Lithium battery broadband impedance spectroscopy test method based on maximum length binary sequence

Abstract

The present invention discloses a lithium battery broadband impedance spectrum test method based on a maximum length binary sequence, comprising the following steps: designing a maximum length binary sequence for broadband impedance measurement; injecting the maximum length binary sequence into the lithium battery through a battery management system, collecting the lithium battery terminal voltage and current during the injection process, and calculating the measured impedance of the lithium battery at different frequencies; establishing a three-dimensional data cloud in the frequency domain using the power spectrum density characteristics of the excitation signal and the real and imaginary parts of the measured impedance; filtering the data points in the three-dimensional data cloud by establishing a statistically based data cleaning automatic selection mechanism to obtain a lithium battery broadband impedance spectrum curve. The present invention injects a maximum length binary sequence into the lithium battery, and uses a data cleaning automatic selection mechanism to filter the three-dimensional data cloud to reduce noise interference, thereby improving the accuracy and stability of broadband impedance measurement of power lithium batteries.

Description

Technical Field

[0001] The present invention belongs to the technical field of power lithium battery applications, and in particular, relates to a lithium battery broadband impedance spectrum testing method based on a maximum length binary sequence. Background Art

[0002] In the field of electric vehicles, power batteries are mainly responsible for providing power for the entire vehicle. With the continuous innovation of power battery technology, the types of power batteries are becoming more and more diverse. Among them, lithium-ion batteries are widely used due to their excellent performance. Lithium batteries use the electrochemical potential of their internal two poles to convert chemical energy into electrical energy. Due to the existence of aging mechanisms, the performance of lithium batteries will gradually decline after recycling, which is prominently manifested in the decrease of capacity and increase of resistance. In order to ensure that lithium batteries can maintain a good operating state, they need to be diagnosed and predicted. Electrochemical impedance spectroscopy technology is a safe disturbance technology used to detect the internal changes of electrochemical systems. This technology measures the impedance of the battery within a certain frequency range and can provide rich electrode kinetic information. It has great application potential in lithium battery SOX estimation, degradation pattern recognition, internal temperature estimation and safety detection.

[0003] The measurement methods of electrochemical impedance spectroscopy mainly include frequency domain measurement methods and time domain measurement methods. The frequency domain measurement method is easy to implement and apply. It is characterized in that the frequency range of the excitation signal is selected and different frequency points are determined. The excitation signals of different frequency points are used in turn for sweep frequency measurement. The amplitude and phase of the same frequency excitation and response signals are analyzed to obtain the frequency response characteristics of the system, and finally the electrochemical impedance spectrum is obtained. This method is a single-frequency excitation sweep frequency measurement. The measured data is accurate, but the test time is long. The impedance spectrum test time can be greatly shortened by superimposing sinusoidal signals, but the wave of the sinusoidal superposition signal requires complex hardware design, which greatly limits the practical use of this method. It is relatively easy to implement the impedance spectrum using a square wave signal, but the frequency domain information contained in the square wave signal is extremely insufficient, so it is difficult to obtain complete impedance spectrum information. In contrast, the pseudo-random binary sequence has a simple structure, low complexity, short measurement time, and high accuracy. The autocorrelation characteristics are similar to white noise. The spectrum is distributed in a wide frequency range. Only one injection is required to obtain the impedance in a wide frequency range. The time required for the test is greatly reduced and the results are accurate.

