A battery state prediction method, system, computer device, and storage medium

Abstract

The present invention provides a battery state prediction method, system, computer device and storage medium. The method is to randomly select a preset proportion of batteries in the target batch of batteries as a test sample set, perform a current switching cycle life test on each battery sample to obtain multiple sets of current mutation data including current time series data, voltage time series data and battery health state corresponding to a preset state of charge segment point sequence, then obtain the battery impedance spectrum at the current mutation point according to the current time series data and voltage time series data in each set of current mutation data, construct a battery state prediction model based on all the battery impedance spectra at the current mutation points and the corresponding battery health states, and obtain the battery state prediction value according to the battery impedance spectrum at the current mutation of the battery to be evaluated obtained and the battery state prediction model. The present invention can effectively improve the efficiency, accuracy and operation convenience of online monitoring of the battery health state at low cost without additional measurement equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of battery state management, and particularly to a battery state prediction method, system, computer device, and storage medium. Background Art

[0002] The battery state is the basis for measuring the battery's storage capacity and power transmission capacity, which reflects the degree of battery aging and can be used for battery life assessment and battery optimization management to avoid potential safety hazards caused by battery failures.

[0003] Existing battery state predictions mainly include state estimation methods based on battery equivalent circuit models and battery health state estimation methods based on data-driven approaches. However, state estimation based on battery equivalent circuit models mainly estimates the state of charge of the battery and depends on the accuracy of the battery equivalent circuit model. It is difficult for the battery equivalent circuit model to directly reflect the battery capacity, and the battery capacity estimation is easily coupled with the state of charge estimation, resulting in greater difficulty in battery state estimation. During the battery aging process, when the battery model changes, it is easy for the estimation deviation to gradually accumulate, even leading to internal circulation or short-board effects in the energy storage system. At the same time, data-driven battery health state estimation mainly directly uses data such as current, voltage, and temperature during battery operation to fit the state data of the current battery through various neural networks. However, the coupling between current and temperature in the actual battery operation data is strong, and the temperature itself cannot have a sudden change, making it easy for the neural network training to have overfitting / underfitting problems. As a result, while consuming a large amount of time and computational costs, the estimation results are still relatively divergent. Although existing methods introducing impedance information can, to a certain extent, solve the application defects of the above data-driven methods, battery impedance must be measured using pulsed current, which generally can only be carried out when the system is idle and is difficult to perform during the real-time operation of the system. There are significant limitations and it is difficult to meet the timeliness requirements of state detection, nor can it be applied to the state monitoring of the current energy storage system energy scheduling scenario. Summary of the Invention

[0004] The purpose of the present invention is to provide a battery state prediction method. By collecting voltage data and current data before and after current mutation points under different battery health states based on current switching cycle life tests, a data-driven method is used in combination with polynomial fitting to construct a battery state prediction model for analyzing the current mutation battery impedance spectrum of the same batch of batteries to be evaluated in actual use to obtain corresponding battery state prediction values. This can not only avoid relying on the battery equivalent circuit model but also avoid problems such as high cost, poor accuracy, and insufficient timeliness in the application of existing data-driven models. It can effectively improve the efficiency, accuracy, and operational convenience of online monitoring of battery health status at low cost without additional measurement equipment.

[0005] To achieve the above object, in view of the above technical problems, it is necessary to provide a battery state prediction method, system, computer device, and storage medium.

[0006] In a first aspect, an embodiment of the present invention provides a battery state prediction method, and the method includes the following steps:

[0007] Randomly select a preset proportion of batteries from the target batch of batteries as a test sample set;

[0008] Perform a current switching cycle life test on each battery sample in the test sample set to obtain multiple sets of current mutation data corresponding to a preset state of charge segment point sequence; the current mutation data includes current time series data and voltage time series data centered on the current mutation moment, and the battery health state corresponding to the current mutation moment;

[0009] According to the current time series data and voltage time series data in each set of current mutation data, obtain the corresponding battery impedance spectrum at the current mutation point;

[0010] Construct a corresponding battery state prediction model according to all the battery impedance spectra at the current mutation points and the corresponding battery health states;

[0011] Obtain the battery impedance spectrum at the current mutation of the battery to be evaluated in the target batch of batteries, and according to the battery impedance spectrum at the current mutation and the battery state prediction model, obtain the battery state prediction value corresponding to the battery to be evaluated.

[0012] Further, the step of performing a current switching cycle life test on each battery sample in the test sample set to obtain multiple sets of current mutation data corresponding to a preset state of charge segment point sequence includes:

[0013] According to the actual operating condition range of the battery sample, set a first charge and discharge current, a second charge and discharge current, and a minimum charge and discharge current; the first charge and discharge current is greater than the second charge and discharge current;

[0014] Perform a cycle life test on each battery sample according to a preset current switching charge and discharge rule and collect the corresponding current mutation data and battery health state until the total cycle discharge charge drops to a preset charge; the preset current switching charge and discharge rule is to alternately use the first charge and discharge current and the second charge and discharge current to perform current rate switching charging on each battery sample according to each state of charge in the preset state of charge segment point sequence until the cut-off voltage is reached, and then switch to constant voltage discharge until the current drops to the minimum charge and discharge current and then stand still for a preset standstill duration.

[0015] Further, the step of obtaining the corresponding battery impedance spectrum at the current mutation point according to the current time series data and voltage time series data in each set of current mutation data includes:

[0016] Perform complex Morlet wavelet transforms on the current time series data and the voltage time series data at different frequencies respectively to obtain the corresponding current wavelet coefficients and voltage wavelet coefficients;

[0017] Based on the current wavelet coefficients and voltage wavelet coefficients at different frequencies, obtain the battery impedance at the corresponding current mutation points;

[0018] Generate the battery impedance spectrum at the current mutation points according to the battery impedance at the current mutation points at different frequencies.

[0019] Further, the step of constructing the corresponding battery state prediction model according to all the battery impedance spectra at the current mutation points and the corresponding battery health states includes:

[0020] Obtain the intersection value of the real axis of the impedance spectrum, the amplitude of the inflection point impedance, and the phase angle value of the inflection point impedance corresponding to each battery impedance spectrum at the current mutation points within different cycle life test periods;

[0021] Based on the intersection value of the real axis of the impedance spectrum, the amplitude of the inflection point impedance, and the phase angle value of the inflection point impedance, obtain the sequence of intersection values of the real axis of the impedance spectrum, the sequence of amplitudes of the inflection point impedance, and the sequence of phase angle values of the inflection point impedance corresponding to the preset state-of-charge segment point sequence within different cycle life test periods;

[0022] Use the sequence of intersection values of the real axis of the impedance spectrum, the sequence of amplitudes of the inflection point impedance, and the sequence of phase angle values of the inflection point impedance as independent variables, and the corresponding battery health state as the dependent variable for polynomial fitting to obtain the battery state prediction model.

