DE112016006166T5 - State estimator and state estimator - Google Patents

Abstract

To accurately estimate an SOC of an energy storage device with a range in which an OCV change amount is small relative to an SOC change amount. A state estimator estimates a state of an energy storage device having a range of small changes and a range of large changes, wherein an OCV change amount relative to an SOC change amount in the range of small changes is relatively small and that in the range of large changes is relatively large. The state estimating device includes: a region determining unit that determines whether the energy storage device is in the range of large changes; and an SOC estimation unit that estimates an SOC representing one of internal states of the energy storage device, wherein the SOC estimation unit executes: estimating the SOC of the energy storage device by a current integration method that generates a current to / integrated by the energy storage device; and an adjustment processing that sets the SOC estimated by the current integration method based on a measured terminal voltage value of the energy storage device and a predicted terminal voltage by an estimation model when the energy storage device is in the range of large changes, wherein the estimation model determines the internal state of the energy storage device. Device estimates.

Description

TECHNICAL PART

The present invention relates to a technique of estimating a state of charge (SOC).

STATE OF THE ART

One of the conventional methods of estimating a SOC of a secondary battery is a current integration method. However, when an SOC is estimated by a current integration method, measurement errors caused by a current sensor are accumulated, which reduces the accuracy in estimating an SOC over time. The following Patent Document 1 describes the improvement of the SOC estimation accuracy by adjusting an SOC estimated by a current integration method using a battery model and a Kalman filter.

DOCUMENTS OF THE PRIOR ART

PATENT DOCUMENTS

Patent Document 1: JP 5074830 B2

SUMMARY OF THE INVENTION

PROBLEMS TO BE SOLVED BY THE INVENTION

Some energy storage devices, such as lithium-ion secondary batteries, have plateau regions in which an amount of change of open circuit voltage (OCV) relative to an SOC change amount is small and OCV is generally constant. When the adjustment using a battery model and a Kalman filter is applied to these energy storage devices having regions in which an OCV change amount is small relative to an SOC change amount, the SOC estimation accuracy can be reduced.

A similar problem occurs when an SOC estimated by a current integration method is set using a battery model and an adaptive digital filter. The present invention has been made in consideration of the above circumstances. An object of the present invention is to accurately estimate an SOC of an energy storage device having a range in which an OCV change amount is small relative to an SOC change amount.

MEANS OF SOLVING THE PROBLEMS

A state estimating device disclosed in this specification estimates a state of an energy storage device having a range of small changes and a range of large changes, wherein an OCV change amount relative to an SOC change amount in the range of small changes is relatively small and that in The range of large changes is relatively large. The state estimating device includes: a region determining unit that determines whether the energy storage device is in the range of large changes; and an SOC estimation unit that estimates an SOC representing one of internal states of the energy storage device, wherein the SOC estimation unit executes: estimating the SOC of the energy storage device by a current integration method that generates a current to / integrated by the energy storage device; and an adjustment processing that sets the SOC estimated by the current integration method based on a measured terminal voltage value of the energy storage device and a terminal voltage predicted by an estimation model when the energy storage device is in the range of large changes, wherein the estimation model indicating the internal state of the Energy storage device estimates.

ADVANTAGES OF THE INVENTION

The state estimators disclosed in this specification can accurately estimate an SOC of an energy storage device having a range in which an OCV change amount is small relative to an SOC change amount.

