State evaluation model training method, and battery health state evaluation method and device
By using real-vehicle data to train the state evaluation model in battery health status assessment, using the LSTM network and the Pearson correlation coefficient method to extract impedance characteristic parameters, the problem that laboratory data cannot accurately evaluate the battery health status under actual operating conditions is solved, and the problem of laboratory data being unable to accurately evaluate the battery health status under actual operating conditions is improved. The accuracy of battery health status assessment and the resource utilization rate of electric heavy trucks are provided.
Patent Information
- Application Number
- CN202411146298.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-20
- Publication Date
- 2024-11-26
AI Technical Summary
The existing battery health status assessment methods mainly rely on laboratory data, resulting in low evaluation accuracy under actual operating conditions and the inability to accurately predict the maximum available capacity and service life of the battery.
By collecting real vehicle data samples (electrochemical impedance spectrum and open circuit voltage value), training the state evaluation model, using the LSTM network and the Pearson correlation coefficient method to extract impedance characteristic parameters, improve the accuracy of battery health status evaluation.
It realizes an accurate assessment of the health status of the battery, improves the accuracy of battery life management and vehicle cost budget, and meets the demand for resource utilization of electric heavy trucks.
Smart Images

Figure CN119024182A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of battery technology, and in particular to a training method for a state assessment model, and a battery health state assessment method and device. Background Art
[0002] Against the backdrop of the increasing shortage of traditional energy, and with the gradual promotion of the country's energy conservation and emission reduction policies, electric vehicles (such as electric heavy trucks) have shown a trend of completely replacing fuel vehicles in various fields. The core technologies of electric vehicles are mainly motors, electronic controls and batteries. Among them, batteries, as the power source of electric vehicles, account for the highest cost of the entire vehicle. The electrochemical system of the battery itself leads to a short service life, and irreversible aging of the battery during the cycle charge and discharge process, which is directly related to the maximum available capacity of the battery. In addition, when the battery ages to a certain extent, it will affect the vehicle's range and even lead to risks such as battery spontaneous combustion.
[0003] According to the national standard GB-32960, when the life state (i.e., battery health state (SOH)) of lithium-ion rechargeable energy storage devices reaches the first threshold (e.g., 80%), and when the SOH of commercial vehicles reaches the second threshold (e.g., 70%), they need to be scrapped. In this way, whether the SOH can be accurately assessed is related to the full life cycle management of the battery, the vehicle cost budget, and the resource utilization rate of the entire vehicle.
[0004] Existing SOH evaluation methods are often based on electrochemical impedance spectroscopy (EIS). However, this method is mainly based on laboratory battery data. Since the laboratory battery data are all full charge and discharge results, the measured EIS is not conducive to the SOH estimation under actual working conditions, which in turn leads to the low accuracy of the final SOH evaluation results. Summary of the invention
[0005] The present invention provides a training method for a state assessment model, a battery health state assessment method and device, which are used to solve the problem that the prior art mainly relies on laboratory battery data. Since the laboratory battery data are all full charge and discharge results, the EIS after measurement is not conducive to the SOH estimation under actual working conditions, which in turn leads to the low accuracy of the final SOH assessment result. By collecting real vehicle data samples (i.e., electrochemical impedance spectrum samples and open circuit voltage value samples) and training the initial state assessment model, a state assessment model with higher accuracy can be obtained, thereby improving the accuracy of the battery health state assessment result corresponding to the battery to be tested.
[0006] In a first aspect, the present invention provides a method for training a state assessment model, comprising the following steps.
[0007] Obtain electrochemical impedance spectrum samples and open circuit voltage value samples of the battery sample in a charged state.
[0008] According to the electrochemical impedance spectrum samples, impedance characteristic parameter samples corresponding to different frequency bands are determined.
[0009] All impedance characteristic parameter samples and the open circuit voltage value samples are input into the initial state assessment model to obtain the predicted battery health state assessment result output by the initial state assessment model.
[0010] According to the predicted battery health status assessment result and the battery health status represented by the battery capacity, the model parameters of the initial status assessment model are updated to obtain a trained status assessment model.
[0011] According to a training method for a state assessment model provided by the present invention, the impedance characteristic parameter samples corresponding to different frequency bands are determined according to the electrochemical impedance spectrum samples, including: determining the target impedance characteristics corresponding to the different frequency bands according to the Nyquist diagram corresponding to the electrochemical impedance spectrum samples; for each target impedance characteristic, sampling the target impedance characteristic to obtain multiple frequency points; determining the impedance values corresponding to the multiple frequency points according to the excitation signal; and determining the impedance characteristic parameter samples corresponding to the frequency band where the target impedance characteristic is located according to all impedance values.
[0012] According to a training method for a state assessment model provided by the present invention, the impedance characteristic parameter samples corresponding to the frequency band where the target impedance feature is located are determined based on all impedance values, including: obtaining a reference impedance value corresponding to the frequency band where the target impedance feature is located; using the Pearson correlation coefficient method to determine the degree of correlation between each impedance value and the reference impedance value, and determining the impedance characteristic parameter samples corresponding to the frequency band where the target impedance feature is located based on the frequency point corresponding to the impedance value corresponding to the maximum correlation degree.
[0013] According to a training method for a state assessment model provided by the present invention, determining the target impedance characteristics corresponding to each of the different frequency bands according to the Nyquist diagram corresponding to the electrochemical impedance spectrum sample includes: dividing the Nyquist diagram corresponding to the electrochemical impedance spectrum sample according to the resistance characteristics of the battery sample to obtain the target impedance characteristics corresponding to each of the different frequency bands.
[0014] According to a training method for a state assessment model provided by the present invention, all impedance characteristic parameter samples and the open circuit voltage value samples are input into an initial state assessment model to obtain a predicted battery health state assessment result output by the initial state assessment model, including: using the initial state assessment model to determine the first implicit information at a previous moment in the initial state assessment model and the cell state information at the previous moment; determining the first input data at the current moment based on all the impedance characteristic parameter samples and the open circuit voltage value samples, and determining the predicted battery health state assessment result in combination with the first implicit information and the cell state information.
