EIS Incentive Signal Optimization Method and Optimization System

Abstract

The present invention belongs to the field of electronic technology and relates to an EIS excitation signal optimization method and an optimization system. The steps of the EIS excitation signal optimization method are as follows: Initial phase generation step: Given i single-sine signals with different frequencies and amplitudes, randomly select phases within a range, and use the phase selection method to generate the initial phase of the mixed-frequency signal according to the amplitudes and random phases of the i single-sine signals; Binary signal approximation step: Convert the initial phase together with the given amplitude and given frequency into a time-domain mixed-frequency signal, and perform binary signal approximation on the time-domain mixed-frequency signal to obtain an approximate phase; Crest factor adjustment step: Optimize the approximate phase according to the genetic algorithm to minimize the crest factor of the mixed-frequency signal, and the mixed-frequency signal with the lowest crest factor is the excitation signal. The present invention can reduce the non-linear response and can accurately and quickly measure the battery EIS when performing battery EIS measurement.

Description

Technical Field

[0001] The present invention belongs to the field of electronic technology, and relates to battery EIS measurement technology. Specifically, it relates to an excitation signal optimization method and an optimization system for battery EIS measurement. Background Art

[0002] The normal operation of marine instrument equipment depends on a safe and stable energy system. Considering the complexity of marine environment operations (low temperature, high pressure, etc.), the reliability requirements for the energy system are further improved. Lithium-ion batteries can be recycled, the battery reaction is reversible, the internal resistance is smaller than that of primary lithium batteries, and the load capacity is higher. Compared with lead-acid batteries and nickel-cadmium batteries, lithium-ion batteries have many advantages such as light weight, high specific energy, low self-discharge rate, and long cycle life. Therefore, lithium-ion batteries have become an indispensable part of the energy systems of marine instrument equipment such as submarines, ships, and underwater robots.

[0003] Lithium-ion batteries will gradually age as the usage cycle increases. The reduction of internal electronic energy and quantity is the main factor causing battery capacity decline microscopically. The capacity decline of lithium-ion batteries is a complex non-linear process coupling physical and chemical reactions, mainly including electrolyte decomposition, changes in the mechanical structure of electrode materials, and the formation of lithium dendrites, etc. It can be summarized as Loss of Lithium Inventory (LLI), Loss of Active Material (LAM), and the thickening of the Solid Electrolyte Interphase (SEI), etc. In order to quantify the capacity attenuation of lithium-ion batteries, researchers introduced the State of Health (SOH) index. SOH is calculated by the ratio of the actual battery capacity to the nominal capacity, and is expressed as:

[0004]

[0005] In the formula, C present is the maximum available capacity of the battery under current conditions, C initial is the initial capacity of a brand-new battery.

[0006] When the State of Health (SOH) of a lithium-ion battery decays to 80%, it is regarded as the End of Life (EOL). At this time, the replacement of lithium-ion batteries and the recycling of waste lithium-ion batteries are required. Recycling waste lithium-ion batteries, as the last link in the use process of lithium-ion batteries, is conducive to realizing the recycling of resources and the cascade utilization of lithium-ion batteries, and can avoid the pollution of the environment by harmful substances in lithium-ion batteries. When the performance of a lithium-ion battery drops to a certain extent, various accidents such as leakage and short circuit may occur.

[0007] As an important part of the energy system of marine instrument equipment, if the health state of a lithium-ion battery is not good, it may lead to failures of marine instrument equipment, data loss or mission failures, etc., bringing serious impacts to aspects such as marine surveys, scientific research and resource development. At the same time, accurate health state assessment provides an information basis for the mission scheduling and equipment maintenance of marine instrument research equipment, avoiding waste of time and cost caused by unnecessary equipment maintenance. Electrochemical Impedance Spectroscopy (EIS) diagnosis method is a non-destructive diagnosis method, which can obtain the internal electrochemical information of the battery while ensuring the integrity of the battery. It can be used for the modeling of the battery equivalent circuit, the estimation of the state of charge and temperature, and the diagnosis of battery health. By applying a small-amplitude sinusoidal current (constant current mode) or voltage (constant voltage mode) within a certain frequency range to the battery, measuring the voltage or current response of the battery, obtaining the complex impedance of the battery, and characterizing it on the complex plane, the electrochemical impedance spectrum of the battery can be obtained.