[0004] Pseudo-random sequences can inject broadband signals into batteries in a very short time, but the power content of pseudo-random sequence frequency domain signals is low, and they are easily interfered by measurement noise when measuring impedance. Impedance filters can filter out the interference of noise on signals to a certain extent, and can be used for smooth measurement and improving signal-to-noise ratio. Moving average filters are a good choice, mainly used to suppress the interference of noise on signals, but the ability of moving average linear filters is limited, and the filtering effect in the frequency domain is relatively poor. As the signal power and signal-to-noise ratio decrease, the impedance filtering results are very prone to deviations. At present, there is still the problem that multi-frequency synthetic signals cannot receive high signal-to-noise ratio responses in all wide frequency bands. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides a lithium battery broadband impedance spectrum testing method based on a maximum length binary sequence. By injecting a maximum length binary sequence into the lithium battery, the three-dimensional data cloud is filtered using a data cleaning automatic selection mechanism to reduce noise interference, thereby improving the accuracy and stability of broadband impedance measurement of power lithium batteries.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: a lithium battery broadband impedance spectrum testing method based on a maximum length binary sequence, specifically comprising the following steps:

[0007] Step 1, design a maximum length binary sequence for broadband impedance measurement;

[0008] Step 2: inject a maximum length binary sequence into the lithium battery through the battery management system, collect the voltage and current of the lithium battery terminal during the injection process, and calculate the measured impedance of the lithium battery at different frequencies;

[0009] Step 3: Create a three-dimensional data cloud in the frequency domain using the power spectrum density characteristics of the excitation signal and the real and imaginary parts of the measured impedance;

[0010] Step 4: Filter the data points in the three-dimensional data cloud by establishing a statistical data cleaning automatic selection mechanism to obtain a broadband impedance spectrum curve of the lithium battery.

[0011] Furthermore, the maximum length binary sequence is generated by a multi-stage linear feedback shift register, and the sequence is composed of a length N L and resolution f r Determine, where the length of the maximum length binary sequence is N L For: N L =2 w -1, w is the order of the shift register, resolution f r Satisfy: f r =f c / N L , fc is the clock frequency of the maximum length binary sequence.

[0012] Furthermore, the calculation process of the measured impedance is:

[0013]

[0014] in, is the impedance at the kth frequency, V(k) is the Fourier change of the lithium battery terminal voltage v(t), and I(k) is the Fourier change of the lithium battery terminal current i(t): n is the number of transformation points, and j represents the imaginary number on the complex plane.

[0015] Furthermore, the kth frequency is kf k , f k =f s / N s , where f s is the sampling frequency, N s is the number of sampled data.

[0016] Furthermore, the power spectral density P MLBS_k (f n ) is calculated as follows:

[0017]

[0018] Among them, I MLBS (f n ) is the maximum length binary sequence after Fourier transformation at frequency f n The amplitude at a is the weighting coefficient, N L is the length of the maximum length binary sequence, n is the number of transformation points, P MLBS (f 1 ) is the maximum length binary sequence after Fourier transformation at frequency f 1 The amplitude at .

[0019] Furthermore, step 4 includes the following sub-steps:

[0020] Step 401: Divide all three-dimensional data clouds into three groups according to the high, medium and low frequency ranges. Each group of data points forms a domain. Traverse all data points in each domain and calculate the average Euclidean distance between the remaining data points in each domain and the data point.

[0021] Step 402: Calculate the average Euclidean distance in each domain in step 401 The mean and variance Δ[d k ], according to the normal distribution probability density curve, using the mean and variance to form the formula Calculate the threshold, filter out the data points whose average Euclidean distance in each field is greater than the threshold, and obtain the broadband impedance spectrum curve of the lithium battery, where N g is the number of groups of the 3D data cloud.

[0022] Furthermore, the rule for dividing the three-dimensional data cloud into three groups in step S401 is: the low frequency range is [0.21 Hz, 10 Hz], the medium frequency range is [10 Hz, 1 kHz], and the high frequency range is [1 kHz, 3.5 kHz].

[0023] Furthermore, the average Euclidean distance between the remaining data points in each field and the data point The calculation process is:

[0024]

[0025] Among them, m is the number of points in the field, u is the index of m, (x k ,y k , z k ) is a data point in the field, x k represents the real part of the measured impedance at the kth frequency, y k represents the imaginary part of the measured impedance at the kth frequency, z k represents the power spectrum density of the measured impedance at the kth frequency; (x u ,y u , z u ) are the remaining data points in the field, x u represents the real part of the measured impedance at the uth frequency, y u represents the imaginary part of the measured impedance at the uth frequency, z u represents the power spectral density of the measured impedance at the uth frequency.

