KR20220128828A - Apparatus and method for estimating battery soh - Google Patents

Abstract

The present invention relates to a device for predicting a state of health (SOH) of a battery, which comprises: a battery information input unit for measuring at least one of voltage, current, capacity, and internal resistance (IR) of a battery; a control unit for predicting SOH_c on the basis of the capacity of the battery after receiving battery information measured by the battery information input unit, predicting SOH_r on the basis of the IR of the battery, and then predicting a final SOH on the basis of the SOH_c and the SOH_r; and a SOH output unit for outputting the SOH predicted by the control unit to a designated device. Therefore, SOH prediction time and complexity are reduced.

Description

Apparatus and method for predicting battery SOH

The present invention relates to an apparatus and method for predicting battery SOH, and more particularly, predicting SOH _c based on the capacity of the battery, predicting SOH _r based on the resistance of the battery, and then predicting the SOH _c (ie, battery capacity It relates to an apparatus and method for predicting battery SOH, enabling the final state of health (SOH) to be predicted based on the SOH for ) and the SOH _r (ie, the SOH for battery resistance growth).

Recently, rechargeable batteries (hereinafter referred to as 'battery') that can be recharged in various fields ranging from portable electronic devices such as smartphones, notebook computers, and PDA's to electric vehicles and energy storage systems (ESS) (eg, lithium batteries) ) is increasing in use.

However, such a battery has a characteristic that the lifespan of the battery is shortened as the continuous use period is longer and the charging/discharging is repeated, and thus the capacity/performance of the battery is gradually reduced.

Therefore, accurately predicting the state of health (SOH) of a battery allows you to identify when the battery needs to be replaced, and accordingly, the system using the battery is stable by minimizing the negative effects that may occur due to the decrease in capacity/performance of the battery. to be able to operate as

However, in the prior art, in predicting the state of health (SOH) of the battery, all data of the voltage and current of the battery (that is, all data measured at a specified time interval from the fully discharged battery to the fully discharged state) of the battery are used. SOH was predicted by calculating capacity or internal resistance (ie, using only either SOH _r or SOH _c ). That is, since the SOH was predicted using all data of voltage and current in the prior art, the computational complexity was high, and this resulted in problems such as decreasing the efficiency of the SOH prediction (estimating) device and increasing the cost.

For example, some have previously used an electrochemical model and an equivalent circuit model (ECM)-based method to estimate the SOH based on the internal resistance (IR) (hereinafter, simply referred to as resistance) growth (or rise) of a battery. A method (such as ECM) for this purpose was proposed, but it required complex battery data analysis to express the operation of the internal resistance (IR), and there was a problem that it could not be accurately generalized to other types of batteries. Although the base method was used, it took a lot of time and required a lot of power, so it was not suitable for real-time applications of embedded systems.

Therefore, there is a need for a SOH prediction method with lower complexity while considering both capacity degradation and internal resistance (IR) increase (growth) even in low-performance embedded systems with less complex calculations.

The background technology of the present invention is disclosed in Korean Patent Publication No. 10-2019-0099904 (published on August 28, 2019, battery management system).

According to one aspect of the present invention, the present invention is created to solve the above problems, and after predicting SOH _c based on the capacity of the battery and predicting SOH _r based on the resistance of the battery, the SOH _c (ie, SOH for battery capacity) and SOH _r (ie, SOH for battery resistance growth) based on the SOH (State of Health) to predict the final SOH (State of Health) to provide a battery SOH prediction device and method to provide it has its purpose

Battery SOH prediction apparatus according to an aspect of the present invention, a battery information input unit for measuring at least one of voltage, current, capacity, and internal resistance (IR) of a battery; After receiving the battery information measured by the battery information input unit, SOH _c is predicted based on the capacity of the battery, SOH _r is predicted based on the internal resistance of the battery, and then SOH _c and SOH _r are used based on the a control unit for predicting the final state of health (SOH); and an SOH output unit for outputting the SOH predicted by the control unit to a designated device.

In the present invention, the control unit is characterized in that it is implemented to perform deep learning.