[0008] Traditional EIS measurements use a single sine wave signal to measure selected frequencies in sequence from a selected frequency range through professional equipment (electrochemical workstation, frequency response analyzer, etc.). The measurement results of this method are accurate, but the experimental process is very slow. To obtain a complete EIS curve, it takes several hours. Shortening the measurement time of EIS helps to apply EIS to practical engineering. Locorotondo et al. used a pseudo-random binary sequence instead of a single sine wave signal. This binary sequence has characteristics similar to white noise signals in the frequency domain and can be used to excite the battery from a wide frequency range, thus accelerating the test process. Wang et al. used wavelet transform to process the excitation current and voltage, converting the frequency domain measurement to the time domain measurement. The results show that except for the low-frequency band, the measurement results of the medium and high-frequency impedance are consistent with the swept-frequency measurement method. Lohmann et al. used multiple signals (rectangular pulse, Gaussian pulse, and Sinc pulse) to excite the battery, reducing the measurement time to 13% of the traditional EIS test. Klotz et al. combined the time domain measurement method with the frequency domain measurement method, which can obtain higher frequency resolution, lower Kramers-Kronig test residuals, and less measurement time compared with the traditional EIS test. Liebhart et al. developed a new framework for on-board impedance spectroscopy of electric vehicles, which only uses the current and voltage signals generated by operating the vehicle to perform passive impedance spectroscopy measurement. The obtained impedance spectroscopy data can be used for the diagnosis of battery-related performance. Temiz et al. used machine learning methods to regenerate the complete battery impedance from a small experimental dataset, avoiding the complete EIS test and saving a large amount of money in terms of time, cost, and repeatability.

[0009] Converting the EIS measurement from the frequency domain to the time domain can shorten the measurement time, but at the same time, it also introduces problems such as frequency leakage and non-linear response, resulting in a decrease in measurement accuracy, which is not conducive to the subsequent interpretation and analysis of EIS data and the estimation of SOH. Therefore, selecting an appropriate time domain excitation signal is an urgent problem to be solved in EIS measurement. Summary of the Invention

[0010] In view of the above problems existing in the prior art, the present invention provides an EIS excitation signal optimization method and an optimization system, which can reduce non-linear response and accurately and quickly measure the battery EIS during battery EIS measurement.

[0011] To achieve the above object, the first aspect of the present invention provides an EIS excitation signal optimization method, the steps of which are as follows:

[0012] Initial phase generation step: Given i single sine wave signals with different frequencies and amplitudes, select random phases within the range of and use the phase selection method according to iGenerate the initial phase of the mixed-frequency signal from the amplitude and random phase of a single sine signal;

[0013] Binary signal approximation step: Convert the initial phase together with the given amplitude and given frequency into a time-domain mixed-frequency signal, and perform binary signal approximation on the time-domain mixed-frequency signal to obtain an approximate phase;

[0014] Crest factor adjustment step: Optimize the approximate phase according to the genetic algorithm to minimize the crest factor of the mixed-frequency signal. The mixed-frequency signal with the lowest crest factor is the excitation signal.

[0015] In some embodiments, the binary signal approximation step further includes:

[0016] Time-domain conversion step: Convert the initial phase, given amplitude, and given frequency together into a time-domain mixed-frequency signal;

[0017] Nonlinear transformation step: Perform a nonlinear transformation on the time-domain mixed-frequency signal to become an approximate square-wave signal, then convert the approximate square-wave signal to the frequency domain through a fast Fourier transform to obtain the frequency-domain phase, and convert the frequency-domain phase, given amplitude, and given frequency back to the time domain through an inverse fast Fourier transform to obtain a time-domain signal;

[0018] Time-domain clipping step: Use a clipping algorithm to binarize the time-domain signal to obtain an approximate phase.

[0019] In some embodiments, in the nonlinear transformation step, an improved Sigmoid function is used for the nonlinear transformation to limit the amplitude of the time-domain mixed-frequency signal within the range of [-0.5, 0.5]. The improved Sigmoid function is expressed as:

[0020]

[0021] where, k is a constant for adjusting the degree of nonlinear transformation.

