State evaluation method for power lithium ion battery of charging facility

Abstract

The invention relates to the technical field of electric power, in particular to a state evaluation method for a power lithium ion battery of a charging facility, which comprises the following steps: collecting charging data in a charging process, and estimating heat distribution in a charging pile; operating data of the power lithium ion battery in the charging / discharging process are collected, and the battery life decline trend of the power lithium ion battery is evaluated; analyzing the heat distribution and the decline trend of the service life of the battery, and determining a mapping relation between the internal temperature change of the power lithium ion battery and the service life of the battery; and determining a state evaluation result of the power lithium ion battery according to the mapping relation. The method has the advantages that the characteristics of the charging pile and the power lithium ion battery are comprehensively and deeply analyzed by setting multiple models, an operator is helped to accurately evaluate the battery state, heat distribution and operation management of the charging pile are optimized, operation efficiency and service quality are improved, and meanwhile information such as battery operation data and the temperature of the charging pile is analyzed.

Description

Technical Field

[0001] The present invention relates to the field of power technology, and particularly to a method for evaluating the state of a power lithium-ion battery of a charging facility. Background Art

[0002] In the prior art, battery state evaluation methods often lack accuracy and real-time performance, and it is difficult to effectively predict the battery life and health status. At the same time, when a charging pile provides fast charging for an electric vehicle, the increase in internal temperature affects the charging efficiency and equipment life, and traditional thermal management methods are difficult to adapt to the heat load brought by fast charging. In addition, the evaluation of the battery life decline trend is crucial for battery maintenance and replacement, but existing methods are difficult to accurately measure and analyze the internal temperature distribution of the battery and its relationship with the life. Summary of the Invention

[0003] In view of the above or existing problems, the present invention is proposed.

[0004] To solve the above technical problems, the present invention provides the following technical solution: A method for evaluating the state of a power lithium-ion battery of a charging facility, which includes collecting charging data during the charging process and estimating the heat distribution inside the charging pile;

[0005] Collecting the operation data of the power lithium-ion battery during the charge / discharge process and evaluating the battery life decline trend of the power lithium-ion battery;

[0006] Analyzing the heat distribution and the battery life decline trend to determine the mapping relationship between the internal temperature change of the power lithium-ion battery and the battery life;

[0007] Determining the state evaluation result of the power lithium-ion battery according to the mapping relationship.

[0008] As a preferred solution of the method for evaluating the state of a power lithium-ion battery of a charging facility of the present invention, wherein: collecting the charging data during the charging process includes,

[0009] Collecting the charging information and the real-time temperature of the charging pile during the charging process.

[0010] As a preferred solution of the method for evaluating the state of a power lithium-ion battery of a charging facility of the present invention, wherein: estimating the heat distribution inside the charging pile includes,

[0011] Clarifying the correlation between the charging information and the charging pile temperature;

[0012] Establishing a thermal model of a DC charging pile, evaluating the dynamic electrothermal characteristics, and estimating the heat distribution inside the charging pile.

[0013] As a preferred embodiment of the method for evaluating the state of a power lithium-ion battery of the charging facility of the present invention, wherein: evaluating the battery life decay trend of the power lithium-ion battery includes,

[0014] Establishing a power lithium-ion battery life model to evaluate the relationship between the capacity decay of the lithium-ion battery and the charging behavior.

[0015] As a preferred embodiment of the method for evaluating the state of a power lithium-ion battery of the charging facility of the present invention, wherein: establishing a power lithium-ion battery life model includes,

[0016] Using a support vector machine to establish the power lithium-ion battery life model.

[0017] As a preferred embodiment of the method for evaluating the state of a power lithium-ion battery of the charging facility of the present invention, wherein: evaluating the relationship between the capacity decay of the lithium-ion battery and the charging behavior includes,

[0018] Collecting a large amount of charging data of lithium-ion batteries;

[0019] Using a recurrent neural network or a long short-term memory network to learn and capture the complex temporal relationship between the battery capacity decay and the charging behavior;

[0020] By training the model and optimizing the network weights, enabling it to more accurately predict the change of the battery capacity with the charging behavior.

[0021] As a preferred embodiment of the method for evaluating the state of a power lithium-ion battery of the charging facility of the present invention, wherein: optimizing the network weights is achieved through the synergistic effect of the backpropagation algorithm and the optimizer.

[0022] As a preferred embodiment of the method for evaluating the state of a power lithium-ion battery of the charging facility of the present invention, wherein: determining the mapping relationship between the internal temperature change of the power lithium-ion battery and the battery life includes,

[0023] Establishing a power lithium-ion battery thermal-life model to determine the relationship between the internal temperature change of the battery and the life.

[0024] As a preferred embodiment of the method for evaluating the state of a power lithium-ion battery of the charging facility of the present invention, wherein: establishing a power lithium-ion battery thermal-life model includes,

[0025] Using Gaussian regression to establish a power lithium-ion battery electrothermal model;

[0026] Using a tree-structured clustering algorithm-based method to establish a life-thermal coupling model of the power lithium-ion battery.

[0027] As a preferred solution of the state evaluation method for the power lithium-ion battery of the charging facility of the present invention, wherein: charging information and real-time temperature of the charging pile during the charging process are collected, including,

[0028] The charging information and the real-time temperature of the charging pile during the charging process are collected in real time through the intelligent power distribution terminal and the communication interface.

