A method, device, and storage medium for estimating the state of charge of a lithium iron phosphate battery

Abstract

The application discloses a lithium iron phosphate battery state of charge estimation method, device, equipment and storage medium. Its characteristics include: collecting the electrochemical impedance spectrum data of the lithium iron phosphate battery to be estimated, decoupling the relaxation time distribution of the electrochemical impedance spectrum data, and obtaining the relaxation time distribution curve; identifying the optimal single feature data in the relaxation time distribution curve based on the local maximum value method; inputting the optimal single feature data into the pre-trained state of charge estimation model to obtain the state of charge estimation result of the lithium iron phosphate battery to be estimated. The problem of overlapping original electrochemical impedance spectrum characteristics and weak anti-interference is solved, the wide temperature robustness and explainability of the state of charge estimation are improved, the data collection time consumption and the algorithm power threshold of large-scale deployment of the power storage system are greatly reduced, and the estimation accuracy and engineering practicability are considered.

Description

Technical Field

[0001] This invention relates to the field of state-of-charge estimation technology for lithium iron phosphate batteries, and particularly to a method, apparatus, device, and storage medium for estimating the state of charge of lithium iron phosphate batteries. Background Technology

[0002] With the rapid development of the electrochemical energy storage industry, lithium iron phosphate (LFP) batteries have become the mainstream technology in the power and energy storage fields due to their advantages of high power density, long cycle life, high safety, and low cost. State of charge (SOC), as a core management parameter of the battery management system (BMS), directly determines the battery's operational safety, lifespan, and efficiency. However, traditional SOC estimation methods typically rely on the battery's stable state and their accuracy is limited by initial numerical errors when identifying and estimating the SOC of lithium iron phosphate batteries. Existing technologies employ high-precision equivalent circuit models and Kalman filtering, using high-order matrix operations, to accurately identify the SOC of lithium iron phosphate batteries, but this requires sufficient computing power. Edge computing platforms or automotive platforms often struggle to meet the computing power requirements of high-precision algorithms, forcing existing evaluation methods to reduce computational accuracy. This results in insufficient accuracy in the flat region of LFP battery SOC estimation, poor real-time measurement performance, and difficulty in meeting the technical requirements of BMS for fast response, low computing power, and high reliability. Summary of the Invention

[0003] This invention provides a method, apparatus, device, and storage medium for estimating the state of charge (SOC) of lithium iron phosphate batteries, thereby solving the problem of high-precision SOC estimation for lithium iron phosphate batteries under low computing power platforms in the prior art.

[0004] According to one aspect of the present invention, a method for estimating the state of charge (SOC) of a lithium iron phosphate battery is provided, comprising: Electrochemical impedance spectroscopy (EIS) data of the lithium iron phosphate battery to be estimated were collected, and the relaxation time distribution of the EIS data was decoupled to obtain the relaxation time distribution curve. Identifying the optimal single feature data in the relaxation time distribution curve based on the local maximum method; The optimal single-feature data is input into a pre-trained state-of-charge estimation model to obtain the state-of-charge estimation result of the lithium iron phosphate battery to be estimated.

[0005] According to another aspect of the present invention, a state-of-charge estimation device for a lithium iron phosphate battery is provided, comprising: The data acquisition module is used to acquire electrochemical impedance spectroscopy data of the lithium iron phosphate battery to be estimated, and to decouple the relaxation time distribution of the electrochemical impedance spectroscopy data to obtain the relaxation time distribution curve. The optimal feature extraction module is used to identify the optimal single feature data in the relaxation time distribution curve based on the local maximum method. The model recognition module is used to input the optimal single feature data into the pre-trained state of charge estimation model to obtain the state of charge estimation result of the lithium iron phosphate battery to be estimated.

[0006] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising: At least one processor; and A memory that is communicatively connected to at least one processor; wherein, The memory stores a computer program that can be executed by at least one processor, such that the at least one processor is able to perform the state-of-charge estimation method for lithium iron phosphate batteries according to any embodiment of the present invention.

[0007] According to another aspect of the present invention, a computer-readable storage medium is provided, which stores computer instructions for causing a processor to execute and implement the state-of-charge estimation method for a lithium iron phosphate battery according to any embodiment of the present invention.

[0008] The technical solution of this invention collects electrochemical impedance spectroscopy (EIS) data of the lithium iron phosphate (LFP) battery to be estimated, decouples the relaxation time distribution of the EIS data to obtain a relaxation time distribution curve, identifies the optimal single feature data in the relaxation time distribution curve based on the local maximum method, and uses the physical features after relaxation time distribution decoupling as model input to effectively overcome the estimation bottleneck of the flat region of the LFP battery OCV-SOC, solve the problems of overlapping features and weak anti-interference of the original EIS, and input the optimal single feature data into the pre-trained state of charge (SOC) estimation model to obtain the SOC estimation result of the LFP battery to be estimated. The use of a lightweight convolutional model can effectively reduce the deployment computing power threshold, adapt to the hardware of low computing power platforms, solve the problem of high-precision SOC estimation of LFP batteries on low computing power platforms in the prior art, improve the wide-temperature robustness and interpretability of SOC estimation, and balance estimation accuracy and engineering practicality.