In the present invention, the control unit collects raw voltage and current data of the battery, performs a pre-processing operation on the collected raw voltage and current to extract the characteristics of the battery health indicator (HI), and dips the extracted features. Predicting SOH _c and SOH _r through running, and predicting SOH based on the predicted SOH _c and SOH _r .

In the present invention, the raw voltage and current are respectively the voltage and current values measured during the first designated initial time of the charging and discharging process.

In the present invention, the preprocessing operation includes data cleaning and normalization, wherein the data cleaning is a process of removing outliers from data to prevent underfitting, and the normalization converts data into values between 0 and 1. It is characterized in that it is an adjustment process.

In the present invention, the SOH _c is defined as the ratio of the initial battery capacity and the rated capacity, and is expressed as Equation 1 below.

(Equation 1)

Here, C _int is the initial value of the battery capacity, and C _k is the capacity in the current cycle.

In the present invention, the SOH _r is defined as the increase ratio of the increase in the internal resistance (IR) of the battery and the increase in the initial internal resistance (IR), and is expressed by Equation 2 below.

(Equation 2)

where IR _k is the internal resistance (IR) of the current cycle, IR _int is the internal resistance of a new battery (IR), and IR _EOL is the internal resistance (IR) at the end of the battery life.

의 연산을 통해 SOH를 예측하는 것을 특징으로 한다.In the present invention, the controller uses a curve fitting method to model the relationship between the SOH, SOH _c and SOH _r , and a polynomial function modeled as in Equation 3 below.

It is characterized in that the SOH is predicted through the operation of .

(Equation 3)

where X is SOH _c , Y is SOH _r , and c _n (n = 1, ..., 6) means the coefficient of the model.

Battery SOH prediction method according to another aspect of the present invention, the control unit collecting the raw voltage and current data of the battery; extracting, by the controller, a characteristic of a battery health indicator (HI) by performing a pre-processing operation on the collected raw voltage and current; predicting, by the controller, SOH _c and SOH _r through deep learning of the extracted features; and predicting, by the controller, SOH based on the predicted SOH _c and SOH _r .

In the present invention, the raw voltage and current are respectively the voltage and current values measured during the first designated initial time of the charging and discharging process.

In the present invention, the preprocessing operation includes data cleaning and normalization, wherein the data cleaning is a process of removing outliers from data to prevent underfitting, and the normalization converts data into values between 0 and 1. It is characterized in that it is an adjustment process.

In the present invention, the SOH _c is defined as the ratio of the initial battery capacity and the rated capacity, and is expressed as Equation 1 below.

(Equation 1)

Here, C _int is the initial value of the battery capacity, and C _k is the capacity in the current cycle.

In the present invention, the SOH _r is defined as the increase ratio of the increase in the internal resistance (IR) of the battery and the increase in the initial internal resistance (IR), and is expressed by Equation 2 below.

(Equation 2)

where IR _k is the internal resistance (IR) of the current cycle, IR _int is the internal resistance of a new battery (IR), and IR _EOL is the internal resistance (IR) at the end of the battery life.

의 연산을 통해 SOH를 예측하는 것을 특징으로 한다.In the present invention, in the step of predicting the SOH, the controller uses a curve fitting method to model the relationship between the SOH, SOH _c and SOH _r , and a polynomial function modeled as in Equation 3 below.

It is characterized in that the SOH is predicted through the operation of .

(Equation 3)

where X is SOH _c , Y is SOH _r , and c _n (n = 1, ..., 6) means the coefficient of the model.

According to one aspect of the present invention, the present invention predicts SOH _c based on the capacity of the battery, predicts SOH _r based on the resistance of the battery, and then the SOH _c (ie, SOH related to the battery capacity) and the By enabling the final state of health (SOH) to be predicted based on SOH _r (that is, SOH for battery resistance growth), the prediction accuracy of SOH is improved and SOH prediction time and complexity are reduced.

In addition, according to another aspect of the present invention, the present invention provides SOH _r and SOH in voltage values of a first designated portion of voltage and current data when using a battery (that is, voltage values measured during an initial time designated in each charging process and discharging process) After extracting _c , it is possible to predict SOH (State of Health) through learning using deep learning, thereby improving the prediction accuracy of SOH and reducing SOH prediction time and complexity.

