Indirect characterization index-based slight expansion prediction and early warning method and system

Abstract

The invention relates to the technical field of battery material performance prediction and monitoring, and discloses a slight expansion prediction and early warning method and system based on an indirect characterization index, and the method comprises the steps: obtaining the indirect characterization index for predicting the slight expansion of a battery material, and carrying out the preprocessing, the indirect characterization indexes comprise a voltage-capacity curve, coulombic efficiency and interface impedance which are obtained by carrying out charge-discharge circulation; inputting the plurality of preprocessed indirect characterization indexes into a set prediction model to train the prediction model; enabling the indirect characterization indexes of the battery material monitored in real time to pass through the trained prediction model to obtain the expansion rate of the battery material in each charge-discharge cycle, carrying out accumulative monitoring on the expansion rate, and carrying out early warning after the accumulative expansion rate exceeds a set first expansion rate threshold value. According to the method and the system, real-time prediction and early warning of slight expansion are realized, and the method and the system have relatively high practical value and application prospect.

Description

Technical Field

[0001] The present invention relates to the technical field of battery material performance prediction and monitoring, and in particular to a slight expansion prediction and early warning method and system based on indirect characterization indicators. Background Art

[0002] Alloy anode materials (such as silicon-based anodes) have attracted widespread attention due to their high theoretical capacity. However, during the charge and discharge process, these materials will undergo volume expansion, resulting in structural damage and performance degradation. Existing technologies mainly focus on the detection and treatment of significant expansion, and there is little research on the prediction and impact of slight expansion (≤5% volume expansion range). Traditional methods have the following shortcomings:

[0003] Slight expansion is difficult to detect: Existing detection methods are difficult to capture the microstructural changes caused by slight volume expansion.

[0004] Lack of correlation model between slight expansion and performance: The effect of slight expansion on cycling performance has not been fully studied, and there is a lack of effective correlation model.

[0005] Inability to achieve real-time monitoring and early warning: Existing methods cannot achieve real-time prediction and early warning of slight expansion using conventional electrochemical test data. Summary of the invention

[0006] The present invention provides a slight expansion prediction and early warning method and system based on indirect characterization indicators. By introducing indirect expansion effect characterization indicators, a correlation model between slight expansion and circulation performance is established to achieve real-time prediction and early warning of slight expansion.

[0007] The present invention is achieved through the following technical solutions:

[0008] A method for predicting and warning slight expansion based on indirect characterization indicators, comprising:

[0009] Obtaining indirect characterization indicators for predicting slight expansion of battery materials and preprocessing them, wherein the indirect characterization indicators include voltage-capacity curves, coulombic efficiency and interface impedance obtained by charge-discharge cycles;

[0010] Inputting a number of pre-processed indirect characterization indicators into a set prediction model to train the prediction model;

[0011] The indirect characterization index of the battery material to be monitored in real time is used by the trained prediction model to obtain the expansion rate of the battery material in each charge and discharge cycle, the expansion rate is cumulatively monitored, and an early warning is issued after the cumulative expansion rate exceeds a set first expansion rate threshold.

[0012] As an optimization, the specific process of preprocessing the voltage-capacity curves, coulombic efficiency and interface impedance is as follows:

[0013] Obtaining a voltage difference corresponding to a charge-discharge cycle and discretized data of each voltage-capacity curve according to the plurality of voltage-capacity curves;

[0014] Obtaining a coulombic efficiency change rate corresponding to a charge-discharge cycle according to the coulombic efficiencies;

[0015] Obtaining the interface impedance change rate corresponding to the charge-discharge cycle according to the plurality of interface impedances;

[0016] The voltage difference, the coulomb efficiency change rate and the interface impedance change rate are normalized.

[0017] As an optimization, the process of obtaining the voltage difference and the discretized data is as follows:

[0018] A1. Discretize the voltage-capacity curve into N discrete data with equal intervals, where N is a positive integer;

[0019] A2, taking the voltage corresponding to the same capacity in the voltage-capacity curve obtained in the current cycle and the voltage-capacity curve obtained in the previous cycle and subtracting them to obtain the voltage difference in the current cycle;

[0020] The coulombic efficiency change rate of the current cycle is the difference between the coulombic efficiency obtained in the current cycle and the coulombic efficiency obtained in the previous cycle divided by the coulombic efficiency obtained in the previous cycle;

[0021] The interface impedance change rate of the current cycle is the difference between the interface impedance obtained in the current cycle and the interface impedance obtained in the previous cycle divided by the interface impedance obtained in the previous cycle.

