A low-cost method and system for generating laboratory life test data for lithium batteries.

Abstract

This invention discloses a low-cost method and system for generating laboratory life test data for lithium batteries. The method involves collecting complete life test data from two identical lithium batteries under two extreme laboratory life test conditions. A weighted average of the measured capacity decay curves of the two lithium batteries over the life test time is used to obtain a generated capacity decay curve. Each data point on each generated capacity decay curve is used as the starting point of the corresponding laboratory discharge curve, and the inflection point corresponding to each data point is predicted. The starting point and inflection point are combined to predict each complete laboratory discharge curve. Based on the predicted discharge curves, the corresponding laboratory charging curve, electrochemical impedance spectroscopy, and relaxation voltage curve are further predicted. This invention also discloses a system for implementing the above method. This invention uses weighted averaging to generate capacity decay curves and predict laboratory data, offering advantages such as simplicity, low cost, diverse data types, and strong controllability.

Description

Technical Field

[0001] This invention pertains to a data generation or enhancement method in the field of batteries, specifically relating to a low-cost method and system for generating laboratory life test data for lithium batteries. Background Technology

[0002] Laboratory lifetime test data for lithium-ion batteries is widely used in the development of new materials, structures, and management strategies, as well as in the study of aging mechanisms. For example, accelerated lifetime testing prediction models trained on laboratory lifetime test data have been applied to accelerate the optimization and screening of fast-charging protocols for lithium-ion batteries, reducing the lifetime test time from over 500 days to 16 days. This significantly reduces the R&D and testing costs of fast-charging protocols for lithium-ion batteries and improves R&D efficiency. However, acquiring laboratory lifetime test data for lithium-ion batteries requires enormous costs, including time, economic, and human resources. This is why currently, the number of lithium-ion batteries included in open-source lithium-ion battery laboratory datasets is at most a few hundred.

[0003] Current methods for generating lithium-ion battery laboratory data mostly employ deep learning models, requiring extensive training with real-world lithium-ion battery lifetime data. This approach is not only complex and costly, but also often only generates one type of laboratory data, such as capacity decay data, discharge data, EIS data, or relaxation voltage data. Furthermore, existing methods cannot generate complete lifetime test data covering the entire battery lifecycle, and they lack control over the lifetime testing time corresponding to the generated data. Therefore, there is an urgent need to propose a simple, low-cost, and controllable method for generating lithium-ion battery laboratory data. Summary of the Invention

[0004] The purpose of this invention is to address the problems in existing technologies by proposing a low-cost method and system for generating laboratory life test data for lithium batteries. This method employs a novel approach, differing from existing methods: it uses complete life test data from only two identical lithium batteries under two extreme laboratory life test conditions. Through weighted averaging and machine learning model prediction, it obtains complete laboratory discharge curve data for the entire life test cycle of the lithium battery. Based on the predicted laboratory discharge curve data, machine learning models can be further used to predict and obtain other types of laboratory data for the entire life test cycle of the lithium battery.

[0005] To achieve the above objectives, this invention proposes a low-cost method for generating laboratory life test data for lithium batteries, comprising the following steps: 1) Collect complete life test data for two identical lithium batteries under two extreme laboratory life test conditions; 2) The measured capacity decay curves of the two lithium batteries over the lifetime test time are weighted and averaged to obtain the generated capacity decay curve; the formula for calculating the weighted average is: ,in, Represents the generated capacity decay curve. This represents a weighted ratio value ranging from [0, 1]. and These represent two measured capacity decay curves obtained from a full life test of two identical lithium batteries. 3) Take each data point on the generated capacity decay curve as the starting point of a corresponding laboratory discharge curve, and predict the inflection point of the discharge curve. 4) Combining the starting point and the inflection point, a complete laboratory discharge curve is predicted and obtained; 5) Based on the complete laboratory discharge curve obtained by prediction, the corresponding laboratory charging curve, laboratory electrochemical impedance spectrum and laboratory relaxation voltage curve are further predicted.

[0006] Preferably, step 1) specifically includes: the laboratory life test conditions include one or more of the following: cyclic aging test, calendar aging test, dynamic load aging test, or mixed aging test; the two extreme laboratory life test conditions refer to: under the premise of keeping other operating condition parameters the same, a single operating condition parameter takes two extreme values; the complete life test data includes the lithium battery laboratory capacity measurement charge-discharge data, laboratory electrochemical impedance spectroscopy data, and laboratory relaxation voltage data corresponding to a specific life test time during the complete life test process for each lithium battery.