[0015] According to a training method for a state evaluation model provided by the present invention, the initial state evaluation model includes: a forget gate, an input gate and an output gate; the initial state evaluation model is used to determine the first implicit information at a previous moment in the initial state evaluation model, and the unit state information at the previous moment, including: using the forget gate to calculate the second input data at the previous moment, and the second implicit information to obtain the state output result at the previous moment, and the second implicit information is the implicit information of the first implicit information at the previous moment; using the input gate to calculate the state output result, the second input data and the second implicit information to obtain the unit state information at the previous moment; using the output gate to calculate the unit state information, the second input data and the third implicit information at the current moment to obtain the first implicit information at the previous moment.
[0016] In a second aspect, the present invention provides a battery health status assessment method, comprising the following steps.
[0017] Obtain the electrochemical impedance spectrum and open circuit voltage value of the battery under test in the charging state.
[0018] The electrochemical impedance spectrum and the open circuit voltage value are input into a state assessment model to obtain a battery health state assessment result output by the state assessment model, wherein the state assessment model is trained based on the training method of the state assessment model described in the first aspect above.
[0019] In a third aspect, the present invention also provides a training device for a state assessment model, comprising the following modules.
[0020] The data acquisition module is used to obtain electrochemical impedance spectrum samples and open circuit voltage value samples of the battery sample in the charging state.
[0021] The model training module is used to determine the impedance characteristic parameter samples corresponding to different frequency bands according to the electrochemical impedance spectrum samples; input all the impedance characteristic parameter samples and the open circuit voltage value samples into the initial state assessment model to obtain the predicted battery health state assessment result output by the initial state assessment model; and update the model parameters of the initial state assessment model according to the predicted battery health state assessment result and the battery health state characterized by the battery capacity to obtain a trained state assessment model.
[0022] In a fourth aspect, the present invention also provides a battery health status assessment device, comprising the following modules.
[0023] The data acquisition module is used to obtain the electrochemical impedance spectrum and open circuit voltage value of the battery under test in the charging state.
[0024] A state assessment module is used to input the electrochemical impedance spectrum and the open circuit voltage value into a state assessment model to obtain a battery health state assessment result output by the state assessment model, wherein the state assessment model is trained based on the training method of the state assessment model described in the first aspect above.
[0025] In a fifth aspect, the present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements a training method for a status assessment model as described in any one of the first aspects above, or a battery health status assessment method as described in the second aspect.
[0026] In a sixth aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a training method for a state assessment model as described in any one of the first aspects above, or a battery health state assessment method as described in the second aspect.
[0027] In the seventh aspect, the present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements the training method of the status assessment model as described in any one of the first aspects above, or the battery health status assessment method as described in the second aspect.
[0028] The training method of the state assessment model, the battery health state assessment method and the device provided by the present invention can obtain a state assessment model with higher accuracy by collecting real vehicle data samples (i.e., electrochemical impedance spectrum samples and open circuit voltage value samples), thereby improving the accuracy of the battery health state assessment result corresponding to the battery to be tested. BRIEF DESCRIPTION OF THE DRAWINGS
[0029] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0030] Figure 1 It is a flow chart of the training method of the state assessment model provided by the present invention.
[0031] Figure 2a It is a numerical correspondence diagram between the Nyquist diagram corresponding to the electrochemical impedance spectroscopy sample provided by the present invention and the equivalent circuit model.
[0032] Figure 2b It is a schematic diagram of the measurement circuit provided by the present invention.
[0033] Figure 2c It is a structural schematic diagram of the initial state evaluation model provided by the present invention.
[0034] Figure 3 It is a flow chart of the battery health status assessment method provided by the present invention.
[0035] Figure 4 It is a structural schematic diagram of a training device for a state assessment model provided by the present invention.
[0036] Figure 5 It is a structural schematic diagram of a battery health status assessment device provided by the present invention.
[0037] Figure 6 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION
[0038] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0039] In order to better understand the embodiments of the present invention, the background technology is described in detail below.
[0040] The existing SOH evaluation method mainly relies on laboratory battery data, and the SOH evaluation is achieved through the measured EIS. Among them, EIS is usually collected by testing fixed working conditions such as the New European Driving Cycle (NEDC) by setting up a measurement circuit in the laboratory. The above-mentioned entire evaluation process has the following disadvantages: 1. Laboratory battery data is usually based on 18650 ternary lithium battery data, which has certain limitations, resulting in the final SOH assessment results lacking in usability, authenticity and universality.
[0041] 2. The battery capacity in the laboratory is relatively small and cannot meet the requirements of the battery management system (BMS) for electric heavy trucks with a capacity of over 300kwh.
[0042] 3. Batteries in the laboratory require long-term testing experiments and cannot meet the real-time requirements of SOH evaluation.
[0043] 4. The laboratory battery data are all full charge and discharge results, which makes the measured EIS not conducive to the SOH estimation under actual working conditions, and thus leads to the low accuracy of the final SOH evaluation results.
[0044] In order to solve the above technical problems, an embodiment of the present invention provides a training method for a state assessment model. The executor of the method can be a training device for the state assessment model or an electronic device. Optionally, the electronic device is a cloud platform / cloud server.
[0045] The following takes an electronic device as an example to explain in detail the training method of the state assessment model provided by the embodiment of the present invention.
[0046] Figure 1 is a flow chart of the training method of the state assessment model provided by the present invention, such as Figure 1 As shown, the method includes the following steps 101 to 104.
[0047] Step 101: Obtain an electrochemical impedance spectrum sample and an open circuit voltage value sample of a battery sample in a charging state.
[0048] The battery sample is a device that converts chemical energy into electrical energy. The embodiments of the present invention mainly involve battery samples (such as lithium iron phosphate battery samples) for electric heavy trucks.
