KR20240061489A - Method of predicting trouble or failure of lithium ion battery contion using electrochemical impedance spectroscopy - Google Patents

Abstract

The present invention is a method of diagnosing failure of a lithium-ion battery in advance under high temperature conditions using electrochemical impedance spectroscopy (EIS). The present invention uses electrochemical impedance spectroscopy using alternating current to detect the lithium-ion battery. extracting the impedance for each cell, and measuring the impedance time series at regular time intervals; Calculating the maximum deviation of the impedance of each cell at each measurement point in time series through a battery management system; and calculating a failure diagnosis preliminary prediction value through the battery management system, and pre-diagnosing a failure of the lithium-ion battery according to the failure diagnosis advance prediction value. The fault diagnosis preliminary prediction value is calculated as SampEn using the following formula, , where, B ^m (r) is the impedance difference or distance between the i and jth battery cells, which are different battery cells, among the plurality of battery cells in the battery, than r, which is an indicator of data volatility. It represents a small number, and A ^m (r) is a different battery among a plurality of battery cells in the battery pack at the next measurement point in time series of the impedance measurement value of the battery cell targeted by B ^m (r). The impedance difference or distance between the i and jth battery cells represents the number smaller than r, which is an indicator of data volatility, It represents the length, and m is characterized as 2.

Description

Method for diagnosing lithium-ion battery failure in advance using electrochemical impedance spectroscopy {METHOD OF PREDICTING TROUBLE OR FAILURE OF LITHIUM ION BATTERY CONTION USING ELECTROCHEMICAL IMPEDANCE SPECTROSCOPY}

The present invention relates to a method for diagnosing a failure of a lithium-ion battery in advance, and more specifically, a method of diagnosing a failure of a lithium-ion battery in advance using electrochemical impedance spectroscopy, especially under high temperature conditions where fires occur frequently. It's about.

As lithium-ion batteries are applied to various applications (EV, ESS, electric railway, unmanned aerial vehicle, UPS, etc.), they are exposed to various environmental conditions. The performance of lithium-ion batteries shows different changes in electrical characteristics depending on temperature conditions. Typically, changes in battery capacity and internal resistance are different. For reference, Figure 1 shows changes in battery capacity and internal resistance, that is, changes in electrical characteristics, depending on the temperature of a lithium-ion battery.

Through the media, cases of thermal runaway occurring due to rapid changes in characteristics of lithium-ion batteries exposed to high temperatures are frequently reported. The thermal runaway process of a lithium-ion battery is a side reaction inside the battery (SEI layer decomposition (>69 degrees Celsius), electrolyte decomposition (>200 degrees Celsius) depending on cell and pack design requirements, environmental conditions due to temperature/humidity, and load conditions. ), etc.), and the decomposition of the SEI layer generates heat and flammable gas through the reaction between the cathode and the electrolyte. An increase in internal pressure due to gas generation causes cell swelling and electrolyte leakage, and a decrease in lithium ion active material causes a decrease in capacity and an increase in internal resistance.

To solve this problem, a representative technology is to monitor material characteristic information of the battery.

As a related conventional technology, Korean Patent No. 10-2449836 (registered on September 27, 2022, title of invention: detection of thermal runaway phenomenon of lithium-ion battery to improve ESS stability and extinguishing system using LN2 (Lithium-ion battery thermal runaway) detection and fire extinguishing system using LN2 for ESS improved reliability)), etc.

In the previously registered patent, "The present invention relates to a detection of thermal runaway precursor phenomenon of lithium-ion batteries to improve ESS stability and a fire extinguishing system using LN2. More specifically, the thermal runaway precursor of ESS facilities composed of lithium-ion batteries A safety-based fire extinguishing system that can detect the phenomenon in advance and rapidly cool the temperature of the lithium-ion battery and the surrounding temperature using LN2 to minimize damage from explosions and fires was disclosed.

