WO2018025350A1 - Estimation device, estimation program, and charging control device - Google Patents

Abstract

The estimation device according to the present invention is characterized by being provided with a storage unit for storing a plurality of model functions of the charging rate and the open-circuit voltage of a chargeable cell, a determination unit for determining a model function from the plurality of model functions in accordance with an electric current status of the cell, a calculation unit for estimating the charging rate and a predicted terminal voltage of the cell through use of a Kalman filter using the model function determined by the determination unit and calculating the difference between the predicted terminal voltage and an actual measured terminal voltage of the cell, and a correction unit for correcting the charging rate on the basis of the calculated difference and the Kalman gain of the Kalman filter.

Description

Estimation device, estimation program, and charge control device

This case relates to an estimation device, an estimation program, and a charge control device.

Secondary batteries such as lithium ion batteries are attracting attention as power storage applications such as electric mobility (electric vehicles, etc.) and stationary power storage systems. In electric mobility applications, a technique for obtaining a charge rate SOC in order to display the remaining travel distance to the driver is desired. Even in a stationary power storage system, obtaining an accurate SOC is important for accurate system control.

The reason why accurate control is desired is to prevent overcharge and overdischarge. When only an inaccurate SOC can be obtained, a large charge / discharge margin must be taken, and it becomes difficult to exhibit the battery's original power storage performance. This increases the number of batteries, leading to an increase in system cost.

The characteristics of the secondary battery are affected by the current status immediately before the secondary battery. In view of this, a technique that considers the current state immediately before the secondary power value in estimating the characteristics of the secondary battery has been disclosed (see, for example, Patent Documents 1 to 3).

International Publication No. 2014-080595 JP2011-151983 JP 2013-105519 A

However, in the above technique, since the SOC is directly obtained from the voltage, there is a possibility that high estimation accuracy cannot be obtained.

An object of one aspect is to provide an estimation device, an estimation program, and a charge control device that can estimate SOC with high estimation accuracy.

In one aspect, the estimation device stores a plurality of model functions of an open-circuit voltage and a charging rate of a rechargeable battery, and calculates a model function from the plurality of model functions according to the current state of the battery. The charging unit and the predicted terminal voltage of the battery are estimated by a determining unit to be determined and a Kalman filter using the model function determined by the determining unit, and a difference between the predicted terminal voltage and the measured value of the terminal voltage of the battery is calculated. A calculation unit for calculating, and a correction unit for correcting the charging rate based on the difference and a Kalman gain of the Kalman filter.

SOC can be estimated with high estimation accuracy.

(A) And (b) is a figure which illustrates the result of having measured the terminal voltage change of the secondary battery experimentally for a long time after stopping an electric current. It is a figure which illustrates the typical method of calculating | requiring an OCV-SOC characteristic curve. (A) And (b) is a figure which illustrates the general method of calculating | requiring an OCV-SOC characteristic curve. (A) And (b) is a figure which illustrates the conventional method of calculating | requiring an OCV-SOC characteristic curve. 1 is a block diagram of an estimation device according to Embodiment 1. FIG. It is a figure which illustrates the equivalent electrical circuit model of a secondary battery. (A) illustrates the terminal voltage-SOC curve and OCV-SOC characteristic curve measured by changing the current, and (b) and (c) are diagrams illustrating how to obtain the OCV when SOC = 50%. . It is a figure which illustrates an OCV-SOC characteristic model function. It is a figure which illustrates the flowchart performed when a parameter determination part determines an OCV-SOC characteristic model function. It is a figure which illustrates about the detail of a Kalman filter. (A) And (b) is a figure which illustrates the result of having determined the sequential charging / discharging tendency. (A) is the case where the reset interval is 100 cycles (10 s), the determination threshold is 3E + 4 mA, the reset multiplier is 0.3, and the charge / discharge tendency is determined, (b) is the reset interval 1000 cycles (100 s), and the determination threshold is Although it is 3E + 5 mA, it is a case where the charge / discharge tendency is determined by resetting the integrated current to zero. It is a figure which illustrates the SOC estimation precision by a Kalman filter when a single OCV-SOC characteristic model function is used and when two OCV-SOC characteristic model functions are switched according to the present embodiment. It is a block diagram for demonstrating an example of the hardware constitutions of an estimation apparatus. It is a figure which illustrates about the estimation system concerning a modification.

Prior to the description of the examples, the characteristics of the rechargeable secondary battery will be described. Since the characteristics of the secondary battery are affected by the current status of the secondary battery (direction, magnitude, time, etc. of the current flowing in the secondary battery), the regularity of the characteristics of the secondary battery is complicated. Therefore, it is difficult to completely reproduce the characteristics of the secondary battery. For this reason, when a single battery characteristic determined by measurement and evaluation under a specific condition is applied to a charging state estimation method, the battery characteristic is often not appropriate as a characteristic of a battery in various usage situations.