[0023] Further, the step of using the sequence of intersection values of the real axis of the impedance spectrum, the sequence of amplitudes of the inflection point impedance, and the sequence of phase angle values of the inflection point impedance as independent variables, and the sequence of corresponding battery health states as the dependent variable for polynomial fitting to obtain the battery state prediction model includes:

[0024] Use the sequence of intersection values of the real axis of the impedance spectrum, the sequence of amplitudes of the inflection point impedance, and the sequence of phase angle values of the inflection point impedance in the previous preset number of test periods as independent variables respectively, and the corresponding battery health state as the dependent variable for polynomial fitting to obtain the corresponding first state prediction model, second state prediction model, and third state prediction model;

[0025] Use the second state prediction model and the third state prediction model as the temperature correction amount and the state-of-charge correction amount respectively to correct the first state prediction model to obtain a polynomial fitting model;

[0026] Adjust the fitting coefficients of the polynomial fitting model iteratively by successively increasing the sequence of intersection values of the real axis of the impedance spectrum, the sequence of amplitudes of the inflection point impedance, the sequence of phase angle values of the inflection point impedance, and the battery health state in the test periods.

[0027] If the correlation coefficient of the fitting curve of the polynomial fitting model after each round of fitting coefficient adjustment iteration update is greater than the preset correlation coefficient threshold, when the preset number of iterations is reached, stop the iteration and use the updated polynomial fitting model as the battery state prediction model;

[0028] If the correlation coefficient of the fitting curve of the polynomial fitting model after the fitting coefficient adjustment iteration update is less than the preset correlation coefficient threshold, stop the iteration, take all the health states before the battery health state value corresponding to the latest added test cycle as the same health state segment, and use the polynomial fitting model corresponding to the previous round of iteration as the piecewise polynomial fitting model of the same health state segment;

[0029] Perform polynomial fitting on the anti-spectrum real axis intersection value sequence, inflection point impedance amplitude sequence, inflection point impedance phase angle value sequence and the corresponding battery health state of the latest added test cycle and each test cycle after the latest added test cycle, construct at least one corresponding piecewise polynomial fitting model, until the preset number of iterations is reached, and summarize all the piecewise polynomial fitting models to obtain the battery state prediction model.

[0030] Further, the piecewise polynomial fitting model is expressed as:

[0031]

[0032] In the formula,

[0033]

[0034]

[0035]

[0036] Among them, represents the predicted value of the battery health state of the i-th health state segment; represents the predicted value of the health state based on the anti-spectrum real axis intersection value within the i-th health state segment; represents the predicted value of the health state based on the inflection point impedance amplitude within the i-th health state segment; represents the predicted value of the health state based on the inflection point impedance phase angle within the i-th health state segment; and are constant coefficients; represents the fitting coefficient in; represents the fitting coefficient in; represents The fitting coefficients in 、 and are positive integer powers representing the i-th health state segment.

[0037] Furthermore, the steps of obtaining the anti-spectrum real-axis intersection value, the inflection point impedance amplitude, and the inflection point impedance phase angle value corresponding to the battery impedance spectrum at each current mutation point include:

[0038] Determine whether there is an impedance point with an impedance phase angle of 0 in the battery impedance spectrum at the current mutation point. If so, obtain the anti-spectrum real-axis intersection value based on the impedance point. Otherwise, obtain the first impedance point and the second impedance point corresponding to the change of the impedance phase angle from a positive value to a negative value in the battery impedance spectrum at the current mutation point;

[0039] Based on the impedance amplitudes and phase angles of the first impedance point and the second impedance point, and according to the principle of similar triangles, obtain the anti-spectrum real-axis intersection value; the anti-spectrum real-axis intersection value is expressed as

[0040]

[0041] where represents the anti-spectrum real-axis intersection value; and respectively represent the impedance amplitude and phase angle of the first impedance point; and respectively represent the impedance amplitude and phase angle of the second impedance point;

[0042] In the order of increasing frequency, obtain the first impedance phase angle mutation point in the battery impedance spectrum at the current mutation point, and use the first impedance phase angle mutation point as the impedance spectrum inflection point;

[0043] Use the impedance amplitude and impedance phase angle value of the impedance spectrum inflection point as the corresponding inflection point impedance amplitude and inflection point impedance phase angle value respectively.

[0044] In a second aspect, an embodiment of the present invention provides a battery state prediction method system, and the system includes:

[0045] A sample acquisition module, configured to randomly select a preset proportion of batteries from the target batch of batteries as a test sample set;

[0046] A data acquisition module, configured to perform a current switching cycle life test on each battery sample in the test sample set to obtain multiple sets of current mutation data corresponding to a preset state of charge segment point sequence; the current mutation data includes current time series data and voltage time series data centered on the current mutation moment, and the battery health state corresponding to the current mutation moment;

[0047] An impedance spectrum acquisition module, configured to obtain a corresponding battery impedance spectrum at current mutation points according to the current time series data and voltage time series data in each group of current mutation data;

[0048] A model construction module, configured to construct a corresponding battery state prediction model according to all the battery impedance spectra at current mutation points and the corresponding battery health states;

[0049] A state prediction module, configured to obtain a battery impedance spectrum at current mutation of a battery to be evaluated in the target batch of batteries, and obtain a corresponding battery state prediction value of the battery to be evaluated according to the battery impedance spectrum at current mutation and the battery state prediction model.

[0050] In a third aspect, an embodiment of the present invention further provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the steps of the above method are implemented.

[0051] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above method are implemented.

[0052] The present invention provides a battery state prediction method, system, computer device, and storage medium. By the method, a preset proportion of batteries are randomly selected from a target batch of batteries as a test sample set, and each battery sample in the test sample set is subjected to a current switching cycle life test. After obtaining multiple groups of current time series data and voltage time series data centered on the current mutation moment corresponding to a preset state of charge segment point sequence, and current mutation data of the battery health state corresponding to the current mutation moment, according to the current time series data and voltage time series data in each group of current mutation data, the corresponding battery impedance spectrum at the current mutation point is obtained, and according to all the battery impedance spectra at the current mutation points and the corresponding battery health states, a corresponding battery state prediction model is constructed, and the battery impedance spectrum at the current mutation of the battery to be evaluated in the target batch of batteries is obtained. According to the battery impedance spectrum at the current mutation and the battery state prediction model, the technical solution of obtaining the battery state prediction value corresponding to the battery to be evaluated is obtained. Compared with the prior art, in this battery state prediction method, by collecting voltage data and current data before and after the current mutation point under different battery health states based on the current switching cycle life test, and then using a data-driven method combined with polynomial fitting to construct a battery state prediction model for analyzing the battery impedance spectrum at the current mutation of the battery to be evaluated in the same batch during actual use to obtain the corresponding battery state prediction value, it can not only avoid relying on the battery equivalent circuit model, but also avoid problems such as high cost, poor accuracy, and insufficient timeliness in the application of the existing data-driven model. It can effectively improve the efficiency, accuracy, and operation convenience of online monitoring of the battery health state at low cost without additional measurement equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 is a schematic flowchart of the battery state prediction method in an embodiment of the present invention;