[0029] Advantages of the present invention: By setting up a variety of models, the present invention comprehensively and deeply analyzes the characteristics of the charging pile and the power lithium-ion battery, helps the operator accurately evaluate the battery state, optimizes the heat distribution and operation management of the charging pile, improves the operation efficiency and service quality. At the same time, by analyzing information such as battery operation data and charging pile temperature, further, the charging pile layout can be optimized to improve the user charging experience, reduce the operation cost, and provide scientific and effective decision-making references for the planning and policy formulation of future electric vehicle charging networks. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings. Among them:

[0031] Figure 1 It is a flow block diagram of the state evaluation method for the power lithium-ion battery of the charging facility.

[0032] Figure 2 It is a schematic diagram of the time series analysis modeling process.

[0033] Figure 3 It is a schematic diagram of the time series analysis modeling process for the thermal characteristics of the charging pile.

[0034] Figure 4 It is a schematic diagram of the support vector machine modeling process.

[0035] Figure 5 It is a contour map of the longitudinal section temperature distribution under different charging rates.

[0036] Figure 6 It is a contour map of the cross-section temperature distribution under different charging rates.

[0037] Figure 7 It is a schematic diagram of the SVM estimated battery SOH curve by the K-CV method.

[0038] Figure 8 It is a schematic diagram of the battery SOH estimation curve after GA optimizes the parameters. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0039] To make the above objects, features, and advantages of the present invention more apparent and understandable, the following detailed description of the specific embodiments of the present invention will be provided in conjunction with the accompanying drawings of the specification.

[0040] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. However, the present invention may be practiced in other ways than those described herein. Those skilled in the art can make similar extensions without departing from the spirit of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0041] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation manner of the present invention. The phrase "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments.

[0042] Embodiment 1

[0043] Refer to Figures 1 to 4 , which is the first embodiment of the present invention. This embodiment provides a method for evaluating the state of a power lithium-ion battery of a charging facility, which includes:

[0044] S1. Collect charging data during the charging process and estimate the heat distribution inside the charging pile.

[0045] Furthermore, collecting charging data during the charging process includes

[0046] Collecting charging information and the real-time temperature of the charging pile during the charging process.

[0047] Furthermore, collecting charging information and the real-time temperature of the charging pile during the charging process includes

[0048] Collecting charging information and the real-time temperature of the charging pile during the charging process in real time through an intelligent power distribution terminal and a communication interface. Among them, the charging information includes, but is not limited to, AC / DC current / voltage and power.

[0049] It should be noted that by using cloud platform and Internet of Things technologies, the charging pile is connected to the charging pile operation platform to achieve remote monitoring and data collection. Furthermore, while integrating the data, it is ensured that the collected information is authorized and compliant to comply with relevant regulations and privacy policies.

[0050] Preferably, through effective real-time data collection, it will be possible to gain an in-depth understanding of various characteristics during the charging process, providing strong support for optimizing the performance of the charging facility and enhancing the user experience.

[0051] Furthermore, estimating the heat distribution inside the charging pile includes

[0052] Clarify the correlation between charging information and the temperature of the charging pile;

[0053] It should be noted that the correlation between charging information and the temperature of the charging pile can be achieved through multi-level data analysis and correlation research methods. Specifically, it includes collecting charging information during the charging process (including current, voltage, charging power, charging time, etc.), ensuring that the acquisition frequency is high enough to capture the dynamic changes during the charging process; performing data cleaning and preprocessing to handle outliers and missing values to ensure the accuracy and consistency of the data. Based on basic statistical analysis, the average value, standard deviation, etc. can be calculated to preliminarily understand the distribution characteristics of the two sets of data;

[0054] Through the thermal model of the charging pile, estimate the internal temperature distribution of the charging pile, correlate the charging time series and the temperature. Further, use the correlation analysis method, the Pearson correlation coefficient to quantify the linear relationship between the charging information and the temperature of the charging pile (through this method, it can be determined whether there is an obvious correlation between them, and further explore the strength and direction of the correlation); through visual analysis, plot the changes of charging information and temperature over time to more intuitively observe the trends and relationships between the two; time series analysis will help to deeply understand the seasonal, trend and other characteristics between the charging information and the temperature;

[0055] Establish a prediction model through regression analysis to determine the impact of charging information on the temperature of the charging pile and identify the main influencing factors (this helps to understand the temperature change trends under different charging conditions). Conduct temperature threshold analysis to determine the operating temperature range of the charging pile and clarify the changes in charging information under different temperature conditions, which helps to optimize the performance of the charging pile. Fully consider other variables that may affect the temperature and control them to ensure that the obtained correlation is caused by the charging information rather than the influence of other external factors.

[0056] Preferably, through such in-depth and comprehensive research and analysis, it can provide substantial data support for the future design and operation of charging piles to better meet user needs and improve the overall performance of the equipment.

[0057] Establish a thermal model for DC charging piles, evaluate the dynamic electro-thermal characteristics, and estimate the internal heat distribution of the charging piles.

[0058] It should be noted that the establishment of the thermal model of the DC charging pile adopts the physical processes of energy transfer and heat conduction, considering multiple aspects such as heat generation, heat radiation, and heat transfer methods of the internal components of the charging equipment. Specifically, use the time series analysis modeling method (refer to the appendix Figure 2) Establish a thermal model for the charging pile. At the same time, the thermal model based on time series also needs to consider the energy transfer process and heat conduction mechanism inside the charging pile. Further, by combining the time series temperature data with physical principles, mathematical expressions for heat generation, transfer, and dissipation can be derived, which can more accurately describe the thermal behavior of the charging pile under different operating conditions. After obtaining the modeling parameters, the model can be verified and optimized to ensure that the model can accurately predict the time series change of the charging pile temperature.