There are two battery characteristics that are generally used for estimation: dynamic characteristics when current is flowing and static characteristics when current is not flowing. Of these, as the static characteristic, a characteristic (hereinafter referred to as OCV-SOC characteristic) between the terminal open circuit voltage (OCV: Open Circuit Voltage) and the state of charge (SOC: State of Charge) is often used. In addition, since the OCV-SOC characteristic is essentially a static characteristic and is considered to be a characteristic unrelated to the usage situation of the secondary battery, a single characteristic has been used regardless of the usage situation. In other words, the battery voltage should converge toward a specific voltage with respect to a specific SOC if a long time has passed since the current stopped flowing under any usage conditions. It is.

FIG. 1A is a diagram illustrating the result of experimentally measuring the voltage change of the secondary battery for a long time after the use of the secondary current is stopped. From the results shown in FIG. 1A, it is apparent that the convergence voltage changes depending on the use situation before stopping the current in a realistic time, and at least the OCV value determined by a general method that can ignore the current situation. Was found not to converge. This indicates that when a voltage value based on a single OCV-SOC characteristic is used for SOC estimation by a Kalman filter, an estimation error occurs due to the current state of the secondary battery.

On the other hand, the OCV-SOC characteristic does not change depending on the current state, but there is also an idea that the dynamic characteristic changes depending on the current state. That is, for example, although the final voltage of the voltage change after the current is stopped does not change, the time to reach the voltage is very long. Based on this idea, there is a conventional technique in which a parameter of dynamic characteristics (for example, an equivalent circuit model used for a Kalman filter), particularly a parameter responsible for a change having a long time constant, is changed depending on the use situation. However, in the case of this method, the value of the parameter changes very greatly, so it is difficult to deal with the characteristics before and after the change only by changing the same parameter. Therefore, if it is an RC equivalent circuit, it will be necessary to add the new RC circuit which bears a very long time constant. This leads to an increase in the amount of calculation because the parameter increases. In particular, in the estimation method using the Kalman filter, since the matrix calculation is performed in the Kalman filter, a large increase in calculation amount occurs due to an increase in parameters. Furthermore, it is considered very difficult to find regularity that changes the time constant parameter according to the current state.

Further, as illustrated in FIG. 1B as a voltage change that converges to the OCV determined by the general method after the usage situation A, the voltage change that can be expressed by the RC equivalent circuit is monotonous. If the final voltage of the voltage change does not change as described above, the voltage change may not be monotonous depending on the current state (after use state B in the figure). To cope with this, it is possible to use a more complex equivalent circuit model in the Kalman filter instead of a simple RC circuit, but avoid the increase in the calculation amount as the calculation becomes complicated (increase in parameters) as described above. I can't.

Here, as an example, a mechanism in which the OCV changes depending on the current state in a lithium ion battery will be described. The battery terminal voltage when current is flowing differs from the open circuit voltage (OCV) simply because there is an apparent internal resistance. When a current flows through the internal resistance, a voltage increase or a voltage drop occurs, and a difference from the OCV occurs. The reason why the terminal voltage changes transiently after the current changes is because the internal resistance apparently changes over time. In order to reproduce this time change, RC circuits are often used in equivalent circuit models.

Here, it can be said that the apparent internal resistance is caused by having the following finite value resulting from the electrochemical reaction inside the battery.
- present in the electrode reaction rate, utilization of reaction rate and negative electrode active material of Li ⁺ moving speed electrode inside the ion conduction velocity and the positive electrode active material through the movement speed-active material interface Li ⁺ ions in the electrolyte solution Electronic conduction velocity of substance itself ・ Ionic conduction velocity in electrolyte

Here, simply, the smaller the speed, the larger the internal resistance. However, it is imagined that the speed is not so high that it reacts instantaneously like an electronic circuit. Further, when any of the above speeds changes with time, the internal resistance changes with time. For example, Li ⁺ ions are taken into the active material of the electrode by charging / discharging, but this reaction occurs immediately after the Li ⁺ ions reach the active material near the surface, but when the surface uptake site is buried, Uptake occurs deep within the active material. At this time, since Li ⁺ ion movement through the active material interface occurs, it takes time until the uptake is completed after Li ⁺ ions reach the active material. In addition, it is conceivable that Li ⁺ ions are also trapped at the active material interface, and thereby the moving speed is gradually reduced. It takes a lot of time to reach a steady state due to a decrease in the moving speed.

Furthermore, in the process in which ions are conducted in the electrolyte and move between the electrodes, time for movement is required. Further, it is known that when ions are conducted in the electrolytic solution, the concentration of positive and negative ions is biased and a concentration distribution is generated. Although this concentration distribution changes with time, it is conceivable that the ion conduction velocity in the electrolytic solution changes due to the concentration distribution.

The above example also shows that it takes time for the internal state of the battery to reach a steady state, and that the change changes with time. In addition to this, it is easily estimated that the above speed or the change with time, in other words, the internal resistance and the change with time are influenced by the internal state of the battery at the time when the current is switched. The apparent internal resistance that determines the terminal voltage is determined by a combination of the plurality of speeds. In addition, since these speeds are determined by the respective states, it is necessary to determine all the states at the time when the current is switched in the model that reproduces the terminal voltage change. However, it is very difficult. Furthermore, it is very difficult to incorporate all of them into an equivalent circuit model strictly, especially if it is a relatively simple model that takes the amount of calculation into account.