[0022] In some embodiments, in the initial phase generation step, the frequency of the single sine signal is from 0.01 Hz to 10 kHz, the amplitude is from 5 to 15 mV, and the frequencies of different single sine signals are fixed multiples of each other.

[0023] In some embodiments, the initial phase is expressed as:

[0024]

[0025] where, ;

[0026] where, is the initial phase of the mixed-frequency signal, is the random phase, nis the number of sine signals that make up the mixed-frequency signal, is the i relative power spectral density of the th i sine signal, M is the number of amplitudes.

[0027] To achieve the above object, a second aspect of the present invention provides an EIS excitation signal optimization system for implementing the EIS excitation signal optimization method described in the first aspect of the present invention. The system includes:

[0028] A setting module for specifying i single sine signals with different frequencies and amplitudes, and selecting random phases within a range;

[0029] An initial phase generation module for generating the initial phase of the mixed-frequency signal according to the amplitudes and random phases of i single sine signals using a phase selection method;

[0030] A binary signal approximation module for converting the initial phase together with the specified amplitude and specified frequency into a time-domain mixed-frequency signal, and performing binary signal approximation on the time-domain mixed-frequency signal to obtain an approximate phase;

[0031] A crest factor adjustment module for optimizing the approximate phase according to a genetic algorithm to minimize the crest factor of the mixed-frequency signal. The mixed-frequency signal with the lowest crest factor is the excitation signal.

[0032] In some embodiments, the binary signal approximation module further includes:

[0033] A time-domain conversion module for converting the initial phase, the specified amplitude, and the specified frequency together into a time-domain mixed-frequency signal;

[0034] A non-linear transformation module for performing non-linear transformation on the time-domain mixed-frequency signal to become an approximate square-wave signal, then converting the approximate square-wave signal to the frequency domain through a fast Fourier transform to obtain a frequency-domain phase, and converting the frequency-domain phase, the specified amplitude, and the specified frequency back to the time domain through an inverse fast Fourier transform to obtain a time-domain signal;

[0035] A time-domain clipping module for binarizing the time-domain signal using a clipping algorithm to obtain an approximate phase.

[0036] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0037] The EIS excitation signal optimization method and optimization system of the present invention introduce the crest factor to evaluate and describe the "sharpness" of the signal in the time domain. The initial phase, given amplitude, and given frequency generated by the phase selection method are converted into a time-domain mixed-frequency signal together, and then binary signal approximation (including non-linear transformation and time-domain clipping) is performed to obtain an approximate phase. Then, the approximate phase is optimized by a genetic algorithm to minimize the crest factor of the overall mixed-frequency signal, so as to achieve the purpose of optimizing the excitation signal. Using the optimized excitation signal of the present invention for battery EIS measurement can reduce the non-linear response of the battery during EIS measurement and improve the accuracy of EIS measurement. Description of the Drawings

[0038] Figure 1 It is a flowchart of the EIS excitation signal optimization method described in the embodiment of the present invention;

[0039] Figure 2 It is a flowchart of the binary signal approximation described in the embodiment of the present invention;

[0040] Figure 3 For different k Sigmoid function schematic diagrams of values in the embodiment of the present invention;

[0041] Figure 4 For different k Crest factor comparison schematic diagrams of values in the embodiment of the present invention;

[0042] Figure 5 It is a non-flat amplitude schematic diagram of sinusoidal modulation in the embodiment of the present invention;

[0043] Figure 6 It is a schematic diagram of the flat amplitude spectrum described in the embodiment of the present invention;

[0044] Figure 7 It is a waveform diagram of the excitation signal before optimization described in the embodiment of the present invention;

[0045] Figure 8 It is a waveform diagram of the excitation signal after optimization described in the embodiment of the present invention;

[0046] Figure 9 It is a structural block diagram of the EIS excitation signal optimization system described in the embodiment of the present invention;

[0047] Figure 10 It is a schematic diagram of the EIS test result measured by using the non-optimized excitation signal through the EIS measurement method based on discrete Fourier transform described in the embodiment of the present invention;

[0048] Figure 11 It is a schematic diagram of the EIS test result measured by using the optimized excitation signal through the EIS measurement method based on discrete Fourier transform described in the embodiment of the present invention;

[0049] Figure 12 Schematic diagram of the K-K analysis residual obtained by using an unoptimized excitation signal through the EIS measurement method based on discrete Fourier transform according to the embodiments of the present invention;

[0050] Figure 13 Schematic diagram of the K-K analysis residual measured by using an unoptimized excitation signal through the EIS measurement method based on discrete Fourier transform according to the embodiments of the present invention.