[0059] Preferably, parameter sensitivity analysis can be carried out through the model to identify the factors that have the most significant impact on the temperature of the charging pile. This provides a guiding direction for subsequent optimization and improvement, such as adjusting the heat dissipation system, optimizing the charging power control strategy, etc. The established thermal model of the DC charging pile not only helps to deeply understand the thermal behavior of the charging pile during actual operation, but also provides a scientific basis for the design, maintenance, and performance optimization of the equipment. Through the application of the model, the temperature distribution of the charging pile under different working conditions can be predicted more effectively, providing strong support for the stable operation and life management of the equipment.

[0060] S2. Collect the operation data of the power lithium-ion battery during the charge / discharge process and evaluate the battery life decay trend of the power lithium-ion battery.

[0061] Furthermore, evaluate the battery life decay trend of the power lithium-ion battery, including

[0062] Establish a power lithium-ion battery life model and evaluate the relationship between the capacity decay of the lithium-ion battery and the charging behavior.

[0063] Furthermore, establish a power lithium-ion battery life model, including

[0064] Use a support vector machine to establish the power lithium-ion battery life model.

[0065] Specifically, by collecting a large amount of actual operation data of power lithium-ion batteries, including parameters such as current, voltage, and temperature during charge and discharge processes, the collected data constitutes a dataset, and each sample is a snapshot of the battery operation state recorded at different time points. Each sample corresponds to a vector, which contains multiple features, and these features form a multi-dimensional feature space. The support vector machine learns the distribution of sample points in this multi-dimensional feature space and searches for a hyperplane that can separate normal operation and life degradation in the non-linear feature space. The training process of the support vector machine adjusts the parameters of the model to better adapt to the distribution of the dataset. The support vector machine learns the distribution of sample points in this multi-dimensional feature space. The training process of the support vector machine adjusts the parameters of the model to better adapt to the distribution of the dataset and constructs a multi-dimensional coordinate system reflecting the battery operation state. This multi-dimensional feature space is the basis for the training and prediction of the support vector machine model, which can more comprehensively capture the complex characteristics of battery life degradation, and the data in this feature space will be used to train and test the support vector machine model.

[0066] Furthermore, through the support vector machine algorithm, the data in the multi-dimensional feature space is mapped into a high-dimensional space. Mapping the data in the multi-dimensional feature space into a high-dimensional space is a core step of the support vector machine (SVM) algorithm. This mapping is accomplished through a so-called kernel function, which can map low-dimensional data into a high-dimensional space and make it easier to separate in the high-dimensional space.

[0067] There are various choices for the kernel function of the support vector machine. One commonly used one is the Radial Basis Function (RBF) kernel function. The formula of the RBF kernel function is:

[0068] K(x i ,x j )=exp(-γ||x i -x j || 2 )

[0069] The support vector machine (SVM) is a brand-new machine learning method based on statistical learning theory. In statistical learning theory, the Structural Risk Minimization (SRM) criterion is often adopted. The goal of the SRM criterion is to improve the generalization ability of the model without restricting the data dimension. While minimizing the sample error, it can also minimize the structural risk. To achieve this goal, according to the principle of maximizing the separation of samples, a classification surface is reasonably selected. SVM is mainly applied to linear classification. In this method, the transformation of the high-dimensional space is used to achieve non-linear classification. By transforming the non-linear problem into a high-dimensional linear problem, the non-linear classification problem can be effectively solved.

[0070] Suppose the linear regression function established in the high-dimensional feature space is as follows:

[0071] f(x) = wF(x) + b

[0072] Among them, f(x) is a non-linear mapping function. Define the e-insensitive loss function:

[0073]

[0074] Among them, f(x) is the predicted value returned by the regression function; y is the corresponding true value. If the difference between f(x) and y is less than or equal to e, the loss is equal to 0.

[0075] Introduce the slack variable x i , And describe the problem of finding w and b in mathematical language, that is

[0076]

[0077]

[0078] Among them, C is the penalty factor. The larger C is, the greater the penalty for samples with training errors greater than e; e specifies the error requirement for regression. The smaller e is, the smaller the error of the regression function.

[0079] After introducing the kernel function, the regression function is:

[0080]

[0081] Among them, only some parameters are not zero, and the corresponding sample x i is the support vector in the problem. The kernel function is:

[0082] K(x i , x j ) = exp(-γ||x i - x j || 2 )

[0083] Specifically, through the application of the RBF kernel function, data points in the original multi-dimensional feature space are mapped into a higher-dimensional vector in the high-dimensional space. This mapping in the high-dimensional space makes the non-linearly separable problem in the original feature space become linearly separable in the high-dimensional space. Such a mapping helps the support vector machine find a hyperplane in the high-dimensional space that can effectively separate the samples of normal operation and life degradation. The separating hyperplane of the support vector machine in the high-dimensional space can be calculated by using the kernel function in the low-dimensional space without directly calculating the inner product in the high-dimensional space. The adjustment of these parameters directly affects the performance of the model. Through methods such as cross-validation, the parameters of the kernel function are adjusted to obtain the best model performance. During the training process, the model will select these support vectors, which are crucial for defining the hyperplane. The support vectors determine the decision boundary of the model, making it as far away as possible from the samples of normal operation and life degradation. The schematic diagram of the support vector machine modeling process is as Figure 4 shown.

[0084] Preferably, through the application of the kernel function, the support vector machine realizes the advantage of processing complex problems in the high-dimensional space, thereby improving the adaptability of the model to non-linear relationships. This mapping process allows the support vector machine to better capture the complex structure of the data. The established life model can better fit the life degradation trend of power lithium-ion batteries through means such as parameter tuning and cross-validation of the support vector machine. The prediction performance of the model is evaluated through verification on the test set to ensure its accuracy and generalization ability in practical applications. The established power lithium-ion battery life model based on the support vector machine can not only predict the battery life state in real-time monitoring, but also provide a scientific basis for battery maintenance and replacement.

