KR20120028000A - A method for the soc estimation of li-ion battery and a system for its implementation - Google Patents

Abstract

PURPOSE: A charged state estimation method of a lithium ion battery and a system for the same are provided to precisely estimate an SOC(State Of Charge) through calculated parameters of an equivalent impedance model by dividing an entire operation region into several operation points. CONSTITUTION: A load(20) and a battery charger(30) are operated by receiving power of a lithium ion battery(10). A temperature sensor(40) measures temperature of the lithium ion battery. A current sensor(45) measures charge and discharge current of the lithium ion battery. A voltage sensor(50) measures voltage between anode and cathode terminals of the lithium ion battery. An SOC(State Of Charge) estimator(60) calculates an open current of the battery and a voltage drop of the equivalent impedance model at a current operation point using an estimation algorithm with respect to a model bank. The SOC estimator estimates an SOC from the calculated open current.

Description

A method for the SOC estimation of Li-ion battery and a system for its implementation}

The present invention relates to a method of estimating the state of charge of a lithium-ion battery (LIB) mounted in a hybrid vehicle or an electric vehicle, and dividing a temperature and a current, which are operating conditions of the battery, into an appropriate number within an allowable range. The state of charge of the battery at any operating point (temperature, current) is measured using current, terminal voltage, and temperature data measured in real time based on the model bank obtained by obtaining the electrical equivalent model at all divided operating points. An algorithm for accurately estimating SOC) and a system for implementing the SOC estimation method in a battery management system (BMS).

Hybrid electric vehicles (HEVs) or electric vehicles (EVs) are equipped with large-capacity secondary batteries as electric energy storage devices, and recently, lithium-ion batteries are prevalent. The electric power stored in the lithium ion battery mounted on the hybrid vehicle or the electric vehicle is used as the driving power of the electric motor to drive the vehicle, and is generated by the power generated by the generator using the waste heat of the internal combustion engine or the power regenerated by the electric motor when decelerating. Is charged. In order to maintain long life and safe use of Li-ion battery, it should be operated within proper charging range of Li-ion battery. To do this, it is necessary to know the remaining charge capacity of the lithium ion battery in real time. However, the remaining charge capacity inside the lithium ion battery has a complex functional relationship between the terminal voltage, the current, the temperature, and the number of charge / discharge cycles of the lithium ion battery. Estimating the state of charge of a lithium ion battery from the information of these measurable variables is called "SOC estimation".

Battery management system for electric vehicle estimates SOC of storage battery by measuring battery temperature, voltage and charge / discharge current. There are four main categories of estimation algorithms: Coulomb counting method for integrating charge and discharge currents, electrochemical modeling techniques for chemical reactions in cells, and cell operation. These include techniques for expressing the dynamic behavior of time and SOC in pure mathematical empirical formulas, and using electrical equivalence models. Since the Coulomb calculation accumulates charge / discharge currents using an integrator, there is a problem that the initial value is inaccurate and the cumulative error becomes very large when used for a long time.

In particular, it may not be appropriate in an environment where charging and discharging are frequently performed, such as HEV / EV. The second and third methods are not easy to apply for estimating SOC in real time. In the conventional methods using the electrical equivalent model, the SOC is estimated based on one equivalent model obtained under a specific operating condition and the experimental compensation value is considered in the region where the temperature is different.

The SOC estimation method, which can guarantee a small error over the entire operating conditions (C-rate, temperature), is a core technology of BMS. In addition to SOC estimation, BMS functions include voltage balancing between battery cells, overcharge protection, and overdischarge protection.

In HEV or EV with regenerative braking function, charging and discharging can be performed at any time. In this case, the SOC is estimated only during charging or discharging mode. Because it may be different from the dynamic characteristics of, it is difficult to accurately estimate the SOC. Moreover, as a general characteristic of Li-ion batteries, a strong nonlinear characteristic appears between open circuit voltage (OCV) and SOC at both ends of the SOC level of 15-20% or less. Precise SOC estimation is difficult.

The present invention has been invented in view of the above problems for SOC estimation, the standard operating range of the temperature and current of the lithium ion battery (for example, the current is 0 ~ 8 C-rate, the cell temperature is -20 ~ 60 ℃) to estimate the SOC precisely. Basically, the method uses an electrical equivalent circuit model between the open circuit voltage (OCV) and the terminal voltage. The SOC level is divided into a region represented by a linear impedance model and a region represented by a nonlinear impedance model. In order to solve the problem that the parameters of the equivalent model change according to the operating conditions such as temperature and current, a method of identifying and using the equivalent model at a plurality of operating points is presented. The intersection point where temperature and current are respectively located on the vertical axis and the horizontal axis of the rectangular coordinate system is divided by the appropriate number to represent one operating point. The division of the operating point of the lithium ion battery may vary according to the manufacturer or model. The electrical equivalent model is identified in advance at all operating points and this set of models is defined as a multiple model for battery voltage drop. It is an object of the present invention to provide a method of estimating the state of charge of a lithium ion battery, which improves the accuracy of SOC estimation by estimating real-time SOC based on this multi-model set.

