A lithium battery health status assessment method based on deep learning

Abstract

The present invention discloses a method for assessing the health status of a lithium battery based on deep learning, and the present invention relates to the technical field of battery assessment. The method comprises the following steps: obtaining the Nyquist curve diagram of a lithium battery through an electrochemical impedance spectroscopy test, and mapping it one by one with internal impedance characteristic parameters to generate a sample image set. A deep learning network model is established based on the sample image set, and a prediction model is generated by training with the Nyquist curve diagram as input and the internal impedance characteristic parameters as labels. The battery to be assessed is tested, and the target Nyquist curve diagram is input to predict the internal impedance characteristic parameters. Combined with the power characteristic parameters, the capacity retention rate index, the cycle efficiency degradation index and the shape expansion influence index are calculated. By comprehensively analyzing the above indicators, a comprehensive health status coefficient is obtained, and compared with the health threshold to judge the health status of the battery. The method improves the assessment accuracy through deep learning and realizes efficient and accurate health status assessment.

Description

Technical Field

[0001] The present invention relates to the technical field of battery evaluation, and in particular to a lithium battery health status evaluation method based on deep learning. Background Art

[0002] As an important component of today's energy storage field, lithium-ion batteries are widely used in consumer electronics, electric vehicles, renewable energy storage and other fields due to their high energy density, excellent cycle performance and long life. However, as the battery is used for a longer time, the health state (SOH) of lithium batteries gradually deteriorates, and its capacity, power output and safety will be significantly affected. Lithium battery health status assessment is not only of great significance to extending battery life and optimizing equipment operating efficiency, but also a key link to ensure equipment operation stability and user safety.

[0003] At present, the health status assessment methods of lithium batteries are mainly divided into two categories: traditional methods based on empirical models and intelligent methods based on data-driven. In traditional methods, battery capacity testing or internal resistance measurement are often used to judge the health status. However, these methods have certain limitations. For example, capacity testing requires a long period of full charge and discharge operation of the battery, which is time-consuming and difficult to adapt to actual application scenarios; and the assessment method based on internal resistance measurement is too dependent on a single parameter and it is difficult to fully reflect the health status of the battery, especially under complex working conditions. Large errors may occur. In addition, due to the complex electrochemical processes inside lithium batteries, the degradation of battery performance is not only related to capacity and internal resistance, but also involves electrolyte decomposition, changes in the thickness of the solid electrolyte interface film (SEI film), and increased lithium ion diffusion resistance. Many factors. Therefore, it is difficult to fully capture the health status of the battery by monitoring a single parameter.

[0004] To overcome the above problems, data-driven methods based on deep learning are becoming a new trend in the field of lithium battery health status assessment. Deep learning algorithms can efficiently process high-dimensional data, extract complex features, and use large-scale data for pattern recognition and prediction. However, in current research, how to combine deep learning technology with electrochemical impedance spectroscopy testing and effectively correlate electrochemical characteristic parameters with health assessment results still faces a technical gap. Therefore, how to achieve a comprehensive analysis and high-precision assessment of the electrochemical characteristics and health status of lithium batteries based on deep learning technology has become a current research hotspot and technical difficulty.

[0005] In the prior art, the publication number CN106353687B discloses a method for evaluating the health status of a lithium battery, the method comprising: determining evaluation factors of the health status of a lithium battery; calculating the initial value of the weight value of the evaluation factor of the health status of a lithium battery; calculating the actual value of the weight value of the evaluation factor of the health status of a lithium battery; and evaluating the health status of a lithium battery. The present invention integrates the terminal voltage change rate, ohmic internal resistance and polarization internal resistance of a lithium battery as evaluation factors, thereby improving the accuracy of the health status evaluation of a lithium battery; adopts a method of simultaneously measuring and calculating the terminal voltage change rate, ohmic internal resistance and polarization internal resistance of a lithium battery after the pulse discharge ends, thereby ensuring the state identity and time consistency of the lithium battery state measurement, thereby improving the accuracy of the evaluation factors of the health status of a lithium battery; and integrates the influence of the standard deviation and average quantity index of the evaluation factors, thereby better reflecting the overall sign variation of the evaluation factors at different levels. However, this method only uses the terminal voltage change rate, ohmic internal resistance and polarization internal resistance as key factors for evaluating the health status. Although these factors are closely related to the health status of the battery, they cannot fully reflect the overall health status of the lithium battery. The health status of lithium batteries is affected by many factors, such as capacity decay, temperature factors, etc. Therefore, evaluating only based on these parameters will reduce the accuracy and effectiveness of the evaluation results.

[0006] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0007] The purpose of the present invention is to provide a lithium battery health status assessment method based on deep learning to solve the problems raised in the above background technology.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] A method for evaluating the health status of a lithium battery based on deep learning, the specific steps include:

[0010] Conducting electrochemical impedance spectroscopy tests on several lithium batteries with known internal impedance characteristic parameters to obtain corresponding Nyquist curves, generating a sample image set based on the obtained Nyquist curves, and mapping each Nyquist curve with the corresponding internal impedance characteristic parameter one by one and storing them in the sample image set, wherein the internal impedance characteristic parameters include electrolyte ohmic impedance, solid electrolyte phase interface film impedance, and diffusion impedance;

[0011] Based on the sample image set, a deep learning network model is established. Several Nyquist curves in the sample image set are used as inputs of the deep learning network model. The internal impedance characteristic parameters corresponding to each curve are used as labels to train the deep learning network model and obtain an internal impedance characteristic parameter prediction model.