[0085] Furthermore, evaluate the relationship between the capacity decay of lithium-ion batteries and the charging behavior, including,

[0086] Collect a large amount of charging data of lithium-ion batteries, where the charging data includes parameters such as current, voltage, temperature during the charging process and real-time records of battery capacity;

[0087] Adopt a recurrent neural network or a long short-term memory network to learn and capture the complex temporal relationship between battery capacity decay and charging behavior;

[0088] By training the model, optimize the network weights so that it can more accurately predict the change of battery capacity with charging behavior.

[0089] Furthermore, optimizing the network weights is achieved through the synergistic effect of the backpropagation algorithm and the optimizer.

[0090] Specifically, first define a loss function that measures the difference between the model output and the actual value. Through forward propagation, the input data is passed through the deep learning model to obtain the model output. Then, by calculating the loss function, the error between the model output and the true value can be obtained. Through the backpropagation algorithm, calculate the gradient of the loss with respect to each network parameter. These gradients indicate the direction of change of the loss function in the parameter space. Select the Adam optimizer and control the update step size of the weights through the learning rate. Finally, by continuously iterating the processes of forward propagation, loss calculation, backpropagation, and weight update, the weights of the deep learning model are gradually adjusted, making the model's predictions approximate the true value and the loss function gradually decrease.

[0091] Preferably, this optimization process can improve the performance of the model, enabling it to better fit the data and have stronger generalization ability, thus providing more accurate and reliable results when evaluating the relationship between lithium-ion battery capacity decay and charging behavior.

[0092] Furthermore, the learning rate directly affects the update step size of the weights and has an important impact on the convergence and performance of the model. To optimize the learning rate, first, it is necessary to understand the impact of learning rate selection on model training. An overly large learning rate may cause the model to diverge during training, while an overly small learning rate may lead to an overly slow convergence rate of the model. The method of optimizing the learning rate is through a learning rate decay strategy. At the initial stage of training, a larger learning rate helps with rapid convergence, and the learning rate is gradually decreased during the training process to ensure that the model is more stable when approaching the optimal solution. The decay strategy can be adjusted according to the number of training epochs or model performance, improving the model's adaptability to complex data distributions. Further, an adaptive learning rate optimization algorithm, RMSprop, can be adopted. These algorithms adaptively adjust the learning rate by considering information from past gradients, thus more flexibly adapting to changes in different parameters. This method typically performs well especially for large-scale, high-dimensional datasets and complex deep neural network structures.

[0093] Preferably, the present invention uses a test set to verify the performance of the deep learning model. By comparing the battery capacity decay trend predicted by the model with the actual measurement results, the prediction accuracy and generalization ability of the model are evaluated. Through algorithm learning, it can more flexibly capture the non-linear relationships in time series data, improving the model's ability to model the complex dynamic relationship between lithium-ion battery capacity decay and charging behavior.

[0094] S3. Analyze the heat distribution and the trend of battery life decline to determine the mapping relationship between the internal temperature change of the power lithium-ion battery and the battery life.

[0095] Furthermore, determining the mapping relationship between the internal temperature change of the power lithium-ion battery and the battery life includes

[0096] Establish a thermal-life model for power lithium-ion batteries to determine the relationship between the internal temperature change of the battery and its life.

[0097] Preferably, the established thermal-life model for power lithium-ion batteries can not only investigate the influence of the number of battery cycles on life under different charge-discharge depths and temperature conditions, but also analyze the cycle efficiency of the battery at different temperatures, that is, the energy lost by the battery in each charge-discharge cycle. Further, deep discharge may lead to the destruction of the solid electrolyte interface and the loss of electrode materials in lithium-ion batteries. By comparing the battery life under shallow discharge and deep discharge conditions at different temperatures, the optimal charge-discharge strategy can be understood. That is, the life characteristics of the battery under different working conditions can be more comprehensively understood, providing a scientific basis for the improvement and application of battery technology.

[0098] Further, establishing a thermal-life model for power lithium-ion batteries includes,

[0099] Establish an electro-thermal model for power lithium-ion batteries using Gaussian regression;

[0100] Specifically, by collecting a large amount of real-time data such as current and voltage of lithium-ion batteries under different charge-discharge conditions, a multi-dimensional feature space is formed. The electro-thermal behavior of the battery is contained in these data. The heat conduction equation is used to describe the heat conduction process inside the battery, a thermal state space equation of the battery interior is constructed, and an advanced filter is used to estimate the internal temperature of the battery.

[0101] Further, using Gaussian Process Regression (GPR) to fit the complex non-linear relationships between features such as current, voltage, and temperature. Gaussian regression can not only fit the mean value of the data, but also provides an estimate of the uncertainty of the model, making the model more reliable. By optimizing the hyperparameters of Gaussian regression, the model can better adapt to the actual data and improve the prediction accuracy. In the process of fitting the mean value of the data, Gaussian regression models the weight of each data point and is committed to capturing the overall trend in the data set. Its core concept is to regard each data point as a sample of a Gaussian distribution, and then form the overall distribution through the superposition of these Gaussian distributions. Such a processing method helps to comprehensively consider the contributions of different data points, making the fitted mean more representative.

[0102] In the actual fitting process, Gaussian regression determines the parameters of the mean function by maximizing the likelihood estimate to best describe the distribution characteristics of the data. At the same time, the method adjusts the weights of the data points, making it have a certain robustness to outliers or noise in the overall fitting process. This statistical fitting method not only focuses on the accuracy of the mean value, but also considers the dispersion of the data, thus more comprehensively reflecting the true distribution of the data.