It is still another object of the present invention to provide a system for implementing the SOC estimation method.

The present invention for performing such an object,

Expressing the relationship between the internal electromotive force of the lithium ion battery, the terminal voltage, and the charge / discharge current using an electrical equivalent circuit, and obtaining impedance Z (I, T, s) differently depending on the operating point.

According to the SOC level, the impedance Z (I, T, s) is classified into two regions represented by the linear transfer function (regions A and B in FIG. 3) and one region represented by the nonlinear function (region C in FIG. 3). . An open circuit test step of generating a multi-model operating point based on temperature and current data in each region and identifying an equivalent circuit model corresponding to the operating point of the lithium ion battery;

By converting the continuum model identified in the open circuit test step into a discrete metric model, a model corresponding to four adjacent operating points is composed of one model bank (for example, the model bank B _i, in FIG. 4). _j consists of four impedance models: Z _{i, j} , Z _{i + 1, j} , Z _{i, j + 1} , Z _{i + 1, j + 1} ), the terminal voltage, current and temperature of the Li-ion battery during operation The equivalent circuit model corresponding to the current state of the lithium ion battery is estimated through the model banks of the multi-model operating point by periodically receiving the data, and the equivalent circuit model obtained by the estimation of the terminal voltage and current data measured at each sample time is obtained. The present invention provides a method for estimating the state of charge of a lithium ion battery, which comprises the steps of calculating the open voltage of the lithium ion battery and calculating the SOC.

According to a preferred embodiment of the present invention can be largely divided into an open circuit test step and a real-time SOC estimation step,

In the open circuit test step, as shown in FIG. 3, the open voltage-SOC correlation curve is classified into a linear model region and a nonlinear model region (region C) according to the SOC level:

As shown in FIG. 4, the temperature and current data are divided into an appropriate number within the allowable range of the lithium ion battery, and the temperature is represented on the vertical axis by the current on the horizontal axis to generate intersection points expressed in matrix form as reference points of the equivalent impedance model. ;

An equivalent circuit display step of displaying a lithium ion battery as an equivalent circuit of open voltage and series impedance in the form of a dependent power supply as a function of the temperature and current data after the multi-model reference point generation step;

Acquiring terminal voltage, OCV, and charge / discharge current data of the battery through an open circuit test to identify the linear transfer function model in the linear region and the nonlinear model in the nonlinear region at the temperature-current operating point, respectively;

Converting the identified continuum model set into a discrete value model in consideration of a sampling time for application to a real-time multi-model SOC estimator to be constructed in a next step;

As shown in FIG. 4, a model bank includes four impedance transfer functions corresponding to four adjacent operating points.

In the real time SOC estimation phase,

A measurement step of measuring terminal voltage of the lithium ion battery, temperature and current data of the lithium ion battery during each operation of the electric vehicle;

Selecting a model bank that surrounds the operating point represented by the temperature and current data of the lithium ion battery measured at each sample time in the measurement step (for example, the current cell temperature Tz and current Ibatz in FIGS. 4 and 5) Model banks Bi, j corresponding to (Tz, Ibatz) are four impedance models containing this operating point)

Estimating OCV and SOC at the operating point by multiple model adaptive estimation (MMAE) using four discrete-valued impedance models of the selected model bank;

If the value of the temperature-current measured at each sample time in accordance with the change in the operating conditions is to include the step of repeating the above process by selecting the model bank including this operating point.

The method of estimating the state of charge of a lithium ion battery according to the present invention and the effect of the system for implementing the method are clearly seen from two viewpoints. Relying on the model obtained at only one operating point when estimating the SOC using the equivalent circuit model is not sufficient to accurately estimate the SOC because the parameters of the equivalent impedance model change with current and temperature. In the present invention, this problem can be solved by dividing the entire operating area into several operating points and calculating the parameters of the equivalent impedance model. In addition, according to the present invention, once the equivalent model of the lithium ion battery is obtained, there is no need to recalculate the model thereafter. .

1 is a schematic block diagram illustrating a configuration of a battery management system including a lithium ion battery for estimating a state of charge of a lithium ion battery according to an exemplary embodiment of the present invention.
FIG. 2 is a circuit diagram modeling a relationship between internal electromotive force, terminal voltage, and current of a lithium ion battery by an equivalent circuit having an open circuit voltage (OCV) and a series impedance of a dependent power source.
3 is identified by classifying the equivalent impedance model of FIG. 2 into regions represented by linear dynamic models (regions A and B) and regions represented by nonlinear static models according to SOC levels due to the inherent characteristics of lithium ion batteries. This graph indicates that it should.
4 illustrates an example of configuring an operating point for obtaining a multi-model required for estimating a state of charge of a lithium ion battery according to the present invention. The temperature is represented on the vertical axis and the C-rate is plotted on the horizontal axis. The location of the meeting point is the reference point for identifying the equivalent model.
5 is a view for explaining the multi-model estimation of the open voltage OCV using a model bank in the present invention.
FIG. 6 illustrates the effect of the present invention and shows how a parameter of the equivalent impedance model of FIG. 2 varies with temperature at the same current for a specific LIB.
FIG. 7 illustrates the effect of the present invention and shows how a parameter of the equivalent impedance model of FIG. 2 varies with current at the same temperature for a specific LIB.
FIG. 8 is a diagram illustrating the effect of the present invention, and comparing and showing the accuracy when the OCV value is estimated by only one model around this operating point when charging and discharging occurs at a certain temperature for a specific LIB.
FIG. 9 is a view illustrating the effect of the present invention through the results of applying the specific model to the estimation method that the OCV can be estimated very precisely even under varying operating conditions.
FIG. 10 is a diagram showing charge / discharge current in the form of PRBS applied to identify an equivalent impedance model in an open circuit test step.
FIG. 11 is a diagram illustrating an open circuit voltage (OCV) and a terminal voltage measured in response to a battery when the charge / discharge current of FIG. 10 is applied to identify an equivalent impedance model.
12 is a principle diagram illustrating a multi-model adaptive SOC estimation proposed in the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