[0049] Electrochemical impedance spectroscopy samples are used to reflect the electrochemical state inside battery samples.
[0050] The open circuit voltage value sample refers to the voltage output by the power supply measured when there is no load in the measurement circuit.
[0051] Exemplarily, the current electric heavy truck samples are divided into 6 groups of vehicle sets according to the accumulated mileage, and each vehicle set includes 10 electric heavy truck samples. The electronic device collects the full charge data of the battery sample of each electric heavy truck sample under the constant current-constant voltage (CC-CV) charging state of the charging pile through the BMS, accumulating 10 times, so that 600 sub-electrochemical impedance spectrum samples and 600 sub-open circuit voltage (OCV) samples can be obtained. At this time, these 600 sub-electrochemical impedance spectrum samples are electrochemical impedance spectrum samples, and these 600 sub-open circuit voltage value samples are open circuit voltage value samples, in preparation for the subsequent training state evaluation model.
[0052] It should be noted that the electrochemical impedance spectrum samples and open circuit voltage value samples are closer to the actual working conditions. Since the number of charge and discharge cycles of the battery samples cannot be quantified, the data label can be defined by cumulative mileage, and the factory-new car is marked as a full battery health state. According to the simulated power point (Maximum Available Power, MAP) of battery cell aging provided by the supplier, 6 points are selected from the cumulative mileage of 0~450000km, corresponding to the battery health state of 70~100% for the whole vehicle, as the basis for selecting electric heavy truck samples.
[0053] In addition, as real vehicle data samples, the electrochemical impedance spectrum samples and open circuit voltage value samples have certain use value, authenticity and universality. At the same time, the battery samples of the electric heavy truck samples can also meet the BMS requirements.
[0054] Optionally, during the charging process of the battery sample, the battery capacity obtained by current integration can be used to characterize the battery health state. In this case, the battery health state characterized by the battery capacity is used in the training set as the true value of SOH.
[0055] It should be noted that the battery health status represented by the battery capacity has obvious energy regeneration phenomenon and is greatly affected by temperature and charge and discharge rate. Therefore, it cannot be directly used to represent the actual health status of the battery.
[0056] Optionally, the electrochemical impedance spectrum sample obtained by the electronic device is obtained based on the following steps: after the measurement circuit (such as the charging plug loop) of the electric heavy truck sample and the main positive addition sampling chip (such as the BMS chip) of the battery box, the initial electrochemical impedance spectrum sample is collected through the sampling chip to reflect the aging, damage or potential safety hazards inside the battery. Then, the initial electrochemical impedance spectrum sample is denoised through a filter (such as a Gaussian filter) to improve the quality of the initial electrochemical impedance spectrum sample and obtain an electrochemical impedance spectrum sample with higher accuracy; then, the electrochemical impedance spectrum sample is sent to the power (Power)-controller area network (P-CAN) through the charging controller local area network (Controller Area Network, CAN) bus and then uploaded to the electronic device through the telematics box (TBOX), so that the electronic device can realize remote monitoring of the battery sample.
[0057] Among them, the charging controller local area network is a vehicle data communication method.
[0058] Exemplarily, Table 1 shows the performance parameters of the BMS chip.
[0059] Table 1
[0060] Among them, AD stands for Analog-to-Digital converter.
[0061] It should be noted that Table 1 lists the performance parameters of some commonly used BMS chips. Among them, the voltage sampling accuracy of the chip model NXPMC33771 is very high, which is more suitable for the measurement of the voltage curve of lithium iron phosphate batteries in the platform area. The external measurement circuit designed based on this chip is also more suitable for the distributed BMS system connected to electric heavy trucks.
[0062] Step 102: Determine impedance characteristic parameter samples corresponding to different frequency bands according to the electrochemical impedance spectrum samples.
[0063] Among them, different frequency bands usually correspond to different electrochemical processes and impedance characteristic parameter samples, which are obtained based on the resistance characteristics of the battery samples. Optionally, the different frequency bands include: low frequency band, medium frequency band and high frequency band.
[0064] In some embodiments, the electronic device determines the impedance characteristic parameter samples corresponding to different frequency bands based on the electrochemical impedance spectrum samples, which may include: the electronic device determines the target impedance characteristics corresponding to different frequency bands based on the Nyquist diagram corresponding to the electrochemical impedance spectrum samples; for each target impedance characteristic, the target impedance characteristic is sampled to obtain multiple frequency points; according to the excitation signal, the impedance value corresponding to each of the multiple frequency points is determined; according to all impedance values, the impedance characteristic parameter sample corresponding to the frequency band where the target impedance characteristic is located is determined.
[0065] When determining the impedance characteristic parameter samples corresponding to different frequency bands, the electronic device can first determine the Nyquist diagram corresponding to the electrochemical impedance spectrum sample, and then divide the Nyquist diagram into frequency bands, so as to more accurately capture the target impedance characteristics corresponding to different frequency bands. For the target impedance characteristics corresponding to each frequency band, since the target impedance characteristics are a continuous value and cannot be effectively calculated, the target impedance characteristics can be sampled to obtain multiple sampling points, that is, multiple frequency points, which simplifies the feature extraction process; then, the electronic device determines the impedance values corresponding to each of the multiple frequency points based on the multiple frequency points combined with the excitation signal, and then determines the impedance characteristic parameter samples corresponding to the frequency band where the target impedance characteristics are located based on all the impedance values. Based on this, each frequency band will correspond to an impedance characteristic parameter sample, that is, as many impedance characteristic parameter samples as there are frequency bands.
[0066] In some embodiments, the electronic device determines the target impedance characteristics corresponding to different frequency bands based on the Nyquist diagram corresponding to the electrochemical impedance spectrum sample, which may include: the electronic device divides the Nyquist diagram corresponding to the electrochemical impedance spectrum sample according to the resistance characteristics of the battery sample to obtain the target impedance characteristics corresponding to different frequency bands.