[0103] Optimizing the hyperparameters of Gaussian regression is a crucial step in ensuring good model performance. The hyperparameter optimization process aims to find a set of parameters that result in the best fit of the Gaussian regression model to the observed data. Two key hyperparameters in Gaussian regression are the length scale and the noise level. The length scale is a parameter in Gaussian regression that determines the rate of change of the function shape. A suitable length scale can adapt to different variation trends in the data. An overly small length scale may lead to overfitting, while an overly large one may lead to underfitting. Therefore, by adjusting the length scale during training, the model can more flexibly adapt to data with different feature scales.

[0104] The noise level, on the other hand, is the degree of observed error or noise considered in the Gaussian regression model. In optimization, adjusting the noise level helps balance the model's fit to the observed data and its resistance to noise. An overly high noise level may cause the model to ignore the true signal, while an overly low one may be overly influenced by outliers. Therefore, by optimizing the noise level, one can better understand the uncertainty of the model and improve its robustness.

[0105] The hyperparameter optimization process usually involves using techniques such as cross-validation, dividing the data into a training set and a validation set, and selecting the optimal hyperparameter combination by comparing the performance of the model on the validation set under different hyperparameters. This process needs to be carried out carefully to avoid overfitting the validation set and resulting in a decline in performance on new data. Additionally, some advanced optimization algorithms, such as Bayesian optimization, can find better hyperparameter combinations with fewer attempts, improving the efficiency of optimization. Hyperparameter optimization is a key step in Gaussian regression modeling, which directly affects the model's fit to the data and its generalization performance. By carefully selecting appropriate length scales and noise levels, a more robust and predictive Gaussian regression model can be established to more accurately analyze and predict the distribution and relationships of complex data.

[0106] Preferably, by adopting the fitting method of Gaussian regression, one can more comprehensively understand the distribution characteristics of the data, including the mean and uncertainty. This statistical modeling method not only provides the overall trend of the data fit but also evaluates the quality of the fit, providing a basis for more accurate data analysis and prediction. Further, it enables the finally established electrothermal model of power lithium-ion batteries to predict the electrothermal performance of the battery under different working conditions. This modeling method based on Gaussian regression helps to deeply understand the electrothermal behavior of the battery and provides reliable means for predicting and monitoring the safety and performance of the battery.

[0107] A life - thermal coupling model of power lithium-ion batteries is established using a tree-structured clustering algorithm, which can more comprehensively consider the correlation between battery life and temperature.

[0108] It should be noted that through the tree - structured clustering algorithm, the battery data is hierarchically grouped according to its characteristics, forming a tree - shaped hierarchical structure. This hierarchical structure reflects the similarity and correlation between different battery characteristics, providing strong support for subsequent model establishment.

[0109] Furthermore, it should be noted that when establishing the life - heat coupling model, the hierarchical characteristics of tree - structured clustering provide a more detailed division of battery groups, enabling the model to more precisely consider the life - heat coupling relationship between different groups. Each hierarchical node represents a battery group with more similar characteristics. By establishing the model on the tree - structured framework, it is possible to capture the variation rules of life and heat coupling between different levels. The advantage of the tree - structured clustering algorithm lies in its ability to discover the potential group structure in the data rather than a pre - defined division. This means that in model establishment, it can better adapt to the actual distribution of battery data without being restricted by prior assumptions. By considering battery groups at different levels, the model can more comprehensively understand the complex relationship between battery life and temperature, thereby more accurately predicting the life performance of batteries under different working conditions. Discovering the potential group structure in the data is a key task, which helps to reveal the internal organizational rules of the data and the potential connections between features. An effective method is to use clustering analysis, and the tree - structured clustering algorithm is a powerful tool that can systematically reveal the hierarchical organization of the data. The tree - structured clustering algorithm constructs a tree - shaped structure by recursively grouping the data, where each node in this structure represents a data set. Such a hierarchical structure not only reflects the organization of the overall data but also enables the observation of group structures at different levels. By calculating similarity metrics, the clustering algorithm aggregates similar data points together to form nodes in the hierarchical structure. Tree - structured clustering can automatically adapt to the potential groups in the data without the need to pre - set the number or structure of the groups. This is particularly important for dealing with data sets with unknown data distributions and complex group structures. During the process of gradually constructing the tree - shaped structure, the clustering algorithm discovers the internal patterns and rules of the data, grouping similar data points into the same group, thus forming a potential group structure. This automatic discovery feature makes tree - structured clustering very valuable for exploring large - scale, high - dimensional data sets. By observing the hierarchical relationships of the tree - structured framework, it is possible to deeply understand the similarities and differences between different groups. The root node of the hierarchical structure represents an overall of the entire data, while the leaf nodes represent the specific subdivided groups. This allows for the exploration of potential group structures at different levels and an understanding of the hierarchical organization of different features in the data.

[0110] Preferably, the established power lithium-ion battery life-thermal coupling model can not only predict the overall battery population, but also provide personalized life-thermal coupling predictions for different levels of battery populations. This method of establishing a model based on tree-structured clustering is expected to provide a more accurate and detailed tool for battery life prediction and a more scientific and reliable basis for battery charging.

[0111] S4. Determine the state evaluation result of the power lithium-ion battery according to the mapping relationship.