1 is a schematic block diagram illustrating a configuration of a battery management system including a lithium ion battery for estimating a state of charge of a lithium ion battery according to an exemplary embodiment of the present invention. Referring to FIG. 1, a lithium ion battery system 100 includes a lithium ion battery 10, a load 20, and a charger 30, and a temperature sensor 40 measuring a temperature T of the lithium ion battery 10. ), A current sensor 45 for measuring the input and output charge and discharge current Ibat of the lithium ion battery 10 and a voltage sensor 50 for measuring the voltage Vt between the terminals of the lithium ion battery. The charge / discharge current of the battery is measured by the current sensor 45, and the voltage sensor 50 measures the change of the lithium ion battery voltage according to the charge / discharge current. The load of the lithium ion battery 10 includes a driving motor, a window wiper, a direction indicator, a lamp, and the like, which are driven and supplied with power, and the generator / charger 30 supplies power to the lithium ion battery 10. The battery management system 70 includes an SOC estimator 60 estimating a state of charge.

The battery management system 70 measures the charge / discharge current, voltage, and temperature of the lithium ion battery at each sample time through a current sensor, a voltage sensor, and a temperature sensor, and selects an arbitrary operating point based on a previously generated multi-model reference point. The corresponding model Z (k) is estimated using a model of four reference operating points that herpes around this operating point. The open circuit voltage (OCV) of the lithium ion battery is calculated from the estimated model Z (k), the output voltage of the voltage sensor, and the lithium ion battery current, and the SOC is estimated from the calculated open voltage.

The SOC estimation method according to the present invention is divided into i) an open circuit test step and ii) a real-time SOC estimation step.

In the open circuit test step, output characteristics are different for each manufacturer or model that manufactures a lithium ion battery, and thus, this is for acquiring a characteristic model used in a real-time estimation step. The open circuit test step may include: generating a multi-model reference point for displaying temperature and current data in a matrix form within an allowable operating range of a lithium ion battery; An equivalent circuit displaying step of displaying the lithium ion battery as an equivalent circuit of series impedance with OCV in the form of a dependent power source; A modeling area dividing step of dividing the SOC curve corresponding to the open voltage of the lithium ion battery into a section represented by a linear dynamic model and a section represented by a nonlinear static model; Acquiring terminal voltage, OCV, and charge / discharge current data of the battery through an open circuit test to identify the linear transfer function model in the linear region and the nonlinear model in the nonlinear region at the temperature-current operating point, respectively; Converting the identified continuum model set into a discrete value model in consideration of a sampling time for application to a real-time multi-model SOC estimator; As shown in FIG. 4, the method includes constructing one model bank with four impedance transfer functions corresponding to four adjacent operating points.

In the real-time SOC estimation step, the measurement step of measuring the terminal voltage of the lithium ion battery, the temperature and current data of the lithium ion battery during each sample time during operation; Selecting a model bank corresponding to an operating point represented by temperature and current data of a lithium ion battery measured at each sample time in the measuring step: adapting a multi-model using four discrete value impedance models of the selected model bank Estimating OCV and SOC at the operating point by multiple model adaptive estimation (MMAE); If the value of the temperature-current measured at each sample time in accordance with the change in the operating conditions includes the step of selecting the model bank including this operating point and performing the above process.

Now, the detailed principle and method of each step will be described.

(1) Electrical equivalent circuit model expressed as a function of operating point of lithium ion battery

The equivalent model of FIG. 2 belongs to a representative model for SOC estimation, but the difference from the conventional method is that the equivalent impedance is expressed as a function of temperature and current. In the conventional method, a model in which Z (s) is a series circuit of R ₁ and {R ₂ C parallel} and constant R and C values is mainly used. In some cases, a structure in which three to four RC parallel circuits are connected in series is also shown, but both are represented by constant R and C parameter values. The equivalent impedance model of FIG. 2 is advantageous when displaying multiple models corresponding to several operating points.

Real-time adaptive estimating techniques are required to represent the dynamic characteristics of lithium ion batteries that vary according to the operating point in the whole operating range. If the variation is not rapid, multi-model adaptation is simpler and more powerful. While the conventional adaptive estimation technique updates the model parameters by using the data acquired at every sample time, the multi-model adaptive estimation method selects a finite number of operating points within the operating range and obtains a model at this operating point beforehand. The model at the operating point of is identified by the linear combination of pre-established models near this operating point.