Due to the mechanism considered above, the Kalman filter using a relatively simple equivalent circuit model and a single OCV-SOC characteristic model has a long time due to the time change of the terminal voltage, particularly the electrochemical speed inside the battery. It is difficult to reproduce changes in time with small errors. As a result, there is a considerable error in the estimation of the SOC.

Here, a method for obtaining the OCV-SOC characteristic on the premise that the OCV-SOC characteristic is single, that is, does not change depending on the current condition of the secondary battery and the current change state will be described. In the first representative method, a chargeable current capacity Cmax is first obtained in advance. As illustrated by the solid line in FIG. 2, SOC is charged from 0% to 100% at a constant current as small as possible (for example, 0.1 C when the current value that can fully charge the battery in 1 hour is 1 C). And determine the charge curve by measuring the voltage change. At this time, the product of the current and time is integrated and divided by Cmax to convert to SOC (%). Further, a discharge curve is obtained by discharging from SOC = 100% to 0% as in the case of charging (broken line in FIG. 2). The above may be carried out from discharge. Finally, a midpoint (dotted line in FIG. 2) between the charge curve and the discharge curve is obtained to obtain a single OCV curve (OCV-SOC characteristic). When creating a table, an OCV value corresponding to an appropriate SOC value is extracted and used.

Next, in the second general method, either charging or discharging is used, but here, charging will be described as an example. As illustrated in FIG. 3A, in the second method as well, the SOC is changed by charging for a predetermined current * time with a small constant current. Here, the current is turned off and the voltage is measured after a predetermined time. The above is repeated at an appropriate SOC interval from 0% to 100%, and the relationship between the measured voltage and the SOC is defined as the OCV-SOC characteristic. As described above, since it takes a very long time for the voltage to converge after the current is turned off, it is conceivable that the measured voltage does not become OCV if the waiting time after the current is turned off is short. Further, when the standby time is lengthened, there is a problem that it takes a very long time to acquire characteristics. As a possibility, even if a long waiting time (for example, 24 hours) that is considered to be realistic, it does not converge to the OCV.

In the second method, it is conceivable that even if the standby time is lengthened, it does not converge to a single OCV. Therefore, as illustrated in FIG. 3B, the second method is performed by charging and discharging, and the midpoint is obtained from the obtained two pseudo OCV-SOC curves in the same manner as the first method. In some cases, the third method is used in which this is the final OCV-SOC curve. In the first method, a current as small as possible is used. However, since the internal temperature of the battery increases as long as the current flows, a change in characteristics due to the influence of temperature is inevitable. According to the third method, this influence can be reduced. It is also possible to shorten the standby time of the second method.

Next, a fourth conventional method that has a plurality of OCV-SOC characteristics depending on the battery current situation or current change situation will be described. For example, as illustrated in FIG. 4A or 4B, a charge / discharge curve is obtained in the same manner as in the first method for obtaining a single characteristic, and OCV = Vt−R · According to I, the OCV during charging and discharging is calculated. Here, R is the internal resistance of the battery and needs to be determined in advance. By making the sign of the current at the time of charging positive and the sign of the current at the time of discharging negative, the OCV at the time of charging is smaller than the voltage of the charging curve by R / I. Conversely, the OCV at the time of discharging is the voltage of the discharging curve. Than R · I. 4A shows a case where the internal resistance R is larger than that shown in FIG. 4B, and the OCV during charging is smaller than the OCV during discharging. This is shown in Patent Document 2. On the other hand, in FIG. 4B, the OCV during charging is larger than the OCV during discharging. In any method, the tendency of charging / discharging is judged, and the SOC is directly obtained from the voltage by switching the correspondence relationship between the two voltages stored in advance and the SOC according to the above tendency.

In the first to third methods, since a single OCV-SOC characteristic is obtained, accurate OCV-SOC characteristics during charging and discharging cannot be obtained. In the fourth method, although two OCV-SOC characteristics are obtained, since the SOC is directly obtained from the voltage, it is not possible to obtain a high estimation accuracy of the SOC.

In the following embodiments, an estimation device, an estimation method, and an estimation program that can obtain high SOC estimation accuracy will be described.

FIG. 5 is a block diagram of the estimation apparatus 100 according to the first embodiment. As illustrated in FIG. 5, the estimation apparatus 100 includes a measurement unit 10, a parameter determination unit 20, a storage unit 30, a calculation unit 40, and an output unit 50. The calculation unit 40 includes a calculation unit 41 and a correction unit 42. The estimation device 100 is incorporated in, for example, a charge control device for a secondary battery. Note that the estimation device 100 may be implemented as a function of a control device such as an electric vehicle or an electric motorcycle, and may control a charging control device for a secondary battery.