However, the above-described conventional patent document simply states that overheating can be detected by measuring the temperature of each unit ESS module consisting of a plurality of units through a temperature sensor, and through prior prediction, failure of the lithium-ion battery in high temperature conditions can be prevented in advance. There is no solution provided regarding how to diagnose.

Republic of Korea Patent No. 10-2449836 (registered on September 27, 2022, title of invention: Lithium-ion battery thermal runaway detection and fire extinguishing system using LN2 and detection of thermal runaway phenomenon of lithium-ion battery to improve ESS stability LN2 for ESS improved reliability))

The present invention was created to solve the above-mentioned problems, and the purpose of the present invention is to provide a method for diagnosing failure of a lithium-ion battery in advance using electrochemical impedance spectroscopy, especially under high temperature conditions where fires occur frequently. The purpose.

A method of diagnosing a failure of a lithium-ion battery in advance under high temperature conditions using electrochemical impedance spectroscopy (EIS) according to a preferred embodiment of the present invention to achieve the above object is an electric current using alternating current. Extracting the impedance for each cell of the lithium-ion battery using chemical impedance spectroscopy, and measuring the impedance in time series at regular time intervals; Calculating the maximum deviation of the impedance of each cell at each measurement point in time series through a battery management system; and calculating a failure diagnosis preliminary prediction value through the battery management system, and pre-diagnosing a failure of the lithium-ion battery according to the failure diagnosis advance prediction value.

Here, the failure diagnosis preliminary prediction value is calculated as SampEn using the following formula, Here, B ^m (r) is the difference in impedance or distance between the i and j battery cells, which are different battery cells, among the plurality of battery cells in the battery, and is smaller than r, which is an indicator of data volatility. Indicates the number, and A ^m (r) is a different battery cell among a plurality of battery cells in the battery pack at the next measurement point in time series of the impedance measurement value of the battery cell targeted by B ^m (r). The impedance difference or distance between the i and jth battery cells represents the number smaller than r, which is an indicator of data volatility, , and m is 2.

In addition, r, which is an indicator of the volatility of the data, is preferably set to less than 0.002.

Additionally, when the failure diagnosis preliminary prediction value exceeds a specific threshold, the battery management system may output a warning or notification that maintenance is necessary.

The method of diagnosing failure of a lithium-ion battery in advance under high temperature conditions using electrochemical impedance spectroscopy according to a preferred embodiment of the present invention is to prevent fires and explosions from occurring in "high temperature environments", especially in the summer when fires occur frequently. It can reflect the characteristics of all batteries and provide a way to diagnose failures in advance to prepare for fire. It provides the effect of being able to see the inside of the battery without physically dismantling the battery at the battery cell level. In other words, the pattern in which the parameter Ri, which is the battery resistance, increases rapidly after a certain cycle at high temperature can be diagnosed in advance through the failure diagnosis prediction value, providing an opportunity for maintenance before a fire, etc. occurs in a high temperature environment.

Figure 1 shows changes in battery capacity and internal resistance, that is, changes in electrical characteristics, depending on the temperature of a lithium-ion battery.
Figure 2 shows analysis of battery characteristics using electrochemical impedance spectroscopy (EIS).
Figure 3 is a diagram explaining the failure diagnosis preliminary prediction value used in the present invention.
Figure 4 shows that when applying the failure diagnosis advance prediction value used in the present invention, failure diagnosis can be predicted in advance more quickly than the conventional anomaly detection technique using a moving average filter.
Figure 5 confirms changes in variation between cells due to battery pack aging.
Figure 6 confirms the change in discharge capacity as the battery pack ages.
Figure 7 shows the results of simulating overdischarge and temperature increase in the minimum cell portion due to imbalance between cells of a high-capacity battery pack.
Figure 8 is a diagram illustrating a method of pre-diagnosing voltage deviation imbalance within a battery pack using the pre-diagnosis imbalance diagnosis value invented in the present invention.
Figure 9 shows a preliminary diagnosis of imbalance in the battery pack based on the imbalance diagnosis preliminary prediction value in the case of r = 0.005.
Figure 10 shows the minimum voltage value of the battery cell in the battery pack at the 187th cycle predicted in Figure 9.
Figure 11 shows a preliminary diagnosis of imbalance in the battery pack based on the imbalance diagnosis preliminary prediction value in the case of r = 0.01.
Figure 12 shows the minimum voltage value of the battery cell in the battery pack at the 228th cycle predicted in Figure 11.

