A method for quickly detecting the SOC of a battery pack for battery replacement

Abstract

The present invention discloses a method for quickly detecting the SOC of a battery pack for battery replacement, which belongs to the field of battery detection technology and includes the following steps: S1, collecting a discharge data set of a battery replacement battery, including the voltage value of each single cell in a historical period; S2, generating a voltage description set according to the discharge data set of the battery replacement battery, and using the voltage description set to obtain a voltage prediction factor; S3, based on the voltage prediction factor, obtaining the remaining power of the battery replacement battery. The extraction of the voltage prediction factor of the present invention can determine the regularization parameter in the prediction model, thereby reducing the computational complexity of the model and improving the prediction speed. By monitoring the remaining power and performance changes of the battery, the abnormal conditions of the battery can be discovered and processed in time, thereby extending the service life of the battery.

Description

Technical Field

[0001] The present invention belongs to the technical field of battery detection, and specifically relates to a method for quickly detecting the SOC of a battery pack for battery replacement. Background Art

[0002] The battery swap mode of new energy vehicles has been widely used and promoted in the commercial vehicle market. As one of the effective ways to solve the range anxiety of electric vehicles, the efficiency and safety of battery swap technology are crucial. However, in the existing battery swap technology, the SOC detection of the battery pack is usually time-consuming and has limited accuracy, which affects the efficiency of the battery swap process and user experience. Therefore, it is of great significance to develop a fast and accurate SOC detection method. Summary of the invention

[0003] In order to solve the above problems, the present invention proposes a method for quickly detecting the SOC of a battery replacement pack.

[0004] The technical solution of the present invention is: a method for quickly detecting the SOC of a battery pack for battery replacement comprises the following steps:

[0005] S1. Collect the discharge data set of the battery replacement, including the voltage value of each single battery over a historical period of time;

[0006] S2. Generate a voltage description set based on the discharge data set of the battery replacement, and use the voltage description set to obtain a voltage prediction factor;

[0007] S3. Based on the voltage prediction factor, the remaining power of the replacement battery is obtained.

[0008] Furthermore, S2 includes the following sub-steps:

[0009] S21, constructing a measurement description matrix for each single cell according to the voltage value of each single cell during the historical period;

[0010] S22, taking the singular values ​​of a plurality of measurement description matrices as a voltage description set;

[0011] S23. Obtain a voltage prediction factor according to the voltage description set.

[0012] The beneficial effect of the above further scheme is: in the present invention, by constructing a measurement description matrix, it is possible to ensure that the data of each single cell is fully recorded, and by extracting the singular values ​​of the measurement description matrix, the high-dimensional matrix can be reduced to a lower dimension while retaining the main features of the voltage data. The singular value decomposition has a certain robustness to slight changes in the voltage data, so the obtained voltage description set is relatively stable, which helps to improve the accuracy of the prediction.

[0013] Furthermore, in S21, the expression of the measurement description matrix X of the single battery is: ; In the formula, u1 represents the voltage value of the single battery at the first historical moment, u t represents the voltage value of the single cell at the tth historical moment, u T represents the voltage value of the single cell at the Tth historical moment, u open represents the open circuit voltage, max(·) represents the maximum value operation, u charge Indicates the charging termination voltage of a single battery, u discharge Indicates the discharge termination voltage of a single cell.

[0014] Further, S23 includes the following sub-steps:

[0015] S231, introduce random indicators for voltage description set;

[0016] S232, selecting an element in the voltage description set that is greater than the random index as a characteristic index;

[0017] S233. Based on the characteristic index, all voltage description sets are input into the XGBoost model to obtain voltage prediction factors.

[0018] The beneficial effect of the above further scheme is: in the present invention, the introduction of random indicators provides a screening benchmark for the elements in the voltage description set, which helps to select features that have an important impact on prediction from a large amount of voltage description data. The introduction of random indicators reduces the subjectivity of artificial feature selection, making the feature selection process more objective and scientific.

[0019] Furthermore, in S231, the calculation formula of the random index σ of the voltage description set is: ; In the formula, τ represents the regularization coefficient, ||·|| represents the L1 norm of the matrix, and X i represents the measurement description matrix of the i-th single cell, and I represents the total number of single cells in the battery replacement battery pack.