The measurement unit 10 measures the current, terminal voltage, and the like of the secondary battery 200 at a predetermined sampling period. The measured current and terminal voltage are referred to as measurement current I and measurement terminal voltage V _OBS . The measurement unit 10 outputs a measurement value to the parameter determination unit 20 and the calculation unit 40 with an ammeter, a voltmeter, or the like, for example.

When the information related to the charging rate (SOC) estimated by the calculation unit 40 is input to the output unit 50 (or when there is a request from the external device 300), the information related to the SOC is output to the external device 300, for example. Output to. External device 300 controls charging / discharging of secondary battery 200 based on the estimated SOC.

The storage unit 30 stores information used for processing in the parameter determination unit 20 and the calculation unit 40. The storage unit 30 stores an OCV (Open Circuit Voltage) -SOC characteristic model function, functions and various parameters used for the Kalman filter, constituent element parameters of an equivalent electric circuit model, calculation parameters for determining them, function parameters, and the like. The OCV-SOC characteristic model function reproduces a graph showing the OCV-SOC characteristic of the secondary battery 200. Examples of various parameters of the Kalman filter include Σv indicating prediction noise, Σw indicating measurement noise, and the like.

First, when the measurement terminal voltage V _OBS and the measurement current I are input from the measurement unit 10, the parameter determination unit 20 acquires the parameters from the storage unit 30, and the constituent element parameters of the equivalent electric circuit model are determined in advance. Calculate using the following formula. In some cases, the component parameters of the equivalent electric circuit model stored in the storage unit 30 are selected.

FIG. 6 is a diagram illustrating an equivalent electric circuit model of the secondary battery 200. As illustrated in FIG. 6, the equivalent electric circuit model is an RC circuit that represents a transient voltage change with respect to a current change, and includes a power supply, a DC resistance R0, and two RC circuits (C1 and R1). , C2 and R2) are connected in series. The RC circuit R1C1 is configured by connecting a resistor R1 and a capacitor C1 in parallel. The RC circuit R2C2 is configured by connecting a resistor R2 and a capacitor C2 in parallel. The parameter determination unit 20 calculates the values of R0, R1, R2, C1, and C2 using a predetermined calculation formula. Alternatively, predetermined values of R0, R1, R2, C1, and C2 are selected.

In the equivalent electric circuit model, a voltage is generated in the power supply by the accumulated power. The voltage generated by this power supply is an open circuit voltage (OCV). The OCV of the power supply varies depending on the SOC. Further, the power source is illustrated assuming that the OCV changes between charging and discharging even if the SOC is the same. For this reason, the power supply has current sources V _{OCV_DC} (SOC) and V _{OCV_CC} (SOC) that represent potential differences OCV that change according to changes in the SOC. Here, the current source V _{OCV_DC} (SOC) represents the potential difference OCV during discharge. A current source V _{OCV_CC} (SOC) represents a potential difference OCV during charging.

The potential difference between both ends of the DC resistor R0 is v0, the potential difference between both ends of the RC circuit R1C1 is v1, and the potential difference between both ends of the RC circuit R2C2 is v2. In this case, the predicted terminal voltage V ⁻ of the secondary battery 200 expressed as the terminal voltage of the equivalent electric circuit model is obtained by using the potential difference OCV, the voltage v 0, the voltage v 1, and the voltage v 2, V ⁻ = OCV ( SOC) -v0-v1-v2.

When the measurement current I and the measurement terminal voltage V _OBS are input from the measurement unit 10, the calculation unit 41 estimates the SOC using a Kalman filter: KF (or an extended Kalman filter: EKF). In the Kalman filter, the arithmetic unit 40 obtains an OCV-SOC characteristic model function, various parameters for KF, etc. from the storage unit 30 and uses each parameter of the equivalent electric circuit model input from the parameter determination unit 20 to calculate the SOC. Perform estimation processing. For each parameter of the equivalent electric circuit model, the parameter determination unit 20 may not use the parameter determination and may obtain and use a predetermined parameter stored in the storage unit 30 in some cases. Note that for each KF calculation step, measurement values are input and parameters are input and determined. Here, the determination of the parameter includes a determination that the parameter of the previous step is used.

Next, the OCV-SOC characteristic model function stored in the storage unit 30 will be described. In this embodiment, an OCV-SOC characteristic model that can reduce an error from an actual characteristic by taking into account a change in characteristics due to the operation of the secondary battery 200 is determined in advance.

Specifically, as illustrated in FIG. 7A, the charge / discharge curve is measured at a constant current with the current changed (3 levels or more), and the OCV-SOC characteristics are determined from the measured current dependence of the charge / discharge curve. To do. FIG. 7B and FIG. 7C illustrate how to obtain the OCV when SOC = 50% (0.5). In each case, a regression curve is obtained from the measured current-terminal voltage data, and the voltage at the intersection of the curve and the 0 A (zero ampere) axis is defined as OCV. Usually, since this current-voltage relationship is not linear, it is desirable to use a curve of a quadratic or higher formula. This is performed a plurality of times while changing the SOC to determine the OCV-SOC characteristic. Thereby, the voltage (OCV) of the current 0A can be obtained with high accuracy.