[0022] As an optimization, the prediction model is a hybrid model composed of CNN+LSTM, and the specific structure includes:

[0023] An input layer, used for inputting the discretized data, voltage difference, coulomb efficiency change rate and interface impedance change rate;

[0024] A feature extraction module, including a CNN network, for extracting features of the discretized data, voltage difference, coulomb efficiency change rate, and interface impedance change rate;

[0025] A timing analysis layer, comprising an LSTM unit, for obtaining a hidden state based on the features of the discretized data, the voltage difference, the coulombic efficiency change rate, and the interface impedance change rate, so as to capture the dynamic changes of the voltage difference, the coulombic efficiency change rate, and the interface impedance change rate;

[0026] A fully connected layer for mapping the hidden state to a predicted value of the expansion rate;

[0027] The output layer is used to output the predicted value of the expansion rate of the current cycle.

[0028] As an optimization, the specific process of training the prediction model is as follows:

[0029] The actual values ​​of the discretized data, voltage difference, coulomb efficiency change rate, interface impedance change rate and expansion rate of the same charge and discharge cycle number are taken as a set of sample data, thereby obtaining a data set formed by multiple sets of sample data;

[0030] Dividing the data set into a training set, a validation set, and a test set according to a set ratio;

[0031] The discretized data, voltage difference, coulomb efficiency change rate, and interface impedance change rate in the training set are used as inputs of the prediction model to obtain a predicted value of the expansion rate, and a loss function is calculated using the corresponding actual value and predicted value in the training set;

[0032] Determine whether to optimize the prediction model according to the loss function. If optimization is required, use the Adam optimizer to optimize the parameters of the prediction model. If optimization is not required, verify the prediction model through the verification set, and then test the prediction model through the test set.

[0033] As an optimization, it also includes optimizing the battery material, interface design, and battery usage conditions according to the predicted value of the expansion rate output by the prediction model.

[0034] As an optimization, a first threshold, a second threshold and a third threshold are set, wherein the first threshold is smaller than the second threshold, the second threshold is smaller than the third threshold, and the third threshold is smaller than the first expansion rate threshold.

[0035] If the predicted value is less than or equal to the first threshold, the optimization measures for the material include:

[0036] Adjusting particle morphology: Using silicon particles that are smaller or more evenly distributed than those used in the materials of the current test batteries as battery materials to reduce the non-uniformity of particle expansion and reduce the structural instability caused by expansion;

[0037] The optimization measures for interface design include:

[0038] Using carbon coating and surface coating methods to improve the interface contact between silicon and electrolyte to increase the mechanical strength and conductivity of the interface, thereby reducing the impact of expansion on battery performance;

[0039] The optimization measures for the battery use conditions include:

[0040] Increase the charge and discharge rate: Increase the charge and discharge rate to the set rate to increase the power output of the battery;

[0041] Optimize temperature control: Increase the operating temperature of the battery to optimize the battery's charge and discharge performance;

[0042] If the predicted value is greater than the first threshold and less than or equal to the second threshold, the optimization measures for the material include:

[0043] Choose silicon particles with more regular or optimized shapes than those used in the materials of the current test cells, or use silicon-carbon composite materials to increase the stability of the material and reduce volume expansion;

[0044] The optimization measures for interface design include:

[0045] Coating a conductive material with a stability higher than a set stability threshold on the surface of the silicon-based negative electrode to reduce the rupture of the SEI film, thereby reducing the impact of expansion on charge and discharge performance;

[0046] The optimization measures for the battery use conditions include:

[0047] Reduce the charging rate to reduce the pressure caused by expansion, thereby delaying the expansion effect;

[0048] Lower the battery operating temperature to avoid high temperature accelerating the expansion process and ensure that the battery materials work in an environment with a set temperature;

[0049] If the predicted value is greater than the second threshold and less than or equal to the third threshold, the optimization measures for the material include:

[0050] Replace the negative electrode material: consider using a negative electrode material with an expansion rate lower than the set second expansion rate threshold, or use a silicon-based composite material with a structural stability higher than the set stability threshold;

[0051] The optimization measures for interface design include:

[0052] Use adhesives or multi-layer interface protection films to ensure that silicon particles do not fall off or break due to expansion during the charge and discharge process;

[0053] The optimization measures for the battery use conditions include:

[0054] Limit the charge and discharge rate to within the set rate threshold;

[0055] The operating temperature of the battery is controlled within the set temperature threshold.

[0056] The present invention also discloses a slight expansion prediction and early warning system based on indirect characterization indicators, which is used to execute the aforementioned slight expansion prediction and early warning method based on indirect characterization indicators, including:

[0057] A collection preprocessing module is used to obtain and preprocess indirect characterization indicators for predicting slight expansion of battery materials, wherein the indirect characterization indicators include voltage-capacity curves, coulomb efficiency and interface impedance obtained by charge-discharge cycles;

[0058] A training module, used for inputting a number of pre-processed indirect characterization indicators into a set prediction model to train the prediction model;

[0059] The monitoring and early warning module is used to obtain the expansion rate of each charge and discharge cycle of the battery material through the trained prediction model using the indirect characterization index of the battery material being monitored in real time, cumulatively monitor the expansion rate, and issue an early warning after the cumulative expansion rate exceeds a set first expansion rate threshold.