[0067] In the process of determining the target impedance characteristics corresponding to different frequency bands, the electronic device first determines the Nyquist plot corresponding to the electrochemical impedance spectrum sample. Since the Nyquist plot has a numerical correspondence with the equivalent circuit model (ECM) corresponding to the battery sample (such as Figure 2a As shown), the electronic device can divide the Nyquist diagram based on the resistance characteristics of the battery sample to obtain the target impedance characteristics corresponding to different frequency bands.
[0068] Specifically, the corresponding relationship is as follows: the high frequency band of the Nyquist diagram is related to the internal resistance of the battery sample, which can be equivalent to the ohmic internal resistance R0; the resistance R dt and capacitor C dtThe electrochemical polarization phenomenon caused by lithium ion insertion and extraction in the battery sample is equivalent to the mid-frequency band of the Nyquist diagram through parallel connection; the concentration polarization phenomenon of the battery is connected through the resistor R wt , capacitor C wt and inductor L wt The series-parallel equivalent is the Weber impedance Z w , corresponding to the low frequency band of the Nyquist diagram.
[0069] For the target impedance characteristics corresponding to different frequency bands, since the high frequency band of the Nyquist diagram is related to the internal resistance of the battery sample, the target impedance characteristics corresponding to the high frequency band only need to be the real characteristic parameters of the filtering point; and the mid-frequency band and the low-frequency band will be affected by the polarization capacitance, so the target impedance characteristics corresponding to the mid-frequency band and the low-frequency band respectively include not only the real characteristic parameters of the filtering point, but also the imaginary characteristic parameters of the filtering point.
[0070] It should be noted that the impedance spectrum test frequency band for battery samples in the embodiment of the present invention is 0.1Hz~10KHz, wherein, based on the Nyquist diagram, the three frequency bands of high frequency band, medium frequency band and low frequency band are divided into a ratio of 1:6:3. Figure 2a It can be seen that the distribution of the mid-frequency band has two semicircles, of which the first semicircle corresponds to the solid electrolyte interface (Solid Electrolyte Interphase, SEI). The formation of the SEI membrane is caused by the irreversible electrochemical decomposition of the electrolyte, which is an electrochemical activation phenomenon. Since the SEI membrane of the lithium iron phosphate battery sample will be formed at the negative electrode as the battery sample ages, after reaching a certain level, the electrolyte molecules cannot pass through the SEI membrane to reach the surface of the active material and react with lithium ions and electrons, thereby inhibiting the further growth of SEI. This phenomenon has little effect on the battery samples involved in the embodiments of the present invention and will not be described in detail here.
[0071] Optionally, the electronic device determines the impedance values corresponding to each of the multiple frequency points according to the excitation signal, which may include: for each frequency point, the electronic device determines the capacitance and resistance of the frequency point according to the excitation signal; and determines the impedance value corresponding to the frequency point according to the capacitance and resistance in combination with the discrete Fourier transform formula.
[0072] Among them, the discrete Fourier transform formula is: Z s = .
[0073] Z s represents the impedance value; R0 represents the ohmic internal resistance obtained by collecting the AC current, and the AC current is the excitation signal; n represents the total number of frequency points; Represents the capacitance of the i-th frequency point, i represents the serial number of the currently sampled frequency point; k represents an integer index; Represents the resistance at the i-th frequency point.
[0074] It should be noted that the electric vehicle sample has its own DC-DC converter, which can convert direct current into battery charging and generate a small alternating current for measuring all impedance values of the battery sample. The alternating current at this time is called an excitation signal. After the excitation signal passes through the battery sample, the electronic device can collect the battery internal resistance, voltage response signal and loop current signal corresponding to the battery sample through the BMS chip, and then measure the circuit (such as Figure 2b As shown in Figure 2), the current response signal is calculated by combining the discrete Fourier transform formula to obtain the impedance values corresponding to multiple frequency points.
[0075] In other words, the electronic device can quickly and completely determine the impedance values corresponding to multiple frequency points through the BMS chip and the measurement circuit.
[0076] In some embodiments, the electronic device determines the impedance characteristic parameter sample corresponding to the frequency band where the target impedance characteristic is located based on all impedance values, which may include: the electronic device obtains the reference impedance value corresponding to the frequency band where the target impedance characteristic is located; the electronic device uses the Pearson correlation coefficient method to determine the degree of correlation between each impedance value and the reference impedance value, and determines the impedance characteristic parameter sample corresponding to the frequency band where the target impedance characteristic is located based on the frequency point corresponding to the impedance value corresponding to the maximum correlation degree.
[0077] For the measured health characteristic data of the electrochemical impedance spectrum samples in different aging stages, it is necessary to further analyze the relationship between the impedance values at different frequency points in each frequency band and the reference impedance values, and then determine the impedance characteristic parameter samples of each frequency band. Specifically, for each frequency band, all impedance values corresponding to the frequency band are obtained, and the reference impedance value corresponding to the frequency band is obtained. The reference impedance value is pre-set before the battery sample leaves the factory; then, the Pearson correlation coefficient method is used to calculate the degree of correlation between each impedance value and the reference impedance value one by one, so that multiple correlation degrees can be obtained; then, the maximum correlation degree is determined from these multiple correlation degrees, and from all frequency points, the impedance corresponding to the maximum correlation degree is determined from the corresponding frequency point, and then the impedance characteristic parameter sample corresponding to the frequency band is determined.
[0078] Optionally, for the ith impedance value among the multiple impedance values corresponding to each frequency band, a correlation calculation formula in the Pearson correlation coefficient method is used to determine the degree of correlation between the ith impedance value and the reference impedance value.
[0079] The correlation calculation formula is: r= .
[0080] r represents the Pearson correlation coefficient, which is used to characterize the degree of correlation between the ith impedance value and the reference impedance value. A positive r indicates a positive correlation, a negative r indicates a negative correlation, and r=0 indicates that there is no correlation between the ith impedance value and the reference impedance value. The maximum value of r is 1.