[0051] In the figure, 1. setting module, 2. initial phase generation module, 3. binary signal approximation module, 31. time domain conversion module, 32. non-linear transformation module, 33. time domain clipping module, 4. crest factor adjustment module. Detailed implementation manners

[0052] Next, the present invention will be specifically described by way of exemplary embodiments. However, it should be understood that, without further elaboration, the elements, structures, and features in one embodiment can also be beneficially incorporated into other embodiments.

[0053] EIS has a wide range of applications in different fields. In the field of electrochemistry, EIS is suitable for measuring the reaction mechanism inside the battery to evaluate the charging and aging status of the battery. For applications in different fields, the EIS measurement frequency range is also different. For the health state estimation of lithium-ion batteries, the EIS measurement frequency usually needs to cover the frequency range from 10 mHz to several kHz. The traditional frequency sweep measurement is slow and cannot meet the requirements of EIS multi-scenario measurement. With the development of digital signal analysis technology, the mixed-frequency signal can be used as an alternative to the single-frequency signal (single sine signal), changing the frequency domain sweep measurement of EIS to a mixed-frequency measurement. The mixed-frequency signal is superimposed by multiple different frequency signals, which can excite multiple frequencies at the same time, greatly shortening the time required for EIS measurement and providing a basis for dynamic in-situ EIS measurement. However, the introduction of the mixed-frequency signal will simultaneously introduce problems of non-linear response and frequency leakage, resulting in poor measurement accuracy. In order to reduce the non-linear response and frequency leakage of the battery during EIS measurement, the present invention provides an EIS excitation signal optimization method and an optimization system, which consider the amplitudes of different frequency components of the mixed-frequency signal, generate an initial phase by combining the phases of different single sine signals, generate a time-domain mixed-frequency signal according to the initial phase, perform non-linear transformation and time-domain clipping processing on the time-domain mixed-frequency signal to obtain an approximate phase, and optimize the approximate phase through a genetic algorithm to obtain a smaller crest factor, thereby reducing the non-linear response of the battery during EIS measurement. The excitation signal obtained based on the above EIS excitation signal optimization method and optimization system. The above EIS excitation signal optimization method and optimization system will be described in detail below with reference to the accompanying drawings.

[0054] SeeFigure 1 , in the embodiment of the first aspect of the present invention, an EIS excitation signal optimization method is provided, and its steps are as follows:

[0055] S1. Initial phase generation step: Given i single sine signals with different frequencies and amplitudes, randomly select phases within range, and use the phase selection method to generate the initial phase of the mixed-frequency signal according to i the amplitudes and random phases of the single sine signals.

[0056] Specifically, in some embodiments, the phase selection method adopts the phase-phase selection method proposed by Schroeder. This method considers the amplitudes of different frequency components of the mixed-frequency signal and controls the crest factor by appropriately combining the phases of the single sine signals. The initial phase is expressed as:

[0057] (1)

[0058] Where ;

[0059] In the formula, is the initial phase of the i th sine signal, is the random phase, n is the number of sine signals that make up the mixed-frequency signal, is the i th sine signal's relative power spectral density, is the i th sine signal's amplitude, M is the number of amplitudes.

[0060] It should be noted that when the amplitudes of each sine signal are constant, formula (1) can be simplified to:

[0061] (2)

[0062] Specifically, although multiple single sine signals can achieve the purpose of simultaneously exciting multiple frequencies by superimposing different sine waves, the non-linear characteristics of the battery itself make the design of this signal not too arbitrary. The amplitude, phase, and frequency of each selected single sine signal will have an impact on the battery EIS measurement results. The different types of batteries determine the frequency selection of multiple single sine signals. Most batteries exhibit certain inductive properties at high frequencies, which mainly depend on the geometric characteristics of the battery. For pouch batteries, no inductive characteristics will appear within the frequency range of several thousand hertz, while for cylindrical or prismatic batteries, the capacitive characteristics at high frequencies may be affected by the inductive characteristics caused by the geometric structure of the collector. When the signal frequency is less than 0.01 Hz, the time required to measure EIS is relatively long, and when the signal frequency is greater than 10 kHz, the measured impedance is the inductive characteristic that we don't care about, and too high a sampling rate also increases the measurement cost. Therefore, to balance the measurement requirements and measurement cost, in some specific embodiments of the invention, the frequency of the single sine signal is 0.01 Hz to 10 kHz.