[0007] Preferably, in step 2), the capacity decay curve over the lifetime test time is expressed as follows: ,in Represents lithium battery capacity. The lifetime test time represents the duration of the life test; the lifetime test time includes one or more of the following: charge / discharge cycles, hours, days, or weeks; the weighting ratio... Each value obtained generates a capacity decay curve.

[0008] Preferably, step 3) specifically involves: the discharge curve being the discharge capacity vs. voltage curve of the lithium battery, i.e., the discharge... Curve; for any laboratory discharge The curve, its starting point is ( , ),in This indicates that the lithium battery is discharging in this condition. The curve corresponds to the lithium battery capacity at the life test time. The initial discharge voltage representing the laboratory life test conditions; the discharge The inflection point of the curve is ( , ),in The inflection point represents the data point on the curve with the largest absolute value of the slope; the slope is calculated using the following formula: ,in, and Indicates laboratory discharge The horizontal coordinate voltage values ​​of adjacent data points on the curve, and > ; will discharge The starting point of the curve ( , Input to the inflection point prediction model, output discharge The inflection point value of the curve; the inflection point prediction model is a machine learning model trained using measured life test data of the same type of lithium battery.

[0009] Preferably, step 4) specifically involves: discharging... The starting point of the curve ( , ) and inflection point ( , Input to laboratory discharge Curve prediction model, outputting complete laboratory discharge The predicted results of the curve; the laboratory discharge The curve prediction model is a machine learning model trained using measured life test data of the same type of lithium battery.

[0010] Preferably, step 5) specifically involves: obtaining the predicted complete laboratory discharge. The curves are respectively input to charge the laboratory. Curve prediction model, laboratory electrochemical impedance spectroscopy (EIS) prediction model, and laboratory relaxation voltage ( The curve prediction model outputs laboratory charging data respectively. Curves, laboratory EIS, and laboratory relaxation The predicted results of the curve; the laboratory charging Curve prediction model, laboratory EIS prediction model and laboratory relaxation The curve prediction models are all machine learning models trained using actual life test data of the same type of lithium battery.

[0011] This invention proposes a low-cost lithium battery laboratory life test data generation system, comprising: The lithium battery life test data acquisition unit is used to acquire complete life test data of two identical lithium batteries under two extreme laboratory life test conditions. Lithium battery capacity degradation The curve data generation unit is used to obtain a generated capacity decay curve by weighted averaging of the measured capacity decay curves of two lithium batteries over the life test time; the formula for calculating the weighted average is as follows: ,in, Represents the generated capacity decay curve. This represents a weighted ratio value ranging from [0, 1]. and These represent two measured capacity decay curves obtained from a full life test of two identical lithium batteries. Laboratory discharge The curve inflection point data prediction unit is used to take each data point on each generated capacity decay curve as the starting point of each corresponding laboratory discharge curve, and predict the inflection point of each data point corresponding to each laboratory discharge curve. Laboratory discharge Curve data prediction unit, used to combine each laboratory discharge The starting point and inflection point of the curve are used to predict and obtain each complete laboratory discharge. curve; The laboratory charging curve, laboratory electrochemical impedance spectroscopy, and laboratory relaxation voltage curve prediction unit is used to further predict the laboratory charging curve, laboratory electrochemical impedance spectroscopy, and laboratory relaxation voltage curve corresponding to each data point on each capacity decay curve based on each complete laboratory discharge curve obtained by prediction.

[0012] The present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described method.

[0013] The present invention proposes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0014] The beneficial effects of this invention are: This invention proposes a low-cost method for generating laboratory life test data for lithium batteries. Compared with existing methods for generating laboratory data for lithium batteries based on deep learning models, this method has the advantages of being simple, low-cost, generating diverse types of data, and having strong controllability of the generated data.

[0015] 1. Simple Method. The method proposed in this invention uses the simplest weighted average method to generate capacity decay curve data, and employs a basic machine learning model to predict and obtain laboratory discharge curve data and other types of laboratory data. Compared with existing methods based on complex deep learning models, this significantly reduces the development difficulty.

[0016] 2. Low cost. The method proposed in this invention only requires collecting complete life test data of two identical lithium batteries under two extreme laboratory life test conditions. Compared with existing methods based on deep learning models, which rely on a large amount of measured life data of lithium batteries, this significantly reduces development costs.