[0009] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 A flowchart of a method for estimating the state of charge of a lithium iron phosphate battery is provided as an embodiment of the present invention; Figure 2 A flowchart illustrating a method for estimating the state of charge of a lithium iron phosphate battery, provided as an embodiment of the present invention; Figure 3 A schematic diagram of the structure of a state-of-charge estimation device for a lithium iron phosphate battery provided in an embodiment of the present invention; Figure 4 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. Detailed Implementation

[0012] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0013] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0014] Figure 1This invention provides a flowchart of a method for estimating the state of charge (SOC) of a lithium iron phosphate (LFP) battery. This embodiment is applicable to technical scenarios requiring rapid identification and estimation of the SOC of LFP batteries on edge computing nodes and low-power platforms. The method can be executed by a LFP battery SOC estimation device, which can be implemented in hardware and / or software and can be configured in an electronic device. Figure 1 As shown, the method includes: S110. Collect electrochemical impedance spectroscopy data of the lithium iron phosphate battery to be estimated, decouple the relaxation time distribution of the electrochemical impedance spectroscopy data, and obtain the relaxation time distribution curve.

[0015] The lithium iron phosphate battery to be estimated can be any lithium iron phosphate battery that requires state-of-charge estimation. For example, the lithium iron phosphate battery to be estimated can be a power battery installed in an electric vehicle, or a lithium iron phosphate battery used in energy storage batteries.

[0016] Optionally, the electrochemical impedance spectroscopy data includes the real part of the impedance, the imaginary part of the impedance, the impedance amplitude, the phase angle, and the battery terminal voltage within a set frequency range. It should be noted that the real part of the impedance data can refer to the real component data of the impedance measured after applying sinusoidal excitation to the lithium iron phosphate battery under estimation using electrochemical measurement equipment, reflecting the battery's ohmic loss and charge transfer loss characteristics. The imaginary part of the impedance data can refer to the imaginary component data of the impedance obtained during the same measurement process, reflecting the battery's capacitance and polarization characteristics. The impedance amplitude data can refer to the impedance magnitude data calculated based on the real and imaginary parts of the impedance, characterizing the overall impedance of the battery. The phase angle data can refer to the angle between the real and imaginary parts of the impedance, reflecting the phase characteristics of the battery impedance. The battery terminal voltage data can refer to the battery terminal voltage data acquired simultaneously during the measurement process.

[0017] Optionally, the set frequency range can be understood as a pre-selected measurement frequency interval. This invention preferentially selects a specific target single frequency point, but it can also be extended to a wide frequency range as needed, and can be adaptively adjusted according to the actual battery evaluation scenario. The target single frequency point can be a pre-selected single frequency point with high-value information.

[0018] Optional relaxation time distribution curves can be DRT (Distribution of Relaxation Times) curves obtained from electrochemical impedance spectroscopy (EIS) data. It should be noted that the specific process for decoupling the relaxation time distribution from EIS data involves splitting the EIS data, performing logarithmic scaling, discretizing the relaxation time domain, constructing a system of linear equations for solving the DRT, regularizing the data, and solving the positive definite normal equations using the QR decomposition method to obtain the basis function coefficients. These coefficients are then substituted into the discretized expression to obtain the final DRT curve.

[0019] Specifically, electrochemical impedance spectroscopy (EIS) data of the lithium iron phosphate battery to be estimated are collected, and the relaxation time distribution of the EIS data is decoupled to obtain the relaxation time distribution curve.

[0020] S120. Identify the optimal single feature data in the relaxation time distribution curve based on the local maximum method.

[0021] The optimal single feature data can be the relaxation time of the optimal electrochemical process that is strongly correlated with the state of charge. It should be noted that the optimal electrochemical process can refer to the electrochemical process with the highest correlation to the change in state of charge; the optimal single feature data is the relaxation time of the electrochemical process with the highest correlation to the change in state of charge. The optimal single feature data is used to effectively separate overlapping electrochemical processes in the original impedance spectrum, eliminate interference from noise and irrelevant processes, accurately reflect the true changes in the state of charge inside the battery, and help overcome the bottleneck of the flat region to achieve high-precision SOC estimation.

[0022] Among them, the local maximum method can be used to detect peaks and extract parameters from the relaxation time distribution curve.

[0023] Specifically, the optimal single feature data in the relaxation time distribution curve is identified based on the local maximum method.

[0024] Optionally, in another optional embodiment of the present invention, the step of identifying the optimal single feature data in the relaxation time distribution curve based on the local maximum method includes: The local maximum method is used to identify at least one independent peak in the relaxation time distribution curve; for each independent peak, the position, height and area features of the independent peak are extracted as candidate relaxation time feature data; all the candidate relaxation time feature data are optimized to determine the optimal single feature data.

[0025] Among them, candidate relaxation time feature data can be independent peaks and independent peak feature data in the relaxation time distribution curve. It should be noted that in the relaxation time distribution curve, each independent peak and independent peak feature data is identified as candidate relaxation time feature data that is correlated with the change of charge state. The optimal single feature data is the relaxation time single feature data of the electrochemical process with the highest correlation to the change of charge state among the candidate relaxation time feature data.

[0026] Among them, an independent peak can be a local peak in the relaxation time distribution curve; it should be noted that an independent peak can be understood as an electrochemical process inside the lithium iron phosphate battery; the position feature can be the position of the peak apex of the independent peak in the relaxation time distribution curve; the height feature can be the DRT distribution function value at the peak apex; and the area feature can be the integral value in the logarithmic domain of the relaxation time.

[0027] Optionally, in this invention, independent peaks of the relaxation time distribution curve are identified, and the position features, height features, and area features corresponding to the independent peaks are extracted and used as candidate relaxation time feature data.

[0028] Optionally, the optimal single feature data can be obtained by selecting the best candidate relaxation time feature data from the multiple identified candidate relaxation time feature data.

[0029] Specifically, at least one independent peak of the relaxation time distribution curve is identified based on the local maximum method; for each independent peak, the position, height, and area features of the independent peak are extracted as candidate relaxation time feature data; all candidate relaxation time feature data are optimized to determine the optimal single feature data.

[0030] S130. Input the optimal single-feature data into the pre-trained state of charge estimation model to obtain the state of charge estimation result of the lithium iron phosphate battery to be estimated.