[0020] Furthermore, in S233, each characteristic index is used as the weight of each leaf node in the XGBoost model, and all voltage description sets are input into the XGBoost model to obtain a voltage prediction factor.

[0021] The beneficial effect of the above further scheme is: in the present invention, the goal of XGBoost is to use a set of decision trees to predict a target variable, and the XGBoost model calculates the weight of each leaf node to determine the contribution of each leaf node to the final prediction result.

[0022] Furthermore, S3 includes the following sub-steps:

[0023] S31, constructing an equivalent circuit model for the battery replacement;

[0024] S32, rounding up the voltage prediction factor to serve as a regularization coefficient of the neural network;

[0025] S33. Input the discharge data set and equivalent circuit model of the battery swap into the neural network to obtain the remaining power of the battery swap.

[0026] The beneficial effect of the above further scheme is: in the present invention, the Thevenin model is a circuit model for simulating battery behavior, which simplifies the complex electrochemical characteristics of the battery into several equivalent circuit elements. Usually, machine learning algorithms (such as neural networks, etc.) are used to model and predict the voltage parameters of the battery, so that the SOC of the battery can be predicted based on the input parameters. This method does not require an in-depth understanding of the internal mechanism of the battery by opening the cover or disassembling the battery pack, and can more accurately predict the voltage response of the battery under different discharge conditions, thereby improving the accuracy of SOC prediction.

[0027] Furthermore, in S31, the Thevenin model is used to construct an equivalent circuit model for the battery replacement battery.

[0028] The beneficial effects of the present invention are as follows: the present invention collects the discharge data set of the battery replacement battery, covering various states of the battery, thereby providing a rich data basis for the subsequent generation of the voltage description set and extraction of the voltage prediction factor; in addition, the extraction of the voltage prediction factor can determine the regularization parameter in the prediction model, thereby reducing the computational complexity of the model and improving the prediction speed. By monitoring the remaining power and performance changes of the battery, the abnormal situation of the battery can be discovered and processed in time, thereby extending the service life of the battery. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 Flowchart of the method for quickly detecting the SOC of a battery swap pack. DETAILED DESCRIPTION

[0030] The embodiments of the present invention will be further described below in conjunction with the accompanying drawings.

[0031] like Figure 1 As shown, the present invention provides a method for quickly detecting the SOC of a battery pack for battery replacement, comprising the following steps:

[0032] S1. Collect the discharge data set of the battery replacement, including the voltage value of each single battery over a historical period of time;

[0033] S2. Generate a voltage description set based on the discharge data set of the battery replacement, and use the voltage description set to obtain a voltage prediction factor;

[0034] S3. Based on the voltage prediction factor, the remaining power of the replacement battery is obtained.

[0035] In this embodiment of the present invention, S2 includes the following sub-steps:

[0036] S21, constructing a measurement description matrix for each single cell according to the voltage value of each single cell during the historical period;

[0037] S22, taking the singular values ​​of a plurality of measurement description matrices as a voltage description set;

[0038] S23. Obtain a voltage prediction factor according to the voltage description set.

[0039] In the present invention, by constructing a measurement description matrix, it is possible to ensure that the data of each single cell is fully recorded. By extracting the singular values ​​of the measurement description matrix, the high-dimensional matrix can be reduced to a lower dimension while retaining the main features of the voltage data. The singular value decomposition has a certain robustness to slight changes in the voltage data, so the obtained voltage description set is relatively stable, which helps to improve the accuracy of the prediction.

[0040] In the embodiment of the present invention, in S21, the expression of the weight description matrix X of the single battery is: ; In the formula, u1 represents the voltage value of the single battery at the first historical moment, u t represents the voltage value of the single cell at the tth historical moment, u T represents the voltage value of the single cell at the Tth historical moment, u open represents the open circuit voltage, max(·) represents the maximum value operation, u charge Indicates the charging termination voltage of a single battery, u discharge Indicates the discharge termination voltage of a single cell.

[0041] In this embodiment of the present invention, S23 includes the following sub-steps:

[0042] S231, introduce random indicators for voltage description set;

[0043] S232, selecting an element in the voltage description set that is greater than the random index as a characteristic index;

[0044] S233. Based on the characteristic index, all voltage description sets are input into the XGBoost model to obtain voltage prediction factors.