Traditional adaptive techniques can be very sensitive to input and output data at every sample time and must be carefully implemented to avoid sensor noise or transient data errors, while multi-model techniques are much simpler because they are linear combinations of models already available. And sound.

것이다. 도 4는 온도를 7개, 전류를 4개로 나누어 총 28 개의 동작점을 얻은 예를 보여주고 있다. 동작점의 수는 모델링 과정에서는 충분히 나누고 각 동작점에서 모델을 식별한 후 그 모델들의 파라미터를 비교하여 모델뱅크를 구성하는 단계에서 축소하는 것이 바람직하다.4 is an example showing a method of setting a finite integer number of reference operating points within an allowable operating range in the step of identifying an electrical equivalent model of a lithium ion battery according to the present invention. In Fig. 2, since the parameters of the equivalent impedance model can be expressed as a function of temperature and current, the operating point is expressed as (temperature, current), the temperature is plotted on the vertical axis, the current is arranged on the horizontal axis, and the intersection is divided by the appropriate number. One reference point

will be. 4 shows an example in which 28 operating points were obtained by dividing the temperature into seven and the current into four. The number of operating points is preferably divided in the modeling process, the model is identified at each operating point, and then reduced in the step of constructing the model bank by comparing the parameters of the models.

For example, in FIG. 4, the operating point where i is 2 and j is 4 represents an operating point of 10 ° C. and a current of 10 A of a lithium ion battery. In FIG. 4, an equivalent impedance model corresponding to each operating point may be expressed as Equation 1, wherein Z _{i and j} (S) have a charge / discharge current Ibat I _i, and the temperature of the battery ( When T) is T _j , it is an impedance, and is expressed as a first or second transfer function in the A and B areas.

<식 1>

<Equation 1>

Estimation of the state of charge of the lithium ion battery according to the present invention is performed in a model in which the lithium ion battery is represented by an electrical equivalent circuit as shown in FIG. 2, where the background principle will be described. In FIG. 2, the capacitor Cc represents the charge capacity of the battery and is defined as in Equation 2.

<식 2>

<Equation 2>

은 충방전 사이클 수에 따라 감쇠특성을 나타내는 인수이다. Where Capacity is the nominal capacity of the lithium ion battery,

Is a factor representing the attenuation characteristic according to the number of charge and discharge cycles.

When SOC of a lithium ion battery is represented by a voltage, it can be represented by Formula 3.

<식 3>

<Equation 3>

Where V _S (0) is the initial voltage at the time of charging or discharging. The value is 4.2V if one Li-ion battery cell is fully charged (SOC 100%), and 3.2V if SOC is 0%. The voltage V _OC at the time of opening of the lithium ion battery cell may be expressed as a voltage controlled voltage source, V _OC (V _S ). V _t (t) and I _bat (t) are terminal voltage and charge / discharge current, respectively. Z (I, T, s) is the equivalent circuit model to be identified at the operating point of temperature and current. In the present invention, the cell self discharge effect is ignored.

Figure 3 shows the OCV (Voc) curve for typical SOC of a lithium ion battery. This SOC curve can be expressed as in Equation 4.

<식 4>

<Equation 4>

는 추정된 개방전압 값,

와

는 식 4의 파라미터이고 전지의 정전류 충/방전 실험데이터로부터 식별한다. 이 때 식 4의 우변 제1항(지수 함수 항)의 계수

는 도 3의 C영역 데이터를 이용하여 식별하고 식 4 우변의 계수

는 도 3의 영역 A와 B의 데이터만을 이용하고 잘 알려진 최소자승오차법(least squared error method)을 이용하여 식별한다.In Equation 4,

Is the estimated open voltage value,

Wow

Is the parameter of equation 4 and is identified from the constant current charge / discharge experimental data of the battery. In this case, the coefficient of the right side first term (exponential function term) of Equation 4

Is identified using the C region data of FIG.

Uses only the data in areas A and B of FIG. 3 and identifies them using the well-known least squared error method.

From Equation 4, if OCV (Voc) is estimated, it can be seen that the corresponding SOC can be determined. The parameter of Equation 4 can be easily obtained by applying the least-squares method after acquiring OCV (Voc) data while discharging or charging a constant current in a battery.

(2) Classification of linear and nonlinear operating regions according to SOC levels of lithium ion batteries

As well as lithium-ion batteries, most secondary batteries have intrinsic characteristics and strong non-linear characteristics that can express the equivalent impedance as a linear transfer function according to SOC level when the relationship between internal electromotive force and terminal voltage is expressed in the model of FIG. There is a section.