[0031] Optionally, the state-of-charge (SOC) estimation model is a trained, regularized, lightweight convolutional neural network model used to quickly output the SOC estimation results for lithium iron phosphate batteries. The SOC estimation result refers to the ratio of the remaining capacity of the lithium iron phosphate battery to its rated total capacity, with a value ranging from 0 to 1, and is used to characterize the remaining usable capacity of the battery.

[0032] Optionally, the state of charge estimation model is trained using a training sample set covering a wide temperature range and all operating conditions, and a lightweight convolutional neural network is used. During the training process, an elastic network regularization constraint is introduced to suppress overfitting.

[0033] Optionally, in this invention, the optimal single feature data is obtained by directly selecting from the relaxation time distribution curve based on a preset feature selection rule.

[0034] Optionally, the preset feature selection rules can be rules specified based on the training sample set. For example, the preset feature selection rules can be feature filtering templates that are pre-calibrated and fixed based on the training sample set, with the preset feature selection rules being calibrated based on the training sample set in the offline stage and the filtering being performed according to the preset feature selection rules in the online stage.

[0035] Specifically, the optimal single-feature data is input into a pre-trained state-of-charge estimation model to obtain the state-of-charge estimation result of the lithium iron phosphate battery to be estimated.

[0036] The technical solution of this invention collects electrochemical impedance spectroscopy (EIS) data of the lithium iron phosphate (LFP) battery to be estimated, decouples the relaxation time distribution of the EIS data to obtain a relaxation time distribution curve, identifies the optimal single feature data in the relaxation time distribution curve based on the local maximum method, and uses the physical features after relaxation time distribution decoupling as model input to effectively overcome the estimation bottleneck of the flat region of the LFP battery OCV-SOC, solve the problems of overlapping features and weak anti-interference of the original EIS, and input the optimal single feature data into the pre-trained state of charge (SOC) estimation model to obtain the SOC estimation result of the LFP battery to be estimated. The use of a lightweight convolutional model can effectively reduce the deployment computing power threshold, adapt to the hardware of low computing power platforms, solve the problem of high-precision SOC estimation of LFP batteries on low computing power platforms in the prior art, improve the wide-temperature robustness and interpretability of SOC estimation, and balance estimation accuracy and engineering practicality.

[0037] Figure 2 This is a flowchart illustrating a method for estimating the state of charge (SOC) of a lithium iron phosphate battery according to an embodiment of the present invention. The relationship between this embodiment and the previous embodiments is that this specifically describes the training process of the SOC estimation model. Figure 2 As shown, the method includes: S210. Obtain the training sample set corresponding to the training lithium iron phosphate battery.

[0038] The training lithium iron phosphate battery can be any type of lithium iron phosphate battery. The training sample set can be a dataset obtained from experiments with lithium iron phosphate batteries.

[0039] Optionally, the training sample set consists of relaxation time single feature data and charge truth labels for the optimal electrochemical process strongly correlated with the state of charge, extracted from the electrochemical impedance spectroscopy data of the training lithium iron phosphate battery. The training sample set also includes the electrochemical impedance spectroscopy data of the training lithium iron phosphate battery corresponding to the relaxation time single feature data, treating the electrochemical impedance spectroscopy data of the training lithium iron phosphate battery and the relaxation time single feature data of the optimal electrochemical process as a set of feature samples. The training sample set includes multiple sets of feature samples and their corresponding charge truth labels.

[0040] Specifically, obtain the training sample set corresponding to the lithium iron phosphate battery.

[0041] Optionally, in another optional embodiment of the present invention, obtaining the training sample set corresponding to the training lithium iron phosphate battery includes: At least one test condition is set for the training lithium iron phosphate battery; under each test condition, electrochemical impedance spectroscopy (EIS) measurement and actual charge measurement are performed on the training lithium iron phosphate battery to determine a set of EIS experimental data and charge label data; relaxation time distribution standardization is performed on each set of EIS experimental data to obtain relaxation time single feature data of the optimal electrochemical process strongly correlated with the state of charge of the lithium iron phosphate battery; a training sample set is constructed based on the relaxation time single feature data and charge label data.

[0042] The test condition can be a set of test conditions for training lithium iron phosphate batteries. It should be noted that, in this invention, the test condition typically consists of temperature and state of charge.

[0043] Optionally, the test conditions set in this invention cover the full temperature range of -15℃ to 40℃, divided into 5 temperature nodes: -15℃, 0℃, 15℃, 25℃, and 40℃, and the temperature control accuracy is ±0.5℃, covering the temperature boundary conditions of the entire scenario.

[0044] Optionally, the test condition settings of this invention include 21 SOC test nodes ranging from 0% to 100% at 5% intervals, namely 0%, 5%, 10%, ..., 95%, and 100%, to obtain a full-range SOC test gradient. A 3600-second rest period is set after each charge and discharge cycle, and the open-circuit voltage change rate is controlled to be less than 10 μV / s to obtain a stable test state after relaxation equilibrium. This has the advantages of eliminating polarization interference and ensuring the accuracy of EIS data.

[0045] Optionally, different test conditions can be obtained from the temperature boundary conditions of the entire scenario and the test gradient of the state of charge across the entire range.

[0046] Optionally, the battery capacity tag data can be the actual battery capacity measured for lithium iron phosphate batteries, describing the percentage of remaining capacity relative to the rated capacity. By directly measuring the actual battery capacity of training lithium iron phosphate batteries under test conditions, the actual battery capacity is obtained and used as the battery capacity tag data for a single test condition.