[0054] Figure 2 is a schematic flowchart of the cycle life test of each battery sample according to a preset current switching charge and discharge rule in an embodiment of the present invention;

[0055] Figure 3 is a schematic diagram of the battery impedance spectrum corresponding to the current mutation point in an embodiment of the present invention;

[0056] Figure 4 is a schematic diagram of the fitting effect of the battery state prediction model in an embodiment of the present invention;

[0057] Figure 5 is a schematic structural diagram of the battery state prediction method system in an embodiment of the present invention;

[0058] Figure 6 is an internal structure diagram of the computer device in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0059] In order to make the objectives, technical solutions, and beneficial effects of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Obviously, the following described embodiments are part of the embodiments of the present invention and are only used to illustrate the present invention, but not to limit the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts fall within the scope of protection of the present invention.

[0060] In one embodiment, as Figure 1 shown, a battery state prediction method is provided, including the following steps:

[0061] S11. Randomly select a preset proportion of batteries from the target batch of batteries as a test sample set; wherein, the target batch of batteries can be understood as any battery cells of the same type with the same production batch as the battery cells to be evaluated; the corresponding preset proportion randomly selected can be set according to actual application requirements and will not be specifically limited here.

[0062] S12. Perform a current switching cycle life test on each battery sample in the test sample set to obtain multiple sets of current mutation data corresponding to a preset state of charge segment point sequence; the current mutation data includes current time series data and voltage time series data centered on the current mutation moment, and the battery health state corresponding to the current mutation moment; wherein, the current switching cycle life test can be understood as a test process in which, during the charge and discharge process of the traditional battery cycle life test, current mutation phenomena are introduced by switching the charging current according to each state of charge point in the preset state of charge segment point sequence, so as to collect the current time series data and voltage time series data centered on the current mutation moment under different battery health states.

[0063] In this embodiment, the current time series data and voltage time series data centered on the current mutation moment can be understood as the current data and voltage data during the constant current charge and discharge process for a certain period of time (such as 1 minute) before and after each current mutation moment. Specifically, the step of performing a current switching cycle life test on each battery sample in the test sample set to obtain multiple sets of current mutation data corresponding to a preset state of charge segment point sequence includes:

[0064] Set the first charge-discharge current, the second charge-discharge current, and the minimum charge-discharge current according to the actual operating condition range of the battery sample; wherein, the actual operating condition range can be understood as the charge-discharge conditions allowed in the normal operating state of the battery sample; in order to introduce current mutation phenomenon during the charging process, in this embodiment, it is preferably set that both the first charge-discharge current I1 and the second charge-discharge current I2 are within the current range between the maximum allowable charge-discharge current and the minimum charge-discharge current, the first charge-discharge current is greater than the second charge-discharge current, and the minimum charge-discharge current is 0.05C, where C is the battery charge-discharge rate.

[0065] Perform a cycle life test on each battery sample according to the preset current switching charge-discharge rule and collect the corresponding current mutation data and battery health state until the total cycle discharge charge amount drops to the preset charge amount; the preset current switching charge-discharge rule is to alternately use the first charge-discharge current and the second charge-discharge current to perform current rate switching charging on each battery sample according to each state of charge in the preset state of charge segment point sequence until the cut-off voltage is reached, and then switch to constant voltage discharge until the current drops to the minimum charge-discharge current and then stand for a preset standstill duration; wherein, the preset charge amount and the preset standstill duration can both be set according to actual application requirements. In this embodiment, it is preferably set that the preset charge amount is 0.7Q nom (Q nom is the initial battery capacity), and the preset standstill duration is set to 3 hours.

[0066] The preset state of charge segment point sequence in this embodiment can be set according to actual application requirements, including multiple different SOC segment points, which can be expressed as SOC = [SOC0, SOC1, SOC2,..., SOCn], and the distance between adjacent state of charge points in this sequence needs to ensure that when the battery is charged and discharged at the first charge-discharge current, the charge-discharge duration is not less than a certain fixed duration t1 (such as 2 minutes), and the starting point of constant voltage charge-discharge in the distance between adjacent state of charge points also needs to meet the interval requirement that the charge-discharge duration is not less than a certain fixed duration t1.

[0067] In actual application, the specific cycle test process of performing a cycle life test on each battery sample according to the preset current switching charge-discharge rule and collecting the corresponding current mutation data and battery health state until the total cycle discharge charge amount drops to the preset charge amount is as Figure 2 shown, including the following steps:

[0068] 1) The cycle starts. Starting from the fully charged state, constant current-constant voltage discharge is carried out at the first charge-discharge current. After the discharge reaches the cut-off voltage, it switches to constant voltage discharge. After the constant voltage discharge until the current decreases to the minimum charge-discharge current (0.05C), it is left standing for a period of time (such as 3 hours). At this time, the SOH of this cycle is considered to be the ratio of the total charge discharged during discharge to the rated charge (SOH = Q / Qnom, which is used to represent the actual capacity attenuation of the battery). Based on this SOH, the charge required for charging corresponding to different SOC segmentation points is calculated.

[0069] 2) After the standing is completed, charging starts. At this time, it is considered that SOC = 0. The second charge-discharge current is used until SOC = SOC0, then it switches to the first charge-discharge current for charging until SOC = SOC1, and then switches back to the second charge-discharge current for charging until SOC = SOC2; and so on. Each time when charging reaches the SOC segmentation point in the preset state of charge segmentation point sequence, the charge-discharge current ratio is switched once until the battery charging cut-off voltage is reached and it switches to constant voltage charging until the current is less than the minimum charge-discharge current. It should be noted that during the cyclic charge-discharge process, each time when charging reaches the SOC segmentation point in the preset state of charge segmentation point sequence and the charge-discharge current ratio is switched once (the first charge-discharge current and the second charge-discharge current are alternately switched at the SOC sequence point), it is necessary to collect the voltage and current during the constant current charge-discharge process for t2 duration (such as 1 minute) before this SOC segmentation point, and the voltage and current during the constant current charge-discharge process for t2 duration (such as 1 minute) after this SOC segmentation point, so as to obtain the current time series data and voltage time series data centered on the current mutation moment in a certain battery health state.