[0012] Conduct an electrochemical impedance spectroscopy test on the lithium battery to be evaluated to obtain a target Nyquist curve diagram, input the target Nyquist curve diagram into the trained internal impedance characteristic parameter prediction model to obtain the internal impedance characteristic parameters of the lithium battery to be evaluated, and at the same time conduct a full charge and discharge test on the lithium battery to be evaluated to record the corresponding power characteristic parameters;

[0013] According to the obtained internal impedance characteristic parameters of the lithium battery to be evaluated, the capacity retention rate index, the cycle efficiency degradation index and the shape expansion influence index are calculated in combination with the corresponding power characteristic parameters, wherein the power characteristic parameters include the start and end time of full charge and discharge, the flowing current, the corresponding voltage and the maximum surface temperature;

[0014] Based on the capacity retention rate index, cycle efficiency degradation index and shape expansion influence index, a comprehensive analysis is conducted to obtain the comprehensive health status coefficient of the lithium battery to be evaluated. The comprehensive health status coefficient of the lithium battery to be evaluated is compared with the lithium battery health threshold value. According to different comparison results, the health status of the lithium battery to be evaluated is judged.

[0015] Furthermore, electrochemical impedance spectroscopy tests are performed on several lithium batteries with known internal impedance characteristic parameters to obtain corresponding Nyquist curves, wherein the steps of performing the electrochemical impedance spectroscopy test include: connecting the positive electrode and the negative electrode of the battery to be tested to the working electrode and the reference electrode end of the electrochemical workstation respectively; applying a frequency signal to perform the test from high frequency to low frequency; setting the AC signal amplitude; setting the DC bias voltage; automatically applying an AC signal to the battery after starting the test, and gradually scanning the set frequency range, while generating a Nyquist curve.

[0016] Furthermore, based on the long short-term memory network model LSTM model, a deep learning network model is established, and the activation function and optimization algorithm are selected, wherein the Tanh function is selected as the activation function, and Adam is selected as the optimization algorithm of the LSTM model; the formula of the Tanh function is:

[0017]

[0018] In the formula, f(r) represents the Tanh function, and the independent variable r represents the weighted sum of the neuron's input, that is, the result of the weighted summation of the input received by the neuron from the previous layer;

[0019] At the same time, the hyperparameters of the LSTM model are set, and the hyperparameters of the LSTM model include: the number of network layers, the number of iterations, the learning rate, the batch size, the number of training times, the batch processing number, and the number of hidden layer neurons;

[0020] The number of network layers is set to 3 layers, the number of iterations is set to 200, the learning rate is set to 0.001, the batch size is set to 32, the number of training times is set to 100, the batch size is set to 256, and the number of hidden layer neurons is 32;

[0021] The input of the trained internal impedance characteristic parameter prediction model is the Nyquist curve diagram of the lithium battery, and the output is the internal impedance characteristic parameter of the lithium battery.

[0022] Furthermore, the capacity retention rate index is calculated based on the obtained internal impedance characteristic parameters of the lithium battery to be evaluated and the corresponding power characteristic parameters, wherein the formula for calculating the capacity retention rate index is:

[0023]

[0024] Where RAF is the capacity retention index, C actual is the actual capacity of the lithium battery to be evaluated, C rated is the rated capacity of the lithium battery to be evaluated, R int is the current internal resistance of the lithium battery to be evaluated, R int0 is the initial internal resistance of the lithium battery to be evaluated, ηc is the Coulomb efficiency of the lithium battery to be evaluated, k 1 is the Coulomb efficiency adjustment constant;

[0025] The actual capacity of the lithium battery to be evaluated is C actual Based on the calculation of power characteristic parameters, the specific calculation formula is as follows:

[0026]

[0027] In the formula, I g (t) represents the charging current at time t during the full charging process, t ga and t gb are the start and end time of the full charging process respectively, y(T) is the temperature correction factor, and the formula for calculating the temperature correction factor y(T) is:

[0028] y(T)=1-β*(T ref -T)

[0029] Where, T ref is the reference temperature, T is the maximum surface temperature of the lithium battery to be evaluated during full charging, and β is the temperature sensitivity coefficient.

[0030] Furthermore, the current internal resistance R of the lithium battery to be evaluated int The calculation is based on the formula:

[0031] R int =R ohm+R sei +R diff

[0032] In the formula, R ohm , R sei and R diff They are the predicted values ​​of electrolyte ohmic impedance, solid electrolyte interface film impedance and diffusion impedance of the lithium battery to be evaluated output by the model respectively;

[0033] The coulombic efficiency ηc of the lithium battery to be evaluated is calculated based on the formula:

[0034]

[0035] In the formula, C charge The fully discharged capacity of the lithium battery to be evaluated is calculated based on the formula:

[0036]

[0037] In the formula, t ha and t hb are the start and end time of the full discharge of the lithium battery to be evaluated, I h (t) is the discharge current at time t during the complete discharge process.