 For example, FIG. 3 shows that A and B regions where linear motion occurs and C regions which are non-linear sections. These characteristics may vary depending on the make and model of the battery. Regions A and B are both linear operating intervals, but are classified to consider cases where the relationship between VOC-SOC is different. After identifying the models in this manner, if the parameters of the model identified in the step of constructing the model bank are similar, the classification of the regions A and B can be reduced. In other words, the three regions of FIG. 3 mean an SOC section that can be represented as a unique dynamic model, and the equivalent impedance modeling of FIG. 2 should also be made in consideration of the operating point independently in this region. According to the present invention, two linear sections (A, B) and one non-linear section (C) are divided based on the curve form of OCV and SOC of the lithium ion battery of FIG. 3. For example, each section of a lithium ion battery model to which the present invention is applied is shown as follows.

Zone A: 3.85 ≤ Voc ≤ 4.06

Area B: 3.62 ≤ Voc ≤ 3.85

Zone C: 3.23 ≤ Voc ≤ 3.62

The internal impedance of the lithium ion battery is represented by N models (N = 28 in this example) corresponding to the operating point of FIG. 4 in the three regions of FIG. 3 of the SOC. The voltage drop between V _OC and the terminal voltage in region A and region B can be modeled as a linear impedance function, but region C cannot be modeled as a linear impedance function because of its strong nonlinearity. Therefore, modeling is divided into each area.

(3) Equivalent Impedance Model of Li-ion Battery

As a result of the present invention, Z (I, T, s) at all specific operating points of region A and region B can be identified as a first or second continuous system transfer function. The equivalent circuit model Z (I, T, S) is defined as

<식 5>

<Equation 5>

에 대한 3차(또는 2차) 다항식으로 모델링될 수 있다. 특히, 영역 C에서는 전지의 비가역성이 무시할 수 없을 만큼 나타나기 때문에 영역 A, B와 달리 식 6의 모델은 충전 모드와 방전 모드를 별개로 하여 구하여야 한다.On the other hand, the equivalent impedance model Z (I, T, Vt) in the C region with the nonlinear static model of FIG.

Can be modeled as a cubic (or quadratic) polynomial for. In particular, since the irreversibility of the battery is not negligible in the region C, unlike the regions A and B, the model of Equation 6 should be obtained separately from the charging mode and the discharging mode.

<식 6>

<Equation 6>

Here, model parameters of Equations 5 and 6 are identified using sampling data of lithium ion battery current I _bat , terminal voltage V _t , and V _OC . Input-output data used for parameter identification can be obtained by applying a test current in the form of pseudo random binary sequence (PRBS) to a lithium ion battery. While maintaining the lithium ion battery at the temperature corresponding to the operating point in the thermo-hygrostat, the current of the operating point size is applied in the PRBS form to obtain the data of V _t and V _OC . The input / output transfer function model can be expressed by Equation 7.

<식 7>

<Equation 7>

Once the data is obtained, parameter identification in Equations 5 and 6 can be obtained in a variety of ways. 10 shows a PRBS test input (charge / discharge current) applied to perform a modeling process for a lithium ion battery of a specific company, and FIG. 11 shows the terminal voltage V _t and the open voltage V _OC , which are output responses. The model parameters thus obtained are shown in Tables 1 to 3.

 Table 1 shows the model parameters in Equation 5 identified at 10A current operating point and seven other temperature operating points in area B. As a result of the present invention, even if a model of the second order or higher is obtained, the error improvement effect is not large as in the case of the first model, and thus it is sufficient to identify the first order.

Model parameters identified at 10 A current operating point and seven other temperature operating points in area B T ℃ a _One a ₀ b _One * e-3 b ₀ * e-3 -20 One 0.0883 10.44 1.248 -10 One 0.2995 6.173 2.478 0 One 0.5803 3.782 3.033 10 One 0.8977 2.376 3.089 24 One 1.351 1.266 2.735 40 One 1.61 0.7225 2.138 60 One 1.588 0.3945 1.435

 Table 2 shows the model parameters of Equation 5 obtained at five current operating points while maintaining the battery cell temperature at -20 ° C.

Model parameters obtained at five current operating points when battery cell temperature is -20 ° C. Current a _One a ₀ b _One * e-3 b ₀ * e-3 1A One 0.099 11.36 1.45 5A One 0.08601 10.91 1.245 10A One 0.0883 10.44 1.248 20A One 0.1746 8.277 1.986 40A One 0.855 6.772 6.772

로 표기하기로 한다.Table 3 shows the polynomial model parameters of Equation 6 obtained in the charge and discharge modes at the 10 A current operating point and the seven temperature operating points in area C, respectively. And subsequently identified model parameters

It is written as.

Polynomial model parameters obtained in charge and discharge modes at 10 A current operating point and 7 temperature operating points in area C, respectively. T ℃ Coefficient of <Equation 6> in the charging mode Coefficient of <Equation 6> in discharge mode c ₃ c ₂ c _One c ₀ c ₂ c _One c ₀ -20 -0.9257 9.4464 -32.0472 36.1577 0.0015 -0.0834 0.2857 -10 -1.116 11.3527 -38.4096 43.2303 0.0011 -0.0811 0.2811 0 -1.1511 11.6293 -39.0776 44.6823 0.001 -0.0807 0.2798 10 -1.182 11.8893 -39.7776 44.2721 0.0009 -0.0806 0.2792 24 -1.1954 11.9766 -39.9105 44.242 0.0009 -0.0805 0.2787 40 -1.1889 11.8792 -39.478 43.641 0.0009 -0.0805 0.2786 60 -1.1809 11.7792 -39.0798 43.1265 0.0009 -0.0805 0.2785

(4) Convert to discrete value model of continuous impedance model

In the three areas A, B, and C of FIG. 3, the modeling process through the open loop test is performed to obtain a model set corresponding to the number of operating points (28 model sets in this example). Since the battery management system 70 of FIG. 1 that executes the SOC estimation is generally implemented in a digital system, the SOC estimation also needs to be represented by a digital algorithm.