[0112] In summary, the beneficial effect of the method for evaluating the state of a power lithium-ion battery of a charging facility according to the present invention is to comprehensively and deeply analyze the characteristics of the charging pile and the power lithium-ion battery by setting multiple models, help the operator accurately evaluate the battery state, optimize the heat distribution and operation management of the charging pile, improve the operation efficiency and service quality, and at the same time through the analysis of battery operation data and charging pile temperature and other information.

[0113] Embodiment 2

[0114] Refer to Figures 5 to 8 , which is the second embodiment of the present invention. This embodiment provides a research and analysis on the dynamic electro-thermal characteristics analysis and safety state quantification of a DC charging pile and a power lithium-ion battery, as follows:

[0115] (1) Thermal characteristics simulation analysis of a typical DC charging pile

[0116] In order to explore the performance of the charging pile in terms of thermal stability, in-depth thermodynamic simulation analysis was carried out. Our goal is to determine the layout of the main heat-generating components on the printed circuit board and design appropriate air-cooling conditions to improve the design rationality of the charging pile module structure and ensure its temperature stability during long-term operation.

[0117] 1) Analysis of heat-generating components

[0118] By simulating the thermal characteristics of the charging pile, a solid foundation was laid for establishing the electro-thermal model of the charging pile. When analyzing the heat-generating components, it was found that the DC charging pile adopts an AC-DC and DC-DC two-stage converter structure. The AC-DC converter converts the input alternating current into a stable direct current output through an interleaved parallel three-phase three-level rectifier. The DC-DC converter then uses a half-bridge three-level logic link control circuit to convert the direct current output by the previous stage into a direct current that meets the charging requirements. These converters contain hundreds of component modules, and their structural layout and connection methods are very complex. The printed circuit board uses multi-layer wiring. If a finite element model is established according to the original model, it will become very complex, and the heating effect of many low-power components on the model is relatively small.

[0119] Thermal conductivity and glass transition temperature of materials

[0120]

[0121]

[0122] The heat generation parameters of the main components are given in the table. Therefore, we simplified the model by removing those components with complex structures and little influence on heat generation and dissipation, such as control chips, small resistors, and small capacitors, etc. At the same time, the copper film, surface mount resistors, surface mount capacitors, and control chips on the PCB board were deleted. The following table lists the heat generation parameters of the main components. The heat generating components in the power module mainly include IGBTs, high-power diodes, DC / DC modules, transformers, and filter inductors, etc. The heat generation rate of these heat generating components is directly related to their heat power consumption. When performing finite element thermal analysis, we assume that the electrical energy loss of the heat generating components is completely converted into heat energy. By considering factors such as the thermal power characteristics, volume size, rated current, and current and voltage when working in the module of each component, we calculated the electrical losses of each heat generating component respectively as the heat generation parameters of the components.

[0123] 2) Thermal simulation calculation and analysis

[0124] The power module model is divided into upper and lower layers. The upper layer adopts a parallel structure of two half-bridge three-level LLC circuits, and the lower layer is a three-phase three-level rectifier circuit. According to the electrical schematic diagram and the difference in the heat generation rate of each heat generating device, combined with PCB wiring techniques, we designed different model schemes by adjusting the position distribution of the main heat generating devices on the PCB. A 3D simplified model of each model scheme was drawn in Creo, and then imported into ANSYS Workbench for mesh generation, boundary condition setting, and load application. Finally, the thermal field distribution diagram of the heat generating components was obtained through simulation calculation.

[0125] The distribution of heat generating components in each scheme is asymmetric. For example, in the 4 different schemes shown, the heat flow fields are different when the right side of the air fluid domain is the inlet and the left side is the outlet (A-direction air cooling condition), and when the left side is the inlet and the right side is the outlet (B-direction air cooling condition). Therefore, we conducted simulation analysis on Scheme 1 and Scheme 2 under the A-direction air cooling condition, while Scheme 3 and Scheme 4 were respectively analyzed under the A and B-direction air cooling conditions. After 6 numerical calculations, we obtained the following simulation data table:

[0126]

[0127]

[0128] It can be seen from the thermal simulation results that the maximum operating temperature of each component is closely related to the location of the main heat-generating components. The more concentrated the heat-generating components are, the higher the operating temperature of the components. This is mainly because the concentration of heat-generating components leads to an increase in the heat generation per unit volume, thereby increasing the maximum temperature of the heat-generating components. Therefore, on the premise of reasonable wiring, the layout positions of the main heat-generating components should be dispersed as much as possible. The air-cooling device has a better heat dissipation effect on the heat-generating components near the air inlet, while the air effect on the components far from the air-cooling port is weaker. By comparing the temperature distributions of different design schemes, we determined that the component distribution in Scheme 3 is relatively reasonable under the air-cooling condition in the B direction, and the maximum temperature of each component is relatively low. In the actual circuit layout design, Scheme 3 can be adopted and air-cooling design can be carried out according to the B-direction conditions.

[0129] (2) Thermal Characteristics Analysis of Power Lithium-Ion Batteries

[0130] During the battery charging process, the discharge rate of the battery can remain stable within a certain range, and the temperature change is relatively stable. However, the working current of the internal power battery is constantly changing. To deeply analyze the relationship between the discharge rate and the temperature change, we selected five different charging rate conditions, namely 1C, 2C, 3C, 4C, and 5C, to study the temperature change of the battery. By exploring the temperature change characteristics inside the battery under different charging rates, we provided theoretical support for establishing the electro-thermal model of the battery. When performing the simulation calculation, we set the initial temperature of the calculation domain of the battery and the external air to 25 °C, and set the outer surface of the external air to a constant temperature of 25 °C. Considering that the thermal conductivity and dynamic viscosity of air change with temperature, we made a table recording the corresponding relationship between the thermal conductivity and dynamic viscosity of air and temperature, and loaded it onto the property values of air. In this way, during the simulation process, we can better reflect the change of air physical property parameters. Finally, we obtained Figure 5 The longitudinal section temperature distribution cloud map under different charging rate conditions and Figure 6 The cross-section temperature distribution cloud map under different charging rate conditions.