[0070] 3) After the charging current is less than the minimum charge-discharge current, charging stops and it is left standing for the preset standing duration, then it returns to step 1) to start the cycle, and the cyclic charge-discharge is continuously carried out until the total discharge charge of the battery in this cycle is reduced to the preset charge.

[0071] It should be noted that considering that in actual applications, the battery capacity correction is uniformly carried out according to the charging or discharging capacity, in this process, a "constant current-constant voltage" charge-discharge process is adopted without current mutation. For example, when it is selected to switch the charging current during the charging process, the current is not switched during the discharging process, and the total discharge charge during the discharging process is used as the battery capacity, and the charge required for charging corresponding to a specific SOC sequence in the next cycle is calculated; when it is selected to switch the discharging current during the discharging process, the current is not switched during the charging process, and the total charge during the charging process is used as the battery capacity, and the discharge charge required for discharging corresponding to a specific SOC sequence in the next cycle is calculated.

[0072] S13. Based on the current time series data and voltage time series data in each group of current mutation data, obtain the corresponding battery impedance spectrum at the current mutation point; among them, the battery impedance spectrum at the current mutation point can be understood as the battery electrochemical impedance spectrum drawn based on the battery impedance obtained by performing wavelet transforms at different frequencies on the current time series data and voltage time series data in the current mutation data. Considering that the impedance spectrum also includes information on other factors affecting the battery health state in addition to the DC internal resistance, which is more conducive to the identification of the actual battery state, in this embodiment, preferably, by analyzing the current time series data and voltage time series data in each group of current mutation data, the battery impedance spectrum at the current mutation point is obtained for constructing the subsequent state prediction data-driven model. This can not only effectively improve the state prediction accuracy but also transform the problem of identifying the energy storage system state into the problem of accurately obtaining the battery electrochemical impedance spectrum online, which helps to simply and efficiently monitor the battery state.

[0073] Specifically, the steps of obtaining the corresponding battery impedance spectrum at the current mutation point based on the current time series data and voltage time series data in each group of current mutation data include:

[0074] Perform complex Morlet wavelet transforms on the current time series data and the voltage time series data at different frequencies respectively to obtain the corresponding current wavelet coefficients and voltage wavelet coefficients; among them, the complex Morlet wavelet transform is used to analyze the amplitude and phase of the signal. The filter corresponding to the symmetric wavelet basis function has the characteristic of linear phase, which can avoid phase distortion. The specific expression is:

[0075]

[0076] Among them, is the complex Morlet wavelet bandwidth parameter, is the complex Morlet wavelet center frequency, is the sampling frequency, is the number of sampling sequence points, is the total number of sampling points; is the complex Morlet wavelet scale factor. The wavelet scale factor a is determined by the wavelet center frequency and the impedance frequency f to be measured. And the battery impedance frequency , by changing the complex Morlet wavelet scale factor the battery impedance at different frequencies can be measured; is the complex Morlet wavelet displacement factor. The wavelet displacement factor b corresponds to the impedance analysis time. The wavelet bandwidth parameter and the wavelet center frequency are selected independently.

[0077] According to the wavelet transform algorithm, the expression of the voltage wavelet coefficient can be obtained as:

[0078]

[0079] Among them, represents the voltage wavelet coefficient; represents the k-th voltage in the voltage time series data; represents taking the conjugate of the complex Morlet wavelet ;

[0080] According to the wavelet transform algorithm, the expression of the current wavelet coefficient is:

[0081]

[0082] Among them, represents the current wavelet coefficient; represents the k-th current in the current time series data.

[0083] According to the current wavelet coefficients and voltage wavelet coefficients of different frequencies, the corresponding battery impedance at the current mutation point is obtained; among them, the battery impedance at the current mutation point is expressed as:

[0084]

[0085] Among them, represents the battery impedance at the current mutation point.

[0086] According to the battery impedance at the current mutation point of different frequencies, the battery impedance spectrum at the current mutation point is generated; that is, according to the battery impedance at the current mutation point at all frequencies, it is plotted to obtain the battery impedance spectrum (EIS) corresponding to a current mutation point as shown in Figure 3 .

[0087] S14. According to all the battery impedance spectra at the current mutation points and the corresponding battery health states, a corresponding battery state prediction model is constructed; among them, the battery state prediction model is preferably obtained by polynomial fitting; specifically, the steps of constructing the corresponding battery state prediction model according to all the battery impedance spectra at the current mutation points and the corresponding battery health states include:

[0088] Obtain the real-axis intersection value of the impedance spectrum, the inflection point impedance amplitude, and the inflection point impedance phase angle value corresponding to each battery impedance spectrum at the current mutation point within different cycle life test periods; among them, the steps of obtaining the real-axis intersection value of the impedance spectrum, the inflection point impedance amplitude, and the inflection point impedance phase angle value corresponding to each battery impedance spectrum at the current mutation point include:

[0089] Judge whether there is an impedance point with an impedance phase angle of 0 in the battery impedance spectrum at the current mutation point. If so, the real-axis intersection value of the impedance spectrum is obtained according to the impedance point. Otherwise, the first impedance point and the second impedance point corresponding to the impedance phase angle changing from a positive value to a negative value in the battery impedance spectrum at the current mutation point are obtained, and the first impedance point is expressed as , and the second impedance point is expressed as , 、 。

[0090] Based on the impedance magnitudes and phase angles of the first impedance point and the second impedance point, and based on the principle of similar triangles, the anti-spectrum real-axis intersection value is obtained; the process of obtaining the anti-spectrum real-axis intersection value is to consider that the impedance spectrum is densely arranged near the intersection with the horizontal axis and this section of the impedance spectrum can be considered linear, and the real-axis intersection value is directly obtained according to the principle of similar triangles, expressed as:

[0091]

[0092] Wherein, represents the anti-spectrum real-axis intersection value; and respectively represent the impedance magnitude and phase angle of the first impedance point; and respectively represent the impedance magnitude and phase angle of the second impedance point;

[0093] According to the order of increasing frequency, the first impedance phase angle mutation point in the battery impedance spectrum of the current mutation point is obtained, and the first impedance phase angle mutation point is used as the impedance spectrum inflection point;

[0094] The impedance magnitude and impedance phase angle value of the impedance spectrum inflection point are respectively used as the corresponding inflection point impedance magnitude and inflection point impedance phase angle value, and are respectively denoted as and 。

[0095] According to the anti-spectrum real-axis intersection value, the inflection point impedance magnitude, and the inflection point impedance phase angle value, an anti-spectrum real-axis intersection value sequence, an inflection point impedance magnitude sequence, and an inflection point impedance phase angle value sequence corresponding to the preset state of charge segment point sequence in different cycle life test periods are obtained; among them, the anti-spectrum real-axis intersection value sequence can be understood as the time-series data obtained by arranging the anti-spectrum real-axis intersection values introduced by the state of charge segment points of the preset state of charge segment point sequence in a cycle life test period in chronological order; similarly, the inflection point impedance magnitude sequence and the inflection point impedance phase angle value sequence can be respectively understood as the time-series data obtained by arranging the inflection point impedance magnitudes and inflection point impedance phase angle values introduced by the state of charge segment points of the preset state of charge segment point sequence in a cycle life test period in chronological order.