[0047] Optionally, in this embodiment, a high-precision true battery charge can be obtained by using a current sensor with an accuracy of ±0.05%, controlling the temperature accuracy of the entire process to ±0.5℃, and allowing each SOC node to stand still until the voltage stabilizes.

[0048] Optionally, the specific process of electrochemical impedance spectroscopy measurement in this invention is as follows: by applying a sinusoidal current excitation with an amplitude of 7A to the battery, 61 logarithmically uniformly distributed frequency points from 0.01 Hz to 10 kHz are set to obtain standardized electrochemical impedance spectroscopy experimental data; the standardized electrochemical impedance spectroscopy experimental data includes impedance real part data, impedance imaginary part data, impedance amplitude data, and phase angle data.

[0049] The complex impedance response contained in the test relationship of electrochemical impedance spectroscopy measurements can describe the impedance characteristics of the battery to sinusoidal excitation. This process is governed by the following equation: in, It is the real part of the impedance parameter, reflecting ohmic loss and charge transfer loss; Z(ω) is the imaginary part of the impedance parameter, reflecting the capacitance and polarization characteristics; Z(ω) is the impedance magnitude parameter, characterizing the overall impedance of the battery; θ(ω) is the phase angle parameter, reflecting the phase characteristics of the battery impedance, and j is the imaginary unit.

[0050] In this embodiment, the phase angle included in the phase angle calculation relationship can be a process describing the angle between the real and imaginary parts of the impedance. This process is controlled by the following equation: in, It is the real part of the impedance parameter. It is the imaginary part of the impedance parameter.

[0051] In this embodiment, the EIS data of 21 SOC nodes under 5 temperature conditions can be structured and organized to obtain 105 sets of electrochemical impedance spectroscopy experimental data and charge label data containing the real part, imaginary part, amplitude, phase angle and terminal voltage of impedance at 61 frequency points. This has the advantage of providing a standardized data source for subsequent DRT decoupling and model training.

[0052] Specifically, at least one test condition is set for the training lithium iron phosphate battery; under each test condition, electrochemical impedance spectroscopy (EIS) measurement and actual charge measurement are performed on the training lithium iron phosphate battery to determine a set of EIS experimental data and charge label data; relaxation time distribution standardization is performed on each set of EIS experimental data to obtain relaxation time single feature data of the optimal electrochemical process strongly correlated with the state of charge of the lithium iron phosphate battery; a training sample set is constructed based on the relaxation time single feature data and charge label data.

[0053] Optionally, in another optional embodiment of the present invention, the relaxation time distribution of each set of electrochemical impedance spectroscopy experimental data is standardized to obtain single-feature data of the relaxation time of the optimal electrochemical process strongly correlated with the state of charge of the lithium iron phosphate battery, including: For each set of electrochemical impedance spectroscopy experimental data, relaxation time distribution analysis was performed on the electrochemical impedance spectroscopy experimental data to determine the relaxation time distribution curve; curve extraction was performed on the relaxation time distribution curve based on the local maximum method to determine at least one candidate relaxation time feature data; all relaxation feature data were optimally screened using the random forest algorithm to determine the optimal relaxation time single feature data of the electrochemical process.

[0054] Optionally, relaxation time distribution analysis can be performed on each set of electrochemical impedance spectroscopy experimental data, specifically as follows: By splitting each set of electrochemical impedance spectroscopy experimental data into the real part of impedance vector, the imaginary part of impedance vector, and the frequency vector, standardized EIS vector data are obtained. Converting frequency and relaxation time to a logarithmic scale yields logarithmically scaled variable data; the logarithmic scale variable included in the variable transformation relationship can describe the mapping process between frequency and relaxation time. This process is governed by the following equation: Where ω is the angular frequency parameter; f is the measurement frequency parameter; and τ is the relaxation time parameter, corresponding to the time constant of a single RC parallel unit.

[0055] The relaxation time distribution transformation process uses the DRT integral equation transformation relation to transform the DRT integral equation into a logarithmic scale form. The logarithmic scale DRT distribution describes the relaxation time distribution transformation process, which is governed by the following equation: in, It is the high-frequency limiting ohmic impedance parameter, taken as the Z' value corresponding to f = 10 kHz; These are the DRT distribution parameters on a logarithmic scale, which are the final solution objectives.

[0056] By transforming the original integral equation into a form with a lower condition number through logarithmic transformation, an optimized DRT solution equation is obtained, which has the advantages of avoiding solution oscillations and improving numerical stability. Furthermore, 610 discrete points are uniformly selected within the logarithmic range of the relaxation time to obtain discrete node data for the DRT solution, ensuring the resolution of the DRT curve. The discrete node positions included in the selection of discrete points can describe the process of relaxation time discretization.

[0057] By substituting the expression of the discrete approximation of the continuous DRT distribution into the integral equation, a system matrix of 2Nf×M dimensions and an observation vector are constructed, resulting in a linear equation system for solving the DRT. This approach has the advantages of transforming the integral problem into a linear algebra problem and facilitating numerical solutions.

[0058] Directly solving the linear equation system of the DRT leads to severe oscillations and non-uniqueness of the solution. To suppress the ill-conditioned nature of the inverse problem, a cost function with first-order Tikhonov regularization is constructed, and its governing equation is: Where λ is the regularization parameter, controlling the smoothness of the solution; L is the parameter of the (M-1)×M dimensional first-order forward difference matrix; A represents the system matrix, and b represents the observation vector. By plotting the L-curve and selecting the λ value at the inflection point as the optimal regularization parameter, and by differentiating the cost function J(x) with respect to x and setting it to zero, we obtain the system of equations that the optimal solution should satisfy: The equation system can be solved using the QR decomposition method to obtain the basis function coefficients, which can then be substituted into the expression of the discrete approximation of the continuous DRT distribution to obtain the final relaxation time distribution curve. This method has the advantages of high numerical stability and excellent solution accuracy.

