Battery soc-soh joint estimation method and system based on EIS rapid detection data

Abstract

The application provides a battery SOC-SOH joint estimation method and system based on EIS rapid detection data driving, which comprises: generating a time-domain multi-frequency composite excitation current signal applied to a battery, obtaining an excitation current and a response voltage, and then rapidly detecting a battery electrochemical impedance spectrum (EIS) through a Fourier transform recursive filtering algorithm; extracting and calculating battery state of charge (SOC)-state of health (SOH) sensitive EIS frequency point characteristic parameters from the battery electrochemical impedance spectrum; establishing a multi-decision tree ensemble model through Bootstrap sampling and feature random selection mechanism; inputting the battery SOC-SOH sensitive EIS frequency point characteristic parameters into the multi-decision tree ensemble model to realize joint estimation of the battery SOC and SOH. Compared with a traditional sweep frequency impedance test method, the single measurement time is shortened to 90 seconds (wide frequency band 0.001 Hz-1000 Hz) or 1 second (medium-high frequency band 1 Hz-1000 Hz), and the problem that the traditional method takes too long to measure and cannot be used for online detection in the dynamic working process of the battery is solved.

Description

Technical Field

[0001] This invention relates to the field of battery intelligent management technology, specifically to a battery SOC-SOH joint estimation method and system driven by EIS rapid detection data. Background Technology

[0002] In recent years, with the transformation of the energy structure and the continuous improvement of new energy sources, batteries have been widely used in energy storage systems and electric vehicles. Among them, lithium-ion batteries, with their advantages of high energy density, low self-discharge rate, and no memory effect, are currently the best choice for electric vehicle (EV) power sources. The battery's state of charge (SOC) reflects the remaining capacity level, while the state of health (SOH) characterizes the degree of performance degradation. These two are strongly coupled, and accurate joint estimation is a core prerequisite for ensuring safe battery operation, optimizing lifespan management, and improving system reliability. To reliably manage battery life and safety, it is necessary to first accurately estimate the battery's state of charge (SOC) and state of health (SOH).

[0003] However, the complex structure, working principle, and performance degradation mechanism of lithium-ion batteries pose significant challenges to the accurate estimation of SOC and SOH. Existing battery state estimation technologies mainly fall into two major categories, both of which have significant technical limitations: The first category is estimation methods based on external electrical characteristics such as voltage, current, and temperature, which are currently the mainstream engineering approach. While these methods can provide SOC and SOH information to some extent, the complex relationships between these external parameters, especially their inability to delve into the internal mechanisms and characteristics of the battery, make it difficult to reliably guarantee their accuracy. Filtering algorithms based on equivalent circuit models (such as patent CN121522484A, which uses a second-order RC model combined with RLS parameter identification with a forgetting factor and dual SRCKF filtering) simulate external battery characteristics through circuit elements. However, these models cannot fully characterize the battery's internal electrode reactions, ion diffusion, SEI film growth, and other degradation mechanisms. Under dynamic operating conditions, wide temperature ranges, and long-term aging scenarios, the model mismatch is severe, resulting in poor generalization of the estimation. Data-driven estimation methods (such as patent CN121208632A, which uses large models to mine time-series features) improve estimation performance by leveraging big data learning capabilities, but they still rely on external parameters such as voltage and current and cannot directly correlate with the battery's internal electrochemical characteristics, making it difficult to fundamentally overcome the estimation accuracy bottleneck.

[0004] The second category is state estimation methods based on electrochemical impedance spectroscopy (EIS). EIS can separate parameters such as ohmic impedance, charge transfer impedance, and diffusion impedance, and has a deterministic mapping relationship with SOC and SOH, making it an important direction for high-precision estimation. However, traditional EIS uses frequency-by-frequency scanning detection, with measurement times reaching tens of minutes, which cannot meet the online detection requirements during battery dynamic operation. Existing EIS estimation methods do not specifically mine the sensitive frequency characteristics of SOC-SOH, and mostly use redundant data across the entire frequency band, resulting in low computational efficiency. Furthermore, they lack a deep fusion scheme between fast EIS detection and data-driven models, making it difficult to achieve real-time joint estimation under dynamic operating conditions.