[0038] Furthermore, the cycle efficiency degradation index is calculated based on the obtained internal impedance characteristic parameters of the lithium battery to be evaluated and the corresponding power characteristic parameters, wherein the cycle efficiency degradation index is calculated based on the formula:

[0039]

[0040] Wherein, EDI is the cycle efficiency degradation index, ηE is the energy efficiency of the lithium battery to be evaluated; the energy efficiency ηE of the lithium battery to be evaluated is calculated based on the formula:

[0041]

[0042] Where V h (t) and V g (t) are the discharge voltage and charge voltage at time t during the complete discharge and complete charge processes, respectively;

[0043] The formula for calculating the shape expansion influence index is:

[0044] PDI=GD 2 *ln(1+T)

[0045] Wherein, PDI is the external expansion impact index, and GD is the maximum expansion volume of the lithium battery to be evaluated during the complete charge and discharge process.

[0046] Furthermore, based on the capacity retention rate index, the cycle efficiency degradation index and the shape expansion influence index, a comprehensive analysis is performed to obtain the comprehensive health coefficient of the lithium battery to be evaluated, wherein the formula for calculating the comprehensive health coefficient is:

[0047]

[0048] In the formula, CHI is the comprehensive coefficient of health status, ω 1 ,ω 2 and ω 3 are the weight coefficients of the shape expansion influence index, cycle efficiency degradation index and capacity retention rate index, respectively, where ω 1 >ω 2 ≥ω 3 And ω 1 ,ω 2 and ω 3 All are greater than 0;

[0049] The comprehensive health status coefficient of the lithium battery to be evaluated is compared with the health threshold of the lithium battery, and the health status of the lithium battery to be evaluated is judged according to different comparison results. The logic for judging the health status of the lithium battery to be evaluated is:

[0050] When CHI≥0.7*yz, the health status of the lithium battery to be evaluated is judged to be excellent, indicating that the lithium battery should be used normally;

[0051] When 0.4*yz≤CHI<0.7*yz, the health status of the lithium battery to be evaluated is judged to be good, indicating that the lithium battery should be repaired or replaced;

[0052] When 0≤CHI<0.4*yz, the health status of the lithium battery to be evaluated is judged to be poor, indicating that the lithium battery cannot be used any further;

[0053] Where yz is the set lithium battery health threshold.

[0054] Compared with the prior art, the present invention has the following beneficial effects:

[0055] First, this scheme significantly improves the extraction and prediction accuracy of electrochemical characteristic parameters of lithium batteries through deep learning modeling of Nyquist curves. The traditional equivalent circuit model fitting method has human subjectivity and computational complexity in parameter extraction, while the deep learning method can automatically extract key features in the curve to achieve efficient prediction of parameters such as electrolyte ohmic impedance, solid electrolyte interface (SEI) film impedance and diffusion impedance. It reduces the impact of human intervention and significantly improves the accuracy and robustness of parameter calculation. Secondly, the electrochemical impedance spectroscopy test is combined with the power characteristic parameters to comprehensively consider health status indicators in multiple dimensions such as capacity retention rate index, cycle efficiency degradation index and shape expansion influence index. It captures the multi-faceted characteristics of the health status of lithium batteries more comprehensively and avoids the possible errors in single parameter evaluation. In addition, the health status assessment method based on the deep learning model in this scheme has extremely high adaptability and scalability. Through the continuous expansion and optimization of the training sample set, the model can adapt to the characteristics of different types of lithium batteries and diversified operating conditions, and has good versatility. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a schematic diagram of the overall method flow of the present invention. DETAILED DESCRIPTION

[0057] 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 specific embodiments.

[0058] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0059] Example:

[0060] See also Figure 1 , the present invention provides a technical solution:

[0061] A method for evaluating the health status of a lithium battery based on deep learning, the specific steps include:

[0062] Step 1: Perform electrochemical impedance spectroscopy tests on several lithium batteries with known internal impedance characteristic parameters to obtain corresponding Nyquist curves, generate a sample image set based on the obtained Nyquist curves, and map each Nyquist curve with the corresponding internal impedance characteristic parameter one by one and store them in the sample image set, wherein the internal impedance characteristic parameters include electrolyte ohmic impedance, solid electrolyte phase interface film impedance and diffusion impedance.

[0063] Electrochemical impedance spectroscopy tests are performed on several lithium batteries with known internal impedance characteristic parameters to obtain corresponding Nyquist curves, wherein the steps of performing the electrochemical impedance spectroscopy test include: connecting the positive electrode and the negative electrode of the battery to be tested to the working electrode and the reference electrode end of the electrochemical workstation respectively; applying a frequency signal to perform the test from high frequency to low frequency; setting the AC signal amplitude; setting the DC bias voltage; automatically applying an AC signal to the battery after starting the test, and gradually scanning the set frequency range, and generating a Nyquist curve at the same time.

[0064] Among them, the electrochemical workstation needs to have equipment with electrochemical impedance testing function, such as an electrochemical workstation or impedance analyzer, and it also needs to have a battery clamp to ensure a firm connection between the lithium battery and the equipment during the test to avoid poor contact.

[0065] Before conducting the electrochemical impedance spectroscopy test, the battery needs to be left to stand for a period of time (usually 1 hour or longer) to allow the electrochemical reaction inside the battery to reach a stable state and eliminate the transient effects caused by charging and discharging; set the amplitude of the AC signal (usually a small signal of 5 to 10mV) to ensure that the internal state of the battery is not disturbed during the test; set the frequency scanning range of the test, generally from high frequency (such as 100kHz) to low frequency (such as 10mHz). The high-frequency region mainly reflects the ohmic impedance of the electrolyte, the medium-frequency region reflects the impedance of the solid electrolyte phase interface film, and the low-frequency region reflects the diffusion impedance; finally, start the electrochemical workstation and run the electrochemical impedance spectroscopy test program. Apply an AC signal to the battery within the frequency range and measure the response to generate a Nyquist curve.