The two OCV-SOC characteristics obtained as shown in FIG. 7A are incorporated into the Kalman filter as OCV-SOC characteristic model functions. Since the characteristic is a curve, it is of course possible to use a function that represents the curve. However, since the characteristic curve is not monotonous, a function that requires a large amount of calculation (time) such as a trigonometric function or an exponential function is generally required. The characteristic curve may be divided into a plurality of SOC regions, and the characteristic curve may be approximated by a linear function in each region.

The method of linear approximation is the same as the case of a single OCV-SOC characteristic, but since there are two characteristics, the number of SOC regions (number of straight lines) is determined for each, and a straight line is determined for each region. Determine the function. As illustrated in FIG. 8, the number and range of SOC regions into which the OCV-SOC characteristic model function is divided, and the approximate linear function can be set separately for each of the two characteristics.

When the characteristic curve is divided into a plurality of SOC regions, and the characteristic curve is approximated by a linear function in each region, the parameter determination unit 20 uses two determination results by the determination method of the current state and the current change state in the Kalman filter. It is determined which one of the OCV-SOC characteristic model functions is used and which region is in the SOC value estimated by the estimation formula. The parameter determination unit 20 selects the above approximate linear function based on the determination result. A method for determining the battery usage status in the Kalman filter will be described later.

FIG. 9 is a diagram illustrating a flowchart executed when the parameter determination unit 20 selects the OCV-SOC characteristic model function. As illustrated in FIG. 9, the parameter determination unit 20 determines whether or not the measured current is equal to or less than a threshold value (step S1). By executing step S1, it can be determined whether or not the secondary battery 200 is passing almost no current.

When it is determined as “Yes” in step S1, the parameter determination unit 20 determines whether or not the state where the measured current is equal to or lower than the threshold value continues for a predetermined period (step S2). If it is determined as “Yes” in step S2, the parameter determination unit 20 selects the OCV-SOC characteristic model function on the discharge side (step S3). The selected OCV-SOC characteristic model function is stored in the storage unit 30.

When it is determined “No” in step S1, the parameter determination unit 20 determines whether or not the state where the measured current exceeds the threshold value continues for a predetermined period (step S4). If it is determined as “Yes” in step S4, the parameter determination unit 20 determines whether the measured current is positive or negative (step S5). If it is determined in step S5 that the measured current is negative, step S3 is executed. When it is determined in step S5 that the measured current is positive, the parameter determination unit 20 selects the charging-side OCV-SOC characteristic model function (step S6). The selected OCV-SOC characteristic model function is stored in the storage unit 30.

When it is determined as “No” in Step S2 or when it is determined as “No” in Step S4, the parameter determination unit 20 determines whether the secondary battery 200 tends to be discharged or charged (Step). S7). If it is determined in step S7 that the secondary battery 200 has a tendency to discharge, step S3 is executed. If it is determined in step S7 that the secondary battery 200 has a charging tendency, step S6 is executed.

Next, an example of a reference for determining whether the secondary battery 200 has a tendency to discharge or a tendency to charge will be described. For example, the parameter determination unit 20 determines the charge / discharge tendency according to the following procedure.
(1) Addition of current value: A tendency is determined by a value obtained by sequentially adding current values.
(2) Reset of addition: Addition is reset at a specific number of additions (for example, 1000 times when the estimation period is 0.1 s).
(3) Processing at reset: The added value is not set to zero. Multiply the previous current addition value by a specific constant (a numerical value less than 1; for example, 0.3), and restart the current addition to that value.
(4) Judgment of tendency: If the value obtained by adding the current while repeating the above reset is larger than a specific threshold (for example, 300 A), it is judged as a charging tendency, and if it is below, it is judged as a discharging tendency.

Here, the integration reset cycle of (2) is a relatively long time. This is because the SOC estimation by the Kalman filter may cause an operation that increases the estimation error for a short time immediately after the switching of the OCV-SOC characteristic model, so that the switching is not performed more frequently than necessary. When resetting in a short period of time, the change in charge / discharge tendency can be determined without delay. However, in actuality, switching of the OCV-SOC characteristic model does not occur in a short time. The estimation accuracy can be increased for a long time.

Also, the reason for multiplying the constant (less than 1) instead of resetting the current integration in (3) to zero is to suppress erroneous switching in a short time by taking over the trend information of the previous reset period. In an operating situation in which the sign of the current flowing through the battery changes frequently, even if the tendency to charge or discharge continues even if the zero reset is performed, the tendency is false (opposite) for a short time. In some cases, a correct tendency is shown by repeating the integration. As described above, in the Kalman filter, since an error increases in a short time immediately after switching, it is preferable to avoid this. On the other hand, although this method can prevent a short-time erroneous determination, a delay occurs in determining that the trend has truly switched. However, switching of the actual OCV-SOC characteristic model does not occur in a short time. In addition, an estimation error due to erroneous selection of the OCV-SOC characteristic model that occurs during the determination delay time is corrected toward the correct estimated value after the correct model is selected (Kalman filter operation). Therefore, in the estimation by the Kalman filter, the accuracy is improved by the determination by the method according to the present embodiment.