list of figures

• 1 FIG. 12 is a schematic diagram showing the structure of a battery pack according to a first embodiment. FIG.
• 2 Fig. 12 is a graph showing the SOC-OCV correlation characteristics of a secondary battery.
• 3 FIG. 12 is a block diagram of a state space model. FIG.
• 4 FIG. 12 is a circuit diagram showing a circuit model that simulates the secondary battery.
• 5 FIG. 10 is a flowchart briefly showing an SOC estimation method using a Kalman filter. FIG.
• 6 FIG. 10 is a flowchart showing a flow of an SOC estimation process. FIG.
• 7A FIG. 12 is a graph showing SOC changes over time when the battery is charged and discharged at a battery temperature of 25 ° C.
• 7B provides an enlarged view of a range of large changes A3 in FIG 7A represents.
• 8th FIG. 12 is a graph showing SOC changes over time when the battery is charged and discharged at a battery temperature of 25 ° C.
• 9 represents a table showing the root-mean-squared values of the SOC values with respect to the true values in different estimation methods.
• 10 Fig. 12 is a graph showing SOC changes with time when the secondary battery is charged and discharged at a battery temperature of 25 ° C in a second embodiment.
• 11 FIG. 12 is a graph showing SOC changes with time when the battery is charged and discharged at a battery temperature of 10 ° C. FIG.
• 12 FIG. 12 is a graph showing SOC changes over time when the battery is charged and discharged at a battery temperature of 0 ° C. FIG.
• 13 Fig. 12 is a table showing the root mean square values of the set SOC values with respect to the true values at different battery temperatures.
• 14 FIG. 10 is a flowchart showing a flow of an SOC estimation process. FIG.
• 15 FIG. 10 is a flowchart showing a flow of an SOC estimation process. FIG.
• 16 Fig. 12 is a graph showing SOC changes with time when the secondary battery is charged and discharged at a battery temperature of 25 ° C in a third embodiment.
• 17 FIG. 10 is a flowchart showing a flow of an SOC estimation method. FIG.
• 18 FIG. 12 is a graph showing SOC changes with time when the secondary battery is charged and discharged at a battery temperature of 25.degree. C. in a fourth embodiment.
• 19 FIG. 12 is a graph showing SOC changes with time when the battery is charged and discharged at a battery temperature of 10 ° C. FIG.
• 20 FIG. 12 is a graph showing SOC changes over time when the battery is charged and discharged at a battery temperature of 0 ° C. FIG.
• 21 Fig. 12 is a table showing the root mean square values of the set SOC values with respect to the true values at different battery temperatures.
• 22 FIG. 10 is a flowchart showing a flow of an SOC estimation process. FIG.
• 23 FIG. 10 is a flowchart showing a flow of an SOC estimation process. FIG.

WAYS FOR CARRYING OUT THE INVENTION

(Overview of the embodiment)

First, an outline of a state estimator disclosed in this embodiment will be described. A state estimator estimates a state of an energy storage device having a range of small changes and a range of large changes wherein an OCV Amount of change relative to an SOC change amount in the range of small changes is relatively small and is relatively large in the range of large changes. The state estimating device includes: a region determining unit that determines whether the energy storage device is in the range of large changes; and an SOC estimating unit that estimates an SOC representing one of internal states of the energy storage device, wherein the SOC estimating unit executes: estimating the SOC of the energy storage device by a current integration method that supplies a current to / from the Energy storage device integrated; and an adjustment processing that sets the SOC estimated by the current integration method based on a measured terminal voltage value of the energy storage device and a terminal voltage predicted by an estimation model when the energy storage device is in the range of large changes, wherein the estimation model determines the internal state of the energy storage device. Device estimates.

When an SOC estimated by a current integration method is set based on an observed terminal voltage of an energy storage device and a terminal voltage predicted by an estimation model estimating an internal state of the energy storage device, the SOC estimation accuracy depends on an OCV change amount relative to a SOC change amount. In particular, in a range of large changes in which an OCV change amount is relatively large relative to an SOC change amount, the SOC estimation accuracy is higher than in a range of small changes in which an OCV change amount relative to an SOC change amount is relatively small is.

In this structure, an SOC estimated by a current integration method based on an observed terminal voltage of an energy storage device and a terminal voltage predicted by an estimation model estimates an internal state of the energy storage device only when the energy storage device is in the range great changes is set. In other words, the setting processing is performed only in the area of large changes in which the SOC estimation accuracy is expected to be improved by the setting. This makes it possible to accurately estimate an SOC of an energy storage device having a range of small changes in which an OCV change amount is small relative to an SOC change amount.

An embodiment of the state estimator should preferably have the following structure.

The range of small changes includes a plateau range in which an OCV change amount relative to an SOC change amount is smaller than a predetermined value. In the plateau area, the setting processing is not performed. Since the adjustment processing in the plateau region is not performed in this structure, reduction of the SOC estimation accuracy by the adjustment can be prevented.

The adjustment is a Kalman filtering that reduces an estimated internal state error of the energy storage device estimated by the estimation model based on an observed terminal voltage of the energy storage device and a terminal voltage predicted by the estimation model. The Kalman filter, which reduces an estimated internal state error of the energy storage device, can accurately adjust an SOC estimated by a current integration method.