[0081] X i represents the i-th impedance value; Represents the mean value of n impedance values; represents the standard deviation of n impedance values; Y i represents the i-th reference impedance value among the reference impedance values; represents the mean value of n reference impedance values; Represents the standard deviation of n reference impedance values.
[0082] The above correlation calculation formula can establish a mapping relationship between two random variables (ie, the i-th impedance value and the reference impedance value).
[0083] It should be noted that the target impedance characteristic corresponding to the high frequency band is only the real characteristic parameter of the filtering point. ; The target impedance characteristics corresponding to the mid-frequency band include the real characteristic parameters of the filter point and the imaginary characteristic parameters ; The target impedance characteristics corresponding to the low frequency band include the real characteristic parameters of the filter point and the imaginary characteristic parameters Based on this, it can be seen that the process of determining the impedance characteristic parameter sample according to the Pearson correlation coefficient method can extract strong correlation features, avoid parameter identification of complex equivalent circuit models, and improve the accuracy of feature extraction.
[0084] Step 103: input all impedance characteristic parameter samples and open circuit voltage value samples into the initial state assessment model to obtain the predicted battery health state assessment result output by the initial state assessment model.
[0085] Among them, the initial state evaluation model is a long short-term memory network (LSTM) model, and the long short-term memory network is a recurrent neural network (RNN), which is used to predict and evaluate the battery health status.
[0086] After obtaining all impedance characteristic parameter samples and open circuit voltage value samples, the electronic device can use all impedance characteristic parameter samples and open circuit voltage value samples as input data, and determine the first input data at the current moment from the input data, input it into the initial state evaluation model, and analyze the first input data through the initial state evaluation model to obtain the predicted battery health status evaluation result corresponding to the battery sample. The whole process can realize the offline prediction of the battery health status, is less affected by abnormal working conditions, and can obtain a highly accurate predicted battery health status evaluation result.
[0087] In some embodiments, the electronic device inputs all impedance characteristic parameter samples and open circuit voltage value samples into an initial state assessment model to obtain a predicted battery health state assessment result output by the initial state assessment model, which may include: the electronic device uses the initial state assessment model to determine the first implicit information at a previous moment in the initial state assessment model, and the cell state information at a previous moment; the electronic device determines the first input data at the current moment based on all impedance characteristic parameter samples and open circuit voltage value samples, and determines the predicted battery health state assessment result in combination with the first implicit information and the cell state information.
[0088] Electronic equipment uses an initial state assessment model (such as Figure 2c As shown in FIG. 1 , in the process of determining the predicted battery health status evaluation result, the first implicit information B at the previous moment can be determined first. N-1 and the unit status information D at the previous moment N-1 , combined with the first input data A at the current moment N , based on the calculation formula Y in the initial positioning solution model N =LSTM(A N , D N-1 , B N-1 ), get the predicted battery health status assessment result Y at the current moment N .
[0089] It should be noted that after the status assessment model is built, since the time series based on the LSTM model will have a lag in prediction, the LSTM model will overemphasize the data information at the previous moment N-1, that is, overemphasize the data information collected by the electronic device in the previous second of the current moment. Based on this, the predicted battery health status assessment result at the current moment requires more repetitions of the prediction result of the previous second.
[0090] In some embodiments, the initial state evaluation model may include: a forget gate, an input gate and an output gate; the electronic device uses the initial state evaluation model to determine the first implicit information at the previous moment in the initial state evaluation model, and the unit state information at the previous moment, which may include: the electronic device uses the forget gate to calculate the second input data at the previous moment, and the second implicit information to obtain the state output result at the previous moment, and the second implicit information is the implicit information of the first implicit information at the previous moment; the electronic device uses the input gate to calculate the state output result, the second input data and the second implicit information to obtain the unit state information at the previous moment; the electronic device uses the output gate to calculate the unit state information, the second input data and the third implicit information at the current moment to obtain the first implicit information at the previous moment.
[0091] Among them, the initial state evaluation model is based on the RNN structure, and three types of valves are added, namely the forget gate, input gate and output gate. At this time, the core logic of the initial state evaluation model is: based on the state of the current cell unit (that is, the long vector neuron conveyor belt in the LSTM model), more useful data information is retained to the next cell unit through the screening, filtering and updating of the valve. Although the time interval of the whole process is long, the data information that is helpful for the subsequent predicted battery health status evaluation results will also be retained, and some invalid data information will be discarded. Finally, the data information that meets the requirements of passing the gate will be passed to the next layer model and participate in the calculation.
[0092] The electronic device uses the initial state evaluation model to determine the first implicit information B N-1 and unit status information D N-1 In the process of feature forgetting, the input information transmitted from the previous node is selectively forgotten. The specific choice is made according to the weight assigned to each parameter in the neural network, forgetting the information with lower relevance and retaining the information with higher relevance. Specifically, the second input data A at the previous moment can be determined from the input data. N-1 ; Then use the forget gate to combine the second input data A N-1 , for the first implicit information B N-1 The second implicit information B at the previous moment N-2 The unnecessary data information in the second implicit information B is discarded. N-2 Read and selectively forget to get the state output result E at the previous moment N-1 .
[0093] In the input memory stage: the input data passed from the previous node is filtered and memorized, focusing on long-term memory preservation of highly relevant implicit information, and short-term memory of less important information in proportion. The whole process is controlled by the gate signal. Specifically, the input gate is used to output the state result E N-1 , the second input data A N-1 and the second implicit information B N-2 Calculate and obtain the unit state information D at the previous moment N-1 Among them, the second implicit information B N-2 is the first implicit information B N-1 The implicit information at the previous moment.
[0094] In the data output stage: the information processed by this node (such as the first implicit information B N-1 ) to make a decision, selectively output the data information in the current state, retain the important implicit information to pass to the next unit, and complete the model closed loop. Specifically, the output gate is used to N-1 , the second input data A N-1 and the third implicit information B at the current moment N Calculate and get the first implicit information B at the previous moment N-1 .