ここで、Ｓ_ｃ，ａは、ＳＯＣ推定の対象である２次電池の蓄電可能容量である。２次電池によりＳ_ｃ，ａは異なる場合がある。また、Ｓ_ｃ，ａは、２次電池の使用仕様と充放電測定により求めることができる。また、Ｓ_ｃ，ａは、温度と劣化により変化するので、測定あるいは推定された２次電池の温度、および測定あるいは推定された劣化度を基に、ＳＯＣ推定周期毎にあるいは定期／不定期に、求めた蓄電可能容量値を適用することができる。 Next, details of the Kalman filter will be described with reference to FIG. FIG. 10 is an explanatory diagram illustrating an example of an SOC estimation process using a Kalman filter. First, the following equation (1) is an example of the state estimated value of the Kalman filter in step k, and is an example of the state estimated values of v1, v2, and SOC. k indicates the number of steps of the Kalman filter. Δt is a time interval in which the Kalman filter is performed, and usually corresponds to a sampling period in which the measurement unit 10 measures the measurement current I and the measurement terminal voltage V _OBS . However, the measurement period and the Kalman filter period do not necessarily coincide.

Here, Sc _{, a} is the chargeable capacity of the secondary battery that is the target of SOC estimation. _{Sc, a} may differ depending on the secondary battery. In addition, _{Sc, a} can be obtained by use specifications and charge / discharge measurement of the secondary battery. In addition, since _{Sc and a} change with temperature and deterioration, based on the measured or estimated secondary battery temperature and the measured or estimated deterioration degree, every SOC estimation period or periodically / irregularly The obtained chargeable capacity value can be applied.

は、ステップｋ－１のカルマンフィルタの補正された状態推定値を示し、以下、１ステップ前の状態推定値という。
［文字２］

は、ステップｋの測定端子電圧と予測端子電圧との差分を示し、以下、差分という。
［文字３］

は、ステップｋのカルマンフィルタの補正前の状態推定値を示し、以下、補正前の状態推定値という。
［文字４］

は、ステップｋのカルマンフィルタの状態推定値の補正値を示し、以下、補正値という。 Next, characters used in the description of FIG. 10 will be described. V _OBS (k) represents the measurement terminal voltage in step k, and is hereinafter referred to as measurement terminal voltage.
[Character 1]

Indicates a corrected state estimated value of the Kalman filter in step k−1, and is hereinafter referred to as a state estimated value one step before.
[Character 2]

Indicates the difference between the measured terminal voltage and the predicted terminal voltage in step k, and hereinafter referred to as the difference.
[Character 3]

Indicates a state estimation value before correction of the Kalman filter in step k, and is hereinafter referred to as a state estimation value before correction.
[Character 4]

Indicates a correction value of the estimated state value of the Kalman filter in step k, and is hereinafter referred to as a correction value.

G (k) represents the Kalman gain of step k. A indicates Jacobian. P (k) represents the error covariance matrix of the estimated value in step k, that is, the accuracy of the estimated value. Σv is a covariance matrix indicating estimated noise. Σw is a covariance matrix indicating measurement noise.

When the measurement current i (k) is input, the calculation unit 41 uses the following equation (2) based on the state estimated value one step before and the measurement current i (k−1) before correction. The estimated state value is calculated (step S11). Here, the measurement current i (k) indicates a case where the measurement current i (k) is input every second, and Δt = 1. Therefore, for example, the relationship of the following formula (3) can be used simply. it can.

Next, as described above, the parameter determination unit 20 uses a plurality of OCV-SOC characteristics stored in the storage unit 30 based on the result of the above equation (2) and the tendency of the measurement current acquired by the measurement unit 10. One OCV-SOC characteristic model function is selected from the model functions (step S12). Next, the calculation unit 41 predicts the predicted terminal voltage V ⁻ according to the following formula (4) from the result of the above formula (2) (step S13).

When the measurement terminal voltage V _OBS is input, the calculation unit 41 calculates the measurement terminal voltage V _OBS and the prediction terminal voltage V ⁻ from the measurement terminal voltage V _OBS and the prediction terminal voltage V ⁻ according to the following equation (5). Is calculated (step S14).

Next, the correction unit 42 calculates Jacobian A using the following equation (6) based on the state estimated value one step before (Step S15).

The correction unit 42 uses the following equation (7) based on the Jacobian A, the one-step previous covariance matrix P (k−1), and the prediction noise Σv, and uses the prior covariance matrix P ⁻ (k). Is calculated (step S16).

The correction unit 42 calculates the Kalman gain G (k) using the following equation (8) based on the prior covariance matrix P ⁻ (k) and the measurement noise Σw (step S17).