The measurement model includes an equivalent circuit model of the energy storage device. The equivalent circuit model includes an OCV, a conductor resistance, a first impedance that simulates a short period polarization of the energy storage device, and a second impedance that simulates a long period polarization of the energy storage device. This model can accurately model the polarization characteristics of the energy storage device. Therefore, an estimated internal state error of the energy storage device can be effectively reduced, and an SOC estimated by a current integration method can be accurately set.

The SOC estimating unit executes the setting processing only when the energy storage device is in the range of large changes and the energy storage device is minimally current-flowing. Since the setting processing in this structure is performed only in the minimum current flow state, an SOC estimated by a current integration method can be accurately set.

The structure includes a temperature detecting unit that detects a temperature of the energy storage device, and the SOC estimating unit executes the setting processing only when the Temperature of the energy storage device is a first temperature or more. In this structure, the setting processing is executed only when the temperature of the energy storage device is a first temperature or more. Therefore, an SOC estimated by a current integration method can be accurately set.

The structure includes a temperature detection unit that detects a temperature of the energy storage device, and the SOC estimation unit does not execute the adjustment processing when the temperature of the energy storage device is a second temperature or less. In this structure, the setting processing is not performed when the temperature of the energy storage device is a second temperature or less. Therefore, a reduction of the SOC estimation accuracy is prevented by the adjustment.

The range determining unit determines whether the energy storage device is in the range of large changes based on an SOC value estimated by the SOC estimating unit. This makes it easy to determine in which area the energy storage device is.

Alternatively, by comparing voltage fluctuation relative to charge-discharge capacity change of the energy storage device, the area determination unit determines with a threshold in which range of small changes and large changes the energy storage device is.

<First Embodiment>

A first embodiment will now be described with reference to FIGS 1 to 9 described.

Structure of the battery pack 20

1 shows the structure of a battery pack 20 according to this embodiment. The battery pack 20 This embodiment is mounted, for example, in an electric vehicle or a hybrid vehicle and supplies electric power to a power source that operates on electric power.

As in 1 shown includes the battery pack 20 a battery group 30 , a current sensor 40 and a battery manager (BM) 50 for managing the battery group 30 , The battery group 30 includes a plurality of secondary batteries 31 connected in series.

The secondary batteries 31 and the current sensor 40 are connected in series via a wiring 35 and further connected to a battery charger 10 mounted in an electric vehicle or a load 10 such as a power source in an electric vehicle or the like.

The battery charger 10 charges the battery pack 30 , The current sensor 40 detects current through the secondary batteries 31 , The current sensor 40 periodically measures a current value of the secondary batteries 31 and sends the measured current value data to a controller 60 ,

The battery manager (BM) 50 includes the controller 60 , a voltage detection circuit 80 and a temperature sensor 95. It should be noted that the secondary batteries 31 represent an example of the energy storage device and the BM 50 an example of the state estimator.

The voltage detection circuit 80 is connected to both ends of each secondary battery 31 via sense lines, and measures a voltage of each secondary battery 31 in response to instructions from the controller 60. The temperature sensor 95 is a contact or non-contact sensor and measures a temperature T [° C] of the secondary batteries 31 ,

The control 60 includes a central processing unit (CPU) 61 , a store 63 and a communication unit 67. The controller 60 determines in which area the secondary batteries 31 are (in step S30 in 6 and estimates an SOC (in steps S20, S40 and S50 in FIG 6 ). It should be noted that the controller 60 an example of the range determining unit and the SOC estimating unit.

The memory 63 stores programs for estimating an SOC, and data for executing the programs such as the SOC-OCV correlation characteristic data stored in 2 and data for determining the range of an SOC. In particular, the memory stores 63 , the data of the SOC Areas in the respective areas of small changes L1 and L2 and in the areas of large changes H1 to H3. The memory 63 also stores the data about the current SOC value.

The communication unit 67 is connected to an electronic control unit (ECU) 100 to communicate with the ECU 100 in a vehicle. The battery pack 20 Also includes an operation unit (not shown) for receiving an input from the user and a display unit (not shown) for displaying the state of the secondary batteries 31 and the same.