[0017] 3. Diverse Data Types Generated. The method proposed in this invention can generate various types of lithium battery laboratory data, including but not limited to capacity decay data, discharge data, EIS and relaxation voltage data, etc. Compared with existing methods based on deep learning models that can only generate a single type of data, this significantly improves the diversity of generated data types.

[0018] 4. Strong controllability of generated data. The method proposed in this invention can control the lifetime testing time corresponding to the generated data and can generate complete laboratory data throughout the entire lifetime testing cycle. Compared with existing methods based on deep learning models, which cannot control the lifetime testing time corresponding to the generated data, this significantly improves the controllability of the generated data.

[0019] The features and advantages of the present invention will be described in detail through embodiments and in conjunction with the accompanying drawings. Attached Figure Description

[0020] Figure 1 This is the overall flowchart of the present invention.

[0021] Figure 2 The measured capacity decay of two identical lithium batteries in this embodiment of the invention was obtained under two extreme laboratory life test conditions. Curve data.

[0022] Figure 3 This embodiment of the invention is based on two measured capacity decays. The generation capacity decays after weighted averaging of curve data into 1000 data points. Curve data.

[0023] Figure 4 This is a laboratory discharge in an embodiment of the present invention. The starting point on the curve ( , ) and inflection point ( , (Diagram)

[0024] Figure 5This is a laboratory discharge in an embodiment of the present invention. Inflection point on the curve ( , Select a schematic diagram.

[0025] Figure 6 This is a laboratory discharge in an embodiment of the present invention. Inflection point on the curve ( , Prediction results (main test set) graph; Figure 7 This is a laboratory discharge in an embodiment of the present invention. Inflection point on the curve ( , Prediction results (second test set) graph.

[0026] Figure 8 This is a complete laboratory discharge in an embodiment of the present invention. The predicted results of the curve are shown in the figure. Detailed Implementation

[0027] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0028] In this embodiment, the selected lithium battery life test dataset contains 124 LFP batteries of LFP chemistry, manufactured by A123, model APR18650M1A, with a rated capacity of 1.1 Ah. These 124 LFP batteries are divided into three subsets: a training set (41 batteries), a main test set (43 batteries), and a secondary test set (40 batteries). The training set is used to select complete life test data from two identical lithium batteries under two extreme laboratory life test conditions, and is used to train laboratory discharge testing. Curve prediction model, laboratory charging Curve prediction model, laboratory EIS prediction model and laboratory relaxation Curve prediction model. The main test set and secondary test set are used to verify the discharge... Curve prediction model, laboratory charging Curve prediction model, laboratory EIS prediction model and laboratory relaxation Predictive performance of the curve prediction model.

[0029] like Figure 1 As shown, the present invention includes the following steps: 1) Collect complete life test data for two identical lithium batteries under two extreme laboratory life test conditions, such as... Figure 2As shown. Laboratory life test conditions include, but are not limited to, various conditions such as cyclic aging test, calendar aging test, dynamic load aging test, and mixed aging test. In this embodiment, the laboratory life test condition is the cyclic aging test condition. The "extreme" in the two extreme laboratory life test conditions refers to taking two extreme values ​​for a single condition parameter while keeping other condition parameters the same. In this embodiment, the condition parameter taking two extreme values ​​is the charging current parameter for two charging stages, namely 2C (10%) C and 6C C, and 4.8C (80%) C and 4.8CC, corresponding to cycle lives of 170 cycles and 1800 cycles, respectively. The complete life test data includes the lithium battery laboratory capacity measurement charge-discharge data, laboratory electrochemical impedance spectroscopy data, and laboratory relaxation voltage data corresponding to a specific life test time during the complete life test process for each lithium battery. In this embodiment, the life test data is the laboratory discharge data for each charge-discharge cycle. Curve data.

[0030] 2) Weighted average the measured capacity decay curves of the two lithium batteries over time to obtain a generated capacity decay curve. Each capacity decay curve represents one lithium battery. Figure 3 As shown, the capacity decay curve over lifetime testing time can be represented as follows. ,in Represents lithium battery capacity. This represents the lifetime testing time. Lifetime testing time includes, but is not limited to, the number of charge / discharge cycles, hours, days, and weeks. The weighted average formula for generating the capacity decay curve is as follows: (1) in, Represents the generated capacity decay curve. This represents a weighted ratio value ranging from [0, 1]. and These represent two measured capacity decay curves obtained from a full life test of two identical lithium batteries. Each time a value is selected, a line can be obtained. Curve data. Theoretically, countless curves can be obtained. Curve data. In this embodiment, 1000 values ​​were selected, and one value was taken every 0.001 within the interval [0, 1]. Therefore, a total of 1000 results were obtained. Curve data.