[0026] Furthermore, the average Euclidean distance in each field The mean The calculation process is: The average Euclidean distance within each field The variance Δ[d k ]The calculation process is: Among them, m is the number of points in the field, and k is the index of m.

[0027] Compared with the prior art, the present invention has the following beneficial effects: the present invention is based on the lithium battery broadband impedance spectrum test method of the maximum length binary sequence, by injecting the maximum length binary sequence, using the three-dimensional cloud data cleaning method to filter the measurement results, so as to achieve rapid measurement of the broadband impedance of the lithium battery. The power spectrum density reflects the power of the signal at different frequencies. A large power spectrum signal indicates that more power is injected into the battery, which can resist noise interference, while the signal with lower power spectrum density contains less information and has a lower possibility of becoming an effective value. By combining the power spectrum density with the battery impedance and removing the impedance with lower power spectrum density, an accurate and effective battery impedance spectrum curve can be obtained. In addition, the maximum length binary sequence proposed by the present invention is a deterministic periodic signal, which can weaken noise interference by increasing the number of cycles. Compared with the pulse signal, the maximum length sequence has a lower excitation amplitude intensity, and the lithium battery is more sensitive to the disturbance amplitude. The smaller the excitation signal amplitude, the higher the test accuracy. The application of the maximum length binary sequence can better play its characteristics and advantages. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the specific embodiments, but do not constitute a limitation of the present invention.

[0029] Figure 1 It is a flow chart of a lithium battery broadband impedance spectrum testing method based on a maximum length binary sequence of the present invention;

[0030] Figure 2 is a Nyquist plot of the original impedance spectrum of a lithium battery injected with a binary sequence of maximum length in an embodiment;

[0031] Figure 3 is a power spectral density diagram of a lithium battery injected with a binary sequence of maximum length in an embodiment;

[0032] Figure 4 is a three-dimensional data cloud diagram of the impedance of the lithium battery in the embodiment;

[0033] Figure 5 It is a statistical data cleaning automatic selection mechanism based on N g =1, schematic diagram of a three-dimensional data cloud cleaning;

[0034] Figure 6 It is a statistical data cleaning automatic selection mechanism based on N g =3 schematic diagram of secondary three-dimensional data cloud cleaning;

[0035] Figure 7 It is a comparison chart of broadband impedance test of the embodiment and the comparative example at an ambient temperature of 25° C. and a lithium battery 50% charged. DETAILED DESCRIPTION

[0036] The technical solution of the present invention is described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific implementation methods referred to herein are only used to illustrate and explain the technical solution of the present invention and do not constitute a limitation of the present invention.

[0037] like Figure 1 The flowchart of the lithium battery broadband impedance spectrum testing method based on the maximum length binary sequence of the present invention is as follows:

[0038] Step 1. Design a maximum length binary sequence for broadband impedance measurement. The maximum length binary sequence is a deterministic periodic signal. The noise interference can be weakened by increasing the number of cycles. Compared with pulse signals, the maximum length sequence has a lower excitation amplitude. Lithium batteries are more sensitive to disturbance amplitude. The smaller the excitation signal amplitude, the higher the test accuracy.

[0039] The maximum length binary sequence in the present invention is generated by a multi-stage linear feedback shift register, and the sequence is composed of a length N L and resolution f r Determine, where the length of the maximum length binary sequence is N L For: N L =2 w -1, w is the order of the shift register, resolution f r Satisfy: f r =f c / N L , f c is the clock frequency of the maximum length binary sequence.