[0060] The present invention also discloses an electronic device, characterized in that it includes at least one processor and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute a method for predicting and warning of slight expansion based on indirect characterization indicators as described above.

[0061] The present invention also discloses a storage medium storing a computer program, characterized in that when the computer program is executed by a processor, the aforementioned slight expansion prediction and early warning method based on indirect characterization indicators is implemented.

[0062] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0063] For the first time, the present invention systematically utilizes indirect characterization indicators such as voltage-capacity curve hysteresis, ICE fluctuation and interface impedance change to construct a prediction model for slight expansion. The slight expansion rate of the battery material is obtained based on the indirect characterization indicators. At the same time, a high-precision slight expansion prediction model is constructed by combining machine learning and deep learning methods, which makes up for the shortcomings of traditional models in predicting slight expansion.

[0064] The method and system of the present invention can realize real-time prediction and early warning of slight expansion, and has high practical value and application prospect.

[0065] The present invention does not require expensive equipment and complicated experiments, and prediction can be achieved only through conventional electrochemical test data, thereby achieving a low-cost and high-efficiency effect.

[0066] The method of the present invention can be applied to a variety of alloy negative electrode materials and has a wide range of applications.

[0067] The method and system of the present invention can be seamlessly integrated with the existing battery management system to improve the intelligence level of the battery. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0069] Figure 1 The present invention is a flowchart of a method for predicting and warning of slight expansion based on indirect characterization indicators. DETAILED DESCRIPTION

[0070] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0071] The present invention aims to provide a method for predicting slight expansion of alloy negative electrode materials based on cycle performance data, by introducing an indirect expansion effect characterization index, establishing a correlation model between slight expansion and cycle performance, and realizing real-time prediction and early warning of slight expansion. Next, the process of the present invention is described in detail through a specific implementation method.

[0072] This embodiment 1 provides a method for predicting and warning slight expansion based on indirect characterization indicators, such as Figure 1 As shown, including:

[0073] S1. Obtaining indirect characterization indicators for predicting slight expansion of battery materials and preprocessing them, wherein the indirect characterization indicators include obtaining a voltage-capacity curve, coulombic efficiency and interface impedance by performing charge and discharge cycles;

[0074] This step is to capture slight expansion features.

[0075] Indirect indicators such as slight hysteresis in the voltage-capacity curve, fluctuation in initial coulombic efficiency (ICE) and change in interface impedance are used to reflect the effect of slight expansion.

[0076] Microhysteresis is the voltage difference (ΔV) in the voltage-capacity curve and is usually in the range of 5-50mV. Hysteresis of 5-30mV usually indicates the first signs of slight swelling.

[0077] In some embodiments, the specific process of preprocessing the voltage-capacity curves, coulombic efficiency and interface impedance is as follows:

[0078] S1.1. Obtaining a voltage difference corresponding to a charge-discharge cycle and discretized data of each voltage-capacity curve according to the plurality of voltage-capacity curves;

[0079] The process of obtaining the voltage difference and the discretized data is as follows:

[0080] A1. Discretize the voltage-capacity curve into N discrete data with equal intervals, where N is a positive integer;

[0081] For example, through constant current charge and discharge testing, the voltage-capacity curve of each cycle is obtained and discretized into 300 equally spaced data points.

[0082] A2. Take the voltage corresponding to the same capacity in the voltage-capacity curve obtained in the current cycle and the voltage-capacity curve obtained in the previous cycle and make a difference to obtain the voltage difference (ΔV) of the current cycle.

[0083] The voltage difference (ΔV) is a voltage hysteresis characteristic, which is the voltage difference between the charging and discharging platforms.

[0084] S1.2, obtaining the coulombic efficiency change rate corresponding to the charge and discharge cycle according to the coulombic efficiencies;

[0085] The coulombic efficiency (ICE) of each cycle was recorded and the coulombic efficiency change rate was calculated.

[0086] The coulombic efficiency change rate of the current cycle is the difference between the coulombic efficiency obtained in the current cycle and the coulombic efficiency obtained in the previous cycle divided by the coulombic efficiency obtained in the previous cycle;

[0087]

[0088] Among them, R ICE,n Indicates the coulombic efficiency change rate (also called coulombic efficiency fluctuation rate) during the nth charge and discharge cycle, ICE n Indicates the coulombic efficiency at the nth charge and discharge cycle, ICE n-1 It represents the coulombic efficiency at the n-1th charge and discharge cycle;

[0089] S1.3, according to the interface impedance (R ct ) to obtain the interface impedance change rate corresponding to the charge and discharge cycle;

[0090] Electrochemical impedance spectroscopy (EIS) tests were performed regularly to extract the interface impedance value.