Hereinafter, preferred embodiments according to the present invention will be described in detail with reference to the attached drawings. Prior to this, the terms or words used in this specification and claims should not be construed as limited to their usual or dictionary meanings, and the inventor should appropriately define the concept of terms in order to explain his or her invention in the best way. Based on the principle of definability, it must be interpreted with meaning and concept consistent with the technical idea of the present invention.

Accordingly, the embodiments described in this specification and the configurations shown in the drawings are only one of the most preferred embodiments of the present invention and do not represent the entire technical idea of the present invention, so at the time of filing this application, various alternatives are available to replace them. It should be understood that equivalents and variations may exist.

Recently, global demand for batteries is expected to exceed 2TWh in 2030, and the transportation sector, which accounts for the largest portion of battery demand, is expected to increase to 85.7% by 2030. As the demand for batteries increases, research on battery status diagnosis and safety is actively underway, but fire accidents in electric vehicles (EV) and energy storage systems (ESS) are being reported both domestically and internationally. .

Causes of fires in electric vehicles and energy storage devices include charging, external collisions, and long-term exposure to environments such as high temperatures. To prevent this, a number of measures have been applied to energy storage devices and electric vehicles, such as reducing the operating SOC range, SPD (Surge protection device), and manufacturing process improvement to increase safety.

However, despite these measures, 18 fire accidents have occurred with Company H's electric vehicle since its launch in 2018 (as of July 2021), with 10 fire accidents occurring in the summer, accounting for more than half. In particular, a failure diagnosis plan is needed to reflect the characteristics of the battery and prepare for fire before a fire or explosion occurs in a "high temperature environment."

In lithium-ion batteries, under high temperature conditions, internal short circuits and venting due to internal gas may occur due to decomposition of the electrolyte and separator. Lithium-ion batteries installed in electric vehicles or energy storage devices are difficult to extinguish in the event of a fire, so battery failure diagnosis methods can be divided into pre- and post-diagnosis.

Pre-diagnosis is a method of diagnosing abnormal signs based on battery characteristics before a fire occurs, and post-diagnosis is aimed at preventing the spread of fire by detecting off-gas, arc, etc. just before a fire occurs. .

The present invention focuses on pre-diagnosis and proposes a diagnostic method that reflects the characteristics of the battery for pre-diagnosis. To diagnose the characteristics of a battery, direct current (DC) and alternative current AC are used to check the characteristics of the battery. In the case of direct current, a method of checking the output characteristics of the battery through changes in voltage by applying pulse current and a method of diagnosing the battery capacity by fully charging and fully discharging are used. This method can diagnose battery aging through electrical characteristics, but it has the disadvantage of not being able to check the internal characteristics of the battery, making it difficult to find outliers before the failure occurs.

Electrochemical impedance spectroscopy (EIS) using alternating current, applied in the present invention, is a Nyquist plot and Bode diagram through the voltage and current output by applying a sine wave to the voltage and current. plot) can be extracted and the internal characteristics of the battery can be checked.

For reference, Figure 2 shows analysis of battery characteristics using alternating current electrochemical impedance spectroscopy (EIS). As shown in FIG. 2, the physical characteristic information of the battery can be expressed using a Nyquist Plot (right side of FIG. 2) and an equivalent circuit model (left side of FIG. 2).