[0095] Optional, the state output result E at the previous moment N-1 The forget gate formula can be used: Calculated.
[0096] Among them, E N-1 The value range is 0~1; ( ) represents the S-type function, which is used to convert the second implicit information B N-2 and the second input data A N-1 Normalize / standardize; W E Represents the weight matrix corresponding to the forget gate; Represents the bias vector corresponding to the forget gate.
[0097] Optional, the unit status information D at the previous moment N-1 The input gate formula can be used: ; ; Calculated.
[0098] Among them, F N-1 Represents the state output result of the input gate at the current moment; W F Represents the weight matrix corresponding to the input gate; represents the bias vector corresponding to the input gate; tanh( ), which can be called the tanh layer, represents the hyperbolic tangent function, which is an activation function that can convert the input value (such as the second implicit information B N-2 ) is mapped to a limited range, which helps prevent the gradient vanishing or exploding problem; represents the unit state vector at the previous moment updated by the tanh layer, and the tanh layer is a memory unit; W D Represents the weight matrix corresponding to the tanh layer; Represents the bias vector corresponding to the tanh layer. D N-2 Indicates unit status information D N-1 The unit status information at the previous moment.
[0099] Optionally, the first implicit information B at the previous moment N-1 The output gate formula can be used: ; Calculated.
[0100] Among them, G N-1 Indicates the state output result of the output gate at the current moment; W G Represents the weight matrix corresponding to the output gate; Represents the bias vector corresponding to the output gate.
[0101] Step 104: Update the model parameters of the initial state assessment model according to the predicted battery health state assessment result and the battery health state represented by the battery capacity to obtain a trained state assessment model.
[0102] It should be noted that after all the impedance characteristic parameter samples corresponding to the battery samples are extracted by the Pearson correlation coefficient method, they are combined with the open circuit voltage value samples mentioned above and the battery health status represented by the battery capacity, and are input into the initial state assessment model as training samples. Compared with the existing EIS estimation measurement for single cells, the process of steps 101-104 is more difficult to estimate based on the battery box of the whole vehicle, and the demand for data volume is high. Therefore, it is necessary to use the LSTM model to learn the features of the training samples, while preventing overfitting, retaining the authenticity of the data to improve the accuracy of the estimation results. Since the LSTM model is a type of recursive memory neural network based on a memory structure, it can record the implicit information of the current node and then pass it to subsequent nodes, so that each node can learn the historical input information, thereby achieving the purpose of long-term memory, and is more suitable for long-term series battery health status estimation, so as to obtain a battery health status assessment result with higher accuracy in the subsequent application stage.
[0103] In addition, in the process of training the initial state assessment model, the above-collected training samples can meet the data volume. In addition, real-time measurements can also be quickly performed when the newly connected vehicle samples are in normal operation and charging.
[0104] In an embodiment of the present invention, an electrochemical impedance spectrum sample and an open circuit voltage value sample of a battery sample in a charging state are obtained; based on the electrochemical impedance spectrum sample, the impedance characteristic parameter samples corresponding to different frequency bands are determined; all impedance characteristic parameter samples and open circuit voltage value samples are input into the initial state assessment model to obtain the predicted battery health state assessment result output by the initial state assessment model; based on the predicted battery health state assessment result and the battery health state characterized by the battery capacity, the model parameters of the initial state assessment model are updated to obtain a trained state assessment model. This method can obtain a state assessment model with high accuracy by collecting real vehicle data samples (i.e., electrochemical impedance spectrum samples and open circuit voltage value samples), and thereby improve the accuracy of the battery health state assessment result corresponding to the battery to be tested.
[0105] It should be noted that the execution subject involved in the embodiment of the present invention may be a battery health status assessment device or an electronic device. Optionally, the electronic device may include: a computer, a mobile terminal, etc.
[0106] The embodiment of the present invention is further described below by taking an electronic device as an example.
[0107] Figure 3 is a flow chart of the battery health status assessment method provided by the present invention, such as Figure 3 As shown, the method includes the following steps 301-302.
[0108] Step 301: Obtain the electrochemical impedance spectrum and open circuit voltage value of the battery to be tested in a charging state.
[0109] Step 302: Input the electrochemical impedance spectrum and the open circuit voltage value into the state assessment model to obtain the battery health state assessment result output by the state assessment model. The state assessment model is based on the above Figure 1 The state assessment model is obtained by training using the training method.
[0110] Optionally, after step 302, the method may further include: the electronic device feeds back the battery health status assessment result to the vehicle-side instrument display.
[0111] In this way, the above process can meet the accuracy and real-time requirements of the battery health status estimation of the battery to be tested in the electric heavy truck.
[0112] In addition, since the status assessment model is trained offline, when online, the electronic device can directly obtain the real-time data of the battery to be tested, namely the electrochemical impedance spectrum and the open circuit voltage value, and combine it with the status assessment model to obtain the battery health status assessment result. The whole process can meet the real-time requirements of battery health status assessment. In addition, the accuracy of the battery health status assessment result is also relatively high.
[0113] In an embodiment of the present invention, an electrochemical impedance spectrum and an open circuit voltage value of the battery to be tested in a charging state are obtained; the electrochemical impedance spectrum and the open circuit voltage value are input into a state assessment model to obtain a battery health state assessment result output by the state assessment model. The state assessment model is based on the above Figure 1 The state assessment model is obtained by training using a training method. The method uses a state assessment model trained with real vehicle data samples (i.e., electrochemical impedance spectrum samples and open circuit voltage value samples), and can obtain a battery health state assessment result of the battery to be tested with high accuracy.
[0114] The training device for the state assessment model provided by the present invention is described below. The training device for the state assessment model described below and the training method for the state assessment model described above can be referred to each other.
[0115] Figure 4 is a structural schematic diagram of a training device for a state assessment model provided by the present invention, such as Figure 4 As shown, the device includes the following data acquisition module 401 and model training module 402.
