binary classification model, in which the battery indicators are deeply detected to further determine whether the retired battery meets the retirement requirements; subsequently, the characteristic parameters of the retired batteries that can be cascaded are input into a trained two-layer multi-classification model, and then through the output category, the market application scenarios that the retired batteries can re-match are determined. Finally, through the recombination of the same type of batteries, they are reused. The present invention reduces the potential safety hazards of cascade utilization batteries, divides the scenarios of cascade utilization of retired batteries, can directly perform matching, reduces the classification steps, improves the efficiency, and provides a solution for the development of the field of retired battery detection and classification.

[0247] Embodiment 2

[0248] Refer to Figure 2 - Figure 4 This is the second embodiment of the present invention. This embodiment provides a classification method for re-matching the applicable scenario of retired batteries. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0249] To verify the effectiveness of the method for classifying applicable scenarios of retired batteries proposed in our solution, the following experiments were conducted. First, 500 retired power batteries were collected from the recycling center of a new energy vehicle manufacturer. These batteries came from different vehicle models and usage years. The experimental team conducted a preliminary screening according to the GB / T34015-2017 standard and recorded the basic information, usage history, and appearance condition of each battery.

[0250] Next, these batteries were deeply detected using professional battery testing equipment, and key parameters such as the maximum voltage, remaining capacity, internal resistance, SOC, and SOH of each battery were measured and recorded. To ensure the accuracy and consistency of the data, each test was repeated three times, and the average value was taken as the final result.

[0251] After the data collection was completed, the research team implemented the two-layer classification model of our solution using the Python programming language and the scikit-learn library. First, 80% of the data was used for training, and 20% was used for testing. In the training of the SVM model, the grid search method was used to optimize the hyperparameters, and finally, the parameter combination of C = 1.0, kernel = 'linear' was selected. For the Softmax regression model, the Adam optimizer was used, the learning rate was set to 0.001, and the number of training epochs was 100.

[0252] To compare the performance differences between the method of our solution and existing technologies, the research team also implemented three common classification methods: a single SVM model, a decision tree model, and a random forest model. All models were trained and evaluated using the same training set and test set.

[0253] After the model training was completed, the research team used the remaining 100 retired batteries as a validation set to conduct practical application tests on each model. The test process was carried out strictly according to the process described in our solution: first, an appearance inspection was conducted, then an electrical performance test was carried out, the data was input into the first-layer SVM model to determine whether it met the conditions for cascade utilization, and finally, the second-layer Softmax regression model was used to classify the specific scenarios for the batteries that met the conditions.

[0254] To comprehensively evaluate the model performance, the research team not only focused on the overall classification accuracy but also calculated the classification precision, recall rate, and F1 score for each specific scenario. At the same time, the calculation time of each method was recorded to evaluate its efficiency in practical applications. In addition, the research team also invited 5 industry experts to conduct a manual review of the classification results to evaluate the rationality and practicality of the model judgment.

[0255] The experimental results are shown in the following table:

[0256]

[0257] Through the analysis of the above experimental results, it can be clearly seen that our proposed method significantly outperforms the existing techniques in multiple key indicators, fully demonstrating its innovation and practical value.

[0258] First of all, in terms of the overall classification accuracy, our proposed method reached 94.5%, which is 5.9 percentage points higher than the second-best random forest model and 9.3 percentage points higher than the commonly used single SVM model. This significant improvement means that in practical applications, this method can more accurately identify the applicable scenarios of retired batteries, greatly reducing the economic losses and safety risks caused by misclassification.

[0259] Secondly, in terms of the classification performance in each specific application scenario, our proposed method also performed excellently. Taking the energy storage scenario as an example, the precision and recall rate of this method reached 96.2% and 95.8% respectively, far higher than other methods. This means that this method can not only accurately identify the batteries suitable for energy storage applications but also maximize the discovery of all potential energy storage batteries, fully leveraging the remaining value of retired batteries.

[0260] It is particularly worth noting that the classification performance of our proposed method is very balanced in each scenario, with an average F1 score of 0.943, indicating that this method achieves a good balance between precision and recall. In contrast, other methods either have a high precision but insufficient recall (such as the single SVM model) or a high recall but low precision (such as the decision tree model). The balanced performance of this method makes it more reliable and stable in practical applications.

[0261] In terms of computational efficiency, although our proposed method adopts a two-layer model structure, its processing speed is still quite remarkable. It only takes 2.8 seconds to process 100 batteries, which is 33.3% faster than the random forest model and only 0.7 seconds slower than the fastest decision tree model. Considering the significantly improved classification accuracy of this method, this additional computational time is completely acceptable.

[0262] Finally, from the perspective of expert evaluation, our proposed method obtained a high score of 4.7 (out of 5), far higher than other methods. This result indicates that this method not only performs well in terms of data but also is recognized by industry experts in practical applications. Experts generally believe that the classification results of this method are more reasonable and more in line with the actual engineering requirements, which is of great significance for improving the economic efficiency and safety of the cascade utilization of retired batteries.