1 is an exemplary diagram showing a schematic configuration of a battery SOH prediction apparatus according to an embodiment of the present invention.
2 is an exemplary view for explaining the reason for predicting SOH based on SOH _c and SOH _r in FIG. 1 .
3 is a flowchart illustrating a battery SOH prediction method according to an embodiment of the present invention.

Hereinafter, an embodiment of an apparatus and method for predicting battery SOH according to the present invention will be described with reference to the accompanying drawings.

In this process, the thickness of the lines or the size of the components shown in the drawings may be exaggerated for clarity and convenience of explanation. In addition, the terms to be described later are terms defined in consideration of functions in the present invention, which may vary according to intentions or customs of users and operators. Therefore, definitions of these terms should be made based on the content throughout this specification.

1 is an exemplary diagram showing a schematic configuration of a battery SOH prediction apparatus according to an embodiment of the present invention.

1 , the battery SOH prediction apparatus according to the present embodiment includes a battery information input unit 110 , a control unit 120 , and an SOH output unit 130 .

The battery information input unit 110 measures at least one of voltage, current, capacity, and internal resistance (IR) of the battery and inputs it to the control unit 120 .

After receiving the voltage and current of the battery, the control unit 120 predicts SOH _c based on the capacity of the battery and predicts SOH _r based on the resistance (ie, internal resistance (IR)) of the battery, A final state of health (SOH) is predicted based on the SOH _c (ie, SOH for battery capacity) and SOH _r (ie, SOH for battery resistance growth).

At this time, the control unit 120 performs deep learning learning.

The SOH output unit 130 outputs the SOH (State of Health) predicted (estimated) by the control unit 120 to a designated device (eg, a display means, an external control device, etc.).

For reference, in the present embodiment, in terms of SOH estimation, a decrease in battery capacity is a major factor in the deterioration of the battery state, and in order to reduce calculation complexity, the decrease in capacity is determined by the voltage values of the first specified part of the discharging process (that is, the charged battery). voltage values measured during the specified initial time in the discharge process of

In this case, the relationship between the first voltage value, the charge/discharge cycle, and the capacity degradation in the battery discharging process (HI _C ) is as shown in FIG. 2A . That is, as shown in (a) of FIG. 2 , the battery capacity decreases as the number of cycles of the discharging process (HI _C ) increases. Therefore, the discharge process (HI _C ) can reflect capacity degradation and can be used to estimate SOH based on capacity.

The increase (or growth) of a battery's internal resistance (IR) is also used as a battery health indicator (HI) (or an indicator of battery health). To reduce computational complexity, the internal resistance (IR) is set to the voltage values of the first specified portion of the charging process (HI _R ) (i.e. voltage values measured during the initial time specified during the charging process of a discharged battery) corresponding to the first voltage value. replaced by a value.

At this time, the relationship between the internal resistance (IR) of the battery and the cycle of the charging process (HI _R ) is as shown in FIG. 2( b ). That is, as shown in (b) of FIG. 2 , the internal resistance IR and the voltage increase according to the number of cycles. Therefore, the charging process (HI _R ) can reflect the increase in internal resistance (IR) and can be used to estimate SOH based on internal resistance (IR).

On the other hand, in the existing SOH estimation method, the learning process of SOH estimation is based on a single type of battery health indicator (HI), which reflects a decrease in capacity or an increase in internal resistance (IR). Increasing the internal resistance (IR) exhibits different behavior and can therefore have different effects on the battery SOH.

Therefore, in this embodiment, the SOH is predicted in a manner that considers both of the above two factors (eg, battery capacity, internal resistance).

That is, the present embodiment provides a method of predicting the SOH in consideration of both the battery capacity and the internal resistance (IR). Hereinafter, the operation of the control unit 120 will be described in more detail with reference to the flowchart of FIG. 3 . .

Referring to FIG. 3 , the controller 120 collects raw voltage and current data ( S101 ).

In addition, the control unit 120 performs a pre-processing operation on the collected raw voltage and current in order to extract the characteristics of the battery health indicator (HI) (S102).