[0116] The data acquisition module 401 is used to acquire electrochemical impedance spectrum samples and open circuit voltage value samples of the battery sample in the charging state.
[0117] The model training module 402 is used to determine the impedance characteristic parameter samples corresponding to different frequency bands according to the electrochemical impedance spectrum samples; input all the impedance characteristic parameter samples and the open circuit voltage value samples into the initial state assessment model to obtain the predicted battery health state assessment result output by the initial state assessment model; update the model parameters of the initial state assessment model according to the predicted battery health state assessment result and the battery health state represented by the battery capacity to obtain a trained state assessment model.
[0118] Optionally, the model training module 402 is specifically used to determine the target impedance characteristics corresponding to the different frequency bands according to the Nyquist diagram corresponding to the electrochemical impedance spectrum sample; for each target impedance characteristic, sample the target impedance characteristic to obtain multiple frequency points; determine the impedance values corresponding to the multiple frequency points according to the excitation signal; and determine the impedance characteristic parameter sample corresponding to the frequency band where the target impedance characteristic is located according to all impedance values.
[0119] Optionally, the model training module 402 is specifically used to obtain a reference impedance value corresponding to the frequency band where the target impedance feature is located; use the Pearson correlation coefficient method to determine the degree of correlation between each impedance value and the reference impedance value, and determine the impedance characteristic parameter sample corresponding to the frequency band where the target impedance feature is located based on the frequency point corresponding to the impedance value corresponding to the maximum correlation degree.
[0120] Optionally, the model training module 402 is specifically used to divide the Nyquist diagram corresponding to the electrochemical impedance spectrum sample according to the resistance characteristics of the battery sample to obtain the target impedance characteristics corresponding to the different frequency bands.
[0121] Optionally, the model training module 402 is specifically used to adopt the initial state evaluation model to determine the first implicit information in the initial state evaluation model at the previous moment, and the unit state information at the previous moment; determine the first input data at the current moment based on all the impedance characteristic parameter samples and the open circuit voltage value samples, and determine the predicted battery health status evaluation result in combination with the first implicit information and the unit state information.
[0122] Optionally, the initial state evaluation model includes: a forget gate, an input gate and an output gate; a model training module 402, which is specifically used to use the forget gate to calculate the second input data at the previous moment, and the second implicit information to obtain the state output result at the previous moment, and the second implicit information is the implicit information of the first implicit information at the previous moment; use the input gate to calculate the state output result, the second input data and the second implicit information to obtain the unit state information at the previous moment; use the output gate to calculate the unit state information, the second input data and the third implicit information at the current moment to obtain the first implicit information at the previous moment.
[0123] The battery health status assessment device provided by the present invention is described below. The battery health status assessment device described below and the battery health status assessment method described above can be referenced to each other.
[0124] Figure 5 is a schematic diagram of the structure of the battery health status assessment device provided by the present invention, such as Figure 5 As shown, the device includes the following data acquisition module 501 and status evaluation module 502.
[0125] The data acquisition module 501 is used to obtain the electrochemical impedance spectrum and open circuit voltage value of the battery to be tested in a charging state.
[0126] The state assessment module 502 is used to input the electrochemical impedance spectrum and the open circuit voltage value into the state assessment model to obtain the battery health state assessment result output by the state assessment model. The state assessment model is based on the above Figure 1 The state assessment model is obtained by training using the training method.
[0127] Figure 6 is a schematic diagram of the structure of the electronic device provided by the present invention, such as Figure 6 As shown, the electronic device may include: a processor 610, a communication interface 620, a memory 630 and a communication bus 640, wherein the processor 610, the communication interface 620 and the memory 630 communicate with each other through the communication bus 640. The processor 610 may call the logic instructions in the memory 630 to execute the training method of the state assessment model, the method comprising: obtaining electrochemical impedance spectrum samples and open circuit voltage value samples of the battery sample in the charging state; determining the impedance characteristic parameter samples corresponding to different frequency bands according to the electrochemical impedance spectrum samples; inputting all impedance characteristic parameter samples and the open circuit voltage value samples into the initial state assessment model to obtain the predicted battery health state assessment result output by the initial state assessment model; updating the model parameters of the initial state assessment model according to the predicted battery health state assessment result and the battery health state represented by the battery capacity to obtain the trained state assessment model. Or, executing a battery health status assessment method, the method comprising: obtaining an electrochemical impedance spectrum and an open circuit voltage value of a battery to be tested in a charging state; inputting the electrochemical impedance spectrum and the open circuit voltage value into a state assessment model to obtain a battery health status assessment result output by the state assessment model, wherein the state assessment model is based on the above Figure 1 The state assessment model is obtained by training using the training method.
[0128] In addition, the logic instructions in the above-mentioned memory 630 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.
[0129] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the training method of the state assessment model provided by the above methods, which includes: obtaining electrochemical impedance spectrum samples and open circuit voltage value samples of battery samples in a charging state; determining impedance characteristic parameter samples corresponding to different frequency bands according to the electrochemical impedance spectrum samples; inputting all impedance characteristic parameter samples and the open circuit voltage value samples into the initial state assessment model to obtain the predicted battery health state assessment result output by the initial state assessment model; updating the model parameters of the initial state assessment model according to the predicted battery health state assessment result and the battery health state characterized by the battery capacity, and obtaining a trained state assessment model. Or, executing the battery health state assessment method provided by the above methods, which includes: obtaining the electrochemical impedance spectrum and open circuit voltage value of the battery to be tested in a charging state; inputting the electrochemical impedance spectrum and the open circuit voltage value into the state assessment model to obtain the battery health state assessment result output by the state assessment model, and the state assessment model is based on the above Figure 1 The state assessment model is obtained by training using the training method.