The correcting unit 42 calculates the covariance matrix P (k) using the following equation (9) based on the Kalman gain G (k) and the prior covariance matrix P ⁻ (k) (step S18). . The correcting unit 42 repeats steps S16 to S18 for each step.

Next, the correction unit 42 uses the following equation (10) based on the calculated difference and the Kalman gain G (k) calculated in step S17 to correct the state estimated value. A value is calculated (step S19).

The correcting unit 42 calculates a state estimated value using the following equation (11) based on the state estimated value before correction calculated in step S11 and the corrected value calculated in step S19 (step S20). ). Here, the state estimated value can also be expressed by the following equation (12).

Based on the state estimated value, the correcting unit 42 calculates the SOC using the following formula (13) in the case of this example (step S21).

As described above, the calculation unit 41 and the correction unit 42 can estimate the SOC every second, for example, by repeating the processes of steps S11 to S21 as the SOC estimation process for each step. In the above Kalman filter, the measured measurements terminal voltage _{V OBS} and SOC, v1, v2 predicted terminal voltage V predicted from the estimated value of ^- using the difference between, and the Kalman gain G, SOC, v1, estimate of v2 Is corrected. By repeating this every step, the estimated values of SOC, v1, and v2 are brought close to the true value.

FIG. 11 (a) and FIG. 11 (b) show the result of determining the sequential charge / discharge tendency by the above method. FIG. 11A shows a change in current (0.1 s interval) and a change in SOC of the battery to be determined. The SOC is calculated by an integration method using a highly accurate ammeter. From the change in the SOC, it can be seen that the charging trend, the discharging tendency, and the charging tendency change from the left.

FIG. 11B shows a determination result by the method according to the present embodiment. The reset interval is 1000 cycles (100 s), the determination threshold is 3E + 5 mA, and the reset multiplier is 0.3. Thus, the change of the tendency can be correctly determined by the method according to the present embodiment, and no erroneous determination is seen while the tendency is maintained. Judgment delay is not significant.

FIG. 12A shows a case where the reset interval is 100 cycles (10 s), the determination threshold is 3E + 4 mA, and the reset multiplier is 0.3. It can be seen that erroneous determination frequently occurs by making the reset interval relatively short. Looking at the determination time, it takes a long time for the correct determination to be made, so that the correct determination is made overall. However, when this determination is applied to the Kalman filter, if the time of erroneous determination is large as shown in this figure, the estimation error caused at the time of erroneous determination cannot be corrected within the correct determination time, and correct estimation cannot be performed.

FIG. 12B shows the case where the reset interval is 1000 cycles (100 s) and the determination threshold is 3E + 5 mA, but the integrated current is reset to zero. Similar to FIG. 12A, erroneous determination frequently occurs.

FIG. 13 is a diagram illustrating the SOC estimation accuracy by the Kalman filter when a single OCV-SOC characteristic model function is used and when two OCV-SOC characteristic model functions are switched according to the present embodiment. In both cases, the OCV-SOC characteristic model function is based on the above-mentioned plurality of linear functions (each of the SOC division regions and the linear function has different coefficients). In the case of using a single OCV-SOC characteristic model function, an estimation error occurs, whereas according to the present embodiment, it is shown that high-precision estimation is possible.

According to the present embodiment, a model function is determined from a plurality of OCV-SOC characteristic model functions according to the current state of the secondary battery 200. Thereby, an appropriate OCV-SOC model function according to the current state can be used. Further, the charging rate and the predicted terminal voltage of the secondary battery 200 are estimated by the Kalman filter using the determined model function, and the difference between the predicted terminal voltage and the measured value of the terminal voltage of the secondary battery 200 is calculated. The charging rate is corrected based on the difference and the Kalman gain of the Kalman filter. In this way, the SOC is corrected for each step by the Kalman filter, instead of directly obtaining the SOC from the voltage using the determined model function. Thereby, the SOC can be estimated with high accuracy.

FIG. 14 is a block diagram for explaining an example of a hardware configuration of the estimation apparatus 100. As illustrated in FIG. 14, the estimation device 100 includes a CPU 101, a RAM 102, a storage device 103, an interface 104, and the like. Each of these devices is connected by a bus or the like. A CPU (Central Processing Unit) 101 is a central processing unit. The CPU 101 includes one or more cores. A RAM (Random Access Memory) 102 is a volatile memory that temporarily stores programs executed by the CPU 101, data processed by the CPU 101, and the like. The storage device 103 is a nonvolatile storage device. As the storage device 103, for example, a ROM (Read Only Memory), a solid state drive (SSD) such as a flash memory, a hard disk driven by a hard disk drive, or the like can be used. The interface 104 is a device that transmits and receives signals to and from an external device. When the CPU 101 executes a program stored in the storage device 103, each unit of the estimation device 100 is realized. Alternatively, an MPU (Micro Processing Unit) or the like may be used instead of the CPU. Alternatively, for example, it may be realized by an integrated circuit such as ASIC (Application Specific Integrated Circuit) or FPGA (Field Programmable Gate Array).