[0005] In summary, existing technologies face three core technical challenges that urgently need to be addressed: traditional estimation methods based on external characteristics fail to consider the internal electrochemical mechanisms of the battery, resulting in insufficient estimation accuracy and robustness under dynamic operating conditions and aging cycles; traditional EIS detection is too time-consuming, making it impossible to achieve rapid online measurement and difficult to adapt to dynamic battery operating scenarios; and existing EIS state estimation methods fail to explore the sensitive frequency characteristics of SOC-SOH, lack a fusion scheme of rapid detection and data-driven models, and cannot achieve high-precision and robust joint estimation of SOC and SOH.

[0006] Therefore, there is an urgent need in the market for a data-driven battery state estimation method based on rapid EIS detection, which can achieve rapid EIS detection through time-domain multi-frequency composite excitation, mine sensitive frequency features and build an integrated learning model to achieve accurate and real-time joint estimation of SOC-SOH from the internal mechanism level of the battery. Summary of the Invention

[0007] To address the shortcomings of existing technologies, the present invention aims to provide a battery SOC-SOH joint estimation method and system based on EIS rapid detection data. The aim is to propose a rapid online detection method for battery EIS, and then combine it with sensitive feature parameters at sensitive frequency points to establish a multi-decision tree ensemble model. By utilizing the deterministic mapping relationship between EIS and the internal mechanism and internal characteristic state of the battery, the accuracy and robustness of the estimation results can be guaranteed.

[0008] A battery SOC-SOH joint estimation method based on EIS fast detection data provided by the present invention includes: Step S1: Generate a time-domain multi-frequency composite excitation current signal and apply it to the battery to obtain the excitation current and response voltage. Then, use the Fourier transform recursive filtering algorithm to quickly detect the battery electrochemical impedance spectroscopy (EIS). Step S2: Extract and calculate the characteristic parameters of the sensitive EIS frequency points of the battery's state of charge (SOC) and state of health (SOH) from the battery's electrochemical impedance spectroscopy; Step S3: Establish a multi-decision tree ensemble model through Bootstrap sampling and random feature selection mechanisms; Step S4: Input the sensitive EIS frequency point feature parameters of the battery's state of charge (SOC) and state of health (SOH) into the multi-decision tree ensemble model to achieve joint estimation of battery SOC and SOH.

[0009] Preferably, the time-domain multi-frequency composite excitation current signal is a discrete binary signal (DIBS). The target signal frequency band is divided into three segments from low to high: low frequency band, mid frequency band, and high frequency band. The parameters of the excitation signal DIBS for each frequency band are set respectively, so that the discrete binary sequence has a spectral distribution corresponding to the target signal, thereby obtaining a DIBS signal containing all set frequency points.

[0010] Preferably, the parameters of the excitation signal DIBS include sequence length N and sampling frequency. Frequency resolution Effective frequency point Clock cycle And the sequence period T.

[0011] Preferably, in step S2, Spearman correlation analysis is used to determine the frequency points where the absolute value of the correlation coefficient between the frequency point characteristic quantity and SOC and SOH at each temperature is >0.3 as the sensitive EIS frequency points of the battery's state of charge (SOC) - state of health (SOH). The battery's state of charge (SOC) and state of health (SOH) sensitive EIS frequency characteristic parameters include the real part of impedance and the impedance modulus.

[0012] Preferably, the multi-decision tree ensemble model is a Coulomb integral random forest algorithm model, including a Coulomb integral channel and a random forest channel; The random forest channel uses the random forest algorithm to predict both battery SOC and SOH. The coulomb integration channel calculates the battery's state of charge (SOC) by integrating the current over time.

[0013] According to the present invention, a battery SOC-SOH joint estimation system driven by EIS fast detection data is provided, comprising: Module M1: Generates a time-domain multi-frequency composite excitation current signal and applies it to the battery to obtain the excitation current and response voltage. Then, through a Fourier transform recursive filtering algorithm, it quickly detects the battery electrochemical impedance spectroscopy (EIS). Module M2: Extracts and calculates the sensitive EIS frequency characteristic parameters of the battery's state of charge (SOC) and state of health (SOH) from the battery's electrochemical impedance spectroscopy; Module M3: Establishes a multi-decision tree ensemble model through Bootstrap sampling and random feature selection mechanisms; Module M4: Inputs the sensitive EIS frequency point characteristic parameters of the battery's State of Charge (SOC) and State of Health (SOH) into the multi-decision tree ensemble model to achieve joint estimation of battery SOC and SOH.