[0131] In the comparative simulation calculations, we made comparisons for five charging rates of 1C, 2C, 3C, 4C, and 5C, and the corresponding calculation times were 3600 seconds, 1800 seconds, 1200 seconds, 900 seconds, and 720 seconds respectively. By observing the result graphs, we can find that under different charging rate conditions, the internal temperature distribution law after the battery charging ends is basically the same. The regions with higher temperatures are mainly concentrated in the position slightly below the center of the battery, and the temperature at the positive extreme is higher than that at the negative extreme. In addition, the temperature inside the battery is significantly higher in the inner layer than in the outer layer, and the internal temperature gradually decreases from the inside to the outside. As the discharge rate increases, the uniformity of the internal temperature distribution of the battery gradually decreases. Additionally, we can also observe that when the discharge rate is greater than 2C, the temperature inside the battery can reach 40°C.

[0132] For improving the heat dissipation ability of the phase change material (PCM), one method is to install heat pipe foam aluminum and aluminum heat sinks in the battery pack. The following table lists the characteristics of different heat transfer media.

[0133]

[0134] The parameters of battery performance are usually very limited, such as the voltage, current, SOC, etc. of the battery / battery pack. Coupling the electrochemical parameters of the battery with the thermal model can effectively predict possible thermal safety accidents.

[0135] (III) Analysis of the Electrothermal Coupling Model for Power Battery Life

[0136] The state of health (SOH) of the battery is the real-time diagnosis and evaluation of the battery attenuation process and is also the key to tracking the actual performance of the battery. SOH can be regarded as a quantitative description of battery degradation and represents the hidden state inside the battery that cannot be directly measured. The methods for diagnosing battery SOH mainly include model-based methods, data-driven methods, and data-model fusion methods. Although model-based methods can well reflect the internal characteristics of the battery, problems such as high modeling complexity or difficulty in identifying model parameters limit their practical applications.

[0137] To find an effective method for estimating the State of Health (SOH), it is first necessary to determine the characteristic quantities that can describe in detail the changes caused by cyclic experiments. With the long-term use of lithium batteries in electric vehicles, the battery gradually ages internally, resulting in a reduction of internal active substances, which in turn affects the battery's stability. When a large current passes through, the voltage at the battery output terminal may jump. Therefore, the health state of the battery can be evaluated by analyzing the voltage change. To study this jump phenomenon, we plotted the relationship between the battery capacity and the number of cycles. It can be observed from the graph that as the number of cycles increases, the change in battery capacity is relatively gentle, showing a downward trend, and the discharged electricity gradually decreases, which is consistent with the characteristics of battery aging. In the experiment, capacity regeneration may occur during individual cycles, which also increases the difficulty of estimating the battery's SOH. When the number of cycles is less than 50, the battery capacity decays slowly. Between 100 and 500 cycles, the available capacity of the battery gradually decreases. However, as the cycling continues, when the number of battery cycles reaches a critical value, the available capacity of the battery will drop sharply. This phenomenon is mainly due to a series of irreversible chemical reactions gradually occurring inside the battery, resulting in a large amount of lithium ions precipitating between the positive and negative electrodes, reducing the number of lithium ions available for the cyclic reaction, and thus leading to a decrease in the available capacity of the battery and an attenuation situation.

[0138] In addition, when using the Support Vector Machine (SVM) for prediction, relevant parameters need to be adjusted to obtain the best prediction accuracy. In the SVM algorithm, the main parameters to be adjusted include the penalty parameter and the kernel function parameter. Common parameter adjustment methods include the Grid Search Method (K-CV method), Genetic Algorithm (GA) parameter optimization, and Particle Swarm Optimization Algorithm (PSO) parameter optimization. The K-CV method divides the original data into K subsets equally. Each subset is used as the validation set in turn, and the remaining K - 1 subsets are used as the training set, so that K models will be obtained. Generally, the value of K is greater than or equal to 2. When the original data is small, the value is 2, and in this paper, we start from 3. This method can effectively avoid the problems of underfitting and overfitting, and the obtained results have a high degree of credibility. The final parameter adjustment results are shown in Figure 7 the schematic diagram of the SVM estimating the battery SOH curve by the K-CV method.

[0139] The genetic algorithm originated from the computer simulation study of biological systems. It began to rise in the 1970s, gradually developed in the 1980s, and reached its peak in the 1990s. As a practical, efficient, and robust optimization technology, the genetic algorithm is widely used in various fields. In machine learning, it can be used to optimize the parameters of the Support Vector Machine (SVM). After being optimized by the genetic algorithm, the SOH results estimated by the SVM are shown in Figure 8 the battery SOH estimation curve after optimizing the parameters by the GA. The following parameters are obtained by comparison:

[0140] Comparison table for finding optimal parameters

[0141]

[0142] After comparing two different methods, it can be clearly seen from the error graph that the error of the K-CV method is smaller than that of the GA algorithm. In addition, in terms of the required time, the calculation time of the K-CV method is the shortest, followed by the GA algorithm.