[0096] Using the sequence of the intersection values on the real axis of the anti-spectrum, the sequence of the inflection point impedance amplitudes, and the sequence of the inflection point impedance phase angles as independent variables, and the corresponding battery health state as the dependent variable for polynomial fitting, the battery state prediction model is obtained; among them, polynomial fitting can use the sequence of the intersection values on the real axis of the anti-spectrum, the sequence of the inflection point impedance amplitudes, and the sequence of the inflection point impedance phase angles as independent variables simultaneously for fitting the battery health state. However, considering that in practical applications, the intersection value on the real axis of the anti-spectrum is a direct influencing factor of the battery health state, while the inflection point impedance amplitude is sensitive to the state of charge of the battery and the inflection point impedance phase angle is sensitive to the battery temperature, that is, it can only be understood as an indirect influencing factor. In this embodiment, it is preferably to use the intersection value on the real axis of the anti-spectrum for state prediction, and then use the state prediction values based on the inflection point impedance amplitude and the inflection point impedance phase angle as the correction amounts for the state of charge and the temperature respectively to improve the accuracy of the battery health state prediction.

[0097] Specifically, the steps of using the sequence of the intersection values on the real axis of the anti-spectrum, the sequence of the inflection point impedance amplitudes, and the sequence of the inflection point impedance phase angles as independent variables, and the corresponding battery health state sequence as the dependent variable for polynomial fitting to obtain the battery state prediction model include:

[0098] Using the sequence of the intersection values on the real axis of the anti-spectrum, the sequence of the inflection point impedance amplitudes, and the sequence of the inflection point impedance phase angles of the previous preset number of test cycles as independent variables respectively, and the corresponding battery health state as the dependent variable for polynomial fitting to obtain the corresponding first state prediction model, second state prediction model, and third state prediction model; among them, the previous preset number of test cycles can be understood as the first few test cycles corresponding to the cyclic test of each battery sample. In practical applications, the first cycle can be selected, or the first three cycles can be selected, etc., which can be set according to actual experience. Correspondingly, the first state prediction model, the second state prediction model, and the third state prediction model can be respectively expressed as:

[0099]

[0100]

[0101]

[0102] Among them, represents the predicted health state value based on the intersection value on the real axis of the anti-spectrum corresponding to the previous preset number of test cycles ; represents the predicted health state value based on the inflection point impedance amplitude corresponding to the previous preset number of test cycles; represents the predicted health state value based on the inflection point impedance phase angle in the i-th health state segment; represents the fitting coefficient in; denote the fitting coefficients in denote the fitting coefficients in; n1, m1, and l1 denote positive integer powers.

[0103] The second state prediction model and the third state prediction model are respectively used as the temperature correction amount and the state of charge correction amount to correct the first state prediction model, and a polynomial fitting model is obtained; wherein, the polynomial fitting model is expressed as:

[0104]

[0105] wherein, denote the predicted values of the battery health state corresponding to the previous preset number of test cycles; and are correction coefficients.

[0106] The obtained through the above steps can be understood as a model with a fitting curve correlation coefficient greater than the preset correlation coefficient threshold obtained by fitting based on the anti-spectrum real axis intersection value sequence, the inflection point impedance amplitude sequence, the inflection point impedance phase angle value sequence, and the battery health state corresponding to the previous preset number of test cycles; wherein, the preset correlation coefficient threshold can be determined according to actual application requirements. For example, if it is set to 0.9545, it is considered that the model fitting effect exceeding this threshold is good and can be used. It should be noted that if a model with a fitting curve correlation coefficient greater than the preset correlation coefficient threshold cannot be obtained by fitting based on the data corresponding to the previous preset number of test cycles, the battery sample with the largest data deviation can be selected, and it is considered that there is a large error in the manufacturing of this battery sample. After excluding the relevant measurement data, refitting can be performed again according to the above steps.

[0107] Considering that the fitting relationships actually presented by the data corresponding to different cycle test periods may be very different. For example, the maximum power exponents and correction coefficients corresponding to the first state prediction model, the second state prediction model, and the third state prediction model will be different, and it is necessary to perform fitting correction by combining the anti-spectrum real axis intersection value sequence, the inflection point impedance amplitude sequence, and the inflection point impedance phase angle value sequence corresponding to different test periods (different health states) through the following method.

[0108] The fitting coefficients of the polynomial fitting model are adjusted iteratively by successively adding the anti-spectrum real axis intersection value sequence, the inflection point impedance amplitude sequence, the inflection point impedance phase angle value sequence, and the battery health state of the test period; when adjusting the fitting coefficients for each additional test period data, it is necessary to consider whether the fitting curve correlation coefficient of the refitted model after adding the new data can still maintain a level greater than the preset correlation coefficient threshold, so as to judge whether the current model framework is applicable to the health state prediction under the new test period.

[0109] If the correlation coefficient of the fitting curve of the polynomial fitting model after each round of iterative update of the fitting coefficient adjustment is greater than the preset correlation coefficient threshold, then when the preset number of iterations is reached, the iteration is stopped and the updated polynomial fitting model is used as the battery state prediction model; that is, each time the data obtained by introducing a new round of current switching cycle life test cycle is used to fine-tune the fitting coefficient of the polynomial fitting model fitted based on the previous preset number of test cycles, it is necessary to confirm whether the correlation coefficient of the fitting curve corresponding to the model after the current coefficient update can be guaranteed to be greater than the preset correlation coefficient threshold. If it can be guaranteed, it is considered that the maximum power number and correction coefficient corresponding to the first state prediction model, the second state prediction model and the third state prediction model in the previous polynomial fitting model are still applicable to the data of the newly introduced cycle, and only the fitting coefficient is different; and so on, until the number of fitting coefficient adjustments for the new round of current switching cycle life test cycle introduced reaches the preset number of iterations, then the iteration can be stopped, and it is considered that all health states are applicable to the same polynomial model framework, and the only difference is that the specific fitting coefficient needs to be adjusted.