[0045] In the present invention, the introduction of random indicators provides a screening benchmark for the elements in the voltage description set, which helps to select features that have an important impact on prediction from a large amount of voltage description data. The introduction of random indicators reduces the subjectivity of artificial feature selection, making the feature selection process more objective and scientific.

[0046] In the embodiment of the present invention, in S231, the calculation formula of the random index σ of the voltage description set is: ; In the formula, τ represents the regularization coefficient, ||·|| represents the L1 norm of the matrix, and X i represents the measurement description matrix of the i-th single cell, and I represents the total number of single cells in the battery replacement battery pack.

[0047] In the embodiment of the present invention, in S233, each characteristic index is used as the weight of each leaf node in the XGBoost model, and all voltage description sets are input into the XGBoost model to obtain a voltage prediction factor.

[0048] In the present invention, the goal of XGBoost is to use a set of decision trees to predict a target variable. The XGBoost model calculates the weight of each leaf node and determines the contribution of each leaf node to the final prediction result.

[0049] In this embodiment of the present invention, S3 includes the following sub-steps:

[0050] S31, constructing an equivalent circuit model for the battery replacement;

[0051] S32, rounding up the voltage prediction factor to serve as a regularization coefficient of the neural network;

[0052] S33. Input the discharge data set and equivalent circuit model of the battery swap into the neural network to obtain the remaining power of the battery swap.

[0053] In the present invention, the Thevenin model is a circuit model used to simulate battery behavior, which simplifies the complex electrochemical characteristics of the battery into several equivalent circuit elements. The voltage parameters of the battery are usually modeled and predicted using machine learning algorithms (such as neural networks, etc.), so that the SOC of the battery can be predicted based on the input parameters. This method does not require an in-depth understanding of the internal mechanism of the battery, and can more accurately predict the voltage response of the battery under different discharge conditions, thereby improving the accuracy of SOC prediction.

[0054] In an embodiment of the present invention, in S31, the Thevenin model is used to construct an equivalent circuit model for the battery replacement battery.

[0055] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A method for quickly detecting the SOC of a battery pack for battery replacement, characterized in that: The following steps are involved: S1. Collect the discharge data set of the battery replacement, including the voltage value of each single battery over a historical period of time; S2. Generate a voltage description set based on the discharge data set of the battery replacement, and use the voltage description set to obtain a voltage prediction factor; S3. Obtaining the remaining power of the replacement battery based on the voltage prediction factor; The S2 comprises the following sub-steps: S21, constructing a measurement description matrix for each single cell according to the voltage value of each single cell during the historical period; S22, taking the singular values ​​of a plurality of measurement description matrices as a voltage description set; S23, obtaining a voltage prediction factor according to the voltage description set; In S21, the expression of the measurement description matrix X of the single battery is: ; In the formula, u1 represents the voltage value of the single battery at the first historical moment, u t represents the voltage value of the single cell at the tth historical moment, u T represents the voltage value of the single cell at the Tth historical moment, u open represents the open circuit voltage, max(·) represents the maximum value operation, u charge Indicates the charging termination voltage of a single battery, u discharge Indicates the discharge termination voltage of a single cell; The S23 comprises the following sub-steps: S231, introduce random indicators for voltage description set; S232, selecting an element in the voltage description set that is greater than the random index as a characteristic index; S233, based on the characteristic index, input all voltage description sets into the XGBoost model to obtain voltage prediction factors; In S231, the calculation formula of the random index σ of the voltage description set is: ; In the formula, τ represents the regularization coefficient, ||·|| represents the L1 norm of the matrix, X i represents the measurement description matrix of the i-th single cell, and I represents the total number of single cells in the battery replacement battery pack; In S233, each characteristic index is used as the weight of each leaf node in the XGBoost model, and all voltage description sets are input into the XGBoost model to obtain a voltage prediction factor.

2. The method for rapid detection of SOC of a battery pack according to claim 1, characterized in that: The S3 comprises the following sub-steps: S31, constructing an equivalent circuit model for the battery replacement; S32, rounding up the voltage prediction factor to serve as a regularization coefficient of the neural network; S33. Input the discharge data set and equivalent circuit model of the battery swap into the neural network to obtain the remaining power of the battery swap.

3. The method for rapid detection of SOC of a battery pack according to claim 2 is characterized in that: In S31, the Thevenin model is used to construct an equivalent circuit model for the battery replacement battery.

Images