[0040] Step 2: inject a maximum length binary sequence into the lithium battery through the battery management system, collect the voltage and current of the lithium battery terminal during the injection process, and calculate the measured impedance of the lithium battery at different frequencies;

[0041] The calculation process of measuring impedance in the present invention is:

[0042]

[0043] in, is the impedance at the kth frequency, V(k) is the Fourier change of the lithium battery terminal voltage v(t), and I(k) is the Fourier change of the lithium battery terminal current i(t): n is the number of transformation points, j represents an imaginary number on the complex plane; the kth frequency in the present invention is kf k , f k =f s / N s , where f s is the sampling frequency, Ns is the number of sampled data.

[0044] Step 3: Use the power spectrum density characteristics of the excitation signal and the real and imaginary parts of the measured impedance in the frequency domain to establish a three-dimensional data cloud; the power spectrum density reflects the power of the signal at different frequencies. A large power spectrum signal indicates that more power is injected into the battery, which can resist noise interference, while a signal with a lower power spectrum density contains less information and is less likely to become an effective value. By combining the power spectrum density with the battery impedance and removing the impedance with a lower power spectrum density, an accurate and effective battery impedance spectrum curve can be obtained.

[0045] In the present invention, the power spectrum density P MLBS_k (f n ) is calculated as follows:

[0046]

[0047] Among them, I MLBS (f n ) is the maximum length binary sequence after Fourier transformation at frequency f n The amplitude at a is the weighting coefficient, N L is the length of the maximum length binary sequence, n is the number of transformation points, P MLBS (f 1 ) is the maximum length binary sequence after Fourier transformation at frequency f 1 The amplitude at .

[0048] Step 4: Filter the data points in the three-dimensional data cloud by establishing a statistical data cleaning automatic selection mechanism to obtain a broadband impedance spectrum curve of the lithium battery, which can reduce noise interference and improve the accuracy and stability of broadband impedance measurement of power lithium batteries; specifically, the following sub-steps are included:

[0049] Step 401: Divide all three-dimensional data clouds into three groups according to the high, medium and low frequency ranges. Each group of data points forms a domain. Traverse all data points in each domain and calculate the average Euclidean distance between the remaining data points in each domain and the data point. Among them, m is the number of points in the field, u is the index of m, (x k ,y k , z k ) is a data point in the field, x k represents the real part of the measured impedance at the kth frequency, y k represents the imaginary part of the measured impedance at the kth frequency, z k represents the power spectrum density of the measured impedance at the kth frequency; (x u ,y u , z u) are the remaining data points in the field, x u represents the real part of the measured impedance at the uth frequency, y u represents the imaginary part of the measured impedance at the uth frequency, z u represents the power spectral density of the measured impedance at the uth frequency. The rule for dividing the three-dimensional data cloud into three groups in the present invention is: the low frequency range is [0.21Hz, 10Hz], the medium frequency range is [10Hz, 1kHz], and the high frequency range is [1kHz, 3.5kHz];

[0050] Step 402: Calculate the average Euclidean distance in each domain in step 401 The mean and variance Δ[d k ], according to the normal distribution probability density curve, using the mean and variance to form the formula Calculate the threshold, filter out the data points whose average Euclidean distance in each field is greater than the threshold, and obtain the broadband impedance spectrum curve of the lithium battery, where N g is the number of groups of the three-dimensional data cloud; the average Euclidean distance in each field in the present invention The mean The calculation process is: The average Euclidean distance within each field The variance Δ[d k ]The calculation process is: Among them, m is the number of points in the field, and k is the index of m.

[0051] Example

[0052] In this embodiment, taking an 18650NMC-based lithium battery with a rated capacity of 3Ah, a voltage range of 2.0V-4.25V, and a signal sampling frequency of 70kHz as an example, a wide-band impedance spectrum curve is obtained by the lithium battery wide-band impedance spectrum testing method based on a maximum length binary sequence of the present invention.