[0063] Specifically, since the frequency distribution with a fixed spacing may cause second-order harmonic distortion. Therefore, in some specific embodiments of the invention, the frequencies of the selected different single sine signals are fixed multiples of each other.

[0064] It should be noted that to ensure the validity of the measured EIS signal, the amplitude of the excitation signal should not be too large, and the battery should be prevented from being polarized to the Tafel region to meet the linear condition. However, it should not be too small either to avoid noise destroying the causality condition and ensure a sufficient signal-to-noise ratio for robust measurement. In some embodiments, the amplitude of the single sine signal is 5 to 15 mV.

[0065] S2. Binary signal approximation step: Convert the initial phase together with the given amplitude and given frequency into a time-domain mixed-frequency signal, and perform binary signal approximation on the time-domain mixed-frequency signal to obtain an approximate phase.

[0066] Specifically, referring to Figure 2 , the binary signal approximation step further includes:

[0067] S21. Time-domain conversion step: Convert the initial phase, given amplitude, and given frequency together into a time-domain mixed-frequency signal, and the time-domain mixed-frequency signal is expressed as:

[0068] (3)

[0069] In the formula, is the time-domain mixed-frequency signal, N is the total number of frequency points, is the amplitude of the i th sine signal, is the initial phase of the i th sine signal, is the frequency of the i th sine signal.

[0070] S22, Nonlinear transformation step: Perform a nonlinear transformation on the time-domain mixing signal to become an approximate square-wave signal, and then convert the approximate square-wave signal to the frequency domain through a fast Fourier transform to obtain the frequency-domain phase. Convert the frequency-domain phase, the given amplitude, and the given frequency to the time domain through an inverse fast Fourier transform to obtain the time-domain signal.

[0071] Specifically, an improved Sigmoid function (see Figure 3 ) is used for the nonlinear transformation to limit the amplitude of the time-domain mixing signal within the range of [-0.5, 0.5]. The improved Sigmoid function is expressed as:

[0072] (4)

[0073] In the formula, k is a constant for adjusting the degree of nonlinear transformation.

[0074] The degree of nonlinear transformation can be determined by changing the k value. For example: Through 10 combinations of different amplitudes, phases, and frequencies, calculate the crest factor after 10 nonlinear transformations to select an appropriate k value. As shown in Figure 4 , the results show that has the best effect.

[0075] S23, Time-domain clipping step: Use a clipping algorithm to binarize the time-domain signal to obtain an approximate phase.

[0076] Specifically, the clipping algorithm performs waveform clipping according to the given clipping threshold. It should be noted that choosing an appropriate clipping threshold is very important for the effect of the clipping algorithm. When the clipping threshold is greater than 90%, the convergence speed may be reduced; when the clipping threshold is less than 70%, the amplitude distribution will change. Therefore, in some embodiments of the present invention, the given clipping threshold is between 70% and 90%.

[0077] Specifically, in some embodiments of the present invention, the clipping algorithm uses an improved sub-optimal threshold selection method proposed by Yang et al., called the variable coefficient clipping algorithm. This clipping algorithm uses a logarithmic function as the clipping function and generates different clipping thresholds according to different numbers of iterations.

[0078] S3, Crest factor adjustment step: Optimize the approximate phase according to the genetic algorithm to minimize the crest factor of the mixing signal. The mixing signal with the lowest crest factor is the excitation signal.

[0079] It should be noted that for a single sine signal, if the approximate impedance range of the battery is known in advance, the amplitude of the excitation signal can be easily controlled. However, for multiple single sine signals, the amplitude and phase of each frequency will affect the total amplitude of the final mixed-frequency signal. Therefore, in order to prevent its peak from being too "sharp", each single sine signal participating in the mixing needs to be modulated to ensure that the final mixed-frequency signal (i.e., the optimized excitation signal) meets the three basic conditions of causality, linearity, and stability of the electrochemical system.