In the present invention, as a method of diagnosing a failure of a lithium-ion battery in advance under high temperature conditions using electrochemical impedance spectroscopy (EIS), first, using electrochemical impedance spectroscopy using alternating current, the lithium ion The impedance for each cell of the battery is extracted, and the impedance is measured time series at regular time intervals.

In addition to the electrochemical impedance spectroscopy described above, the fault diagnosis preliminary prediction value used in the present invention is calculated as SampEn using the following formula. Here, the sample entropy model is used.

Figure 3 is a diagram explaining the failure diagnosis preliminary prediction value used in the present invention. The maximum deviation of the impedance of each cell at each measurement point is calculated in time series through a battery management system or vehicle maintenance system inside or outside the vehicle. Through the method shown in FIG. 3, a battery management system or vehicle maintenance system calculates a failure diagnosis preliminary prediction value and pre-diagnoses a failure of the lithium-ion battery according to the failure diagnosis preliminary prediction value.

Referring to FIG. 3, looking more specifically at calculating the failure diagnosis advance prediction value according to the present invention, the failure diagnosis advance prediction value is sample entropy defined as a technique for quantifying the regularity and irregularity of the pattern of time series data. Use .

Here, the imbalance diagnosis prior prediction value can be calculated with SampEn using the following formula. Predictions are made based on unsupervised learning.

Here, B ^m (r) is the impedance difference or distance between the i and jth battery cells, which are different battery cells, among the plurality of battery cells in the battery pack, than r, which is an indicator of data volatility. Indicates a small number.

A ^m (r) is the i and j th battery cells that are different from each other among the plurality of battery cells in the battery pack at the time-serially next measurement point of the impedance measurement time of the battery cell targeted by B ^m (r). The impedance difference or distance between the battery cells represents the number smaller than r, which is an indicator of data volatility.

X is a variable, and here represents the impedance of the battery cell extracted using electrochemical impedance spectroscopy (EIS), but the temperature of the battery cell in the battery pack (Tcell), the current of the battery pack (Ipack), and the battery cell It can be various measurement values of the battery, including voltage (Vcell). N represents the window size, and m is 2.

If the predicted failure diagnosis value calculated in this way exceeds a certain threshold, it is reported that there is a risk of failure, and the battery management system or vehicle maintenance system inside or outside the vehicle is provided to users of electric vehicles using batteries, etc. A warning or notification that (MRO) is required can be output. Here, the specific threshold may be 1 (one).

This type of fault diagnosis advance prediction value uses previous data to calculate the probability of a new pattern every time. The closer the A/B value is to 1, the closer the fault diagnosis advance prediction value is to 0, and when abnormal data occurs, that is, normal data It was designed to have the feature that the number of impedance differences between in-cells smaller than r is reduced, and the A/B value approaches 0, so the advance prediction value for fault diagnosis increases. Preferably, if the failure diagnosis prior prediction value is 1 or more, the failure is designed to be predicted in advance.

Figure 4 shows that when applying the failure diagnosis advance prediction value used in the present invention, failure diagnosis can be predicted in advance more quickly than the conventional anomaly detection technique using a moving average filter.

As shown in Figure 4, when using the failure diagnosis advance prediction value of the present invention, when r = 0.0001, the 135th cycle failure diagnosis can be predicted in advance, and when r = 0.0002, the 145th cycle failure diagnosis can be predicted in advance. You can confirm that it is possible. On the other hand, the anomaly detection technique using a moving average filter required approximately the 163rd cycle to confirm deviation from the trend line and diagnose a failure.

In other words, the pattern in which the parameter Ri, which is the battery resistance, increases rapidly after a certain cycle at high temperature can be diagnosed in advance before actual failure through the failure diagnosis prediction value (sample entropy value), before a fire, etc. occurs in a high temperature environment. Provides an opportunity for maintenance.

(Other Embodiments)

The above-described failure diagnosis advance prediction value can also be used to predict imbalance between cells using other variables. Below, other embodiments will be described.