[0053] like Figure 2 is the Nyquist diagram of the original impedance spectrum of the lithium battery injected with the maximum length binary sequence in this embodiment, Figure 2 As shown in the figure, the high-frequency region of the lithium battery impedance spectrum is contaminated by noise, and the shape of the normal impedance is only visible in the low-frequency range, making it difficult to directly obtain an effective impedance spectrum. Figure 3 is the power spectrum density diagram of the lithium battery injected with the maximum length binary sequence in this embodiment, Figure 3 As shown in the figure, the power spectrum density of the maximum length binary sequence gradually decays with the increase of frequency, and the power spectrum density fluctuates between similar frequencies. A large power spectrum density signal indicates that more power is injected into the battery. A three-dimensional data cloud is established using the power spectrum density information and the real and imaginary parts of the measured impedance for further data cleaning. Figure 4 3D data cloud diagram of the lithium battery impedance in this embodiment. The sparseness of the three-dimensional cloud diagram shows the power spectrum intensity of each measured impedance. The power spectrum density of the data points in the three-dimensional data cloud is used to process the impedance calculation results, such as Figure 4 As shown in the figure, the bright spots in the grayscale image have higher power than the dark spots. Most of the dark spots deviate from the main data set and are more likely to be outliers in the electrochemical impedance spectroscopy measurement. The bright spots with high power spectrum intensity are concentrated in the middle of the data set, which is consistent with the basic shape of the impedance spectrum. Figure 5 It is a statistical data cleaning automatic selection mechanism based on N g =1, a schematic diagram of a 3D cloud data cleaning, Figure 5 It can be seen that the result of cleaning all the data points in the 3D data cloud uniformly is not ideal. The reference value is a smooth curve. In contrast, the curve after cleaning the 3D cloud data still reflects a lot of useless impedance information. In order to clean the data more efficiently, the data points in the 3D data cloud are divided into three groups according to the frequency range characteristics, such as Figure 6 It is a statistical data cleaning automatic selection mechanism based on N g =3 secondary 3D cloud data cleaning diagram, by Figure 6 As shown, after all data points in the three-dimensional data cloud are divided into three groups for cleaning, only 102 useful points are left, indicating that the noise data and low power spectral density measurements are effectively removed after the three-dimensional cloud cleaning.

[0054] Comparative Example

[0055] In this embodiment, a 18650NMC-based lithium battery with a rated capacity of 3Ah, a voltage range of 2.0V-4.25V, and a signal sampling frequency of 70kHz is taken as an example.

[0056] Step 1: Design a maximum length binary sequence for broadband impedance measurement;

[0057] Step 2: Inject the maximum length binary sequence designed in step 1 into the 18650NMC-based lithium battery through the battery management system. During the signal injection process, collect the terminal voltage and current at both ends of the battery, and calculate the measured impedance of the lithium battery at different frequencies;

[0058] Step 3: Filter the measured impedance in the frequency domain using a sliding average window filtering method to obtain a broadband impedance spectrum curve.

[0059] like Figure 7 It is a comparison chart of broadband impedance test of the embodiment and the comparative example at an ambient temperature of 25°C and a battery state of 50% charge. Compared with the sliding average window filtering method, the filtering method proposed by the present invention is obviously closer to the reference value. Although the method of the present invention has some deviation in the low-frequency area, Figure 7 No outliers were found in the impedance spectrum shown. In contrast, the sliding average window filtering method had larger errors and multiple outliers in the mid- and high-frequency ranges.

[0060] The above are only preferred embodiments of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should be regarded as the protection scope of the present invention.

Claims

1. A lithium battery broadband impedance spectrum testing method based on maximum length binary sequence, characterized in that: The specific steps include: Step 1, design a maximum length binary sequence for broadband impedance measurement; Step 2: inject a maximum length binary sequence into the lithium battery through the battery management system, collect the voltage and current of the lithium battery terminal during the injection process, and calculate the measured impedance of the lithium battery at different frequencies; Step 3: Create a three-dimensional data cloud in the frequency domain using the power spectrum density characteristics of the excitation signal and the real and imaginary parts of the measured impedance; Step 4: Filter the data points in the three-dimensional data cloud by establishing a statistical data cleaning automatic selection mechanism to obtain a broadband impedance spectrum curve of the lithium battery.