[0066] Step 2: Based on the sample image set, a deep learning network model is established. Several Nyquist curve graphs in the sample image set are used as inputs of the deep learning network model. The internal impedance characteristic parameters corresponding to each curve graph are used as labels to train the deep learning network model and obtain an internal impedance characteristic parameter prediction model.

[0067] Based on the long short-term memory network model LSTM model, a deep learning network model is established, and the activation function and optimization algorithm are selected. The Tanh function is selected as the activation function, and Adam is selected as the optimization algorithm of the LSTM model; the formula of the Tanh function is:

[0068]

[0069] In the formula, f(r) represents the Tanh function, and the independent variable r represents the weighted sum of the neuron's input, that is, the result of the weighted summation of the input received by the neuron from the previous layer;

[0070] At the same time, the hyperparameters of the LSTM model are set, and the hyperparameters of the LSTM model include: the number of network layers, the number of iterations, the learning rate, the batch size, the number of training times, the batch processing number, and the number of hidden layer neurons;

[0071] The number of network layers is set to 3 layers, the number of iterations is set to 200, the learning rate is set to 0.001, the batch size is set to 32, the number of training times is set to 100, the batch size is set to 256, and the number of hidden layer neurons is 32;

[0072] The input of the trained internal impedance characteristic parameter prediction model is the Nyquist curve diagram of the lithium battery, and the output is the internal impedance characteristic parameter of the lithium battery.

[0073] The Nyquist curve is drawn based on electrochemical impedance spectroscopy (EIS) data. Each point corresponds to an impedance value (a complex number consisting of real and imaginary parts) at a specific frequency. The points on the curve are arranged in order of frequency, so it can be regarded as a time series data with a certain order dependence. The shape of the Nyquist curve reflects the dynamic response characteristics of the electrochemical system at different frequencies. LSTM can capture these dynamic changes and associate them with target parameters (such as ohmic impedance or diffusion impedance). The impedance data in the Nyquist curve may have complex nonlinear relationships, such as mutual influence between different frequency bands. LSTM has strong nonlinear expression capabilities, and its deep neural network structure can model these complex relationships well. When predicting internal impedance parameters, LSTM can learn the nonlinear mapping relationship between frequency and impedance parameters based on the overall shape of the curve.

[0074] At the same time, the number of points of the Nyquist curve may vary depending on the test frequency range or resolution, resulting in inconsistent lengths of input curve data. LSTM naturally supports processing sequence data of different lengths, which makes the model more flexible.

[0075] Step 3: Perform an electrochemical impedance spectroscopy test on the lithium battery to be evaluated to obtain a target Nyquist curve diagram, input the target Nyquist curve diagram into the trained internal impedance characteristic parameter prediction model to obtain the internal impedance characteristic parameters of the lithium battery to be evaluated, and at the same time, perform a full charge and discharge test on the lithium battery to be evaluated and record the corresponding power characteristic parameters.

[0076] The steps for obtaining the start and end time of full charge and discharge are as follows: Start time: when the charge or discharge process starts, record the timestamp; Start of charging: the battery is connected to the current through the battery tester or DC power supply, the voltage gradually rises, and the start time is recorded; Start of discharging: the battery is connected to the electronic load device or tester, the current starts to flow out, and the time is recorded.

[0077] End of charging: Usually ends when the battery voltage reaches the maximum charging voltage (such as 4.2V) and the current drops to the set threshold (such as 0.05C), and usually ends when the battery voltage drops to the minimum discharge voltage (such as 2.5V). Most battery test instruments will automatically record the time data of charging and discharging. Commonly used equipment includes professional battery testing equipment such as Neware, Ametek, and Arbin.

[0078] The battery tester has built-in high-precision sensors that can record current and voltage in real time. The tester usually records current and voltage data at a high sampling frequency (such as 1Hz or even higher) to generate corresponding time series curves (such as voltage-time curve, current-time curve), which are displayed in real time through the test software interface. The data can also be exported for subsequent analysis to obtain the current flowing through and the corresponding voltage.

[0079] The method of obtaining the maximum surface temperature includes: fixing the thermocouple on the battery surface and monitoring the temperature change in real time. The data of the thermocouple can be connected to the computer through the data acquisition module; infrared temperature measuring instrument: non-contact measurement of the temperature distribution on the battery surface can reduce the interference of contact. The temperature sensor or infrared measuring instrument can record the temperature change over time through the data acquisition software, record the temperature change curve, and mark the highest temperature during the charging and discharging process.

[0080] Step 4: Based on the obtained internal impedance characteristic parameters of the lithium battery to be evaluated, the capacity retention rate index, cycle efficiency degradation index and shape expansion influence index are calculated in combination with the corresponding power characteristic parameters. The power characteristic parameters include the start and end time of full charge and discharge, the flowing current, the corresponding voltage and the maximum surface temperature.