[0110] If the correlation coefficient of the fitting curve of the polynomial fitting model after iterative update of the fitting coefficient adjustment is less than the preset correlation coefficient threshold, the iteration is stopped, and all health states before the battery health state value corresponding to the latest added test cycle are taken as the same health state segment, and the polynomial fitting model corresponding to the previous round of iteration is taken as the piecewise polynomial fitting model of the same health state segment; that is, when the correlation coefficient of the fitting curve obtained by introducing the data of a new round of current switching cycle life test cycle for fitting coefficient adjustment is less than the preset correlation coefficient threshold, it is considered that the health state corresponding to this test cycle is no longer suitable for using the previously obtained model framework, and it is necessary to re-select the polynomial power and polynomial fitting coefficient starting from this test cycle.

[0111] Perform polynomial fitting on the impedance spectrum real axis intersection value sequence, inflection point impedance amplitude sequence, inflection point impedance phase angle value sequence and the corresponding battery health status of the latest added test cycle and each test cycle after the latest added test cycle, and construct at least one corresponding piecewise polynomial fitting model until the preset number of iterations is reached, and summarize all the piecewise polynomial fitting models to obtain the battery status prediction model; it should be noted that when the battery status is fitted for the latest added test cycle and its subsequent data according to the above method steps, a second or more health status segmentation points may appear, then two or more piecewise polynomial fitting models will be obtained, which will not be repeated here.

[0112] In this embodiment, the piecewise polynomial fitting model in the battery state model when there are health state segmentation points can be expressed as:

[0113]

[0114] In the formula,

[0115]

[0116]

[0117]

[0118] where, represents the predicted value of the battery health state for the i-th health state segment; represents the predicted value of the health state within the i-th health state segment based on the intersection value of the anti-spectrum real axis ; represents the predicted value of the health state within the i-th health state segment based on the inflection point impedance amplitude ; represents the predicted value of the health state within the i-th health state segment based on the inflection point impedance phase angle value ; and are constant coefficients; represents the fitting coefficient in; represents the fitting coefficient in; represents the fitting coefficient in; , and represent the positive integer powers of the i-th health state segment.

[0119] The data-driven model provided in this embodiment is obtained by polynomial fitting. It is not only simple to operate and convenient for online training, but also effectively improves the accuracy of battery health state prediction by introducing the inflection point impedance amplitude and the inflection point impedance phase angle value as the correction amounts of the state of charge and temperature respectively to correct the state prediction value based on the intersection value of the anti-spectrum real axis. It also adaptively corrects the fitting model by fine-tuning the fitting coefficients by successively adding data under different health states, effectively ensuring the reliability of the battery state prediction corresponding to different health state stages.

[0120] S15. Obtain the current mutation battery impedance spectrum of the battery to be evaluated in the target batch of batteries, and obtain the battery state prediction value corresponding to the battery to be evaluated according to the current mutation battery impedance spectrum and the battery state prediction model; where, the battery to be evaluated in the target batch of batteries can be understood as the battery used online in the energy storage system; the acquisition of the current mutation battery impedance spectrum corresponding to the battery to be evaluated can be achieved by the following method:

[0121] The active power required by the system can be increased, the active power required by the system can be reduced, the system can be started from a shutdown state to a constant active power state, or the system can be triggered from a constant active power state to a shutdown state for a current mutation test by means of a power conversion system converter (PCS) connected to the battery system; alternatively, the current mutation test can be triggered by controlling the active or passive equalization circuit inside the battery system from a non-operating state to an operating state or the battery equalization circuit from an operating state to a non-operating state.

[0122] After the current mutation is induced by the converter or the equalization circuit, the corresponding current time-series data and voltage time-series data centered on the current mutation moment are collected, and the corresponding current mutation battery impedance spectrum is obtained according to the aforementioned method for obtaining the battery impedance spectrum at the current mutation point. By substituting the real-axis intersection value, the inflection point impedance amplitude, and the inflection point impedance phase angle value obtained from the current mutation battery impedance spectrum into the corresponding battery state prediction model, the corresponding battery state prediction value can be obtained.

[0123] It should be noted that the method of state identification based on the current mutation point data is particularly suitable for the current energy scheduling form of energy storage systems; the power scheduling instructions of existing energy storage systems are often in minutes, that is, the output power remains unchanged for at least a minute. Since the voltage of lithium batteries is stable in the normal operating range and can be regarded as constant current charge and discharge, the current and voltage time-series data corresponding to the corresponding current mutation can be obtained when the power instruction changes, providing reliable data support for battery state assessment and analysis, and having high practicality.

[0124] In the technical solution provided by the embodiment of the present invention, a preset proportion of batteries are randomly selected from the target batch of batteries as a test sample set, and each battery sample in the test sample set is subjected to a current switching cycle life test. After obtaining multiple groups of current time series data and voltage time series data centered on the current mutation moment, as well as current mutation data of the battery health state corresponding to the preset state of charge segment point sequence, according to the current time series data and voltage time series data in each group of current mutation data, the corresponding battery impedance spectrum at the current mutation point is obtained. Then, according to all the battery impedance spectra at the current mutation points and the corresponding battery health states, the corresponding battery state prediction model is constructed. In addition, the battery impedance spectrum at the current mutation of the battery to be evaluated in the target batch of batteries is obtained, and according to the battery impedance spectrum at the current mutation and the battery state prediction model, the battery state prediction value corresponding to the battery to be evaluated is obtained. This technical solution not only ensures the simplicity and convenience of state prediction through a data-driven method that directly corresponds to the battery health state based on statistical analysis of battery sample test data (without full charge and discharge), but also in online applications, only the original converter or battery equalization circuit in the battery energy storage system can be used to trigger current mutation and collect relevant voltage and current data based on short constant current pulse measurement to obtain the battery impedance spectrum at the current mutation for evaluating the battery state. This can avoid increasing the test cost due to additional test equipment. Moreover, by constructing the battery state prediction model through polynomial fitting combined with adaptive correction, it is efficient, accurate, convenient for online training, and has higher practicality, effectively ensuring the reliability of battery state prediction corresponding to different health state stages.

[0125] In addition, to verify the effectiveness of the battery state prediction method provided by the present invention, a 34Ah ternary lithium-ion battery single cell (the cathode active material is Li(Ni0.5Co0.2Mn0.3)O2, and the anode active material is graphite) is placed in a 45°C constant temperature oven. Using the BTS-Neware battery test system, with charge and discharge rates of 1C and 0.5C, charge and discharge cut-off voltages of 2.75 - 4.2V, and setting the SOC segmentation points to 80% SOC, 60% SOC, 40% SOC, 20% SOC, etc., the constant current charge and discharge cycle life test is carried out on it according to the aforementioned test method of the present invention. And based on the analysis of the collected current mutation data, the following are the first prediction model, the second prediction model, and the third prediction model, and the corresponding fitting effects are as Figure 4 shown. The correlation coefficient of the fitting degree of the polynomial fitting model (R 2 ) is as high as 97%, and the maximum estimation deviation does not exceed 4%.