[0091] The interface impedance change rate of the current cycle is the difference between the interface impedance obtained in the current cycle and the interface impedance obtained in the previous cycle divided by the interface impedance obtained in the previous cycle.

[0092]

[0093] Represents the interface impedance change rate R of the nth charge and discharge cyclect_rate , represents the interfacial impedance of the nth charge-discharge cycle, represents the interface impedance of the n-1th charge-discharge cycle.

[0094] S1.4. Normalize the voltage difference, coulomb efficiency change rate and interface impedance change rate.

[0095] Normalizing all input data (voltage difference, coulomb efficiency change rate, and interface impedance change rate) to [0,1] can improve the convergence speed of subsequent prediction model training.

[0096] At the same time, set the timing sliding window:

[0097] A sliding window covering the past 10 cycles is constructed as the LSTM input to ensure that dynamic changes are captured.

[0098] S2, inputting a number of pre-processed indirect characterization indicators into a set prediction model to train the prediction model;

[0099] This step is the construction and training of the model. Specifically, it is to establish a correlation model between slight expansion and the above-mentioned indirect characterization indicators to achieve quantitative prediction of slight expansion.

[0100] First, the present invention adopts CNN+LSTM hybrid model to predict slight expansion, and the model structure is as follows:

[0101] 2.1. Input layer, used to input the discretized data, voltage difference, coulomb efficiency change rate and interface impedance change rate;

[0102] That is, there are two types of inputs of the present invention:

[0103] Input 1: Discrete point data of voltage-capacity curve.

[0104] Input 2: Cycle performance timing data (ΔV, R ICE , R ct_rate ).

[0105] Input 1 provides the microscopic characteristics of the voltage-capacity curve, and input 2 provides the macroscopic dynamic trend of the cycling performance. Input 1 and input 2 complement each other and can jointly improve the comprehensive prediction ability of the prediction model for slight expansion rates.

[0106] 2.2. A feature extraction module, including a CNN network, for extracting features of the discretized data, voltage difference, coulomb efficiency change rate, and interface impedance change rate;

[0107] The CNN network includes the following structure:

[0108] Convolutional Layer:

[0109] Convolution kernel size: 3×1

[0110] Number of convolution kernels: 64

[0111] Activation function: ReLU

[0112] Max Pooling Layer:

[0113] Pooling window size: 2×1

[0114] The purpose is to reduce dimension and extract features.

[0115] 2.3, a timing analysis layer, including an LSTM unit, for obtaining a hidden state based on the features of the discretized data, the voltage difference, the coulombic efficiency change rate, and the interface impedance change rate, so as to capture the dynamic changes of the voltage difference, the coulombic efficiency change rate, and the interface impedance change rate;

[0116] Specifically, the structure of the LSTM unit includes:

[0117] Number of units: 128;

[0118] Activation function: Tanh (input, output), Sigmoid (forgetting).

[0119] The purpose is to output hidden states and capture the dynamic changes of time series data.

[0120] 2.4. A fully connected layer is used to map the hidden state to the predicted value of the expansion rate; that is, the hidden state output from the LSTM unit is mapped to the predicted value of the slight expansion rate.

[0121] 2.5. Output layer, used to output the predicted value of the expansion rate of the current cycle.

[0122] Finally, the predicted values ​​from the first charge-discharge cycle to the current charge-discharge cycle are accumulated to obtain the cumulative expansion rate.

[0123] The purpose is to output the cumulative expansion rate for early warning analysis.

[0124] The specific process of training the prediction model is as follows:

[0125] The actual values ​​of the discretized data, voltage difference, coulomb efficiency change rate, interface impedance change rate and expansion rate of the same number of charge and discharge cycles are taken as a set of sample data, thereby obtaining a data set formed by multiple sets of sample data; the voltage difference, coulomb efficiency change rate, interface impedance change rate are collectively referred to as cycle performance data.

[0126] In this embodiment, the ratio of the training set, the validation set, and the test set is 70:15:15.

[0127] Dividing the data set into a training set, a validation set, and a test set according to a set ratio;

[0128] The discretized data, voltage difference, coulomb efficiency change rate, and interface impedance change rate in the training set are used as inputs of the prediction model to obtain a predicted value of the expansion rate, and a loss function is calculated using the corresponding actual value and predicted value in the training set;

[0129] According to the loss function, it is determined whether to optimize the prediction model. If optimization is required, the parameters of the prediction model are optimized using the Adam optimizer. If optimization is not required, the prediction model is verified by the verification set and then tested by the test set. The model parameters are updated using the Adam optimizer, and the learning rate is set to 0.001.

[0130] The loss function uses the mean square error (MSE) to calculate the error between the predicted value and the actual value of the expansion rate:

[0131]

[0132] n is the number of sample data in the test set, y i are the predicted value and actual value of the i-th sample data input into the prediction model respectively.