[0059] Optionally, for the calculated relaxation time distribution curve, independent peaks in the DRT curve are identified using the local maximum method. Three types of features—position, height, and area—are extracted for each peak, and these are used as candidate relaxation time feature data. Each candidate relaxation time feature data typically corresponds to the relaxation time range of the independent peak on the relaxation time distribution curve. By mapping the relaxation time range to the corresponding internal electrochemical processes of the battery, the correlation between the internal electrochemical processes of each battery and the state of charge (SOC) is obtained.

[0060] In this embodiment, N can be constructed by integrating electrochemical impedance spectroscopy data, relaxation time distribution decoupling characteristics, and auxiliary characteristics. sample A multi-source feature set of ×(4Nf+3Np+1) dimensions yields a complete feature library covering the original measurement data and decoupled physical features, providing comprehensive candidate features for subsequent feature selection. In this embodiment, 244-dimensional electrochemical impedance spectroscopy data can be obtained by extracting the real part, imaginary part, amplitude, and phase angle of impedance at each frequency point, preserving the integrity of the original measurement information; 12-dimensional relaxation time distribution decoupling features can be obtained by extracting the position, height, and area of ​​each independent peak, preserving physically meaningful relaxation process information; and 1-dimensional auxiliary features can be obtained by introducing the battery terminal voltage as an auxiliary feature, supplementing the macroscopic state information of the battery.

[0061] Specifically, for each set of electrochemical impedance spectroscopy experimental data, relaxation time distribution analysis is performed on the electrochemical impedance spectroscopy experimental data to determine the relaxation time distribution curve; curve extraction is performed on the relaxation time distribution curve based on the local maximum method to determine at least one candidate relaxation time feature data; and the optimal relaxation time single feature data of the optimal electrochemical process is determined by using the random forest algorithm to perform optimal screening of all relaxation feature data.

[0062] Optionally, in another optional embodiment of the present invention, the optimal relaxation time feature data of the electrochemical process is determined by optimally filtering all candidate relaxation time feature data using a random forest algorithm, including: The Gini importance of each candidate relaxation time feature data was calculated using the random forest regression algorithm with the charge label data as the target variable, and the contribution ranking of each candidate relaxation time feature data was obtained. The candidate relaxation time feature data with the highest contribution ranking was determined as the relaxation time single feature data of the optimal electrochemical process.

[0063] The contribution ranking can be a ranking of the contribution scores of candidate relaxation time feature data to the battery state of charge estimation. It should be noted that by quantifying the Gini importance into a contribution score to represent the relevance of candidate relaxation time feature data to the battery state of charge estimation, it has the advantages of significantly reducing feature dimensionality, reducing computational load, and ensuring estimation accuracy.

[0064] Optionally, the Gini importance of each candidate relaxation time feature data can be calculated using the random forest algorithm to identify the contribution of the candidate relaxation time feature data to the battery state of charge estimation.

[0065] Specifically, the Gini importance of each candidate relaxation time feature data is calculated using the random forest regression algorithm with the electric charge label data as the target variable, and the contribution ranking of each candidate relaxation time feature data is obtained; the candidate relaxation time feature data with the highest contribution ranking is determined as the relaxation time single feature data of the optimal electrochemical process.

[0066] S220. Input each set of feature samples and the corresponding true charge value label of the training sample set into the original lightweight convolutional neural network for training, and obtain the training loss and validation loss.

[0067] Optionally, for the feature samples and their corresponding charge truth labels, the electrochemical impedance spectroscopy data of the feature samples are normalized by min-max to map the data to the [0,1] interval, so as to eliminate dimensional differences and improve the convergence speed of the model.

[0068] Optionally, after normalizing each set of feature samples and their corresponding charge truth labels in the training sample set, normalized electrochemical impedance spectroscopy data, relaxation time single feature data of the optimal electrochemical process, and charge truth labels can be obtained.

[0069] Optionally, the present invention can also support two types of standardized datasets constructed from electrochemical impedance spectroscopy data and charge truth labels or relaxation time single feature data of the optimal electrochemical process and charge truth labels, in order to support different algorithm verification needs.

[0070] Optionally, in this embodiment, the quality of the generated dataset can be guaranteed through multi-dimensional verification of label consistency, feature integrity, distribution uniformity, and abnormal sample rate, which has the advantage of providing a reliable data foundation for subsequent model training.

[0071] Optionally, after obtaining the normalized training sample set, the training sample set is divided into a training set, a validation set, and a test set in a 7:2:1 ratio. Each set of feature samples and its corresponding ground truth battery value label in the training set are used as training data for model training.

[0072] Optionally, the original lightweight convolutional neural network can be a pre-built lightweight convolutional neural network. The original lightweight convolutional neural network consists of convolutional layers, fully connected layers, and an output layer. The convolutional layers and fully connected layers use the ReLU activation function to enhance nonlinear fitting ability, and the output layer uses the Sigmoid activation function to map the output to the [0,1] interval. During the training process, elasticity regularization is introduced to suppress model overfitting, the Adam optimizer is used to perform the parameter update process, and an early stopping mechanism is added to prevent model overfitting, thereby improving training efficiency and model generalization ability.

[0073] Optionally, the training loss and validation loss are the loss differences between the output of each set of feature samples during forward propagation and the ground truth battery label during training; and the loss differences between the output of each set of feature samples during forward propagation and the ground truth battery label during validation. The training loss is used to calculate the gradients of the weights and bias parameters of each layer, and the Adam optimizer is used to update the model parameters along the gradient descent direction. The validation loss is used to evaluate the model performance of the original lightweight convolutional neural network during training.