[0130] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the training method of the state assessment model provided by the above methods, the method comprising: obtaining electrochemical impedance spectrum samples and open circuit voltage value samples of battery samples in a charging state; determining impedance characteristic parameter samples corresponding to different frequency bands according to the electrochemical impedance spectrum samples; inputting all impedance characteristic parameter samples and the open circuit voltage value samples into the initial state assessment model to obtain the predicted battery health state assessment result output by the initial state assessment model; updating the model parameters of the initial state assessment model according to the predicted battery health state assessment result and the battery health state characterized by the battery capacity, to obtain a trained state assessment model. Or, executing the battery health state assessment method provided by the above methods, the method comprising: obtaining the electrochemical impedance spectrum and open circuit voltage value of the battery to be tested in a charging state; inputting the electrochemical impedance spectrum and the open circuit voltage value into the state assessment model to obtain the battery health state assessment result output by the state assessment model, the state assessment model is based on the above Figure 1 The state assessment model is obtained by training using the training method.
[0131] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.
[0132] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.
[0133] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A training method for a state assessment model, characterized in that: include: Obtaining electrochemical impedance spectrum samples and open circuit voltage value samples of the battery sample in a charged state; Determining impedance characteristic parameter samples corresponding to different frequency bands according to the electrochemical impedance spectrum samples; Inputting all impedance characteristic parameter samples and the open circuit voltage value samples into an initial state assessment model to obtain a predicted battery health state assessment result output by the initial state assessment model; According to the predicted battery health status assessment result and the battery health status represented by the battery capacity, the model parameters of the initial status assessment model are updated to obtain a trained status assessment model.
2. The training method of the state assessment model according to claim 1, characterized in that: Determining impedance characteristic parameter samples corresponding to different frequency bands according to the electrochemical impedance spectrum samples includes: Determining target impedance characteristics corresponding to the different frequency bands according to the Nyquist diagram corresponding to the electrochemical impedance spectroscopy sample; For each target impedance feature, the target impedance feature is sampled to obtain multiple frequency points; according to the excitation signal, the impedance values corresponding to each of the multiple frequency points are determined; according to all the impedance values, the impedance characteristic parameter sample corresponding to the frequency band where the target impedance feature is located is determined.
3. The training method of the state assessment model according to claim 2, characterized in that: The step of determining, based on all impedance values, an impedance characteristic parameter sample corresponding to the frequency band where the target impedance characteristic is located comprises: Obtaining a reference impedance value corresponding to the frequency band where the target impedance characteristic is located; The Pearson correlation coefficient method is used to determine the degree of correlation between each impedance value and the reference impedance value, and the impedance characteristic parameter sample corresponding to the frequency band where the target impedance characteristic is located is determined according to the frequency point corresponding to the impedance value corresponding to the maximum correlation degree.
4. The training method of the state assessment model according to claim 2, characterized in that: Determining the target impedance characteristics corresponding to the different frequency bands according to the Nyquist diagram corresponding to the electrochemical impedance spectrum sample includes: According to the resistance characteristics of the battery sample, the Nyquist diagram corresponding to the electrochemical impedance spectrum sample is divided to obtain the target impedance characteristics corresponding to the different frequency bands.
5. The training method of the state assessment model according to any one of claims 1 to 4, characterized in that: The step of inputting all impedance characteristic parameter samples and the open circuit voltage value samples into the initial state evaluation model to obtain the predicted battery health state evaluation result output by the initial state evaluation model includes: Using the initial state evaluation model, determining the first implicit information at a previous moment in the initial state evaluation model and the unit state information at the previous moment; The first input data at the current moment is determined according to all the impedance characteristic parameter samples and the open circuit voltage value samples, and the predicted battery health status evaluation result is determined in combination with the first implicit information and the cell status information.
6. The training method of the state assessment model according to claim 5, characterized in that: The initial state evaluation model includes: a forget gate, an input gate and an output gate; the initial state evaluation model is used to determine the first implicit information at the previous moment in the initial state evaluation model and the unit state information at the previous moment, including: The forget gate is used to calculate the second input data at the previous moment and the second implicit information to obtain the state output result at the previous moment, where the second implicit information is the implicit information of the first implicit information at the previous moment; Using the input gate, calculating the state output result, the second input data and the second implicit information to obtain the unit state information at the previous moment; The output gate is used to calculate the unit state information, the second input data and the third implicit information at the current moment to obtain the first implicit information at the previous moment.
7. A battery health status assessment method, characterized in that: include: Obtain the electrochemical impedance spectrum and open circuit voltage value of the battery under test in the charging state; The electrochemical impedance spectrum and the open circuit voltage value are input into a state assessment model to obtain a battery health status assessment result output by the state assessment model, wherein the state assessment model is trained based on the training method of the state assessment model described in any one of claims 1 to 6 above.
8. A training device for a state assessment model, characterized in that: include: A data acquisition module is used to obtain electrochemical impedance spectrum samples and open circuit voltage value samples of the battery sample in a charging state; A model training module, used to determine impedance characteristic parameter samples corresponding to different frequency bands according to the electrochemical impedance spectrum samples; All impedance characteristic parameter samples and the open circuit voltage value samples are input into the initial state assessment model to obtain a predicted battery health state assessment result output by the initial state assessment model; according to the predicted battery health state assessment result and the battery health state represented by the battery capacity, the model parameters of the initial state assessment model are updated to obtain a trained state assessment model.
9. A battery health status assessment device, characterized in that: include: A data acquisition module is used to obtain the electrochemical impedance spectrum and open circuit voltage value of the battery under test in a charging state; A state assessment module is used to input the electrochemical impedance spectrum and the open circuit voltage value into a state assessment model to obtain a battery health state assessment result output by the state assessment model, wherein the state assessment model is trained based on the training method of the state assessment model described in any one of claims 1 to 6 above.
10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, it implements the training method of the status assessment model as described in any one of claims 1 to 6, or the battery health status assessment method as described in claim 7.
Citation Information
Cited By
Battery state prediction method and system, computer equipment and storage medium
CN119395562A
Battery lithium precipitation quality determination method and module, and model training method and module
CN119782745A
Battery short circuit fault diagnosis method and system based on impedance spectrum
CN120610181A