Here, the raw voltage and current refer to voltage and current values measured during the first designated initial time of the charging and discharging process, respectively.

and the preprocessing operation includes data cleaning and normalization, wherein the data cleaning is a process of removing outliers from the data to prevent under-fitting, and the normalization is the process of converting the data into values between 0 and 1. It is an adjustment process.

When the pre-processing operation is completed, the control unit 120 extracts the characteristics (eg, HI _C , HI _R ) of the battery health indicator (HI) from the data on which the pre-processing operation is completed ( S103 ).

And the control unit 120 uses the extracted features (eg, HI _C , HI _R ) through deep learning (eg, RNN, LSTM, etc.) to SOH _c (ie, SOH related to battery capacity) and SOH _r (ie, Predict (estimate) (SOH) related to battery resistance growth (S104, S105).

Here, the SOH _c is defined as the ratio of the initial battery capacity to the rated capacity.

So SOH _c is 100% for a new battery (the battery capacity is equal to the nominal capacity) and 0% when the capacity is reduced to 80% of the nominal capacity.

The SOH _c is expressed as in Equation 1 below.

Here, C _int is the initial value of the battery capacity, and C _k is the capacity in the current cycle.

Meanwhile, the SOH _r is defined as a ratio of an increase in the internal resistance (IR) of the battery and an increase in the initial internal resistance (IR), and is expressed as Equation 2 below.

where IR _k is the internal resistance (IR) of the current cycle, IR _int is the internal resistance of a new battery (IR), and IR _EOL is the internal resistance (IR) at the end of the battery life.

Also, the control unit 120 predicts SOH based on the predicted (estimated) SOH _c and SOH _r ( S106 and S107 ).

At this time, the estimated SOH is treated as a variable with two components, SOH _c and SOH _r , because SOH _c and SOH _r have a significant effect on SOH.

를 사용하여 모델링 할 수 있다. 즉, 아래의 수학식 3의 연산을 통해 SOH가 예측(추정)된다.For example, a curve fitting method may be used to model the relationship between the SOH, SOH _c , and SOH _r , and a polynomial function as shown in Equation 3 below

can be modeled using That is, SOH is predicted (estimated) through the operation of Equation 3 below.

where X is SOH _c , Y is SOH _r , and c _n (n = 1, ..., 6) means the coefficient of the model.

As described above, in this embodiment, SOH _c is predicted based on the capacity of the battery, and _{SOH r} is predicted based on the resistance of the battery _. , SOH related to battery resistance growth) can be used to predict the final state of health (SOH), thereby improving the prediction accuracy of the SOH and reducing the SOH prediction time and complexity.

In addition, this embodiment extracts SOH _r and SOH _c from the voltage values of the first specified part of the voltage and current data when using the battery (that is, the voltage values measured during the initial time specified in each charging process and discharging process), and then dips them By making it possible to finally predict the State of Health (SOH) through learning using learning, it has the effect of improving the prediction accuracy of the SOH and reducing the SOH prediction time and complexity.

As described above, the present invention has been described with reference to the embodiment shown in the drawings, but this is merely an example, and various modifications and equivalent other embodiments are possible therefrom by those skilled in the art. will understand the point. Therefore, the technical protection scope of the present invention should be defined by the following claims. Implementations described herein may also be implemented as, for example, a method or process, an apparatus, a software program, a data stream, or a signal. Although discussed only in the context of a single form of implementation (eg, discussed only as a method), implementations of the discussed features may also be implemented in other forms (eg, as an apparatus or program). The apparatus may be implemented in suitable hardware, software and firmware, and the like. A method may be implemented in an apparatus such as, for example, a processor, which generally refers to a computer, a microprocessor, a processing device, including an integrated circuit or programmable logic device, or the like. Processors also include communication devices such as computers, cell phones, portable/personal digital assistants ("PDA") and other devices that facilitate communication of information between end-users.