[0074] Specifically, each set of feature samples and its corresponding ground truth label of the battery power are input into the original lightweight convolutional neural network for training, and the training loss and validation loss are obtained.

[0075] S230. Iteratively update the weights and bias parameters of the original lightweight convolutional neural network, and repeatedly execute the steps of inputting each set of feature samples and corresponding true charge values ​​of the training sample set into the original lightweight convolutional neural network for training, and obtaining the training loss and validation loss, until the obtained validation loss no longer decreases or the number of iterations reaches the set threshold, and determine the state of charge estimation model.

[0076] Optionally, if the validation loss fails to achieve the preset performance, the weights and bias parameters of the original lightweight convolutional neural network are iteratively updated. This involves repeatedly inputting each set of feature samples from the training sample set along with their corresponding ground truth battery values ​​into the original lightweight convolutional neural network for training, and obtaining the training loss and validation loss. The preset performance can be defined as the validation loss being less than a pre-set loss threshold.

[0077] The criteria for stopping the model training and validation process can be either that the validation loss stops decreasing or the number of iterations reaches a certain threshold. Specifically, "the validation loss stops decreasing" can mean that the validation loss stops decreasing for multiple consecutive iterations; "the number of iterations reaches a certain threshold" can mean that the number of training iterations reaches a certain threshold. Training stops when either condition is met.

[0078] Optionally, during the training of the original lightweight convolutional neural network, the regularization parameter for decoupling the relaxation time distribution, the basis function width, the number of convolutional kernels, the number of neurons in the fully connected layer, and the regularization coefficient of the elastic network are respectively tuned and learned. The model generated under each parameter configuration is used for state-of-charge estimation under various operating conditions across the entire temperature range from -15℃ to 40℃, considering multiple challenging scenarios including wide-temperature operating conditions and the OCV (Open Circuit Voltage) flat region. By comparing the mean absolute error and single-sample inference delay under different parameter values, the optimal combination of parameter values ​​is determined. For example, experiments of this invention show that the regularization parameter for decoupling the relaxation time distribution is 10. -3 The basis function width is 0.1. The original lightweight convolutional neural network has 64 kernels in the first layer, 32 kernels in the second layer, 64 and 32 neurons in the fully connected layers, and a regularization coefficient of 10 for the elastic network. -4 At this point, the overall performance of the original lightweight convolutional neural network reaches its optimal level. Further increasing the parameter values ​​will only increase computation and memory requirements, while the improvement in estimation accuracy will be negligible. The optimal relaxation time range for decoupling the relaxation time distribution is determined to be 1-20s. This range is consistent with the dynamic process of lithium-ion solid-phase diffusion in LFP cathodes and can effectively separate the core electrochemical processes strongly correlated with the state of charge.

[0079] Optionally, when training the state-of-charge estimation model using the original lightweight convolutional neural network, this invention can further optimize the parameters of the entire process of the state-of-charge estimation model. First, the regularization parameter for relaxation time distribution decoupling, the basis function width, the number of convolutional kernels, the number of neurons in the fully connected layer, and the range of values ​​for the elastic network regularization coefficient are determined by the control variable method. Second, by comparing the mean absolute error, coefficient of determination, and single-sample inference delay under different parameter combinations, the parameter combination with the best overall performance is selected. Finally, the state-of-charge estimation model is reconstructed and trained using the optimal parameters to comprehensively improve its estimation performance and engineering adaptability.

[0080] During the optimization process, controlled variable experiments and multi-index comprehensive evaluation were used to iteratively adjust key parameters, balancing estimation accuracy, computational complexity, and real-time requirements. Experimental results show that the proposed method maintains the state-of-charge estimation error of the LFP battery OCV flat region below 2.75% under a wide temperature range of -15℃ to 40℃. Specifically, the weights and bias parameters of the original lightweight convolutional neural network are iteratively updated. The process involves repeatedly inputting each set of feature samples from the training sample set and the corresponding ground truth labels of the charge values ​​into the original lightweight convolutional neural network for training, and obtaining the training loss and validation loss. This process continues until the validation loss no longer decreases or the number of iterations reaches a set threshold, thus determining the state of charge estimation model.

[0081] S240. Collect electrochemical impedance spectroscopy data of the lithium iron phosphate battery to be estimated, decouple the relaxation time distribution of the electrochemical impedance spectroscopy data, and obtain the relaxation time distribution curve.

[0082] S250. Identify the optimal single feature data in the relaxation time distribution curve based on the local maximum method.

[0083] S260. Input the optimal single-feature data into the pre-trained state of charge estimation model to obtain the state of charge estimation result of the lithium iron phosphate battery to be estimated.

[0084] The technical solution of this invention collects electrochemical impedance spectroscopy (EIS) data of the lithium iron phosphate (LFP) battery to be estimated, decouples the relaxation time distribution of the EIS data to obtain a relaxation time distribution curve, identifies the optimal single feature data in the relaxation time distribution curve based on the local maximum method, and uses the physical features after relaxation time distribution decoupling as model input to effectively overcome the estimation bottleneck of the flat region of the LFP battery OCV-SOC, solve the problems of overlapping features and weak anti-interference of the original EIS, and input the optimal single feature data into the pre-trained state of charge (SOC) estimation model to obtain the SOC estimation result of the LFP battery to be estimated. The use of a lightweight convolutional model can effectively reduce the deployment computing power threshold, adapt to the hardware of low computing power platforms, solve the problem of high-precision SOC estimation of LFP batteries on low computing power platforms in the prior art, improve the wide-temperature robustness and interpretability of SOC estimation, and balance estimation accuracy and engineering practicality.