SOC-OCV Correlation Characteristics of Secondary Batteries 31

The secondary batteries 31 are iron phosphate lithium ion batteries in which iron phosphate lithium (LiFePO4) is used for positive active material and graphite for negative active material.

2 shows the SOC-OCV correlation characteristics of the secondary batteries 31. The horizontal axis shows an SOC [%] and the vertical axis shows an OCV [V]. An SOC represents a ratio of a remaining capacity to a full charge capacity. An OCV represents an open circuit voltage of the secondary batteries 31 represents.

As in 2 shown have the secondary batteries 31 a plurality of charge areas, including areas of small changes in which an OCV change amount relative to an SOC change amount is relatively small, and areas of large changes in which an OCV change amount relative to an SOC change amount is relatively large.

In particular, the secondary batteries have 31 two areas of small changes L1 and L2 and three areas of big changes H1, H2 and H3. As in 2 As shown, the range of small changes L1 is in the range of SOC values from 31 [%] to 62 [%]. The range of small changes L2 is in the range of SOC values from 68 [%] to 97 [%]. The region of small changes L1 includes a plateau region in which an OCV change amount relative to an SOC change amount is very small and OCV values are generally constant by 3.3 [V]. Similarly, the region of small changes L2 includes a plateau region in which OCV values are generally constant around 3.34 [V]. The plateau area is defined as a range in which an OCV change amount relative to an SOC change amount is 2 [mV /%] or less.

The first range of large changes H1 is in the range of SOC values greater than 62 [%] and less than 68 [%], and between the two ranges of small changes L1 and L2. The second range of large changes H2 is in the range of SOC values less than 31 [%] and in the range of SOC values lower than the SOC values in the range of small changes L1. The third range of large changes H3 is in the range of SOC values greater than 97 [%] and in the range of SOC values higher than the SOC values in the range of small changes L2. In the first to third regions of large changes H1 to H3, the OCV change amounts are larger relative to the SOC change amounts (the inclination of the graph in FIG 2 is larger) than in the areas of small changes L1 and L2.

SOC estimation process using the Kalman filter

<Description of the Kalman Filter>

wobei x(k) einen internen Zustand des Systems darstellt; f, b_u und b Systemmatrizen darstellen; u(k) eine Eingabe darstellt; und v(k) Systemrauschen darstellt. $$y\left(k\right)=h\left(x\left(k\right)\right)+w\left(k\right),$$

wobei y(k) eine Ausgabe darstellt, h eine Systemmatrix darstellt und w(k) Beobachtungsrauschen darstellt. 3 shows a state space model, where u (k) represents an input and y (k) represents an output. The following expression 1 represents a state equation of in 3 and the following expression 2 represents an output equation of a space-state model. $$x\left(k+1\right)=f\left(x\left(k\right)\right)+b_{u}u\left(k\right)+b\nu \left(k\right).$$

where x (k) represents an internal state of the system; f, b _u and b represent system matrices; u (k) represents an input; and v (k) represents system noise. $$y\left(k\right)=H\left(x\left(k\right)\right)+w\left(k\right).$$

where y (k) represents an output, h represents a system matrix and w (k) represents observation noise.

wobei x(k) einen geschätzten Wert von x(k) darstellt und x(k) einen geschätzten Fehler von x(k) darstellt. $$J\left(k\right)=\left(E\left[{\tilde{x}}^2\left(k\right)\right]\right)$$

$$\widehat{x}\left(k\right)=\text{arg }\underset{x\left(k\right)}{\text{min}}J\left(k\right)$$

$$\widehat{x}\left(k\right)={\widehat{x}}^{-}\left(k\right)+g\left(k\right)\tilde{y}\left(k\right),$$

wobei x̂(k) einen geschätzten Wert von x(k) darstellt, x̂(k) eine Vorausschätzung von x(k) darstellt, g(k) eine Kalman-Verstärkung darstellt und ỹ(k) einen vorhergesagten Fehler von y(k) darstellt.A Kalman filter represents an algorithm for determining an estimated value of an internal state x (k) of a system based on a time series of observed values {y (i) = 1, 2, 3, k}. In particular, a Kalman filter sets Algorithm for determining an optimal estimate of the internal state x (k) by minimizing a weighting function J (k). The weighting function J (k) represents a mean squared error that can be used to evaluate an estimated error of a state or difference between the true internal state x (k) and an estimated value is used (See the following expressions 3 to 5). The symbol (^) in an internal state x (k) shows that the value represents an estimated value and the symbol (~) in an internal state x (k) shows that the value represents an estimated error of a state. $$\tilde{x}\left(k\right)=x\left(k\right)-\widehat{x}\left(k\right).$$