로 치환하여 구한다.Therefore, it is necessary to convert the continuous system model into a discrete value model. Although the discretization model may be directly obtained in the modeling process, it is advantageous to obtain a continuous model that does not need to be remodeled even if the sampling time changes because the sampling time may vary depending on the battery and the BMS. When the sampling time of the BMS temperature, current, and terminal voltage signal is T _s , the discrete value model Equation 8 of Equation 5 can be obtained by using the Tustin approximation method. In other words, the complex variable s in Equation 5

It is obtained by substituting for.

<식 8>

<Equation 8>

The discrete value model of the equivalent impedance of the C region is shown in Equation 9 by substituting the terminal voltage for a sample value from Equation 6.

<식 9>

<Equation 9>

일 때, 등가임피던스 양단의 전압강하는 다음 식 10으로 계산되고

일 때는 식 11로 계산된다.If the charge or discharge current flowing through the battery is measured at every sample time

, The voltage drop across the equivalent impedance is calculated by

Is calculated by Equation 11.

<식 10>

<Equation 10>

<식 11>

<Equation 11>

(5) Definition and construction of model banks for MMAE

In order to accurately estimate the OCV, the MMAE technique is applied to the electrical equivalent circuit model of the lithium ion battery because the model parameters change according to the values of temperature and current defined as operating points. In the MMAE, the model corresponding to each operating point is identified in advance, and a model set (112 model sets of 28 models in each region in the present embodiment) is obtained. Will be estimated.

:온도 T_z, 전류 I_batz)을 둘러싸는 4개의 기준동작점 모델 셋 {Z_{i, j} , Z_{i, j +1}, Z_{i+1, j} , Z_{i+1, j +1}}을 모델뱅크 B_{i, j} 라 정의 한다. 이 정의에 의하면 모델뱅크 B_{i+1, j+1} 은 4개의 모델 기준점들의 임피던스 모델 셋{Z_{i+1, j+1} , Z_{i+1, j +2}, Z_{i+2, j+1} , Z_{i+2, j +2}}를 나타낸다. 본 발명에서는 MMAE에 전체 모델 셋 중에서 임의의 동작점을 포함하는 하나의 모델뱅크가 선택되고 이 모델뱅크에 포함된 4 개의 모델 ( 동작범위 가장자리 부근에서는 2개 또는 1개의 모델 셋)에 의해 OCV를 추정하는데 사용한다. 예를 들어, 도 4에서 동작점

(온도 7 ℃, 전류 7 A)에서 동작점

(온도 18 ℃, 전류 17 A)로 변하였다고 하면 모델뱅크는

에서

가 된다. In this step, it is a question of which finite number of models are used for the MMAE estimation. In the present invention, as shown in FIG.

Four reference operating point models (Z _{i, j} , Z _{i, j +1} , Z _{i + 1, j} , Z _{i + 1, j +1} } that surround the temperature T _z , current I _batz ) Defined as banks B _{i and j} . According to this definition, model banks B _{i + 1 and j + 1} are impedance model sets of four model reference points {Z _{i + 1, j + 1} , Z _{i + 1, j +2} , Z _{i + 2, j + 1} , Z _{i + 2, j +2} }. In the present invention, one model bank including an arbitrary operating point is selected in the MMAE, and the OCV is selected by four models (two or one model set near the edge of the operating range) included in the model bank. Used to estimate For example, the operating point in FIG.

Operating point at (temperature 7 ℃, current 7A)

If the temperature is changed to 18 ° C and 17A, the model bank

in

Becomes

(6) Open Voltage (OCV) Estimation Algorithm Using Multi-Model Adaptive Estimation

를 추정하는 알고리즘이다. 기본 가정은 하나의 모델뱅크 영역에서는 A,B 영역 뿐 만 아니라 C영역에서도 선형적이라고 보는 것이다. 이러한 가정아래 리튬이온전지의 OCV추정알고즘을 한 예로서 현재 동작점이 도 5의

에 있다고 하면 이 동작점에 대응하는 등가임피던스의 전압강하(V_z)의 추정값은 다음 식 11처럼 모델뱅크

에 속하는 4개의 모델의 전압강하의 선형결합(linear combination)으로 나타낼 수 있다. Given the multiple model sets and model banks defined above, and the current values of temperature and current from the BMS are measured and entered, the remaining problem is the open circuit (OCV) from the four impedance models of each model bank by MMAE.

Is an algorithm for estimating. The basic assumption is that one model bank is linear in C as well as A and B. Under this assumption, the current operating point of the OCV estimation algorithm of the lithium ion battery is shown in FIG.