[0133] In each iteration, the sliding window data is input, the model weights are updated, and the training and validation errors are recorded to prevent overfitting.

[0134] When the prediction model is verified by the validation set, the discretized data, voltage difference, coulomb efficiency change rate, and interface impedance change rate in the validation set are used as inputs of the prediction model to obtain the predicted value of the expansion rate, and then the predicted value is compared with the corresponding actual value.

[0135] Similarly, when the test set verifies the prediction model, the discretized data, voltage difference, coulomb efficiency change rate, and interface impedance change rate in the test set are used as inputs of the prediction model to obtain the predicted value of the expansion rate, and then the predicted value is compared with the corresponding actual value.

[0136] S3. The indirect characterization index of the battery material to be monitored in real time is used by the trained prediction model to obtain the expansion rate of the battery material in each charge and discharge cycle, the expansion rate is cumulatively monitored, and an early warning is issued after the cumulative expansion rate exceeds a set first expansion rate threshold.

[0137] The process is to predict the trend of slight expansion by real-time monitoring of cycle performance data (voltage difference, coulomb efficiency change rate, interface impedance change rate), provide early warning, and guide material design and battery usage optimization.

[0138] This process inputs the cycle performance data into the trained prediction model, predicts the slight expansion rate of each cycle in real time, and then accumulates these predicted slight expansion rates to obtain the current total expansion rate Cumulative_Rate, that is:

[0139]

[0140] Here, the first expansion rate threshold is a safety threshold. For example, when the cumulative expansion rate is higher than 10% (safety threshold), an early warning signal is triggered.

[0141] In some embodiments, the method further includes optimizing battery materials, interface design, and battery usage conditions according to the predicted value of the expansion rate output by the prediction model.

[0142] That is, optimize material design (such as adjusting particle morphology, interface design) or battery usage conditions (such as reducing the rate, optimizing temperature) based on the predicted results.

[0143] More specifically, in this embodiment, a first threshold, a second threshold and a third threshold can be set, wherein the first threshold is smaller than the second threshold, the second threshold is smaller than the third threshold, and the third threshold is smaller than the first expansion rate threshold, that is, the third threshold is the closest to the first expansion rate threshold.

[0144] If the predicted value is less than or equal to the first threshold, it can be understood that the slight expansion is small and the expansion rate is low.

[0145] The optimization measures for materials include:

[0146] Adjusting particle morphology: Using smaller or more evenly distributed silicon particles than those used in the materials of the current test cells (i.e., using smaller or evenly distributed silicon particles) as battery materials to reduce the non-uniformity of particle expansion and reduce the structural instability caused by expansion;

[0147] The optimization measures for interface design include:

[0148] Improve the interface contact between silicon and electrolyte using methods such as carbon coating and surface coating to increase the mechanical strength and conductivity of the interface, thereby reducing the impact of expansion on battery performance;

[0149] The optimization measures for the battery use conditions include:

[0150] Increase the charge and discharge rate: Increase the charge and discharge rate to the set rate to increase the battery's power output; since slight expansion has little effect, the charge and discharge rate can be appropriately increased to increase the battery's power output.

[0151] Optimize temperature control: Properly increase the operating temperature of the battery to optimize the battery's charge and discharge performance and avoid reducing battery efficiency due to low temperature;

[0152] If the predicted value is greater than the first threshold and less than or equal to the second threshold, that is, the slight expansion is larger, the expansion rate is closer to the safety threshold.

[0153] The optimization measures for materials include:

[0154] Choose silicon particles with more regular or optimized shapes than those used in the materials of the current test cells, or use silicon-carbon composite materials to increase the stability of the material and reduce volume expansion;

[0155] Here, the optimization of the particle structure is specifically to adopt a porous silicon structure: by designing a porous structure, an internal buffer space can be provided for volume expansion, while increasing the wettability of the electrolyte and improving lithium ion transmission.

[0156] Optimization measures for interface design (enhanced interface design) include:

[0157] Coating a conductive material with a stability higher than a set stability threshold on the surface of the silicon-based negative electrode (i.e., coating a conductive material with higher stability on the surface of the silicon-based negative electrode) to reduce the rupture of the SEI film, thereby reducing the impact of expansion on the charge and discharge performance;

[0158] The optimization measures for the battery use conditions include:

[0159] Reduce the charging rate, especially in the early stages of charging, to reduce the pressure caused by expansion and thus delay the expansion effect;

[0160] Lower the battery operating temperature (i.e. appropriately lower the battery operating temperature) to avoid high temperature accelerating the expansion process and ensure that the battery material works in an environment with a set temperature, i.e. ensure that the material works in a milder environment;

[0161] If the predicted value is greater than the second threshold and less than or equal to the third threshold, the expansion is very close to the safety threshold and a large expansion occurs;

[0162] The optimization measures for materials include:

[0163] Replace the negative electrode material: consider using a negative electrode material with an expansion rate lower than the set second expansion rate threshold, or use a silicon-based composite material with a structural stability higher than the set stability threshold;

[0164] Consider using negative electrode materials with lower expansion rates, such as graphite / silicon composites, or further developing silicon-based composites with higher structural stability (e.g., silicon-carbon composites).