(Modification)
FIG. 15 is a diagram illustrating an estimation system according to a modification. In each of the above examples, the parameter determination unit 20 and the calculation unit 40 obtain measurement values such as current values and terminal voltages from the measurement unit 10. On the other hand, a server having the functions of the parameter determination unit 20 and the calculation unit 40 may acquire measurement data from the measurement unit 10 through a telecommunication line. For example, the server includes the CPU 101, the RAM 102, the storage device 103, the interface 104, and the like illustrated in FIG. 14 and realizes functions as the parameter determination unit 20 and the calculation unit 40.

In each of the above examples, the storage unit 30 functions as an example of a storage unit that stores a plurality of model functions of the open-circuit voltage and the charging rate of a rechargeable battery. The parameter determination unit 20 functions as an example of a determination unit that determines a model function from a plurality of model functions in accordance with the battery current status. The calculation unit 41 estimates a battery charging rate and a predicted terminal voltage by a Kalman filter using the model function determined by the determination unit, and calculates a difference between the predicted terminal voltage and the measured value of the battery terminal voltage. It serves as an example. The correction unit 42 functions as an example of a correction unit that corrects the charging rate based on the difference and the Kalman gain of the Kalman filter.

Although the embodiments of the present invention have been described in detail above, the present invention is not limited to such specific embodiments, and various modifications and changes can be made within the scope of the gist of the present invention described in the claims. It can be changed.

DESCRIPTION OF SYMBOLS 10 Measurement part 20 Parameter determination part 30 Memory | storage part 40 Calculation part 50 Output part 100 Estimation apparatus 200 Secondary battery 300 External apparatus

Claims (15)

A storage unit for storing a plurality of model functions of an open voltage and a charging rate of a rechargeable battery;
A determining unit that determines a model function from a plurality of the model functions according to the current state of the battery,
By a Kalman filter using the model function determined by the determination unit, a charging unit and a predicted terminal voltage of the battery are estimated, and a calculation unit that calculates a difference between the predicted terminal voltage and an actual value of the battery terminal voltage;
An estimation device comprising: a correction unit that corrects the charging rate based on the difference and a Kalman gain of the Kalman filter. The plurality of model functions are determined based on a plurality of voltages virtually determined as voltages at zero ampere using charge / discharge curve data measured by changing the current of the battery. The estimation apparatus according to claim 1, characterized in that: The estimation device according to claim 1 or 2, wherein the determination unit determines the model function according to a direction of current of the battery. The estimation apparatus according to any one of claims 1 to 3, wherein the determination unit switches the plurality of model functions based on presence or absence of current of the battery and a change in the current. The estimation device according to claim 4, wherein the determination unit switches the plurality of model functions based on a current integrated value of the battery that is initialized at a predetermined period. 6. The estimation apparatus according to claim 5, wherein the determining unit initializes the current integrated value by multiplying the current integrated value by a numerical value less than 1. The estimation apparatus according to any one of claims 4 to 6, wherein the determination unit switches the plurality of model functions based on a magnitude relationship between the current integrated value and a threshold value. On the computer,
A process of determining a model function according to the current state of the battery from a plurality of model functions of an open voltage and a charging rate of the rechargeable battery,
A process for estimating a charging rate and a predicted terminal voltage of the battery by a Kalman filter using the determined model function, and calculating a difference between the predicted terminal voltage and an actual value of the terminal voltage of the battery;
An estimation program for executing the process of correcting the charging rate based on the difference and a Kalman gain of the Kalman filter. The plurality of model functions are determined based on a plurality of voltages virtually determined as voltages at zero ampere using charge / discharge curve data measured by changing the current of the battery. 9. The estimation program according to claim 8, wherein The estimation program according to claim 8 or 9, wherein, in the determining process, the model function is determined in accordance with a direction of current of the battery. The estimation program according to any one of claims 8 to 10, wherein, in the determining process, the plurality of model functions are switched based on presence / absence of a current of the battery and a change in the current. 12. The estimation program according to claim 11, wherein, in the determining process, the plurality of model functions are switched based on a current integrated value of the battery that is initialized at a predetermined cycle. 13. The estimation program according to claim 12, wherein, in the determining process, the current integrated value is initialized by multiplying the current integrated value by a numerical value less than 1. The estimation program according to any one of claims 11 to 13, wherein, in the determining process, the plurality of model functions are switched based on a magnitude relationship between the integrated current value and a threshold value. A storage unit that stores a plurality of model functions of an open-circuit voltage and a charging rate of a rechargeable battery, a determination unit that determines a model function from a plurality of the model functions according to the current state of the battery, and the determination unit The Kalman filter using the model function determined by the calculation unit estimates a charging rate and a predicted terminal voltage of the battery, and calculates a difference between the predicted terminal voltage and the measured value of the terminal voltage of the battery, and the difference A correction unit that corrects the charging rate based on the Kalman gain of the Kalman filter;
And a control device that performs charge / discharge control of the battery based on the charge rate corrected by the correction unit.

Images