[0014] Preferably, the time-domain multi-frequency composite excitation current signal is a discrete binary signal (DIBS). The target signal frequency band is divided into three segments from low to high: low frequency band, mid frequency band, and high frequency band. The parameters of the excitation signal DIBS for each frequency band are set respectively, so that the discrete binary sequence has a spectral distribution corresponding to the target signal, thereby obtaining a DIBS signal containing all set frequency points.

[0015] Preferably, the parameters of the excitation signal DIBS include sequence length N and sampling frequency. Frequency resolution Effective frequency point Clock cycle And the sequence period T.

[0016] Preferably, in module M2, Spearman correlation analysis is used to determine the frequency points where the absolute value of the correlation coefficient between the frequency point characteristic quantity and SOC and SOH at each temperature is >0.3 as the sensitive EIS frequency points of the battery's state of charge (SOC) - state of health (SOH). The battery's state of charge (SOC) and state of health (SOH) sensitive EIS frequency characteristic parameters include the real part of impedance and the impedance modulus.

[0017] Preferably, the multi-decision tree ensemble model is a Coulomb integral random forest algorithm model, including a Coulomb integral channel and a random forest channel; The random forest channel uses the random forest algorithm to predict both battery SOC and SOH. The coulomb integration channel calculates the battery's state of charge (SOC) by integrating the current over time.

[0018] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention uses time-domain multi-frequency composite excitation (DIBS signal) and Goertzel algorithm to quickly detect EIS. Compared with the traditional swept frequency impedance test method, it shortens the single measurement time to within 90 seconds (wide frequency band 0.001Hz~1000Hz) or within 1 second (mid-to-high frequency band 1Hz~1000Hz), solving the problem that the traditional method has too long measurement time and cannot be detected online during the dynamic operation of the battery.

[0019] 2. This invention achieves accurate intrinsic joint estimation of SOC-SOH by relying on the impedance characteristic parameters at the sensitive EIS frequency point of the battery state of SOC-SOH and the deterministic mapping relationship between EIS and the internal mechanism of the battery, compared with the existing estimation methods that rely on external parameters such as voltage and current. This significantly improves the accuracy and robustness of the estimation results.

[0020] 3. This invention constructs a multi-decision tree ensemble model based on Coulomb integral and random forest, combined with a channel switching mechanism. Compared with a single model, this model not only suppresses overfitting but also adapts to high-rate charging and discharging scenarios, ensuring the stability and reliability of battery state estimation under different operating conditions. Attached Figure Description

[0021] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a schematic diagram of the working method of the present invention. Detailed Implementation

[0022] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0023] Example 1 According to the present invention, a battery SOC-SOH joint estimation method based on EIS fast detection data is provided, such as... Figure 1 As shown, it includes: Step S1: Rapid Measurement of Battery Electrochemical Impedance Spectroscopy (EIS) Based on Time-Domain Multi-Frequency Composite Excitation. A time-domain multi-frequency composite excitation current signal is generated and applied to the battery to obtain the excitation current and response voltage. Then, the battery electrochemical impedance spectroscopy (EIS) is rapidly detected using a Fourier transform recursive filtering algorithm. The time-domain multi-frequency composite excitation current signal is a discrete binary signal (DIBS). The target signal frequency band is divided into three segments from low to high: low frequency, mid frequency, and high frequency. The parameters of the DIBS excitation signal for each frequency segment are set, ensuring that the discrete binary sequence has a spectral distribution corresponding to the target signal, thus obtaining a DIBS signal containing all set frequency points. The sampling frequency of each of the three frequency bands is at least twice the upper limit of the frequency band. The parameters of the excitation signal DIBS include the sequence length N and the sampling frequency. Frequency resolution Effective frequency point Clock cycle The sequence period T. Since the DIBS signal applied to the battery includes multiple frequency points across different frequency bands, it eliminates the time-consuming problem caused by frequency-by-frequency scanning measurements required by traditional EIS measurement methods, thus achieving rapid measurement of battery EIS over a wide frequency range. The Fourier transform recursive filtering algorithm is the Goertzel algorithm, which includes the following steps: Step S1.1: Adjust the excitation current and response voltage Sampling is performed separately to obtain discrete sequences. and ,in ; Step S1.2: Determine the frequency points of interest k The discrete sequences of excitation current and response voltage are substituted into the second-order recursive formula of the Goertzel algorithm for recursion, and the spectrum values ​​are calculated respectively.