[0081] The capacity retention rate index is calculated based on the obtained internal impedance characteristic parameters of the lithium battery to be evaluated and the corresponding power characteristic parameters, wherein the formula for calculating the capacity retention rate index is:

[0082]

[0083] Where RAF is the capacity retention index, C actual is the actual capacity of the lithium battery to be evaluated, C rated is the rated capacity of the lithium battery to be evaluated, R intis the current internal resistance of the lithium battery to be evaluated, R int0 is the initial internal resistance of the lithium battery to be evaluated, ηc is the Coulomb efficiency of the lithium battery to be evaluated, k 1 is the Coulomb efficiency adjustment constant;

[0084] It should be noted that the capacity retention index RAF is used to comprehensively consider the current actual capacity and internal resistance information of the lithium battery to be evaluated, and indicates the capacity of the lithium battery. The larger the capacity retention index RAF value is, the better the capacity of the lithium battery is maintained, the closer it is to the rated capacity, and the healthier the battery is.

[0085] The actual capacity of the lithium battery to be evaluated is C actual , reflects the capacity decay of the battery during use. Capacity is one of the most direct indicators of battery health. After long-term use, the capacity of the battery will gradually decay. Under normal circumstances, the actual capacity of the battery decreases with the increase of the number of cycles. Therefore, the current actual capacity of the lithium battery is C actual The higher it is, the closer it is to the rated capacity and the healthier the battery is, so C actual It is proportional to the capacity retention index and is expressed by the ratio to the rated capacity of the lithium battery to be evaluated. Indicates a proportional relationship.

[0086] The current internal resistance R of the lithium battery to be evaluated int It refers to the internal resistance of the battery in its current state, reflecting the loss of the battery in overcoming electrochemical reactions and ion transmission during the charge and discharge process. Changes in internal resistance are usually a precursor to capacity decay. Increased internal resistance will reduce the output power and energy efficiency of the battery. Increased internal resistance will cause the battery voltage to drop faster, limiting the operating voltage range of the battery, thereby affecting the battery life. Increased internal resistance will also increase heat generation, affecting the thermal stability of the battery and possibly causing safety problems. Therefore, the larger the internal resistance, the worse the health of the battery. Therefore, the current internal resistance R of the lithium battery to be evaluated is int It is inversely proportional to the capacity retention index RAF, through the exponential function It indicates an inverse relationship. The effect of internal resistance change on battery performance is nonlinear. Therefore, the exponential form used in the formula is more in line with the actual attenuation law of battery performance. It also indicates that when the internal resistance increases, the capacity retention rate index RAF decreases significantly.

[0087] The coulombic efficiency ηc of the lithium battery to be evaluated refers to the energy utilization efficiency of the battery in a complete charge and discharge cycle. The reduction of coulombic efficiency indicates that the side reactions (such as electrolyte decomposition, SEI film formation, and byproduct deposition) are more serious, which directly affects the battery capacity retention ability. Low coulombic efficiency may cause overcharge or over-discharge risks, aggravate battery performance degradation and even cause accidents. Therefore, the larger the coulombic efficiency ηc, the better the battery health status. Therefore, the coulombic efficiency ηc is proportional to the capacity retention rate index RAF, which can be expressed in the quadratic form (1-ηc) 2 Characterizes a proportional relationship, used to emphasize that as Coulombic efficiency decreases, its impact on battery health will be exacerbated.

[0088] Among them, the rated capacity C of the lithium battery to be evaluated rated The rated capacity refers to the amount of electricity that a lithium battery can store under standard test conditions. It is usually provided by the manufacturer, who will calibrate the rated capacity C of the battery. rated , usually in "mAh" or "Ah", recorded in the battery's technical specification. The initial internal resistance R of the lithium battery to be evaluated int0 , is the internal resistance of the battery when it is just shipped or put into use for the first time, and it represents the baseline state of the battery. The initial internal resistance of lithium batteries is usually tested by the manufacturer during quality control and recorded in the technical specification. The unit of internal resistance is usually "mΩ".

[0089] Coulomb efficiency adjustment constant k 1 The weight of this item is introduced to allow adjustment to adapt to different working conditions, and the general value range is 0 to 1.

[0090] The actual capacity of the lithium battery to be evaluated is C actual Based on the calculation of power characteristic parameters, the specific calculation formula is as follows:

[0091]

[0092] In the formula, I g (t) represents the charging current at time t during the full charging process, t ga and t gb are the start and end time of the full charging process respectively, y(T) is the temperature correction factor, and the formula for calculating the temperature correction factor y(T) is:

[0093] y(T)=1-β*(T ref -T)

[0094] Where, T ref is the reference temperature, T is the maximum surface temperature of the lithium battery to be evaluated during full charging, and β is the temperature sensitivity coefficient.

[0095] Temperature has a significant effect on battery capacity. Low temperature (such as 0°C or lower) slows down the diffusion of lithium ions in the electrolyte, resulting in a decrease in capacity. High temperature (over 45°C) may accelerate side reactions, causing rapid aging of the battery. Therefore, the reference temperature T ref The battery capacity is corrected, with the reference temperature T ref Generally, it is 25°C. The temperature sensitivity coefficient β can be set according to the temperature tolerance of different lithium batteries and combined with expert experience. The value range is generally between 0 and 0.65.