where x (k) represents an estimated value of x (k) and x (k) represents an estimated error of x (k). $$J\left(k\right)=\left(e\left[{\tilde{x}}^2\left(k\right)\right]\right)$$

$$\widehat{x}\left(k\right)=\text{bad}\underset{x\left(k\right)}{\text{min}}J\left(k\right)$$

$$\widehat{x}\left(k\right)={\widehat{x}}^{-}\left(k\right)+G\left(k\right)\tilde{y}\left(k\right).$$

where x (k) represents an estimated value of x (k), x (k) represents a prediction of x (k), g (k) represents a Kalman gain, and ỹ (k) represents a predicted error of y (k) represents.

An estimated value of an internal state x (k) determined using the Kalman filter may be obtained by a preliminary estimation of the internal state x (k) (the first term on the right side of Expression 6) and a correction term (the second one Term on the right side of expression 6).

wobei P (k) eine Voraus-Kovarianzmatrix (advance covariance matrix) eines geschätzten Fehlers von x(k) darstellt, σ² eine Varianz von w(k) darstellt, und c^T(k) eine transponierte Matrix von c(k) darstellt. $$c^T\left(k\right)=\frac{\partial h\left(x\right)}{\partial x}|{}_{x={\widehat{x}}^{-}\left(k\right)}$$

The pre-estimate of an internal state x (k) represents a predicted estimate of the internal state x (k) at time k based on the data available until time k-1. The correction term is used to set the predicted estimated state x Value of the first term is used and can be expressed by the product of a Kalman gain g (k) and a predicted error of an output y (k). The Kalman gain g (k) represents a value for minimizing the covariance of an estimated error of an internal state x (k) (an error between the true value and the estimated value), and can be calculated based on, for example, the orthogonality principle (See the following expression 7). $$G\left(k\right)=\frac{P^{-}\left(k\right)c\left(k\right)}{c^T\left(k\right)P^{-}\left(k\right)c\left(k\right)+{\sigma }^2}.$$

where P (k) represents an advance covariance matrix of an estimated error of x (k), σ ^{2 represents} a variance of w (k), and c ^T (k) represents a transposed matrix of c (k) , $$c^T\left(k\right)=\frac{\partial H\left(x\right)}{\partial x}|{}_{x={\widehat{x}}^{-}\left(k\right)}$$

<Systematization of secondary batteries>

The secondary batteries 31 can be considered as a system in which an input represents a current I and an output a terminal voltage U _L , and which can be expressed by the equation of state of expression 1 and the output equation of expression 2. In this embodiment the secondary batteries 31 using the equivalent circuit model of in 4 shown secondary batteries 31 systematized. In particular, the secondary batteries 31 expressed by an OCV representing the electromotive force (DC source); a conductor resistance R _{0 representing} a resistance of a current collector, an electrolyte, and the like; a first impedance Z1; and a second impedance Z2.

Since an OCV changes with an SOC change, an OCV can be expressed as a function of an SOC. The first impedance Z1 represents a parallel connection of a first resistor R1 and a first capacitor C1. The second impedance Z2 represents a parallel connection of a second resistor R2 and a second capacitor C2.

The time constant (τ = R × C) of the first impedance Z1 differs from the time constant (τ = R × C) of the second impedance Z2. In particular, the time constant τ1 of the first impedance Z1 is smaller than the time constant τ2 of the second impedance Z2. The first impedance Z1 simulates a fast response of the secondary batteries 31, that is, a short-period polarization voltage of the secondary batteries 31. The second impedance Z2 simulates a slow response of the secondary batteries 31 that is, a long period polarization voltage of the secondary batteries 31 , An equivalent circuit model M represents a model having an internal state (SOC, U1, U2, R0) of the secondary batteries 31 estimates and illustrates an example of the estimation model of the present invention.