[0165] Optimization measures for interface design (improving interface protection) include:

[0166] Use adhesives or multi-layer interface protection films to ensure that silicon particles do not fall off or break due to expansion during the charge and discharge process;

[0167] The optimization measures for the battery use conditions include:

[0168] Strictly control the charge and discharge rate: limit the charge and discharge rate within the set rate threshold, limit the charge and discharge rate, especially use a lower rate during the charging process to slow down the expansion rate and prevent excessive expansion of the battery;

[0169] Strict temperature control: Keep the battery's operating temperature within the set temperature threshold. Optimize temperature control to ensure that the battery operates within a lower temperature range to prevent expansion and affect battery life.

[0170] Next, the method and effects of the present invention are described through a specific experimental design.

[0171] 1. Experimental design:

[0172] 1.1. Experimental objectives: To verify the ability of the CNN+LSTM model to predict the slight expansion of silicon-based negative electrode materials, analyze the prediction accuracy of the model under different cycle performance data (such as voltage hysteresis, coulombic efficiency fluctuation, and interface impedance change), and compare with the control group data.

[0173] 1.2 Experimental materials and conditions: Silicon-based negative electrode material (particle size 100 nanometers, surface coated with carbon layer); capacity 1.0 mAh / cm 2 , constant current charge and discharge rate is 0.5C; the electrolyte is 1M LiPF6 in EC / DEC (1:1v / v); the test temperature is 25℃, and the number of cycles is 0 to 50 times.

[0174] 1.3. Data collection: record the voltage-capacity curve of each cycle, calculate the charge-discharge platform voltage difference (ΔV); measure the initial coulomb efficiency (ICE) and interface impedance (R ct ).

[0175] The experimental data includes one experimental group and two control groups:

[0176] Experimental group: including ΔV, ICE fluctuation and R ct Changing data.

[0177] Control group 1: Ignore ICE fluctuation data and use only ΔV and R ct .

[0178] Control group 2: only ΔV data were used.

[0179] 2. Data and feature extraction:

[0180] Actual data of the experimental group (partial cycle performance data):

[0181] Cycle times ΔV(mV) ICE(%) <![CDATA[R ct (Oh)]]> Volume expansion rate (%) 1 25 85.0 120 0.5 5 27 84.5 123 0.8 10 30 84.2 125 1.2 20 35 83.8 130 1.8 30 42 83.5 135 2.4 50 50 83.0 145 3.6

[0182] Control group data:

[0183] Control group 1 (ignoring ICE fluctuation data) and control group 2 (using only ΔV data):

[0184] Control group 1 (partial cycle performance data):

[0185] Cycle times ΔV(mV) <![CDATA[R ct (Oh)]]> Volume expansion rate (%) 1 25 120 0.5 5 27 123 0.8 10 30 125 1.3 20 35 130 2.1 30 42 135 3 50 50 145 4.2

[0186] Control group 2 (partial cycle performance data):

[0187] Cycle times ΔV(mV) Volume expansion rate (%) 1 25 0.5 5 27 0.9 10 30 1.5 20 35 2.5 30 42 3.5 50 50 4.8

[0188] 3. Model training and prediction:

[0189] 3.1 Model training:

[0190] Input characteristics: ΔV,R ICE ,R ct The time series data of R_{ct}; the target value is the actual measured volume expansion rate.

[0191] Model settings: CNN is used to extract voltage hysteresis features, and LSTM is used to process time series data. The optimizer is Adam, the learning rate is 0.001, and the loss function is mean square error (MSE).

[0192] It should be noted that the CNN model extracts the complex characteristic patterns including ΔV in the voltage-capacity curve, while the direct input of ΔV is provided as explicit time series data, which can jointly improve the predictive ability of the prediction model as mentioned above.

[0193] Prediction results:

[0194] Cycle times Measured expansion rate (%) Predicted expansion rate of experimental group (%) Control group 1 (%) Control group 2 (%) 1 0.5 0.52 0.55 0.60 5 0.8 0.78 0.85 0.92 10 1.2 1.18 1.25 1.40 20 1.8 1.82 2.00 2.25 30 2.4 2.38 2.65 3.10 50 3.6 3.58 4.00 4.90

[0195] 4. Analysis of experimental results:

[0196] Prediction accuracy: The prediction error of the experimental group model is within ±3%, which has high accuracy. The control group 1 has a large error, ignoring ICE fluctuations, resulting in an underestimated cumulative effect. The control group 2 only uses ΔV, which has a significantly larger error and cannot effectively capture the complex effects of slight expansion.