[0263] Generally speaking, our proposed solution effectively combines the high-efficiency binary classification ability of SVM and the flexible multi-classification characteristics of Softmax regression through an innovative double-layer model structure. While ensuring high accuracy, it also achieves fast processing and scenario segmentation. This method not only improves the utilization efficiency of retired batteries but also provides a reliable decision-making tool for battery recycling enterprises, contributing to the sustainable development of the new energy vehicle industry. In the future, with the increase in data volume and further optimization of the model, this method is expected to play an important role in a wider range of application scenarios.

[0264] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A classification method for rematching applicable scenarios for retired batteries, characterized in that: including constructing a new dataset that meets the requirements of second-life utilization by collecting and organizing the data on the second-life utilization of retired batteries studied by previous scholars training a one-layer SVM model after preprocessing the dataset exporting the trained SVM model to a file and applying it to the classification system determining the applicable scenarios for retired batteries, and establishing a dataset for two-layer model training by collecting the battery usage requirements in the national standards for each scenario training a two-layer Softmax regression model after preprocessing this dataset exporting the trained two-layer Softmax regression model to a file and applying it to the classification system sending the retired battery to be tested into the system, and performing appearance detection on the retired battery to be classified through a high-definition camera immediately sending the retired battery that has passed the appearance detection into the battery parameter detection link, which is mainly used to detect various parameters of the retired battery, including battery performance and reliability after passing through the battery detection system link, collecting all the detection parameters of the retired battery, inputting these parameters into the one-layer SVM model for binary classification, and performing second-life utilization judgment immediately inputting the detection parameters of the retired battery that meets the requirements of second-life utilization into the two-layer Softmax regression model for multi-classification sorting the retired batteries to the corresponding applicable scenarios according to the classification results 2. The classification method for re-matching applicable scenarios for retired batteries according to claim 1, wherein: collecting data feature values including the basic information, usage history, performance parameters, and safety of retired batteries, and the corresponding data categories are divided into meeting the requirements of second-life utilization and not meeting the requirements of second-life utilization. Subsequently, data cleaning is performed on the dataset to handle missing values and outliers when training the one-layer SVM model after preprocessing the dataset, the specific implementation of the one-layer SVM model adopts the following steps Data preparation: Collecting, including but not limited to, the basic information, usage history, performance parameters, and safety of retired batteries. Subsequently, data cleaning is performed on the dataset to handle missing values and outliers Data cleaning: Cleaning the data to remove invalid or abnormal data points Feature selection: Selecting the most relevant features from the integrated dataset for input to the one-layer SVM model Using the Z-score standardization method to scale the features to the same scale. For each feature, the standardized feature is expressed as Among them, X j ′ represents the standardized feature; X j represents the feature; μ j represents the average value of feature X j , and σ j represents the standard deviation of feature X j ; Subsequently, performing correlation analysis between features, calculating the correlation coefficient matrix between features to evaluate the relationship between numerous features. The correlation coefficient represents the degree of correlation between feature i and j Among them, r ij represents the correlation coefficient, that is, the degree of correlation between features i and j; x ik , x jk respectively represent the values of feature i and feature j in the k-th sample; represents the sample mean of features i and j; n represents the number of samples, that is, the total number of observations in the dataset; k represents the sample index, a loop variable from 1 to n; According to the correlation coefficient matrix, retaining the features with higher correlation. Setting a threshold of 0.5, that is, only retaining the features with a correlation coefficient greater than 0.

5. Through the results of correlation analysis, sorting the features according to the proportion of importance, and then selecting relatively important features to form a subset of the dataset, so as to be used for the input of the SVM model. Selecting the remaining capacity, maximum voltage, and internal resistance as the main feature vectors, then the feature vector of each battery is expressed as X=(x1,x2,x3)=(remaining capacity, maximum voltage, internal resistance) Constructing a one-layer SVM model Performing binary classification, that is, linearly separable. At this time, the objective function of this SVM model is expressed as Among them, W is the weight vector, b is the bias term, C is the penalty parameter, and T represents the transpose of the W weight vector; in order to allow some feature points to violate the classification boundary, ε is added i is the slack variable; The constraint conditions of the above objective function are expressed as y i (W T X i +b)≥1 - ε i , i = 1, 2, ..., n where y i is the true class label of the i-th sample, which is used to train a one-layer SVM model to learn how to determine whether the battery is suitable for cascade utilization based on the characteristic parameters of the remaining capacity, maximum voltage, and internal resistance of the battery; Since the data is linearly separable, there is no need to use a complex kernel function. A simple linear kernel function is sufficient. Therefore, the decision function of this SVM model is as follows: f(x) = sign(W T X + b) Classification decision: After the model is trained, for any new battery feature vector x j , this SVM model is used to determine whether the battery meets the basic conditions for cascade utilization; If f(x)' > 0, the battery meets the requirements for cascade utilization and enters the subsequent scenario judgment and classification. If f(x)' < 0, the battery does not meet the requirements for cascade utilization and is thus removed for metal recycling. In a one-layer SVM model, retired batteries whose output results do not meet the cascade utilization conditions will be directly removed and will not enter the next step.