Figure 5 confirms the change in variation between cells depending on battery pack aging, and Figure 6 confirms the change in discharge capacity according to battery pack aging.

As shown in Figures 5 and 6, the present applicant confirmed that as the voltage difference between cells increases, a rapid decrease in capacity occurs. In Figures 5 and 6, it can be seen that the point at which the inter-cell voltage deviation occurs is the 156th cycle, and the point at which the minimum voltage of 2.7V or less is reached is the 230th cycle.

In Figure 6, in the case of the 270th cycle of aging, 118.5 Ah was expected, but it was confirmed that it was actually 116.3 Ah. As the battery pack ages, the slope of the discharge capacity changes from the point when an imbalance between cells in the battery pack occurs, and the performance of the battery pack decreases from the point when the voltage difference between cells occurs.

Figure 7 shows the results of simulating overdischarge and temperature increase in the minimum cell portion due to imbalance between cells of a high-capacity battery pack. As shown in Figure 7, at the time before inter-cell variation occurs (150 cycles), the minimum cell voltage is (3.108V) and temperature deviation (6.97°C), but at the time when inter-cell variation is most severe (270 cycles), the minimum cell voltage is (1.713V) and temperature deviation (7.26℃).

Figure 8 is a diagram illustrating a method of pre-diagnosing voltage deviation imbalance within a battery pack using the imbalance diagnosis prior prediction value invented in the present invention.

As shown in FIG. 8, the voltage (Vcell) of the battery cell in the battery pack is measured time-series at regular time intervals, that is, measured time-series for each of a plurality of battery cells in the battery pack, and at each measurement point. Calculate the maximum deviation of the voltage of multiple battery cells within a battery pack. Here, the voltage of the battery cell (Vcell) is used as an example, but the maximum time-series deviation of various measurement values of the battery, including the temperature of the battery cell in the battery pack (Tcell) and the current (Ipack) of the battery pack, can be used.

Next, a pre-prediction value for imbalance diagnosis, which will be described later, is calculated, and the imbalance is pre-diagnosed according to the pre-prediction value for imbalance diagnosis. Here, the imbalance is predicted in advance based on unsupervised learning through the variation between cells in the battery pack. If it exceeds a certain threshold, usually 1, it is considered that there is a risk of imbalance and maintenance (MRO) is required. It is possible to provide warnings or notifications to the user.

As already described, referring to FIG. 3, looking more specifically at calculating the imbalance diagnosis advance prediction value used in the high-capacity battery pack imbalance advance diagnosis method according to the present invention, the imbalance diagnosis advance prediction value is the time series data. It utilizes sample entropy, which is defined as a technique to quantify regularity and irregularity in patterns.

Here, the imbalance diagnosis prior prediction value can be calculated with SampEn using the following formula. Predictions are made based on unsupervised learning.

Here, B ^m (r) is the difference or distance in voltage (Vcell) of the i and jth battery cells, which are different battery cells, among the plurality of battery cells in the battery pack, and is an indicator of data volatility. It represents a number smaller than r.

A ^m (r) is a different battery cell i among a plurality of battery cells in the battery pack at the time-series immediately following measurement time of the voltage (Vcell) of the battery cell targeted by B ^m (r). , indicates the number of differences or distances in the voltage (Vcell) of the jth battery cell that are smaller than r, which is an indicator of data volatility.

X is a variable, and here represents the voltage of the battery cell (Vcell), but as described above, it can be various measurement values of the battery, including the temperature of the battery cell in the battery pack (Tcell) and the current (Ipack) of the battery pack. N represents the window size, and m is 2.

If the pre-predicted imbalance diagnosis value calculated in this way exceeds a certain threshold, it is considered that there is a risk of imbalance occurring, and the battery management system warns or notifies users of electric vehicles using the battery that maintenance (MRO) is required. can be output. Here, the specific threshold may be 1.