[0024] in, The angular frequency corresponding to the frequency point of interest. For rotation factor, Indicates excitation current or response voltage Discrete sequence in the first k The spectrum of the point frequency is as follows: and , Intermediate variables introduced into the second-order recurrence formula of the Goertzel algorithm; Step S1.3: Calculate the impedance at each frequency point. Repeat the above steps to obtain the final complete impedance spectrum. .

[0025] Step S2: Extract and calculate the sensitive EIS frequency point characteristic parameters of the battery's state of charge (SOC) versus state of health (SOH) from the battery's electrochemical impedance spectroscopy, and construct the input feature vector for the battery state estimation model. Spearman correlation analysis is used to determine the frequency points where the absolute value of the correlation coefficient between the frequency point characteristic quantity and SOC and SOH at each temperature is >0.3 as the sensitive EIS frequency points of the battery's SOC versus SOH. Since the low-frequency, high-frequency, and mid-frequency regions of the battery's EIS mainly reflect the diffusion process of lithium ions in the electrode material and concentration polarization, they are highly sensitive to changes in the kinetics of internal electrochemical reactions. Therefore, the impedance characteristics in the low-frequency region better reflect changes in the internal reaction kinetics of the battery. The impedance changes in the low-frequency region are more pronounced during changes in SOC and SOH, thus the sensitive frequency points are usually low-frequency points. The sensitive EIS frequency point characteristic parameters of the battery's SOC versus SOH include the real part of the impedance and the impedance magnitude.

[0026] Step S3: Establish a multi-decision tree ensemble model through Bootstrap sampling and random feature selection. The multi-decision tree ensemble model is a Coulomb integral random forest algorithm model, including a Coulomb integral channel and a random forest channel. Wherein: The random forest channel employs the random forest algorithm to simultaneously predict battery SOC and SOH. Several subsets of samples are randomly selected from the original training sample set using the Bootstrap sampling method, and each subset is used to train an independent decision tree model. Since each decision tree uses different training samples, the dependence of a single model on the data distribution is effectively reduced. During the decision tree growth process, a feature random selection mechanism is used; instead of using all input features when splitting nodes, a subset of features is randomly selected from all features to participate in the optimal split, further reducing the correlation between different decision trees. This effectively suppresses overfitting while maintaining the model's expressive power.

[0027] During model training, the dataset was first divided for separate use in model training and performance evaluation. Experimental data obtained at different temperatures and under different SOH and SOC conditions were used. To ensure the model could fully learn the impedance variation patterns under different operating conditions and to avoid the impact of uneven training sample distribution, the data was divided for each temperature condition and each SOH and SOC condition. Specifically, from all impedance data obtained under the same temperature and SOH / SOC conditions, 80% of the samples were randomly selected as the training set, and the remaining 20% ​​as the test set. This ensured that the training and test data maintained consistency in temperature and SOH / SOC distribution, making the model training process more stable and allowing for a more objective evaluation of the model's predictive ability under different operating conditions. Then, the random forest model was trained using the training set data. Based on extensive testing and analysis, it is recommended to set the number of decision trees in the random forest to 300 to ensure the model can fully learn the nonlinear relationships in the sample data; the maximum tree depth should be set to 15 to control the complexity of a single decision tree, thereby avoiding overfitting of the training data; and the minimum number of samples for node splits should be set to 3 to ensure sufficient samples are available for computation during node splitting. Furthermore, to ensure the reproducibility of the model training results, the random seed should be set to 42, and multi-core processors should be used for parallel computing to improve model training efficiency.

[0028] The coulomb integration channel calculates the battery's state of charge (SOC) by integrating the current over time.

[0029] When the ampere-hour integral value during battery EIS measurement exceeds 0.2% of the battery's rated capacity, the SOC result output adopts the output of the Coulomb integral channel; otherwise, the output of the random forest channel is adopted, and the output value of the Coulomb integral channel is updated accordingly to ensure its accuracy and adapt to high-rate charge and discharge scenarios (because high-rate charge and discharge, such as above 1C, can affect the battery within tens of seconds, causing it to fail to meet the EIS measurement requirements).