[0096] The current internal resistance R of the lithium battery to be evaluated int The calculation is based on the formula:

[0097] R int =R ohm +R sei +R diff

[0098] In the formula, R ohm , R sei and R diff They are the predicted values ​​of electrolyte ohmic impedance, solid electrolyte interface film impedance and diffusion impedance of the lithium battery to be evaluated output by the model respectively;

[0099] The coulombic efficiency ηc of the lithium battery to be evaluated is calculated based on the formula:

[0100]

[0101] In the formula, C charge The fully discharged capacity of the lithium battery to be evaluated is calculated based on the formula:

[0102]

[0103] Where, t ha and t hb are the start and end time of the full discharge of the lithium battery to be evaluated, I h (t) is the discharge current at time t during the complete discharge process.

[0104] According to the obtained internal impedance characteristic parameters of the lithium battery to be evaluated, the cycle efficiency degradation index is calculated and generated in combination with the corresponding power characteristic parameters, wherein the formula for calculating the cycle efficiency degradation index is:

[0105]

[0106] Where EDI is the cycle efficiency degradation index, and ηE is the energy efficiency of the lithium battery to be evaluated.

[0107] The cycle efficiency degradation index EDI characterizes the battery cycle efficiency by comprehensively considering the solid electrolyte phase interface film impedance, energy efficiency and Coulomb efficiency. The higher the cycle efficiency degradation index EDI is, the higher the efficiency of the battery energy utilization and the better the health of the battery.

[0108] Where, the solid electrolyte interface film impedance prediction value R sei , represents the impedance of the SEI film. The SEI film is a passivation film formed on the surface of the battery negative electrode due to the decomposition of the electrolyte. Its existence can protect the negative electrode from further side reactions, but the thickness of the film increases with the increase of cycles, resulting in an increase in impedance. The predicted value of the solid electrolyte interface film impedance R sei The increase of R will directly increase the ion transport resistance in the electrochemical reaction, reducing the energy and power output efficiency of the battery. sei It may also cause excessive heat inside the battery, further deteriorating other performances, so the solid electrolyte interface film impedance prediction value R sei The larger the value, the worse the battery health status is. Therefore, it is inversely proportional to the cycle efficiency degradation index EDI. Indicates an inverse relationship, when R sei The larger the value is, the significantly smaller the cycle efficiency degradation index EDI is, indicating that the battery health state is rapidly deteriorating.

[0109] T represents the highest surface temperature of the lithium battery during full charging. Temperature is a key factor affecting the performance of lithium batteries. High temperature will intensify the side reactions inside the battery (such as electrolyte decomposition, SEI film growth, etc.), thereby accelerating capacity decay and increasing internal resistance. High temperature environment may also destroy the structural integrity of active materials and even cause thermal runaway, which seriously affects the safety of the battery. Therefore, the larger the T, the worse the health of the battery. Therefore, T is inversely proportional to EDI. The inverse relationship is expressed by the logarithm ln(1+T) in the denominator, describing the nonlinear effect of temperature on battery performance. Especially at high temperatures, the acceleration effect of temperature increase on chemical reactions is more obvious, but as the temperature gradually increases, the effect gradually decreases.

[0110] The energy efficiency ηE of the lithium battery to be evaluated refers to the ratio of the energy output to the energy input during a cycle. (1-ηE) reflects the degree of energy loss. The larger the value, the lower the energy efficiency. Therefore, (1-ηE) is inversely proportional to the cycle efficiency degradation index. The coulomb efficiency is similar to the energy efficiency and will not be elaborated here.

[0111] The formula for calculating the energy efficiency ηE of the lithium battery to be evaluated is:

[0112]

[0113] Where V h (t) and Vg (t) are the discharge voltage and charge voltage at time t during the complete discharge and complete charge processes, respectively;

[0114] The formula for calculating the shape expansion influence index is:

[0115] PDI=GD 2 *ln(1+T)

[0116] Wherein, PDI is the external expansion impact index, and GD is the maximum expansion volume of the lithium battery to be evaluated during the complete charge and discharge process.

[0117] Among them, PDI is the shape expansion influence index. The larger the value, the greater the deformation of the battery during use, the greater the risk of use, and therefore the worse the health of the battery.

[0118] The maximum expansion volume GD of the lithium battery to be evaluated during the complete charge and discharge process. During the charge and discharge process of the lithium battery, the negative electrode material undergoes lithium insertion and delithiation reactions, and the electrode may change in volume. At the same time, side reactions (such as electrolyte decomposition, SEI film growth, etc.) and gas generation will also cause the internal pressure of the battery to increase, causing the shape to expand. The expansion of the battery may affect the contact between the electrode and the electrolyte, reduce the efficiency of the electrochemical reaction, and thus affect the energy density and cycle life. Excessive expansion may damage the battery casing and even cause internal short circuits, making the battery at risk of fire or explosion. Therefore, the larger the maximum expansion volume GD, the greater the battery safety hazard and the worse the health status. Therefore, GD is proportional to PDI, and the impact of the expansion volume on battery performance and safety performance is not a linear relationship. When the expansion volume is small, its impact on performance may be limited; but when the expansion volume exceeds a certain critical value, the consequences may deteriorate sharply (such as internal short circuits, casing ruptures, etc.). Through the square term GD 2 The setting can better reflect the nonlinear characteristics of the impact of expansion degree on battery performance.