[0080] A commonly used index for evaluating and describing the "sharpness" of a signal in the time domain is the Crest Factor (CF). This index indicates the amplitude consumed by the signal when introducing energy into the system. The larger the value, the "sharper" the signal. The crest factor is defined as the ratio of the peak value of the signal to its effective value. For the time-domain mixed-frequency signal within the set time interval T , the crest factor is expressed as:

[0081] (5)

[0082] In the formula, , by adjusting the initial phase of the sine signal in formula (3), the crest factor of the overall mixed-frequency signal can be changed to achieve the purpose of optimizing the excitation signal.

[0083] Specifically, the problem of approximate phase optimization is actually a crest factor optimization problem. For the crest factor optimization problem, in the genetic algorithm, all genes on the chromosome represent the initial phases of the individual single sine signals that make up the synthesized mixed-frequency signal (i.e., the multi-sine signal) . Using the tournament selection method, the individual with the highest fitness (i.e., the lowest crest factor) is selected from the candidate chromosomes for inheritance. The two selected chromosomes are randomly crossed with a certain probability to produce offspring, and finally, the phase angles of each offspring are changed through probability-based mutation. Repeat the above steps until the maximum number of iterations or a suitable crest factor is reached. The entire optimization problem can be described as: .

[0084] In the above EIS excitation signal optimization method of the embodiment of the present invention, the phase selection method is used to generate the initial phase of the mixed-frequency signal according to the amplitudes and random phases of i single sine signals, convert the initial phase into the time-domain form, and then perform binaryization through nonlinear transformation (including fast Fourier transform and inverse fast Fourier transform) and time-domain clipping to obtain the approximate phase. The approximate phase is optimized according to the genetic algorithm to minimize the crest factor of the mixed-frequency signal to obtain the excitation signal. Using the optimized excitation signal of the present invention for battery EIS measurement can reduce the nonlinear response of the battery during EIS measurement and improve the accuracy of EIS measurement.

[0085] To evaluate the optimization effect of the above excitation signal optimization method on the crest factor, the performance is verified using the simulated generated spectrum. According to the amplitudes of the single-frequency signals that make up the multi-sine signal, the simulated spectrum is divided into a flat amplitude spectrum and a non-flat amplitude spectrum. A flat amplitude spectrum means that the amplitude of each single-frequency signal participating in the mixing is the same, while the non-flat amplitude spectrum is not necessarily the same.

[0086] It should be noted that the accuracy of the crest factor calculation determines the overall quality of the signal. To ensure that the calculation result is approximately equal to the true value, the sampling rate must satisfy the Shannon sampling theorem, that is, the sampling rate is greater than twice the highest frequency. The sampling rate setting condition shown in the following formula (6) is adopted. In this way, at least 10 sampling points are included in each period of the highest frequency, fully ensuring the accuracy of the root mean square value and the maximum value of the signal.

[0087] (6)

[0088] In the formula, is the sampling rate, is the highest frequency.

[0089] For the simulated spectrum used for testing, five methods are used to optimize the crest factor, namely: (1) approximate binarization; (2) traditional genetic algorithm; (3) the excitation signal optimization method proposed by the present invention; (4) randomly generate phases; (5) Schroeder phase selection method. Each algorithm runs multiple times, and its minimum value, maximum value and average value are recorded for comparison.

[0090] For the non-flat amplitude spectrum, it only needs to ensure that the maximum and minimum values of the amplitude are inconsistent. In the embodiment of the present invention, a sine wave modulation method is used to generate the amplitude value of the non-flat amplitude spectrum, as shown in formula (7).

[0091] (7)

[0092] In the formula, i is the number of frequencies. The form of the amplitude of the non-flat amplitude spectrum modulated by sine is as Figure 5 shown.

[0093] Table 1

[0094]

[0095] Referring to Table 1, the randomness of the initial population of the traditional genetic algorithm may lead to an accidental minimum value comparable to that of the improved algorithm, but in terms of the overall average value, the excitation signal optimization method proposed by the present invention is significantly more superior.