[0197] Capturing the cumulative effect: The experimental group combines ΔV, R ICE Fluctuation and Rct_rate , which can accurately predict the cumulative trend of the expansion rate. The control group model lacks key parameters and cannot accurately characterize the long-term cumulative effect of expansion.

[0198] The above experiments show that the slight expansion prediction method based on the CNN+LSTM hybrid model of the present invention can efficiently capture the indirect expansion effect in the cycle performance data and provide accurate expansion rate prediction results. Compared with the control group that ignores key parameters, the prediction accuracy of the experimental group model is significantly improved, which fully verifies the innovation and practicality of the present invention.

[0199] Embodiment 2 further discloses a slight expansion prediction and early warning system based on indirect characterization indicators, which is used to execute a slight expansion prediction and early warning method based on indirect characterization indicators described in the embodiment, including:

[0200] A collection preprocessing module is used to obtain and preprocess indirect characterization indicators for predicting slight expansion of battery materials, wherein the indirect characterization indicators include voltage-capacity curves, coulomb efficiency and interface impedance obtained by charge-discharge cycles;

[0201] A training module, used for inputting a number of pre-processed indirect characterization indicators into a set prediction model to train the prediction model;

[0202] The monitoring and early warning module is used to obtain the expansion rate of each charge and discharge cycle of the battery material through the trained prediction model using the indirect characterization index of the battery material being monitored in real time, cumulatively monitor the expansion rate, and issue an early warning after the cumulative expansion rate exceeds a set first expansion rate threshold.

[0203] Example 3 also discloses an electronic device, characterized in that it includes at least one processor and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute a method for predicting and warning of slight expansion based on indirect characterization indicators as described in Example 1.

[0204] Embodiment 4 further discloses a storage medium storing a computer program, characterized in that when the computer program is executed by a processor, the method for predicting and warning of slight expansion based on indirect characterization indicators described in Embodiment 1 is implemented.

[0205] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for predicting and warning slight expansion based on indirect characterization indicators, characterized in that: include: Obtaining indirect characterization indicators for predicting slight expansion of battery materials and preprocessing them, wherein the indirect characterization indicators include voltage-capacity curves, coulombic efficiency and interface impedance obtained by charge-discharge cycles; Inputting a number of pre-processed indirect characterization indicators into a set prediction model to train the prediction model; The indirect characterization index of the battery material to be monitored in real time is used by the trained prediction model to obtain the expansion rate of the battery material in each charge and discharge cycle, the expansion rate is cumulatively monitored, and an early warning is issued after the cumulative expansion rate exceeds a set first expansion rate threshold.

2. A method for predicting and warning slight expansion based on indirect characterization indicators according to claim 1, characterized in that: The specific process of preprocessing the voltage-capacity curves, coulombic efficiency and interface impedance is as follows: Obtaining a voltage difference corresponding to a charge-discharge cycle and discretized data of each voltage-capacity curve according to the plurality of voltage-capacity curves; Obtaining a coulombic efficiency change rate corresponding to a charge-discharge cycle according to the coulombic efficiencies; Obtaining the interface impedance change rate corresponding to the charge-discharge cycle according to the plurality of interface impedances; The voltage difference, the coulomb efficiency change rate and the interface impedance change rate are normalized.

3. A method for predicting and warning slight expansion based on indirect characterization indicators according to claim 2, characterized in that: The process of obtaining the voltage difference and discretized data is as follows: A1. Discretize the voltage-capacity curve into N discrete data with equal intervals, where N is a positive integer; A2, taking the voltage corresponding to the same capacity in the voltage-capacity curve obtained in the current cycle and the voltage-capacity curve obtained in the previous cycle and subtracting them to obtain the voltage difference in the current cycle; The coulombic efficiency change rate of the current cycle is the difference between the coulombic efficiency obtained in the current cycle and the coulombic efficiency obtained in the previous cycle divided by the coulombic efficiency obtained in the previous cycle; The interface impedance change rate of the current cycle is the difference between the interface impedance obtained in the current cycle and the interface impedance obtained in the previous cycle divided by the interface impedance obtained in the previous cycle.

4. The method for predicting and warning slight expansion based on indirect characterization indicators according to claim 1 is characterized in that: The prediction model is a hybrid model composed of CNN+LSTM, and its specific structure includes: An input layer, used for inputting the discretized data, voltage difference, coulomb efficiency change rate and interface impedance change rate; A feature extraction module, including a CNN network, for extracting features of the discretized data, voltage difference, coulomb efficiency change rate, and interface impedance change rate; A timing analysis layer, comprising an LSTM unit, for obtaining a hidden state based on the features of the discretized data, the voltage difference, the coulombic efficiency change rate, and the interface impedance change rate, so as to capture the dynamic changes of the voltage difference, the coulombic efficiency change rate, and the interface impedance change rate; A fully connected layer for mapping the hidden state to a predicted value of the expansion rate; The output layer is used to output the predicted value of the expansion rate of the current cycle.