[0143] When establishing the life-thermal coupling model, the method of tree-structured clustering is adopted. Its hierarchical characteristics provide a more detailed division for the battery population. This hierarchical structure allows the model to more carefully consider the life and thermal coupling relationships between different populations. Each hierarchical node represents a battery population, and they have more similar characteristics. By constructing the model on the tree structure, the changing rules of life and thermal coupling between different levels can be captured. At the same time, the advantage of the tree-structured clustering algorithm is that it can discover the potential population structure in the data without setting the division criteria in advance. This means that when establishing the model, we can better adapt to the actual distribution of battery data without being restricted by prior assumptions. By considering the battery populations at different levels, the model can more comprehensively understand the complex relationship between battery life and temperature, and thus more accurately predict the life performance of the battery under different working conditions. Discovering the potential population structure in the data is a crucial task, which helps to reveal the internal organizational rules of the data and the potential connections between features. As a powerful tool, the tree-structured clustering algorithm can systematically reveal the hierarchical organization of the data, thus providing important support for establishing a more accurate and detailed battery life-thermal coupling model.

[0144] The tree-structured clustering algorithm gradually groups the data to form a hierarchical tree structure, where each node represents a data set. This structure not only reflects the organization of the overall data but also enables us to clearly observe the population structure at different levels. By calculating the similarity measure between data points, the clustering algorithm clusters similar data points together to form nodes in the hierarchical structure.

[0145] Dendrogram-based clustering can automatically adapt to the potential groups in the data without the need to pre-set the number or structure of the groups. This is particularly important for dealing with datasets with unknown data distributions and complex group structures. During the process of gradually constructing the tree structure, the clustering algorithm discovers the inherent patterns and regularities in the data and groups similar data points into the same group, thus forming a potential group structure. This characteristic of automatic discovery makes dendrogram-based clustering very valuable for the exploration of large-scale and high-dimensional datasets. By observing the hierarchical relationships in the dendrogram, we can gain a deeper understanding of the similarities and differences between different groups. The root node of the dendrogram represents the overall situation of the entire data, while the leaf nodes represent more specific groups. This structure enables us to explore the potential group structure at different levels and understand the hierarchical organization of different features in the data.

[0146] The finally established life - thermal coupling model of power lithium-ion batteries can not only predict the overall battery group, but also provide personalized life - thermal coupling predictions for battery groups at different levels. This method of model establishment based on dendrogram-based clustering is expected to provide a more accurate and detailed tool for battery life prediction and a more scientific and reliable basis for battery charging.

[0147] In summary, the present invention optimizes the layout of charging piles, improves the user charging experience, reduces the operation cost, and provides a scientific and effective decision-making reference for the planning and policy-making of future electric vehicle charging networks.

[0148] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A method for evaluating the status of a power lithium-ion battery of a charging facility, characterized in that: include, Collect charging data during the charging process and estimate the heat distribution inside the charging pile; Collect the operating data of the power lithium-ion battery during the charging / discharging process and evaluate the battery life decay trend of the power lithium-ion battery; Analyzing the heat distribution and the battery life decay trend to determine a mapping relationship between the internal temperature change of the power lithium-ion battery and the battery life; A state evaluation result of the power lithium-ion battery is determined according to the mapping relationship.

2. The method for evaluating the status of a power lithium-ion battery of a charging facility according to claim 1, characterized in that: The collecting of charging data during the charging process includes: Collect charging information and real-time temperature of the charging pile during the charging process.

3. The method for evaluating the status of a power lithium-ion battery of a charging facility according to claim 2, characterized in that: The estimating the heat distribution inside the charging pile includes: Clarify the correlation between charging information and charging pile temperature; A thermal model of a DC charging pile is established to evaluate the dynamic electrothermal characteristics and estimate the internal heat distribution of the charging pile.

4. The method for evaluating the status of a power lithium-ion battery of a charging facility according to claim 3, characterized in that: The evaluation of the battery life degradation trend of the power lithium-ion battery includes: Establish a power lithium-ion battery life model and evaluate the relationship between lithium-ion battery capacity decay and charging behavior.

5. The method for evaluating the status of a power lithium-ion battery of a charging facility according to claim 4, characterized in that: The power lithium-ion battery life model is established. include, Support vector machine is used to establish the life model of power lithium-ion battery.

6. The method for evaluating the status of a power lithium-ion battery of a charging facility according to claim 5, characterized in that: The method for evaluating the relationship between the capacity decay and the charging behavior of a lithium-ion battery comprises: Collect charging data from a large number of lithium-ion batteries; Use recurrent neural networks or long short-term memory networks to learn and capture the complex temporal relationship between battery capacity decay and charging behavior; By training the model, the network weights are optimized so that it can more accurately predict the change in battery capacity as a function of charging behavior.

7. The method for evaluating the status of a power lithium-ion battery of a charging facility according to claim 6, characterized in that: The optimization of network weights is achieved through the synergy of the back-propagation algorithm and the optimizer.

8. The method for evaluating the status of a power lithium-ion battery of a charging facility according to claim 7, characterized in that: The determining of the mapping relationship between the internal temperature change of the power lithium-ion battery and the battery life includes: A thermal-life model for power lithium-ion batteries is established to determine the relationship between the internal temperature change and life of the battery.

9. The method for evaluating the status of a power lithium-ion battery of a charging facility according to claim 8, characterized in that: The method of establishing a thermal-life model for a power lithium-ion battery includes: Use Gaussian regression to establish the electrothermal model of power lithium-ion batteries; The life-thermal coupling model of power lithium-ion batteries is established using a tree-structured clustering algorithm.

10. The method for evaluating the status of a power lithium-ion battery of a charging facility according to any one of claims 2 to 9, characterized in that: The charging information and real-time temperature of the charging pile during the charging process are collected. include, The charging information during the charging process and the real-time temperature of the charging pile are collected in real time through the intelligent power distribution terminal and communication interface.

Images