[0085] Figure 3 This is a schematic diagram of a state-of-charge estimation device for a lithium iron phosphate battery provided in an embodiment of the present invention. Figure 3 As shown, the device includes: a data acquisition module 310, an optimal feature extraction module 320, and a model recognition module 330; wherein, The data acquisition module 310 is used to acquire the electrochemical impedance spectroscopy data of the lithium iron phosphate battery to be estimated, and to decouple the relaxation time distribution of the electrochemical impedance spectroscopy data to obtain the relaxation time distribution curve. The optimal feature extraction module 320 is used to identify the optimal single feature data in the relaxation time distribution curve based on the local maximum method. The model recognition module 330 is used to input the optimal single feature data into the pre-trained state of charge estimation model to obtain the state of charge estimation result of the lithium iron phosphate battery to be estimated.

[0086] The technical solution of this invention collects electrochemical impedance spectroscopy (EIS) data of the lithium iron phosphate (LFP) battery to be estimated, decouples the relaxation time distribution of the EIS data to obtain a relaxation time distribution curve, identifies the optimal single feature data in the relaxation time distribution curve based on the local maximum method, and uses the physical features after relaxation time distribution decoupling as model input to effectively overcome the estimation bottleneck of the flat region of the LFP battery OCV-SOC, solve the problems of overlapping features and weak anti-interference of the original EIS, and input the optimal single feature data into the pre-trained state of charge (SOC) estimation model to obtain the SOC estimation result of the LFP battery to be estimated. The use of a lightweight convolutional model can effectively reduce the deployment computing power threshold, adapt to the hardware of low computing power platforms, solve the problem of high-precision SOC estimation of LFP batteries on low computing power platforms in the prior art, improve the wide-temperature robustness and interpretability of SOC estimation, and balance estimation accuracy and engineering practicality.

[0087] Optionally, the optimal feature extraction module 320 is specifically used for: Identify at least one independent peak in the relaxation time distribution curve based on the local maximum method; For each independent peak, the position, height, and area features of the independent peak are extracted as candidate relaxation time feature data. The optimal single feature data is determined by optimizing all the candidate relaxation time feature data.

[0088] The device further includes: a training sample acquisition module, a model training module, and a model iteration module; wherein... The training sample acquisition module is used to acquire the training sample set corresponding to the training lithium iron phosphate battery; wherein, the training sample set includes multiple sets of feature samples and the corresponding true value label of the power of each feature sample; The model training module is used to input each set of feature samples and the corresponding ground truth labels of the battery power into the original lightweight convolutional neural network for training, and to obtain the training loss and validation loss. The model iteration module is used to iteratively update the weights and bias parameters of the original lightweight convolutional neural network. It iteratively executes the steps of inputting each set of feature samples and corresponding true charge values ​​of the training sample set into the original lightweight convolutional neural network for training, and obtaining the training loss and validation loss, until the obtained validation loss no longer decreases or the number of iterations reaches a set threshold, thus determining the state of charge estimation model.

[0089] Optionally, the training sample acquisition module is specifically used for: Set at least one test condition corresponding to the training lithium iron phosphate battery; Under each test condition, electrochemical impedance spectroscopy (EIS) measurements and actual charge measurements were performed on the trained lithium iron phosphate batteries to determine a set of EIS experimental data and charge label data. The relaxation time distribution of each set of electrochemical impedance spectroscopy experimental data was standardized to obtain the relaxation time single feature data of the optimal electrochemical process with strong correlation to the state of charge of lithium iron phosphate battery. A training sample set was constructed based on relaxation time single feature data and power tag data.

[0090] Optionally, the training sample acquisition module is also specifically used for: For each set of electrochemical impedance spectroscopy experimental data, relaxation time distribution analysis was performed to determine the relaxation time distribution curve. Curve extraction of the relaxation time distribution curve is performed based on the local maximum method to determine at least one candidate relaxation time feature data. The optimal relaxation time single feature data for the electrochemical process was determined by using the random forest algorithm to perform optimal screening of all relaxation feature data.

[0091] Optionally, the training sample acquisition module is also specifically used for: The Gini importance of each candidate relaxation time feature data was calculated using the random forest regression algorithm with the electricity tag data as the target variable, and the contribution ranking of each candidate relaxation time feature data was obtained. The candidate relaxation time feature data with the highest contribution ranking is determined as the single feature data of the relaxation time of the optimal electrochemical process.

[0092] The state-of-charge estimation device for lithium iron phosphate batteries provided in this embodiment of the invention can execute the state-of-charge estimation method for lithium iron phosphate batteries provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0093] Figure 4 A schematic diagram of an electronic device 10 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their patterns are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0094] like Figure 4 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0095] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0096] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as the state-of-charge estimation method for lithium iron phosphate batteries.

[0097] In some embodiments, the state-of-charge (SOC) estimation method for lithium iron phosphate batteries may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded into and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the SOC estimation method for lithium iron phosphate batteries described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the SOC estimation method for lithium iron phosphate batteries by any other suitable means (e.g., by means of firmware).

[0098] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0099] Computer programs used to implement the methods of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the patterns / operations specified in the flowcharts and / or block diagrams to be implemented. The computer programs can be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0100] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0101] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0102] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0103] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS servers, such as high management difficulty and weak business scalability.

[0104] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0105] This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the state-of-charge estimation method steps for lithium iron phosphate batteries as provided in any embodiment of the present invention. The method includes: Electrochemical impedance spectroscopy (EIS) data of the lithium iron phosphate battery to be estimated were collected, and the relaxation time distribution of the EIS data was decoupled to obtain the relaxation time distribution curve. Identifying the optimal single feature data in the relaxation time distribution curve based on the local maximum method; The optimal single-feature data is input into a pre-trained state-of-charge estimation model to obtain the state-of-charge estimation result of the lithium iron phosphate battery to be estimated.