This kind of imbalance diagnosis advance prediction value uses previous data to calculate the probability of a new pattern every time. The closer the A/B value is to 1, the closer the imbalance diagnosis advance prediction value is to 0, and when abnormal data occurs, that is, normal data It was designed to have the feature that the number of voltage differences between in-cells smaller than r decreases, and the A/B value approaches 0, so the preliminary prediction value for imbalance diagnosis increases. Preferably, if the imbalance diagnosis prior prediction value is greater than 1, the imbalance is designed to be predicted in advance.

Figure 9 shows a preliminary diagnosis of imbalance in the battery pack based on the imbalance diagnosis preliminary prediction value in the case of r = 0.005, and Figure 10 shows the minimum voltage value of the battery cell in the battery pack at the 187th cycle predicted in Figure 9.

As shown in Figures 9 and 10, the imbalance diagnosis advance prediction value (SampEn) exceeds the threshold of 1 at the 187th cycle before the imbalance of the battery pack worsens, and the lower limit voltage of each cell of the battery (2.7V) is 3.046V. ), it was confirmed that it was possible to diagnose voltage imbalance within the battery pack before it occurred.

Figure 11 shows a preliminary diagnosis of imbalance in the battery pack based on the imbalance diagnosis preliminary prediction value in the case of r = 0.01, and Figure 12 shows the minimum voltage value of the battery cell in the battery pack at the 228th cycle predicted in Figure 11.

As shown in FIG. 11, it can be confirmed that the imbalance diagnosis advance prediction value (SampEn) exceeds the threshold of 1 at the 228th cycle before the imbalance of the battery pack worsens. As shown in Figure 12, it was confirmed that the voltage imbalance in the battery pack could be diagnosed before the lower limit voltage (2.7V) of each cell of the battery was reached at 2.705V at the 228th cycle.

As shown in FIGS. 9 to 12 , it is possible to confirm the possibility of pre-diagnosing imbalance before the cell minimum voltage is reached due to the voltage difference between battery cells in the battery pack.

As described above, although the present invention has been described with limited examples and drawings, the present invention is not limited thereto, and the technical idea of the present invention and the following will be understood by those skilled in the art to which the present invention pertains. Of course, various modifications and variations are possible within the scope of equivalence of the claims to be described.

Claims (4)

A method of diagnosing failure of a lithium-ion battery in advance under high temperature conditions using electrochemical impedance spectroscopy (EIS),
Extracting the impedance of each cell of the lithium-ion battery using electrochemical impedance spectroscopy using alternating current, and measuring the impedance in time series at regular time intervals;
Calculating the maximum deviation of the impedance of each cell at each measurement point in time series through a battery management system; and
Comprising: calculating a failure diagnosis preliminary prediction value through the battery management system, and pre-diagnosing a failure of the lithium-ion battery according to the failure diagnosis advance prediction value;
How to diagnose lithium-ion battery failure in advance.
According to claim 1,
The fault diagnosis preliminary prediction value is calculated as SampEn using the following formula,

Here, B ^m (r) is the impedance difference or distance between the i and jth battery cells, which are different battery cells, among the plurality of battery cells in the battery, and is smaller than r, which is an indicator of data volatility. Indicates the number,
A ^m (r) is the i and j th battery cells that are different from each other among the plurality of battery cells in the battery pack at the next measurement point in time series of the impedance measurement value of the battery cell targeted by B ^m (r). The impedance difference or distance between the battery cells represents the number smaller than r, which is an indicator of the volatility of the data.
X is a variable, here representing the impedance of the battery cell,
N represents the data length of the variable, and m is 2.
How to diagnose lithium-ion battery failure in advance.
According to claim 2,
r, the volatility index of the data, is set to be less than 0.002,
How to diagnose lithium-ion battery failure in advance.
According to claim 3,
When the failure diagnosis preliminary prediction value exceeds a certain threshold, the battery management system outputs a warning or notification that maintenance is necessary,
How to diagnose lithium-ion battery failure in advance.