[0126]

[0127]

[0128]

[0129] In the formula, , and represent the first prediction model, the second prediction model, and the third prediction model.

[0130] It should be noted that although the steps in the above flow chart are sequentially shown according to the indication of the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders.

[0131] In one embodiment, as Figure 5 shown, a battery state prediction method system is provided, and the system includes:

[0132] A sample acquisition module 1, configured to randomly select a preset proportion of batteries from the target batch of batteries as a test sample set;

[0133] A data acquisition module 2, configured to perform a current switching cycle life test on each battery sample in the test sample set to obtain multiple groups of current mutation data corresponding to a preset state of charge segment point sequence; the current mutation data includes current time series data and voltage time series data centered on the current mutation moment, and the battery health state corresponding to the current mutation moment;

[0134] An impedance spectrum acquisition module 3, configured to obtain a corresponding battery impedance spectrum at the current mutation point according to the current time series data and voltage time series data in each group of current mutation data;

[0135] A model construction module 4, configured to construct a corresponding battery state prediction model according to all the battery impedance spectra at the current mutation points and the corresponding battery health states;

[0136] A state prediction module 5, configured to obtain the battery impedance spectrum at the current mutation of the battery to be evaluated in the target batch of batteries, and obtain a corresponding battery state prediction value of the battery to be evaluated according to the battery impedance spectrum at the current mutation and the battery state prediction model.

[0137] For the specific limitations of the battery state prediction method system, reference can be made to the limitations of the battery state prediction method in the above text, and the corresponding technical effects can also be equivalently obtained, which will not be elaborated here. Each module in the above battery state prediction method system can be implemented in whole or in part by software, hardware, and their combination. The above modules can be embedded in the processor of the computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to the above modules.

[0138] Figure 6 The internal structure diagram of a computer device in an embodiment is shown. The computer device may specifically be a terminal or a server. As Figure 6 shown, the computer device includes a processor, a memory, a network interface, a display, a camera, and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. The computer program, when executed by the processor, implements the battery state prediction method. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device may be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the housing of the computer device, or an external keyboard, a touchpad, or a mouse, etc.

[0139] Those of ordinary skill in the art can understand that Figure 6 the structure shown in

[0140] is only a block diagram of some structures related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied. The specific computing device may include more or fewer components than those shown in the figure, or combine some components, or have the same component arrangement.

[0141] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the steps of the above method are implemented.

[0142] In summary, a battery state prediction method, system, computer device, and storage medium provided by embodiments of the present invention not only directly correspond to the data-driven method of battery health state through statistical analysis based on battery sample test data (without full charge and discharge), ensuring the simplicity and convenience of state prediction, but also in online applications, only the original converter or battery equalization circuit in the battery energy storage system is required to trigger current mutation and collect relevant voltage and current data based on short constant current pulse measurement to obtain the current mutation battery impedance spectrum for evaluating the battery state. This can avoid increasing the test cost due to additional test equipment. Additionally, by constructing a battery state prediction model through polynomial fitting combined with adaptive correction, it is efficient, accurate, convenient for online training, and has higher practicality, effectively ensuring the reliability of battery state prediction corresponding to different health state stages.

[0143] Each embodiment in this specification is described in a progressive manner. For parts that are the same or similar in each embodiment, they can be referred to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0144] The above-described embodiments only represent several preferred implementation manners of the present invention. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be pointed out that for those of ordinary skill in the art in this technical field, without departing from the technical principle of the present invention, several improvements and substitutions can be made, and these improvements and substitutions should also be regarded as the protection scope of the present invention. Therefore, the protection scope of the invention patent should be subject to the protection scope of the claims.

Claims

1. A battery status prediction method, characterized in that: The method comprises the following steps: Randomly select a preset proportion of batteries from the target batch of batteries as a test sample set; Performing a current switching cycle life test on each battery sample in the test sample set to obtain multiple sets of current mutation data corresponding to a preset state of charge segment point sequence; the current mutation data includes current time series data and voltage time series data centered on the current mutation moment, and a battery health state corresponding to the current mutation moment; According to the current time series data and voltage time series data in each group of current mutation data, the corresponding battery impedance spectrum of the current mutation point is obtained; According to the battery impedance spectra of all current mutation points and the corresponding battery health status, a corresponding battery status prediction model is constructed; By triggering a current mutation through a power management system converter connected to the battery system or a balancing circuit inside the battery system, a current mutation battery impedance spectrum of a battery to be evaluated in the target batch of batteries is obtained, and a battery state prediction value corresponding to the battery to be evaluated is obtained according to the current mutation battery impedance spectrum and the battery state prediction model; the battery to be evaluated is a battery used online in the energy storage system; The step of constructing a corresponding battery state prediction model according to the battery impedance spectra of all current mutation points and the corresponding battery health status includes: Obtain the impedance spectrum real axis intersection value, inflection point impedance amplitude and inflection point impedance phase angle value corresponding to each current mutation point battery impedance spectrum in different cycle life test cycles; the inflection point impedance amplitude and inflection point impedance phase angle value are respectively the impedance amplitude and impedance phase angle value of the impedance spectrum inflection point in the corresponding current mutation point battery impedance spectrum; According to the real axis intersection value of the impedance spectrum, the inflection point impedance amplitude value and the inflection point impedance phase angle value, a sequence of real axis intersection value of the impedance spectrum, a sequence of inflection point impedance amplitude value and a sequence of inflection point impedance phase angle value corresponding to the preset state of charge segment point sequence in different cycle life test periods are obtained; The impedance spectrum real axis intersection value sequence, the inflection point impedance amplitude sequence and the inflection point impedance phase angle value sequence are used as independent variables, and the corresponding battery health state is used as the dependent variable for polynomial fitting to obtain the battery state prediction model; the battery state prediction model is constructed by fitting based on the principle of introducing the inflection point impedance amplitude and the inflection point impedance phase angle value as the correction amount of charge and temperature respectively to correct the state prediction value based on the impedance spectrum real axis intersection value; the battery state prediction model includes at least one piecewise polynomial fitting model; the piecewise polynomial fitting model is expressed as: in, represents the battery health state prediction value of the i-th health state segment; Represents the intersection value of the real axis based on the anti-spectrum in the i-th health state segment Predicted value of health status; Indicates the impedance amplitude based on the inflection point in the i-th health state segment Predicted value of health status; Indicates the impedance phase angle value based on the inflection point in the i-th health state segment Predicted value of health status; and is a constant coefficient.