If is, the estimated value of the voltage drop (V _z ) of the equivalent impedance corresponding to this operating point is model bank as

It can be expressed as a linear combination of voltage drops of four models belonging to.

<Equation 11>

here

<식 12>

<Equation 12>

<식 13>

<Equation 13>

는 온도에 대한 가중인수(weighting factor)이고,

는 전류에 대한 가중인수이다.

와

의 선택은 인접한 두 기준 동작점 사이에서의 상대적 거리에 비례해서 결정한다. 예를 들어, 온도 동작점 T_z = 7℃면, 인접 기준 동작점 온도가 0℃ 와 10 ℃이므로

로 선택된다. 마찬가지로 전류 동작점

이면, 인접 기준동작점 전류가 1 A 와 10 A이므로

로 선택된다. 식 11에서 계수

는

와

값으로부터 계산된다. 식 11로 현재 동작점에서의 전압강하가 추정되었다면 개방전압의 추정값

은 도 2로부터 식 14에 의해 결정된다.

Is the weighting factor for temperature,

Is the weighted number of currents.

Wow

The choice of is determined proportionally to the relative distance between two adjacent reference operating points. For example, if the temperature operating point T _z = 7 ° C, the adjacent reference operating point temperatures are 0 ° C and 10 ° C.

Is selected. Similarly current operating point

If the adjacent reference operating point current is 1 A and 10 A

Is selected. Coefficient in Equation 11

Is

Wow

It is calculated from the value. If the voltage drop at the current operating point is estimated by Eq. 11, then the estimated open voltage

Is determined by equation (14) from FIG.

<식 14>

<Equation 14>

는 전지의 단자 간 전압으로 매 샘플시간 마다 측정되는 값이고

는 식 11에 의해 계산된 등가임피던스의 전압강하이다.here

Is the voltage between terminals of the battery and is measured every sample time

Is the voltage drop of the equivalent impedance calculated by Eq.

12 is a block diagram showing an open circuit voltage (OCV) estimation algorithm using the multi-model adaptive estimation technique.

In conclusion, since the open-circuit voltage Voc of the lithium ion battery is estimated, the SOC can be estimated from the equation for calculating the SOC of Equation 4, or the SOC is calculated using the SOC experimental curve corresponding to the open voltage of the lithium ion battery of FIG. It can also be estimated.

FIG. 6 shows how the parameters of the equivalent impedance model identified for the temperature-current permissible operating range of a specific lithium ion battery are different at -20 to 60 ° C. at 1/4 C-rate, and FIG. Shows the model parameter value according to C-rate at -20 ℃. This demonstrates that the model obtained at only one operating point may not be sufficient when estimating the SOC using the equivalent circuit model of FIG. 2. On the other hand, the method proposed in the present invention can solve this problem because it considers the whole operating area divided into several operating points. FIG. 8 shows the measured value Voc of OCV and four adjacent operating points (1A, 0 ° C) when charged to 8 A for 40 minutes and then discharged to 8 A for 50 minutes while maintaining the temperature of LIB at 7 ° C. (1A , 10 ℃), (10A, 0 ℃) and (10A, 10 ℃) show the results of estimating OCV by impedance model together to show this effect more clearly.

In addition, FIG. 9 shows the measured value of OCV when charging and discharging current of 18 A was carried out while charging at 8 A while maintaining the temperature of LIB at 7 degreeC until 140 minutes, and maintaining the temperature of LIB at 20 degreeC after 140 minutes. And OCV values estimated by the method proposed in the present invention. Even when the C-rate and the temperature were changed, the maximum estimation error was 5.3 mV, showing the precision that it was less than 2%.

In the present invention, once the equivalent model of a different lithium ion battery (LIB) is obtained for each manufacturer, there is no need to obtain the equivalent model again afterwards, and the calculation is simple since there are always four models used at every sample time. And easy to implement.

Claims (9)