5. The method for predicting and warning slight expansion based on indirect characterization indicators according to claim 2 is characterized in that: The specific process of training the prediction model is as follows: The actual values ​​of the discretized data, voltage difference, coulomb efficiency change rate, interface impedance change rate and expansion rate of the same charge and discharge cycle number are taken as a set of sample data, thereby obtaining a data set formed by multiple sets of sample data; Dividing the data set into a training set, a validation set, and a test set according to a set ratio; The discretized data, voltage difference, coulomb efficiency change rate, and interface impedance change rate in the training set are used as inputs of the prediction model to obtain a predicted value of the expansion rate, and a loss function is calculated using the corresponding actual value and predicted value in the training set; Determine whether to optimize the prediction model according to the loss function. If optimization is required, use the Adam optimizer to optimize the parameters of the prediction model. If optimization is not required, verify the prediction model through the verification set, and then test the prediction model through the test set.

6. The method for predicting and warning slight expansion based on indirect characterization indicators according to claim 1 is characterized in that: The method also includes optimizing the battery material, interface design, and battery use conditions according to the predicted value of the expansion rate output by the prediction model.

7. A method for predicting and warning slight expansion based on indirect characterization indicators according to claim 6, characterized in that: A first threshold, a second threshold, and a third threshold are set, wherein the first threshold is smaller than the second threshold, the second threshold is smaller than the third threshold, and the third threshold is smaller than the first expansion rate threshold, If the predicted value is less than or equal to the first threshold, the optimization measures for the material include: Adjusting particle morphology: Using silicon particles that are smaller or more evenly distributed than those used in the materials of the current test batteries as battery materials to reduce the non-uniformity of particle expansion and reduce the structural instability caused by expansion; The optimization measures for interface design include: Using carbon coating and surface coating methods to improve the interface contact between silicon and electrolyte to increase the mechanical strength and conductivity of the interface, thereby reducing the impact of expansion on battery performance; The optimization measures for the battery use conditions include: Increase the charge and discharge rate: Increase the charge and discharge rate to the set rate to increase the power output of the battery; Optimize temperature control: Increase the operating temperature of the battery to optimize the battery's charge and discharge performance; If the predicted value is greater than the first threshold and less than or equal to the second threshold, the optimization measures for the material include: Choose silicon particles with more regular or optimized shapes than those used in the materials of the current test cells, or use silicon-carbon composite materials to increase the stability of the material and reduce volume expansion; The optimization measures for interface design include: Coating a conductive material with a stability higher than a set stability threshold on the surface of the silicon-based negative electrode to reduce the rupture of the SEI film, thereby reducing the impact of expansion on charge and discharge performance; The optimization measures for the battery use conditions include: Reduce the charging rate to reduce the pressure caused by expansion, thereby delaying the expansion effect; Lower the battery operating temperature to avoid high temperature accelerating the expansion process and ensure that the battery materials work in an environment with a set temperature; If the predicted value is greater than the second threshold and less than or equal to the third threshold, the optimization measures for the material include: Replace the negative electrode material: consider using a negative electrode material with an expansion rate lower than the set second expansion rate threshold, or use a silicon-based composite material with a structural stability higher than the set stability threshold; The optimization measures for interface design include: Use adhesives or multi-layer interface protection films to ensure that silicon particles do not fall off or break due to expansion during the charge and discharge process; The optimization measures for the battery use conditions include: Limit the charge and discharge rate to within the set rate threshold; The operating temperature of the battery is controlled within the set temperature threshold.

8. A system for predicting and warning slight expansion based on indirect characterization indicators, used to implement a method for predicting and warning slight expansion based on indirect characterization indicators as claimed in any one of claims 1 to 7, characterized in that: include: A collection preprocessing module is used to obtain and preprocess indirect characterization indicators for predicting slight expansion of battery materials, wherein the indirect characterization indicators include voltage-capacity curves, coulomb efficiency and interface impedance obtained by charge-discharge cycles; A training module, used for inputting a number of pre-processed indirect characterization indicators into a set prediction model to train the prediction model; The monitoring and early warning module is used to obtain the expansion rate of each charge and discharge cycle of the battery material through the trained prediction model using the indirect characterization index of the battery material being monitored in real time, cumulatively monitor the expansion rate, and issue an early warning after the cumulative expansion rate exceeds a set first expansion rate threshold.

9. An electronic device, characterized in that: It includes at least one processor and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute a method for predicting and warning of slight expansion based on indirect characterization indicators as described in any one of claims 1 to 7.

10. A storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, a method for predicting and warning slight expansion based on indirect characterization indicators according to any one of claims 1 to 7 is implemented.

Images