2. The battery status prediction method according to claim 1, characterized in that: The step of performing a current switching cycle life test on each battery sample in the test sample set to obtain multiple sets of current mutation data corresponding to a preset state of charge segment point sequence includes: According to the actual operating condition range of the battery sample, a first charge and discharge current, a second charge and discharge current and a minimum charge and discharge current are set; the first charge and discharge current is greater than the second charge and discharge current; According to the preset current switching charge and discharge rule, a cycle life test is performed on each battery sample and corresponding current mutation data and battery health status are collected until the total discharge charge in the cycle drops to a preset charge amount; the preset current switching charge and discharge rule is to use the first charge and discharge current and the second charge and discharge current alternately to perform current rate switching charging on each battery sample according to each charge state in the preset charge state segment point sequence until the cut-off voltage is reached, and then switch to constant voltage discharge until the current drops to the minimum charge and discharge current and then stand for a preset standstill time.

3. The battery status prediction method according to claim 1, characterized in that: The step of obtaining the battery impedance spectrum at the corresponding current mutation point according to the current time series data and the voltage time series data in each group of current mutation data comprises: Performing complex Morlet wavelet transforms at different frequencies on the current time series data and the voltage time series data respectively to obtain corresponding current wavelet coefficients and voltage wavelet coefficients; According to the current wavelet coefficient and voltage wavelet coefficient of different frequencies, the corresponding battery impedance of the current mutation point is obtained; The current mutation point battery impedance spectrum is generated according to the current mutation point battery impedance at different frequencies.

4. The battery status prediction method according to claim 1, characterized in that: The step of performing polynomial fitting using the impedance spectrum real axis intersection value sequence, the inflection point impedance amplitude sequence and the inflection point impedance phase angle value sequence as independent variables and the corresponding battery health state sequence as the dependent variable to obtain the battery state prediction model comprises: The impedance spectrum real axis intersection value sequence, the inflection point impedance amplitude sequence and the inflection point impedance phase angle value sequence of the previously preset test cycle number are used as independent variables, and polynomial fitting is performed with the corresponding battery health state as the dependent variable to obtain the corresponding first state prediction model, second state prediction model and third state prediction model; Using the second state prediction model and the third state prediction model as a temperature correction amount and a state of charge correction amount respectively to correct the first state prediction model, so as to obtain a polynomial fitting model; Iteratively adjusting the fitting coefficients of the polynomial fitting model by sequentially increasing the impedance spectrum real axis intersection value sequence, the inflection point impedance amplitude sequence, the inflection point impedance phase angle value sequence and the battery health state of the test cycle; If the correlation coefficient of the fitting curve of the polynomial fitting model after each round of fitting coefficient adjustment iteration is greater than the preset correlation coefficient threshold, then when the preset number of iterations is reached, the iteration is stopped and the updated polynomial fitting model is used as the battery state prediction model; If the fitting curve correlation coefficient of the polynomial fitting model after iterative update of the fitting coefficient adjustment is less than the preset correlation coefficient threshold, the iteration is stopped, and all health states before the battery health state value corresponding to the latest added test cycle are taken as the same health state segment, and the polynomial fitting model corresponding to the previous round of iteration is taken as the piecewise polynomial fitting model of the same health state segment; The impedance spectrum real axis intersection value sequence, the inflection point impedance amplitude value sequence, the inflection point impedance phase angle value sequence and the corresponding battery health status of the latest added test cycle and each test cycle after the latest added test cycle are polynomially fitted to construct at least one corresponding piecewise polynomial fitting model until the preset number of iterations is reached, and all piecewise polynomial fitting models are aggregated to obtain the battery state prediction model.

5. The battery status prediction method according to claim 1, characterized in that: The health status prediction values ​​based on the real axis intersection value of the impedance spectrum, the inflection point impedance amplitude and the inflection point impedance phase angle value in the health status segment in the piecewise polynomial fitting model are respectively expressed as: in, Represents the intersection value of the real axis based on the anti-spectrum in the i-th health state segment Predicted value of health status; Indicates the impedance amplitude based on the inflection point in the i-th health state segment Predicted value of health status; Indicates the impedance phase angle value based on the inflection point in the i-th health state segment Predicted value of health status; express The fitting coefficients in ; express The fitting coefficients in ; express The fitting coefficients in ; , and The positive integer power representing the i-th health status segment.

6. The battery status prediction method according to claim 1, characterized in that: The steps of obtaining the impedance spectrum real axis intersection value, the inflection point impedance amplitude value and the inflection point impedance phase angle value corresponding to the battery impedance spectrum at each current mutation point include: Determine whether there is an impedance point with an impedance phase angle of 0 in the battery impedance spectrum at the current mutation point, and if so, obtain the real axis intersection value of the impedance spectrum according to the impedance point, and vice versa, obtain the first impedance point and the second impedance point corresponding to the impedance phase angle changing from a positive value to a negative value in the battery impedance spectrum at the current mutation point; According to the impedance amplitude and phase angle of the first impedance point and the second impedance point, based on the principle of similar triangle, the real axis intersection value of the impedance spectrum is obtained; the real axis intersection value of the impedance spectrum is expressed as in, Represents the intersection value of the real axis of the anti-spectrum; and Respectively represent the impedance amplitude and phase angle of the first impedance point; and represent the impedance amplitude and phase angle of the second impedance point respectively; Obtaining, in order from low to high frequency, a first impedance phase angle mutation point in the current mutation point battery impedance spectrum, and taking the first impedance phase angle mutation point as an impedance spectrum inflection point; The impedance amplitude and the impedance phase angle value of the inflection point of the impedance spectrum are respectively used as the corresponding inflection point impedance amplitude and inflection point impedance phase angle value.

7. A battery status prediction system, characterized in that: Applying the battery state prediction method according to claim 1, the system comprises: A sample acquisition module, used to randomly select a preset proportion of batteries from a target batch of batteries as a test sample set; A data acquisition module, used to perform a current switching cycle life test on each battery sample in the test sample set, and obtain multiple sets of current mutation data corresponding to a preset state of charge segment point sequence; the current mutation data includes current time series data and voltage time series data centered on the current mutation moment, and a battery health state corresponding to the current mutation moment; An impedance spectrum acquisition module is used to obtain a battery impedance spectrum at a corresponding current mutation point according to the current time series data and voltage time series data in each group of current mutation data; A model building module is used to build a corresponding battery state prediction model based on the battery impedance spectra of all current mutation points and the corresponding battery health status; The state prediction module is used to obtain the current mutation battery impedance spectrum of the battery to be evaluated in the target batch of batteries by triggering the current mutation through the power management system converter connected to the battery system or the balancing circuit inside the battery system, and obtain the battery state prediction value corresponding to the battery to be evaluated according to the current mutation battery impedance spectrum and the battery state prediction model; the battery to be evaluated is a battery used online in the energy storage system.

8. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images