3. A classification method for re - matching applicable scenarios for retired batteries according to claim 2, characterized in that: The dataset is preprocessed and then used to train a two-layer Softmax regression model. Train the two-layer Softmax regression model and continuously optimize the model accuracy through parameter tuning. Use the Z-score standardization method to standardize the features to ensure that all features have the same scale. The standardized features are represented as: Among them, X ij represents the j-th feature of the i-th sample, and μ j represents the average value of the j-th feature, and σ j represents the standard deviation of the j-th feature; SoftMax regression model: Initialize the weight matrix: W ∈ R d×K where W is the weight matrix, belonging to the real number field, with a dimension of d×K ; d is the number of features, K is the number of classes; R is a real number; b represents the bias vector: b ∈ R K where R is a real number; Forward propagation: After calculating the linear combination of the features of each input class and the model weights, add the bias vector. The expression at this time is: z i = W i T x + b i where W is the model weight, b represents the bias vector, i represents the class, and x represents the feature; SoftMax function: Use the SoftMax function to convert the result of the linear combination into a normalized probability distribution: where p(y = i|x) represents the probability that the sample belongs to class i given the input x, and K is the total number of classes; Loss function: Cross-entropy loss function: For each sample, the loss function can be expressed as: L i = -log(p(y = y i |x i )) Among them, L i represents the loss function of the i-th sample; Total loss function: Based on the entire training set, the total loss function is the average of all cross-entropy functions, that is, the average of the losses of all samples: where L represents the average of the losses of all samples; N is the total number of samples in the training dataset, that is, the number of retired batteries participating in the training; Backward propagation: Calculate the gradients: Calculate the gradients of the loss function with respect to each weight W ij and bias: Among them; W ij represents each individual weight; b i represents the bias value; δ(y k , j) is an indicator function, (y k , i) is also an indicator function, x ki represents the i-th eigenvalue in the k-th sample, and is 1 when y k = i, otherwise 0; Parameter update: Gradient descent: Use the gradient descent method to update the weights and biases: where η is the learning rate.

4. A classification method for re-matching an applicable scenario for retired batteries according to claim 3, characterized in that: The trained two-layer Softmax regression model can classify the scenarios of retired batteries that meet the cascade utilization requirements, that is, re-match the applicable scenarios of the retired batteries, and then enter the market again to determine the data type of the data feature set in the dataset, facilitating the subsequent output of classification results.

5. A classification method for re-matching an applicable scenario for retired batteries according to claim 4, characterized in that: It is detected that the appearance has no scratches, deformations, damages, or leakage, the positive and negative electrodes have no rust, and the markings are correct and clear. Those that meet the conditions enter the next step, while those that do not meet the conditions are removed and not reused. Through the battery detection system, which is mainly used to detect various parameters of retired batteries. The parameters include two aspects: battery performance and reliability. Collect all the detection parameters of the retired battery. After inputting these parameters into the one-layer SVM model, retired batteries whose output results do not meet the cascade utilization conditions will be directly removed and will not enter the next step. Among them, the range of battery performance detection parameters includes charge and discharge voltages and capacities at different temperatures, charge and discharge energies at different rates, DC internal resistance, storage performance, standard cycle life, SOC calibration, peak power, and energy efficiency. The range of reliability detection parameters includes electrical insulation, flame retardancy, thermal management, and high temperature and high humidity.

6. The classification method for re-matching an applicable scenario for retired batteries according to claim 5, wherein: The detection parameters of the retired batteries that meet the requirements for cascade utilization are input into a two-layer Softmax regression model, and the classification results of the retired batteries can be output, so as to obtain the scenarios suitable for the cascade utilization of the batteries, and then enter the market for the second time; After passing through the battery detection system, the detection parameters of the retired battery are collected and numbered, and the parameters are input through a one-layer SVM model to screen the batteries for cascade utilization; The detection parameters of the retired batteries that pass the screening are input into a two-layer Softmax regression model, so as to obtain the re-matching results of the numbered retired batteries, and then re-enter the market.

7. A classification system for rematching applicable scenarios for retired batteries, based on any one of the classification methods for rematching applicable scenarios for retired batteries described in claims 1 to 6, characterized in that: It also includes, A construction module that constructs a new dataset that meets the requirements for cascade utilization by collecting and organizing the data of previous scholars' research on the cascade utilization of retired batteries; A training module that preprocesses the dataset and then trains a one-layer SVM model; An output module that exports the trained SVM model to a file and applies it to the classification system; A matching module that determines the scenarios suitable for the retired batteries, and establishes a dataset for training a two-layer Softmax regression model by collecting the battery usage requirements of each scenario in the national standard; A processing module that preprocesses the dataset and then trains a two-layer Softmax regression model; A classification module that exports the trained two-layer Softmax regression model to a file and applies it to the classification system.

8. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that: When the processor executes the computer program, it implements the steps of a classification method for re-matching suitable scenarios for retired batteries according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of a classification method for re-matching suitable scenarios for retired batteries according to any one of claims 1 to 6.

Images