[0119] Temperature is an important factor affecting expansion. Increased temperature will intensify the chemical and physical changes inside the battery, leading to increased expansion. Therefore, temperature not only directly affects expansion, but may also indirectly amplify its negative impact through other factors acting on expansion.

[0120] Among them, during the complete charge and discharge process, the maximum expansion volume of the lithium battery to be evaluated is obtained in the following way: using high-precision displacement measuring equipment, measuring the displacement changes on the surface of the battery casing, and calculating the expanded volume in combination with the geometric dimensions of the battery; a high-precision laser displacement sensor can monitor the expansion displacement of the battery surface at different positions during the charge and discharge process in real time, and calculate the expansion volume based on the displacement changes in combination with the battery shape structure (such as a plane shape or a cylindrical shape); or using three-dimensional scanning technology to record the changes in the surface contour of the battery before and after charge and discharge, and then calculate the expansion volume.

[0121] Step 5: Based on the capacity retention rate index, cycle efficiency degradation index and shape expansion impact index, a comprehensive analysis is performed to obtain the comprehensive health status coefficient of the lithium battery to be evaluated, and the comprehensive health status coefficient of the lithium battery to be evaluated is compared with the lithium battery health threshold value. According to different comparison results, the health status of the lithium battery to be evaluated is judged.

[0122] Based on the capacity retention rate index, cycle efficiency degradation index and shape expansion impact index, a comprehensive analysis is conducted to obtain the comprehensive health status coefficient of the lithium battery to be evaluated. The formula for calculating the comprehensive health status coefficient is:

[0123]

[0124] In the formula, CHI is the comprehensive coefficient of health status, ω 1 ,ω 2 and ω 3 are the weight coefficients of the shape expansion influence index, cycle efficiency degradation index and capacity retention rate index, respectively, where ω 1 >ω 2 ≥ω 3 And ω 1 ,ω 2 and ω 3 All are greater than 0;

[0125] Among them, since the greater the shape expansion impact index, the greater the danger of using the battery, and the cycle efficiency degradation index and the capacity retention rate index are only differences in battery performance, the weight coefficient of the shape expansion impact index is the largest, and the cycle efficiency degradation index directly indicates the efficiency of battery use, and the capacity retention rate index indicates the use of the battery through the size of the capacity, so the weight coefficient of the cycle efficiency degradation index is set slightly larger than the weight coefficient of the capacity retention rate index, and is finally set to ω 1 >ω 2 ≥ω 3 And ω 1 ,ω 2 and ω 3 Both are greater than 0.

[0126] The comprehensive health status coefficient of the lithium battery to be evaluated is compared with the health threshold of the lithium battery, and the health status of the lithium battery to be evaluated is judged according to different comparison results. The logic for judging the health status of the lithium battery to be evaluated is:

[0127] When CHI≥0.7*yz, the health status of the lithium battery to be evaluated is judged to be excellent, indicating that the lithium battery should be used normally;

[0128] When 0.4*yz≤CHI<0.7*yz, the health status of the lithium battery to be evaluated is judged to be good, indicating that the lithium battery should be repaired or replaced;

[0129] When 0≤CHI<0.4*yz, the health status of the lithium battery to be evaluated is judged to be poor, indicating that the lithium battery cannot be used any further;

[0130] Where yz is the set lithium battery health threshold.

[0131] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0132] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0133] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0134] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A lithium battery health status assessment method based on deep learning, characterized in that: The specific steps include: Conducting electrochemical impedance spectroscopy tests on several lithium batteries with known internal impedance characteristic parameters to obtain corresponding Nyquist curves, generating a sample image set based on the obtained Nyquist curves, and mapping each Nyquist curve with the corresponding internal impedance characteristic parameter one by one and storing them in the sample image set, wherein the internal impedance characteristic parameters include electrolyte ohmic impedance, solid electrolyte phase interface film impedance, and diffusion impedance; Based on the sample image set, a deep learning network model is established. Several Nyquist curves in the sample image set are used as inputs of the deep learning network model. The internal impedance characteristic parameters corresponding to each curve are used as labels to train the deep learning network model and obtain an internal impedance characteristic parameter prediction model. Conduct an electrochemical impedance spectroscopy test on the lithium battery to be evaluated to obtain a target Nyquist curve diagram, input the target Nyquist curve diagram into the trained internal impedance characteristic parameter prediction model to obtain the internal impedance characteristic parameters of the lithium battery to be evaluated, and at the same time conduct a full charge and discharge test on the lithium battery to be evaluated to record the corresponding power characteristic parameters; According to the obtained internal impedance characteristic parameters of the lithium battery to be evaluated, the capacity retention rate index, the cycle efficiency degradation index and the shape expansion influence index are calculated in combination with the corresponding power characteristic parameters, wherein the power characteristic parameters include the start and end time of full charge and discharge, the flowing current, the corresponding voltage and the maximum surface temperature; Based on the capacity retention rate index, cycle efficiency degradation index and shape expansion impact index, a comprehensive health status comprehensive coefficient of the lithium battery to be evaluated is obtained by comprehensive analysis, and the comprehensive health status coefficient of the lithium battery to be evaluated is compared with the lithium battery health threshold value, and the health status of the lithium battery to be evaluated is judged according to different comparison results; The formula for calculating the comprehensive coefficient of health status is: In the formula, CHI is the comprehensive coefficient of health status, ω1, ω2 and ω3 are the weight coefficients of the shape expansion influence index, cycle efficiency degradation index and capacity retention index respectively, where ω1>ω2≥ω3 and ω1, ω2 and ω3 are all greater than 0, EDI is the cycle efficiency degradation index, RAF is the capacity retention index, and PDI is the shape expansion influence index; The comprehensive health status coefficient of the lithium battery to be evaluated is compared with the health threshold of the lithium battery, and the health status of the lithium battery to be evaluated is judged according to different comparison results. The logic for judging the health status of the lithium battery to be evaluated is: When CHI≥0.7*yz, the health status of the lithium battery to be evaluated is judged to be excellent, indicating that the lithium battery should be used normally; When 0.4*yz≤CHI<0.7*yz, the health status of the lithium battery to be evaluated is judged to be good, indicating that the lithium battery should be repaired or replaced; When 0≤CHI<0.4*yz, the health status of the lithium battery to be evaluated is judged to be poor, indicating that the lithium battery cannot be used any further; Where yz is the set lithium battery health threshold.