2. A lithium battery broadband impedance spectrum testing method based on maximum length binary sequence according to claim 1, characterized in that: The maximum length binary sequence is generated by a multi-stage linear feedback shift register, and the sequence consists of a length N L and resolution f r Determine, where the length of the maximum length binary sequence is N L For: N L =2 w -1, w is the order of the shift register, resolution f r Satisfy: f r =f c / N L , f c is the clock frequency of the maximum length binary sequence.

3. The method for testing a lithium battery broadband impedance spectrum based on a maximum length binary sequence according to claim 1, characterized in that: The calculation process of the measured impedance is: in, is the impedance at the kth frequency, V(k) is the Fourier change of the lithium battery terminal voltage v(t), and I(k) is the Fourier change of the lithium battery terminal current i(t): n is the number of transformation points, and j represents the imaginary number on the complex plane.

4. A lithium battery broadband impedance spectrum testing method based on maximum length binary sequence according to claim 3, characterized in that: The kth frequency is kf k , f k =f s / N s , where f s is the sampling frequency, N s is the number of sampled data.

5. The method for testing a lithium battery broadband impedance spectrum based on a maximum length binary sequence according to claim 1, characterized in that: The power spectral density P MLBS_k (f n ) is calculated as follows: Among them, I MLBS (f n ) is the maximum length binary sequence after Fourier transformation at frequency f n The amplitude at a is the weighting coefficient, N L is the length of the maximum length binary sequence, n is the number of transformation points, P MLBS (f1) is the amplitude of the maximum length binary sequence at frequency f1 after Fourier transform.

6. The method for testing a lithium battery broadband impedance spectrum based on a maximum length binary sequence according to claim 1, characterized in that: Step 4 includes the following sub-steps: Step 401: Divide all three-dimensional data clouds into three groups according to the high, medium and low frequency ranges. Each group of data points forms a domain. Traverse all data points in each domain and calculate the average Euclidean distance between the remaining data points in each domain and the data point. Step 402: Calculate the average Euclidean distance in each domain in step 401 The mean and variance Δ[d k ], according to the normal distribution probability density curve, using the mean and variance to form the formula Calculate the threshold, filter out the data points whose average Euclidean distance in each field is greater than the threshold, and obtain the broadband impedance spectrum curve of the lithium battery, where N g is the number of groups of the 3D data cloud.

7. A lithium battery broadband impedance spectrum testing method based on maximum length binary sequence according to claim 6, characterized in that: The rule for dividing the three-dimensional data cloud into three groups in step S401 is: the low frequency range is [0.21 Hz, 10 Hz], the medium frequency range is [10 Hz, 1 kHz], and the high frequency range is [1 kHz, 3.5 kHz].

8. The method for testing a lithium battery broadband impedance spectrum based on a maximum length binary sequence according to claim 6, characterized in that: The average Euclidean distance between the remaining data points in each field and the data point The calculation process is: Among them, m is the number of points in the field, u is the index of m, (x k ,y k , z k ) is a data point in the field, x k represents the real part of the measured impedance at the kth frequency, y k represents the imaginary part of the measured impedance at the kth frequency, z k represents the power spectrum density of the measured impedance at the kth frequency; (x u ,y u , z u ) are the remaining data points in the field, x u represents the real part of the measured impedance at the uth frequency, y u represents the imaginary part of the measured impedance at the uth frequency, z u represents the power spectral density of the measured impedance at the uth frequency.

9. The method for testing a lithium battery broadband impedance spectrum based on a maximum length binary sequence according to claim 6, characterized in that: The average Euclidean distance within each field The mean The calculation process is: The average Euclidean distance within each field The variance Δ[d k ]The calculation process is: Among them, m is the number of points in the field, and k is the index of m.

Images