110: battery information input unit
120: control unit
130: SOH output unit

Claims (14)

a battery information input unit for measuring at least one of voltage, current, capacity, and internal resistance (IR) of the battery;
After receiving the battery information measured by the battery information input unit, SOH _c is predicted based on the capacity of the battery, SOHr is predicted based on the internal resistance of the battery, and finally based on the SOH _c and the SOHr Control unit for predicting SOH (State of Health); and
Battery SOH prediction device comprising a; SOH output unit for outputting the SOH predicted by the control unit to a designated device.
According to claim 1, wherein the control unit,
Battery SOH prediction device, characterized in that implemented to perform deep learning.
According to claim 1, wherein the control unit,
Collecting the raw voltage and current data of the battery, performing a preprocessing operation on the collected raw voltage and current to extract the characteristics of the battery health indicator (HI),
Predict SOH _c and SOH _r using the extracted features through deep learning,
Battery SOH prediction device, characterized in that for predicting SOH based on the predicted SOH _c and SOH _r .
4. The method of claim 3, wherein the raw voltage and current are:
Battery SOH prediction device, characterized in that the voltage and current values are measured during the first specified initial time of the charging and discharging process, respectively.
The method of claim 3, wherein the pre-processing operation,
including data cleaning and normalization;
The data cleaning is a process of removing outliers from data to prevent underfitting, and the normalization is a process of adjusting the data to a value between 0 and 1.
According to claim 3, wherein SOH _c is,
It is defined as a ratio of the initial battery capacity to the rated capacity, and is characterized in that it is expressed as in Equation 1 below.
(Equation 1)

Here, C _int is the initial value of the battery capacity, and C _k is the capacity in the current cycle.
The method of claim 3, wherein the SOH _r is,
A battery SOH prediction device, characterized in that it is defined as the increase ratio of the increase in the internal resistance (IR) of the battery and the increase in the initial internal resistance (IR), and is expressed as Equation 2 below.
(Equation 2)

where IR _k is the internal resistance (IR) of the current cycle, IR _int is the internal resistance of a new battery (IR), and IR _EOL is the internal resistance (IR) at the end of the battery life.
According to claim 3, wherein the control unit,
A curve fitting method is used to model the relationship between the SOH, SOH _c and SOH _r , and a polynomial function modeled as in Equation 3 below

Battery SOH prediction device, characterized in that predicting the SOH through the operation of.
(Equation 3)

where X is SOH _c , Y is SOH _r , and c _n (n = 1, ..., 6) means the coefficient of the model.
Collecting, by the controller, raw voltage and current data of the battery;
extracting, by the controller, a characteristic of a battery health indicator (HI) by performing a pre-processing operation on the collected raw voltage and current;
predicting, by the controller, SOH _c and SOH _r through deep learning of the extracted features; and
and predicting, by the controller, SOH based on the predicted SOH _c and SOH _r .
10. The method of claim 9, wherein the raw voltage and current are:
A method for predicting battery SOH, characterized in that the voltage and current values are measured during the first specified initial time of the charging and discharging process, respectively.
10. The method of claim 9, wherein the pre-processing operation,
including data cleaning and normalization;
The data cleaning is a process of removing outliers from data to prevent underfitting, and the normalization is a process of adjusting the data to a value between 0 and 1.
The method of claim 9, wherein SOH _c is,
It is defined as a ratio of the initial battery capacity to the rated capacity, and is expressed as Equation 1 below.
(Equation 1)

Here, C _int is the initial value of the battery capacity, and C _k is the capacity in the current cycle.
10. The method of claim 9, wherein the SOH _r is,
The battery SOH prediction method, characterized in that it is defined as the increase ratio of the increase in the internal resistance (IR) of the battery and the increase in the initial internal resistance (IR), and is expressed as Equation 2 below.
(Equation 2)

where IR _k is the internal resistance (IR) of the current cycle, IR _int is the internal resistance of a new battery (IR), and IR _EOL is the internal resistance (IR) at the end of the battery life.
10. The method of claim 9, wherein in predicting the SOH,
The control unit is
A curve fitting method is used to model the relationship between the SOH, SOH _c and SOH _r , and a polynomial function modeled as in Equation 3 below

Battery SOH prediction method, characterized in that predicting the SOH through the operation of.
(Equation 3)

where X is SOH _c , Y is SOH _r , and c _n (n = 1, ..., 6) means the coefficient of the model.

Images