2. The method for evaluating the health status of a lithium battery based on deep learning according to claim 1, characterized in that: Electrochemical impedance spectroscopy tests are performed on several lithium batteries with known internal impedance characteristic parameters to obtain corresponding Nyquist curves, wherein the steps of performing the electrochemical impedance spectroscopy test include: connecting the positive electrode and the negative electrode of the battery to be tested to the working electrode and the reference electrode end of the electrochemical workstation respectively; applying a frequency signal to perform the test from high frequency to low frequency; setting the AC signal amplitude; setting the DC bias voltage; automatically applying an AC signal to the battery after starting the test, and gradually scanning the set frequency range, and generating a Nyquist curve at the same time.

3. The method for evaluating the health status of a lithium battery based on deep learning according to claim 2, characterized in that: Based on the long short-term memory network model LSTM model, a deep learning network model is established, and the activation function and optimization algorithm are selected. The Tanh function is selected as the activation function, and Adam is selected as the optimization algorithm of the LSTM model; the formula of the Tanh function is: In the formula, f(r) represents the Tanh function, and the independent variable r represents the weighted sum of the neuron's input, that is, the result of the weighted summation of the input received by the neuron from the previous layer; At the same time, the hyperparameters of the LSTM model are set, and the hyperparameters of the LSTM model include: the number of network layers, the number of iterations, the learning rate, the batch size, the number of training times, the batch processing number, and the number of hidden layer neurons; The number of network layers is set to 3 layers, the number of iterations is set to 200, the learning rate is set to 0.001, the batch size is set to 32, the number of training times is set to 100, the batch size is set to 256, and the number of hidden layer neurons is 32; The input of the trained internal impedance characteristic parameter prediction model is the Nyquist curve diagram of the lithium battery, and the output is the internal impedance characteristic parameter of the lithium battery.

4. The method for evaluating the health status of a lithium battery based on deep learning according to claim 1, characterized in that: The formula for calculating the capacity retention index is: Where RAF is the capacity retention index, C actual is the actual capacity of the lithium battery to be evaluated, C rated is the rated capacity of the lithium battery to be evaluated, R int is the current internal resistance of the lithium battery to be evaluated, R int0 is the initial internal resistance of the lithium battery to be evaluated, ηc is the Coulomb efficiency of the lithium battery to be evaluated, and k1 is the Coulomb efficiency adjustment constant; The actual capacity of the lithium battery to be evaluated is C actual Based on the calculation of power characteristic parameters, the specific calculation formula is as follows: In the formula, I g (t) represents the charging current at time t during the full charging process, t ga and t gb are the start and end time of the full charging process respectively, y(T) is the temperature correction factor, and the formula for calculating the temperature correction factor y(T) is: y(T)=1-β*(T ref -T) Where, T ref is the reference temperature, T is the maximum surface temperature of the lithium battery to be evaluated during full charging, and β is the temperature sensitivity coefficient.

5. A lithium battery health status assessment method based on deep learning according to claim 4, characterized in that: The current internal resistance of the lithium battery to be evaluated is R int The calculation is based on the formula: R int =R ohm +R sei +R diff In the formula, R ohm , R sei and R diff They are the predicted values ​​of electrolyte ohmic impedance, solid electrolyte interface film impedance and diffusion impedance of the lithium battery to be evaluated output by the model respectively; The coulombic efficiency ηc of the lithium battery to be evaluated is calculated based on the formula: In the formula, C charge The fully discharged capacity of the lithium battery to be evaluated is calculated based on the formula: Where, t ha and t hb are the start and end time of the full discharge of the lithium battery to be evaluated, I h (t) is the discharge current at time t during the complete discharge process.

6. A lithium battery health status assessment method based on deep learning according to claim 5, characterized in that: The formula for calculating the cycle efficiency degradation index is: Wherein, EDI is the cycle efficiency degradation index, ηE is the energy efficiency of the lithium battery to be evaluated; the energy efficiency ηE of the lithium battery to be evaluated is calculated based on the formula: Where V h (t) and V g (t) are the discharge voltage and charge voltage at time t during the complete discharge and complete charge processes, respectively; The formula for calculating the shape expansion influence index is: PDI=GD 2 *ln(1+T) Wherein, PDI is the external expansion impact index, and GD is the maximum expansion volume of the lithium battery to be evaluated during the complete charge and discharge process.

Images