[0030] Step S4: Input the battery's State of Charge (SOC)-State of Health (SOH) sensitive EIS frequency point characteristic parameters into the multi-decision tree ensemble model to achieve joint estimation of battery SOC and SOH. Specifically, the SOC-SOH sensitive EIS frequency point characteristic parameters, battery current, running time, etc., are input into the multi-decision tree ensemble model to jointly estimate battery SOC and SOH. The results are shown in Table 1 below, and the estimation results are significantly better than other methods: Table 1

[0031] This invention aims to address the problem that traditional battery EIS measurement methods cannot meet the requirements of online real-time measurement. It discovers the sensitive EIS frequency characteristic parameters of battery state of charge (SOC) to state of health (SOH), and by constructing a multi-decision tree ensemble model, it can obtain accurate, stable, and robust multi-dimensional battery state estimation results, providing a new technical method for battery system management state monitoring and health assessment.

[0032] Example 2 This invention also provides a battery SOC-SOH joint estimation system driven by EIS rapid detection data. The battery SOC-SOH joint estimation system driven by EIS rapid detection data can be implemented by executing the process steps of the battery SOC-SOH joint estimation method driven by EIS rapid detection data. That is, those skilled in the art can understand the battery SOC-SOH joint estimation method driven by EIS rapid detection data as a preferred embodiment of the battery SOC-SOH joint estimation system driven by EIS rapid detection data.

[0033] According to the present invention, a battery SOC-SOH joint estimation system driven by EIS fast detection data is provided, comprising: Module M1: Generates a time-domain multi-frequency composite excitation current signal and applies it to the battery to obtain the excitation current and response voltage. Then, using a Fourier transform recursive filtering algorithm, it quickly detects the battery's electrochemical impedance spectroscopy (EIS). The time-domain multi-frequency composite excitation current signal is a discrete binary signal (DIBS). The target signal frequency band is divided into three segments—low-frequency, mid-frequency, and high-frequency—from low to high. Parameters for each frequency segment's DIBS excitation signal are set, ensuring the discrete binary sequence has a spectral distribution corresponding to the target signal, thus obtaining a DIBS signal containing all set frequency points. The parameters of the DIBS excitation signal include sequence length N, sampling frequency, etc. Frequency resolution Effective frequency point Clock cycle And the sequence period T.

[0034] Module M2: Extracts and calculates the EIS frequency characteristic parameters sensitive to the state of charge (SOC) versus state of health (SOH) from the battery's electrochemical impedance spectroscopy. In Module M2, Spearman correlation analysis is used to determine the frequency points where the absolute value of the correlation coefficient between the frequency characteristic quantity and SOC and SOH at each temperature is >0.3 as the EIS frequency points sensitive to the SOC versus SOH of the battery. The characteristic parameters of the EIS frequency points sensitive to the SOC versus SOH of the battery include the real part of the impedance and the impedance magnitude.

[0035] Module M3: Establishes a multi-decision tree ensemble model through Bootstrap sampling and random feature selection mechanisms. The multi-decision tree ensemble model is a Coulomb integral random forest algorithm model, including a Coulomb integral channel and a random forest channel. The random forest channel uses the random forest algorithm to simultaneously predict the battery's SOC and SOH. The Coulomb integral channel calculates the battery's SOC by integrating the current over time.

[0036] Module M4: Inputs the sensitive EIS frequency point characteristic parameters of the battery's State of Charge (SOC) and State of Health (SOH) into the multi-decision tree ensemble model to achieve joint estimation of battery SOC and SOH.

[0037] Those skilled in the art will understand that, besides implementing the system and its various devices, modules, and units provided by this invention in the form of purely computer-readable program code, the same functions can be achieved entirely through logical programming of the method steps, making the system and its various devices, modules, and units of this invention function in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, the system and its various devices, modules, and units provided by this invention can be considered as a hardware component, and the devices, modules, and units included therein for implementing various functions can also be considered as structures within the hardware component; alternatively, the devices, modules, and units for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0038] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. A battery SOC-SOH joint estimation method driven by EIS rapid detection data, characterized in that, include: Step S1: Generate a time-domain multi-frequency composite excitation current signal and apply it to the battery to obtain the excitation current and response voltage. Then, use the Fourier transform recursive filtering algorithm to quickly detect the battery electrochemical impedance spectroscopy (EIS). Step S2: Extract and calculate the characteristic parameters of the sensitive EIS frequency points of the battery's state of charge (SOC) and state of health (SOH) from the battery's electrochemical impedance spectroscopy; Step S3: Establish a multi-decision tree ensemble model through Bootstrap sampling and random feature selection mechanisms; Step S4: Input the sensitive EIS frequency point feature parameters of the battery's state of charge (SOC) and state of health (SOH) into the multi-decision tree ensemble model to achieve joint estimation of battery SOC and SOH.