Create reference operating point based on temperature and current data of lithium ion battery, SOC curve divided according to SOC size is displayed as linear dynamic model, A and B area with high SOC area and middle SOC area and low SOC area An open circuit test step of identifying equivalent circuit models corresponding to the reference operating points, respectively, in a section divided into a C region represented by a nonlinear static model having a?
After the open circuit test step, up to four models around the operating point are defined as model banks from the equivalent circuit model set, and the terminal voltage, current, and temperature data of the lithium ion battery are periodically input during operation. The model bank was selected to include the operating point, and the open-circuit voltage of the lithium-ion battery was calculated and estimated by substituting the open-circuit voltage (OCV) estimation algorithm that invented the terminal voltage and current data measured at each sample time. Method of estimating the state of charge of a real-time lithium-ion battery to estimate the SOC by using a. The method of claim 1, wherein the open circuit test step,
The electrical dynamic characteristics model of the lithium ion battery is represented by the electromotive force in the form of a subordinate power supply (equivalent to the OVC) and the equivalent circuit of the series impedance, and the temperature and current operating points can be treated as multiple models corresponding to the multiple operating points. A function expression step of expressing a function;
A reference operating point generation step of dividing a temperature and current operating point into an appropriate number within an allowable range of operating conditions of the lithium ion battery and displaying temperature and current data in a matrix form;
The SOC curve corresponding to the open voltage of the Li-ion battery is divided into two sections (A and B regions), a high SOC region and an intermediate SOC region represented by a linear dynamic model, and a nonlinear static model. A modeling area partitioning step of dividing into one section (region C) which is the displayed low SOC region;
The equivalent impedance of the linear section corresponding to the temperature and current operating points divided into finite pieces for each of the linear and nonlinear regions is represented by a second or first order linear transfer function model, and the equivalent impedance of the nonlinear section is the third order with respect to the open voltage. Or identifying an equivalent impedance model represented by a second order polynomial;
In order to identify the parameters of the equivalent impedance model, an appropriate charge / discharge current is applied to the battery in the form of pseudo binary random number (PRBS) for each region, the battery's open voltage (OCV) and terminal voltage are measured, and these data and identification algorithm are used. An equivalent impedance model parameter identification step of obtaining a continuous system equivalent impedance model;
In each region of the modeling domain division step, a modeling process is performed through the open circuit test to obtain a set of continuous system models corresponding to the number of operating points considered, and the continuous system model is considered in consideration of a given sampling time to estimate a real-time SOC. A model conversion step of converting to a discrete value model; And
And an open voltage estimating step of estimating an open voltage based on a model bank which is a model assembly around an arbitrary operating point from a model set already identified in the open circuit test. The method of claim 1, wherein the real-time SOC estimation step,
A measurement step of measuring terminal voltage of the lithium ion battery, temperature and current data of the lithium ion battery during each operation of the electric vehicle;
In the measuring step, the temperature and current data measured at each sample time are recognized as the operating point at the present time, one model bank surrounding the operating point is selected, and an open voltage is obtained using four models included in the model bank. An OCV estimating step of estimating (OCV);
A voltage drop estimating step of calculating an estimated value of the voltage drop V _z of the equivalent impedance corresponding to the operating point at the present time through the OCV estimating step as a linear combination of voltage drops of four models belonging to a model bank; And
And a SOC estimating step of determining an estimate of the open voltage using the voltage drop of the equivalent impedance at the current operating point calculated in the step of calculating the voltage drop estimate and estimating the SOC from the open voltage value. How to estimate the state of charge. The method of claim 3, wherein the OCV estimating step defines four models corresponding to the vertices of the small quadrangle in the temperature and current operating point matrix from the calculated discrete value equivalent impedance model as one model bank, and an arbitrary operating point. A method of estimating the state of charge of a lithium ion battery, characterized in that only these model banks are used among the set of models when estimating the voltage drop at the equivalent impedance at. The equivalent impedance model Z (I, T, S) in the regions A and B having the linear dynamic model in the open circuit test step is the first or second order according to claim 1 Method of estimating the state of charge of a lithium ion battery, characterized in that defined by the transfer function of.

<Equation 1>
In Equation 1, I is the charge and discharge current of the lithium ion battery, T is the temperature of the lithium ion battery,

Wow

Is the parameter identified from the constant current charge / discharge experimental data of the battery. The equivalent impedance model Z (I, T, Vt) in the region C having the nonlinear dynamic model in the open circuit test step is given by Equation 2 below.

Method of estimating the state of charge of a lithium ion battery, characterized in that defined as the second or third polynomial of.

<Equation 2>
In Equation 2, I is the charge and discharge current of the lithium ion battery, T is the temperature of the lithium ion battery,

Is the parameter of the polynomial. The method of claim 2, wherein the step of identifying the equivalent impedance model parameter is applied to the battery in the form of pseudo binary random number (PRBS) while maintaining the respective operating point temperature and measuring the open voltage and the terminal voltage in response. Equivalent model parameter identification method of a lithium ion battery, characterized in that to obtain a continuous system model parameter. The method of claim 1 or 3, wherein the open voltage value

Method for estimating the state of charge of a lithium ion battery, characterized in that the calculation of SOC is performed by the following <Equation 3>.

<Equation 3>
In <Equation 3>,

Is the estimated open voltage value,

Wow

Is identified from the cell's constant current charge / discharge experimental data, where

Identifies using the C area data,

Identifies using only data in areas A and B. A load 20 and a charger 30 operating by receiving power from the lithium ion battery 10;
A temperature sensor 40 measuring a temperature T of the lithium ion battery 10;
A current sensor 45 measuring charge and discharge current Ibat of the lithium ion battery 10;
A voltage sensor 50 measuring a voltage Vt between the cathode and the anode terminals of the lithium ion battery; And
The charging and discharging current, voltage and temperature data of the lithium ion battery are measured at each sample time through the current sensor, the voltage sensor, and the temperature sensor to define temperature and current values as operating points at the present point of time. Select up to four reference operating point models as model banks and use the estimated algorithm for this model bank to determine the voltage drop of the equivalent impedance model of the current operating point and the open voltage of the battery.

The system for estimating the state of charge of a lithium ion battery comprising a battery management system (70) comprising a SOC estimator (60) for calculating and estimating SOC from the calculated open circuit voltage.

Images