[0106] The computer storage medium of this invention can be any combination of one or more computer-readable media. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0107] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of sending, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0108] Program code contained on a computer-readable medium may be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0109] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0110] Those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computing device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0111] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0112] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for estimating the state of charge (SOC) of a lithium iron phosphate battery, characterized in that, include: Electrochemical impedance spectroscopy (EIS) data of the lithium iron phosphate battery to be estimated were collected, and the relaxation time distribution of the EIS data was decoupled to obtain the relaxation time distribution curve. Identifying the optimal single feature data in the relaxation time distribution curve based on the local maximum method; The optimal single-feature data is input into a pre-trained state-of-charge estimation model to obtain the state-of-charge estimation result of the lithium iron phosphate battery to be estimated.

2. The method according to claim 1, characterized in that, Identifying the optimal single feature data in the relaxation time distribution curve based on the local maximum method, including: Identify at least one independent peak in the relaxation time distribution curve based on the local maximum method; For each independent peak, the position, height, and area features of the independent peak are extracted as candidate relaxation time feature data. The optimal single feature data is determined by optimizing all the candidate relaxation time feature data.

3. The method according to claim 1, before inputting the optimal single-feature data into the pre-trained state-of-charge estimation model to obtain the state-of-charge estimation result of the lithium iron phosphate battery to be estimated, further includes: Obtain the training sample set corresponding to the lithium iron phosphate battery; wherein, the training sample set includes multiple sets of feature samples and the corresponding ground truth labels of the power capacity of the feature samples; Each set of feature samples and its corresponding ground truth label of the battery power is input into the original lightweight convolutional neural network for training, and the training loss and validation loss are obtained. The weights and bias parameters of the original lightweight convolutional neural network are iteratively updated. The process of inputting each set of feature samples and corresponding true charge values ​​of the training sample set into the original lightweight convolutional neural network for training and obtaining the training loss and validation loss is repeated until the validation loss no longer decreases or the number of iterations reaches a set threshold, thus determining the state of charge estimation model.

4. The method according to claim 3, characterized in that, Obtain the training sample set corresponding to the lithium iron phosphate battery, including: Set at least one test condition corresponding to the training lithium iron phosphate battery; Under each test condition, electrochemical impedance spectroscopy (EIS) measurements and actual charge measurements were performed on the trained lithium iron phosphate batteries to determine a set of EIS experimental data and charge label data. The relaxation time distribution of each set of electrochemical impedance spectroscopy experimental data was standardized to obtain the relaxation time single feature data of the optimal electrochemical process with strong correlation to the state of charge of lithium iron phosphate battery. A training sample set was constructed based on relaxation time single feature data and power tag data.

5. The method according to claim 4, characterized in that, The relaxation time distribution of each set of electrochemical impedance spectroscopy experimental data was standardized to obtain single-feature data of the relaxation time of the optimal electrochemical process strongly correlated with the state of charge of lithium iron phosphate batteries, including: For each set of electrochemical impedance spectroscopy experimental data, relaxation time distribution analysis was performed to determine the relaxation time distribution curve. Curve extraction of the relaxation time distribution curve is performed based on the local maximum method to determine at least one candidate relaxation time feature data. The optimal relaxation time single feature data for the electrochemical process was determined by using the random forest algorithm to perform optimal screening of all relaxation feature data.

6. The method according to claim 5, characterized in that, The optimal relaxation time feature data for the electrochemical process was determined by using a random forest algorithm to filter all candidate relaxation time feature data. This data includes: The Gini importance of each candidate relaxation time feature data was calculated using the random forest regression algorithm with the electricity tag data as the target variable, and the contribution ranking of each candidate relaxation time feature data was obtained. The candidate relaxation time feature data with the highest contribution ranking is determined as the single feature data of the relaxation time of the optimal electrochemical process.

7. A state-of-charge estimation device for a lithium iron phosphate battery, characterized in that, include: The data acquisition module is used to acquire electrochemical impedance spectroscopy data of the lithium iron phosphate battery to be estimated, and to decouple the relaxation time distribution of the electrochemical impedance spectroscopy data to obtain the relaxation time distribution curve. The optimal feature extraction module is used to identify the optimal single feature data in the relaxation time distribution curve based on the local maximum method. The model recognition module is used to input the optimal single feature data into the pre-trained state of charge estimation model to obtain the state of charge estimation result of the lithium iron phosphate battery to be estimated.

8. The apparatus according to claim 7, characterized in that, Also includes: The system comprises a training sample acquisition module, a model training module, and a model iteration module; among which, The training sample acquisition module is used to acquire the training sample set corresponding to the training lithium iron phosphate battery; wherein, the training sample set includes multiple sets of feature samples and the corresponding true value label of the power of each feature sample; The model training module is used to input each set of feature samples and the corresponding ground truth labels of the battery power into the original lightweight convolutional neural network for training, and to obtain the training loss and validation loss. The model iteration module is used to iteratively update the weights and bias parameters of the original lightweight convolutional neural network. It iteratively executes the steps of inputting each set of feature samples and corresponding true charge values ​​of the training sample set into the original lightweight convolutional neural network for training, and obtaining the training loss and validation loss, until the obtained validation loss no longer decreases or the number of iterations reaches a set threshold, thus determining the state of charge estimation model.

9. An electronic device, characterized in that, Electronic devices include: At least one processor; and A memory that is communicatively connected to at least one processor; wherein, The memory stores a computer program that can be executed by at least one processor, such that the at least one processor is able to perform the method for estimating the state of charge of a lithium iron phosphate battery according to any one of claims 1-6.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing a processor to execute the method for estimating the state of charge of a lithium iron phosphate battery according to any one of claims 1-6.

Images