[0031] 3) Using each data point on each generated capacity decay curve as the starting point of the corresponding laboratory discharge curve, predict the inflection point of each data point for each laboratory discharge curve, such as... Figure 4 As shown. The discharge curve is the discharge capacity vs. voltage curve of the lithium battery, abbreviated as discharge. Curve. For any discharge... The curve, its starting point is ( , ),in This represents the lithium battery discharging in this circuit. The curve corresponds to the lithium battery capacity at the life test time. The initial discharge voltage represents the laboratory life test conditions. In this embodiment, The value is 2.1 V. Discharge. The inflection point of the curve is ( , ),in Represents an inflection point. At any discharge point... On a curve, the inflection point is the data point with the largest absolute value of the slope, such as... Figure 5 As shown. The formula for calculating the slope is as follows. (2) in, and Indicates laboratory discharge The horizontal coordinate voltage values ​​of adjacent data points on the curve, and > .

[0032] Discharge The starting point of the curve ( , Input to the inflection point prediction model, output discharge The inflection point value of the curve. The inflection point prediction model is a priori pre-trained machine learning model, which is trained using measured lifespan test data of the same type of lithium battery. In this embodiment, the inflection point prediction model is an RF regression model. The test results on the main test set (43 batteries) and the secondary test set (40 batteries) are as follows: Figure 6 , Figure 7 As shown.

[0033] 4) By combining the starting point and inflection point of each laboratory discharge curve, predict and obtain each complete laboratory discharge curve. The discharge... The starting point of the curve ( , ) and inflection point ( , Input to laboratory discharge Curve prediction model, outputting complete laboratory discharge The predicted results of the curve, such as Figure 8 As shown. Laboratory discharge. The curve prediction model is a machine learning model trained using actual life test data of the same type of lithium battery. In this embodiment, laboratory discharge... The XGBoost regression model was used for curve prediction. The test results on the main test set (43 batteries) and the secondary test set (40 batteries) are shown in Table 1.

[0034] Table 1 Laboratory discharges on the main test set and secondary test set Evaluation metrics for curve prediction results 5) Based on each complete laboratory discharge curve obtained from the prediction, further predict the laboratory charge curve, laboratory electrochemical impedance spectroscopy, and laboratory relaxation voltage curve corresponding to each data point on each capacity decay curve. Then, use the predicted complete laboratory discharge curves... The curves are respectively input to charge the laboratory. Curve prediction model, laboratory electrochemical impedance spectroscopy (EIS) prediction model, and laboratory relaxation voltage ( The curve prediction model outputs laboratory charging data respectively. Curves, laboratory EIS, and laboratory relaxation The predicted results of the curve. Laboratory charging. Curve prediction model, laboratory EIS prediction model and laboratory relaxation The curve prediction models are all machine learning models trained using actual life test data of the same type of lithium battery.

[0035] The above embodiments are illustrative of the present invention and are not intended to limit the present invention. Any simple modifications to the present invention are within the scope of protection of the present invention.

Claims

1. A low-cost method for generating laboratory life test data for lithium batteries, characterized in that, Includes the following steps: 1) Collect complete life test data for two identical lithium batteries under two extreme laboratory life test conditions; 2) The measured capacity decay curves of the two lithium batteries over the lifetime test time are weighted and averaged to obtain the generated capacity decay curve; the formula for calculating the weighted average is: ,in, Represents the generated capacity decay curve. This represents a weighted ratio value ranging from [0, 1]. and These represent two measured capacity decay curves obtained from a full life test of two identical lithium batteries. 3) Take each data point on the generated capacity decay curve as the starting point of a corresponding laboratory discharge curve, and predict the inflection point of the discharge curve. 4) Combining the starting point and the inflection point, a complete laboratory discharge curve is predicted and obtained; 5) Based on the complete laboratory discharge curve obtained by prediction, the corresponding laboratory charging curve, laboratory electrochemical impedance spectrum and laboratory relaxation voltage curve are further predicted.