[0096] For the flat amplitude spectrum, it only needs to ensure that the amplitudes of each frequency component are the same. The present invention selects 0.015 as the amplitude of the single-frequency signal, and the corresponding amplitude value of each frequency is asFigure 6 as shown

[0097] Table 2

[0098]

[0099] As can be seen from Table 1 and Table 2, for both flat amplitude spectra and non-flat amplitude spectra, the excitation signal optimization method proposed by the present invention can achieve the crest factor with the lowest overall mean level, solving the problem that the genetic algorithm of the random population is prone to falling into local minima. All five methods tested have better performance under flat amplitude. This is because the different amplitudes increase the difficulty of phase selection, making the overall crest factor of the non-flat amplitude spectrum relatively high.

[0100] Figure 7 and Figure 8 is a comparison between the signal optimized by the excitation signal optimization method proposed by the present invention and the random phase signal in an experiment selected from the flat amplitude spectrum. As can be seen from the yellow rectangular part in the figure, the "sharpness" of the optimized signal is significantly reduced, and the peak value of the signal is less than 0.15V, which can meet the requirements of battery pseudo-linearity.

[0101] See Figure 9 , in the second aspect embodiment of the present invention, there is provided an EIS excitation signal optimization system for implementing the EIS excitation signal optimization method described in the first aspect embodiment of the present invention. The system includes:

[0102] Setting module 1, for giving i single sine signals with different frequencies and amplitudes, and selecting random phases within range;

[0103] Initial phase generation module 2, for using the phase selection method to generate the initial phase of the mixed-frequency signal according to the amplitudes and random phases of i single sine signals;

[0104] Binary signal approximation module 3, for converting the initial phase together with the given amplitude and given frequency into a time-domain mixed-frequency signal, and performing binary signal approximation on the time-domain mixed-frequency signal to obtain an approximate phase;

[0105] Crest factor adjustment module 4, for optimizing the approximate phase according to the genetic algorithm to make the crest factor of the mixed-frequency signal the lowest, and the mixed-frequency signal with the lowest crest factor is the excitation signal.

[0106] In some embodiments, continue to refer to Figure 9 , the binary signal approximation module 3 further includes:

[0107] The time-domain conversion module 31 is used to convert the initial phase, the given amplitude, and the given frequency together into a time-domain mixed-frequency signal;

[0108] The non-linear transformation module 32 is used to perform non-linear transformation on the time-domain mixed-frequency signal to become an approximate square-wave signal, and then convert the approximate square-wave signal to the frequency domain through fast Fourier transform to obtain the frequency-domain phase. The frequency-domain phase, the given amplitude, and the given frequency are converted to the time domain through inverse fast Fourier transform to obtain a time-domain signal;

[0109] The time-domain clipping module 33 is used to perform binaryization on the time-domain signal using a clipping algorithm to obtain an approximate phase.

[0110] In the EIS excitation signal optimization system of the embodiment of the present invention, the phase selection method is used to generate the initial phase of the mixed-frequency signal according to the amplitudes and random phases of i single-sine signals, convert the initial phase into a time-domain form, and then perform binaryization through non-linear transformation (including fast Fourier transform and inverse fast Fourier transform) and time-domain clipping to obtain an approximate phase. The approximate phase is optimized according to the genetic algorithm to minimize the crest factor of the mixed-frequency signal to obtain the excitation signal. Using the optimized excitation signal of the present invention for battery EIS measurement can reduce the non-linear response of the battery during EIS measurement and improve the accuracy of EIS measurement.

[0111] In order to verify the effect of the optimized excitation signal in the EIS test, an EIS measurement method based on discrete Fourier transform is used for the EIS test. Figure 10 and Figure 11 are respectively the comparisons of the EIS curves measured by the unoptimized excitation signal and the optimized excitation signal with the true EIS curve.

[0112] From Figure 10 it can be seen that the EIS measured using the unoptimized excitation signal is affected by the non-linear characteristics of the battery, resulting in obvious offset errors between each frequency point and the true value, which affects the smoothness of the entire EIS curve. As Figure 11 shown, the optimized signal reduces the crest factor, makes the battery operate in the linear region, the overall EIS curve is relatively smooth, and the offset is low. In order to specifically measure the magnitude of this error, the K-K analysis is continued to evaluate the error. The K-K analysis residuals of the EIS curves of the unoptimized excitation signal and the optimized EIS curve are respectively as Figure 12 and Figure 13 shown. The K-K analysis residual of the unoptimized excitation signal can reach up to 4%, and the residual of the optimized excitation signal does not exceed 0.2%. The error is mainly distributed in the low-frequency band, and this error performance may be due to insufficient frequency resolution in the low-frequency band.