2. The battery SOC-SOH joint estimation method based on EIS fast detection data as described in claim 1, characterized in that, The time-domain multi-frequency composite excitation current signal is a discrete binary signal (DIBS). The target signal frequency band is divided into three segments from low to high: low frequency, mid frequency, and high frequency. The parameters of the excitation signal DIBS for each frequency band are set, and the discrete binary sequence has a spectral distribution corresponding to the target signal, thus obtaining a DIBS signal containing all set frequency points.

3. The battery SOC-SOH joint estimation method based on EIS rapid detection data as described in claim 2, characterized in that, The parameters of the excitation signal DIBS include sequence length N and sampling frequency. Frequency resolution Effective frequency point Clock cycle And the sequence period T.

4. The battery SOC-SOH joint estimation method based on EIS fast detection data as described in claim 1, characterized in that, In step S2, Spearman correlation analysis is used to determine the frequency points where the absolute value of the correlation coefficient between the frequency point characteristic quantity and SOC and SOH at each temperature is >0.3 as the sensitive EIS frequency points of the battery's state of charge (SOC) - state of health (SOH). The battery's state of charge (SOC) and state of health (SOH) sensitive EIS frequency characteristic parameters include the real part of impedance and the impedance modulus.

5. The battery SOC-SOH joint estimation method based on EIS fast detection data as described in claim 1, characterized in that, The multi-decision tree ensemble model is a Coulomb integral random forest algorithm model, including a Coulomb integral channel and a random forest channel; The random forest channel uses the random forest algorithm to predict both battery SOC and SOH. The coulomb integration channel calculates the battery's state of charge (SOC) by integrating the current over time.

6. A battery SOC-SOH joint estimation system driven by EIS rapid detection data, characterized in that, include: Module M1: Generates a time-domain multi-frequency composite excitation current signal and applies it to the battery to obtain the excitation current and response voltage. Then, through a Fourier transform recursive filtering algorithm, it quickly detects the battery electrochemical impedance spectroscopy (EIS). Module M2: Extracts and calculates the sensitive EIS frequency characteristic parameters of the battery's state of charge (SOC) and state of health (SOH) from the battery's electrochemical impedance spectroscopy; Module M3: Establishes a multi-decision tree ensemble model through Bootstrap sampling and random feature selection mechanisms; Module M4: Inputs the sensitive EIS frequency point characteristic parameters of the battery's State of Charge (SOC) and State of Health (SOH) into the multi-decision tree ensemble model to achieve joint estimation of battery SOC and SOH.

7. The battery SOC-SOH joint estimation system based on EIS fast detection data driven according to claim 6, characterized in that, The time-domain multi-frequency composite excitation current signal is a discrete binary signal (DIBS). The target signal frequency band is divided into three segments from low to high: low frequency, mid frequency, and high frequency. The parameters of the excitation signal DIBS for each frequency band are set, and the discrete binary sequence has a spectral distribution corresponding to the target signal, thus obtaining a DIBS signal containing all set frequency points.

8. The battery SOC-SOH joint estimation system based on EIS fast detection data as described in claim 7, characterized in that, The parameters of the excitation signal DIBS include sequence length N and sampling frequency. Frequency resolution Effective frequency point Clock cycle And the sequence period T.

9. The battery SOC-SOH joint estimation system based on EIS fast detection data driven according to claim 6, characterized in that, In module M2, Spearman correlation analysis is used to determine the frequency points where the absolute value of the correlation coefficient between the frequency point characteristic quantity and SOC and SOH at each temperature is >0.3 as the sensitive EIS frequency points of the battery's state of charge (SOC) - state of health (SOH). The battery's state of charge (SOC) and state of health (SOH) sensitive EIS frequency characteristic parameters include the real part of impedance and the impedance modulus.

10. The battery SOC-SOH joint estimation system based on EIS fast detection data driven according to claim 6, characterized in that, The multi-decision tree ensemble model is a Coulomb integral random forest algorithm model, including a Coulomb integral channel and a random forest channel; The random forest channel uses the random forest algorithm to predict both battery SOC and SOH. The coulomb integration channel calculates the battery's state of charge (SOC) by integrating the current over time.

Images