2. The method for generating low-cost lithium battery laboratory life test data according to claim 1, characterized in that, Step 1) specifically refers to the following: the laboratory life test conditions include one or more of the following: cyclic aging test, calendar aging test, dynamic load aging test, or mixed aging test; the two extreme laboratory life test conditions refer to the following: under the premise of keeping other operating parameters the same, one operating parameter takes two extreme values; the complete life test data includes the lithium battery laboratory capacity measurement charge and discharge data, laboratory electrochemical impedance spectroscopy data, and laboratory relaxation voltage data corresponding to a specific life test time during the complete life test process for each lithium battery.

3. The method for generating low-cost lithium battery laboratory life test data according to claim 1, characterized in that, In step 2), the capacity decay curve over the lifetime test time is expressed as follows: ,in Represents lithium battery capacity. The lifetime test time represents the duration of the life test; the lifetime test time includes one or more of the following: charge / discharge cycles, hours, days, or weeks; the weighting ratio... Each value obtained generates a capacity decay curve.

4. The method for generating low-cost lithium battery laboratory life test data according to claim 1, characterized in that, Step 3) specifically involves: the discharge curve being the discharge capacity vs. voltage curve of the lithium battery, i.e., the discharge... Curve; for any laboratory discharge The curve, its starting point is ( , ),in This indicates that the lithium battery is discharging in this condition. The curve corresponds to the lithium battery capacity at the life test time. The initial discharge voltage representing the laboratory life test conditions; the discharge The inflection point of the curve is ( , ),in The inflection point represents the data point on the curve with the largest absolute value of the slope; the slope is calculated using the following formula: ,in, and Indicates laboratory discharge The horizontal coordinate voltage values ​​of adjacent data points on the curve, and > ; will discharge The starting point of the curve ( , Input to the inflection point prediction model, output discharge The inflection point value of the curve; the inflection point prediction model is a machine learning model trained using measured life test data of the same type of lithium battery.

5. The method for generating low-cost lithium battery laboratory life test data according to claim 1, characterized in that, Step 4) specifically involves: discharging... The starting point of the curve ( , ) and inflection point ( , Input to laboratory discharge Curve prediction model, outputting complete laboratory discharge The predicted results of the curve; the laboratory discharge The curve prediction model is a machine learning model trained using measured life test data of the same type of lithium battery.

6. The method for generating low-cost lithium battery laboratory life test data according to claim 1, characterized in that, Step 5) specifically involves: obtaining the predicted complete laboratory discharge. The curves are respectively input to charge the laboratory. Curve prediction model, laboratory electrochemical impedance spectroscopy (EIS) prediction model, and laboratory relaxation voltage ( The curve prediction model outputs laboratory charging data respectively. Curves, laboratory EIS, and laboratory relaxation The predicted results of the curve; the laboratory charging Curve prediction model, laboratory EIS prediction model and laboratory relaxation The curve prediction models are all machine learning models trained using actual life test data of the same type of lithium battery.

7. A low-cost lithium battery laboratory life test data generation system, characterized in that, include: The lithium battery life test data acquisition unit is used to acquire complete life test data of two identical lithium batteries under two extreme laboratory life test conditions. Lithium battery capacity decay The curve data generation unit is used to obtain a generated capacity decay curve by weighted averaging of the measured capacity decay curves of two lithium batteries over the life test time; the formula for calculating the weighted average is as follows: ,in, Represents the generated capacity decay curve. This represents a weighted ratio value ranging from [0, 1]. and These represent two measured capacity decay curves obtained from a full life test of two identical lithium batteries. Laboratory discharge The curve inflection point data prediction unit is used to take each data point on each generated capacity decay curve as the starting point of each corresponding laboratory discharge curve, and predict the inflection point of each data point corresponding to each laboratory discharge curve. Laboratory discharge Curve data prediction unit, used to combine each laboratory discharge The starting point and inflection point of the curve are used to predict and obtain each complete laboratory discharge. curve; The laboratory charging curve, laboratory electrochemical impedance spectroscopy, and laboratory relaxation voltage curve prediction unit is used to further predict the laboratory charging curve, laboratory electrochemical impedance spectroscopy, and laboratory relaxation voltage curve corresponding to each data point on each capacity decay curve based on each complete laboratory discharge curve obtained by prediction.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method 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 a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images