[0113] In summary, precise EIS signals can be obtained by performing EIS measurements using the optimized excitation signal.

[0114] The above embodiments are used to explain the present invention rather than to limit the present invention. Any modifications and changes made to the present invention within the spirit and scope of the claims of the present invention fall within the protection scope of the present invention.

Claims

1. An EIS excitation signal optimization method, characterized in that The steps are as follows: Initial phase generation step: Given i single sine signals with different frequencies and amplitudes, select random phases within , and use the phase selection method to generate the initial phase of the mixed-frequency signal according to the amplitudes and random phases of i single sine signals; Binary signal approximation step: Convert the initial phase together with the given amplitude and given frequency into a time-domain mixed-frequency signal, and perform binary signal approximation on the time-domain mixed-frequency signal to obtain an approximate phase; Crest factor adjustment step: Optimize the approximate phase according to the genetic algorithm to minimize the crest factor of the mixed-frequency signal. The mixed-frequency signal with the lowest crest factor is the excitation signal; The binary signal approximation step further includes: Time-domain conversion step: Convert the initial phase, given amplitude, and given frequency together into a time-domain mixed-frequency signal; Nonlinear transformation step: Perform a nonlinear transformation on the time-domain mixed-frequency signal to become an approximate square-wave signal, then convert the approximate square-wave signal to the frequency domain through the fast Fourier transform to obtain the frequency-domain phase, and convert the frequency-domain phase, given amplitude, and given frequency back to the time domain through the inverse fast Fourier transform to obtain the time-domain signal; Time-domain clipping step: Use a clipping algorithm to binarize the time-domain signal to obtain an approximate phase.

2. The EIS excitation signal optimization method according to claim 1, wherein In the nonlinear transformation step, an improved Sigmoid function is used for nonlinear transformation to limit the amplitude of the time-domain mixed-frequency signal within the range of [-0.5, 0.5]. The improved Sigmoid function is expressed as: wherein, k is a constant for adjusting the degree of non-linear transformation.

3. The EIS excitation signal optimization method according to claim 1, characterized in that, In the initial phase generation step, the frequency of the single sine wave signal is from 0.01 Hz to 10 kHz, and the amplitude is from 5 to 15 mV. The frequencies of different single sine wave signals are fixed multiples of each other.

4. The EIS excitation signal optimization method according to claim 1, characterized in that The initial phase is expressed as: Among them, ; Wherein, is the initial phase of the mixed-frequency signal, is the random phase, n is the number of sine signals constituting the mixed-frequency signal, is the i th relative power spectral density of the sine signal, is the i th amplitude of the sine signal, M is the number of amplitudes.

5. An EIS excitation signal optimization system for implementing the EIS excitation signal optimization method according to any one of claims 1 to 4, characterized in that, including: A setting module for providing i single sine signals with different frequencies and amplitudes and selecting random phases within a range; An initial phase generation module, which is used to generate the initial phase of the mixed-frequency signal according to the amplitude and random phase of i single sine signals by using the phase selection method; A binary signal approximation module for converting the initial phase together with the given amplitude and given frequency into a time-domain mixed-frequency signal, and performing binary signal approximation on the time-domain mixed-frequency signal to obtain an approximate phase; A crest factor adjustment module for optimizing the approximate phase according to the genetic algorithm to minimize the crest factor of the mixed-frequency signal. The mixed-frequency signal with the lowest crest factor is the excitation signal; The binary signal approximation module further includes: A time-domain conversion module for converting the initial phase, given amplitude, and given frequency together into a time-domain mixed-frequency signal; A nonlinear transformation module for performing a nonlinear transformation on the time-domain mixed-frequency signal to become an approximate square-wave signal, then converting the approximate square-wave signal to the frequency domain through the fast Fourier transform to obtain the frequency-domain phase, and converting the frequency-domain phase, given amplitude, and given frequency back to the time domain through the inverse fast Fourier transform to obtain the time-domain signal; A time-domain clipping module for using a clipping algorithm to binarize the time-domain signal to obtain an approximate phase.

Images