The above equivalent circuit model M is mathematically expressed by the following expression 9. $$U_L\left(t\right)=OCV\left(SOC\left(t\right)\right)+R_{0}I\left(t\right)+{\mathrm{<figure-callout\; id="1"\; label="R"\; filenames="DE112016006166T5\_0018.png,DE112016006166T5\_0023.png"\; state="\{\{state\}\}">R</figure-callout>}}_1\times I\left(t\right)\times \left(1-e^{-\frac{t}{\tau 1}}\right)+R2\times I\left(t\right)\times \left(1-e^{-\frac{t}{\tau 2}}\right)$$

$$x\left(k+1\right)=\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& \left(1-\frac{\Delta t}{R_1{C}_1}\right)& 0& 0\\ 0& 0& \left(1-\frac{\Delta t}{R_2{C}_2}\right)& 0\\ 0& 0& 0& 1\end{array}\right]\times \left[\begin{array}{c}SOC\left(k\right)\\ U_1\left(k\right)\\ U_2\left(k\right)\\ R_0\left(k\right)\end{array}\right]+\left[\begin{array}{c}\frac{\Delta t}{C_{batt}}\\ \frac{1}{C_1}\\ \frac{1}{C_2}\\ 0\end{array}\right]\times i\left(k\right)+b\nu \left(k\right)$$

$$y\left(k\right)=u_L\left(k\right)=OCV\left(SOC\left(k\right)\right)+U_1\left(k\right)+U_2\left(k\right)+R_0\left(k\right)i\left(k\right)+w\left(k\right),$$

wobei u_L(k) eine Anschlussspannung darstellt, i(k) einen Strom darstellt, U₁(k) ein Spannungsabfall in Z1 darstellt, und U₂(k) ein Spannungsabfall in Z2 darstellt.When the expression 9 is discretized, the following equation of state (Expression 11) and the following output equation (Expression 12) with respect to an internal state x (k) of the secondary batteries become 31 where x (k) = SOC (k), U ₁ (k), U ₂ (k), R ₀ (k). $$x\left(k\right)=\left[\begin{array}{c}SOC\left(k\right)\\ U_1\left(k\right)\\ U_2\left(k\right)\\ R_0\left(k\right)\end{array}\right]$$

$$x\left(k+1\right)=\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& \left(1-\frac{\Delta t}{R_1{C}_1}\right)& 0& 0\\ 0& 0& \left(1-\frac{\Delta t}{R_2{C}_2}\right)& 0\\ 0& 0& 0& 1\end{array}\right]\times \left[\begin{array}{c}SOC\left(k\right)\\ U_1\left(k\right)\\ U_2\left(k\right)\\ R_0\left(k\right)\end{array}\right]+\left[\begin{array}{c}\frac{\Delta t}{C_{batt}}\\ \frac{1}{{\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="DE112016006166T5\_0018.png,DE112016006166T5\_0023.png"\; state="\{\{state\}\}">C</figure-callout>}}_1}\\ \frac{1}{C_2}\\ 0\end{array}\right]\times i\left(k\right)+b\nu \left(k\right)$$

$$y\left(k\right)=u_L\left(k\right)=OCV\left(SOC\left(k\right)\right)+U_1\left(k\right)+U_2\left(k\right)+R_0\left(k\right)i\left(k\right)+w\left(k\right).$$

where u _L (k) represents a terminal voltage, i (k) represents a current, U ₁ (k) represents a voltage drop in Z ₁ , and U ₂ (k) represents a voltage drop in Z ₂ .

wobei SÔC(k) ein geschätzter Wert von SOC(k) darstellt, SÔC^.(k) eine Vorausschätzung von SOC(k) darstellt, g(k) eine Kalman-Verstärkung darstellt, und ũ_L(k) ein vorhergesagter Fehler von u(k) darstellt.An estimated SOC value may be determined by using a Kalman filter, particularly an extended Kalman filter, on the system expressed by the equation of state of expression 11 and the output equation of expression 12. In other words, as shown in the following Expression 13, a pre-estimate of an SOC value may be set by adding the product of a Kalman gain g (k) and a predicted error of a terminal voltage U _L (k) to the estimated SOC value , $$S\widehat{O}C\left(k\right)=S\widehat{O}{C}^{-}\left(k\right)+G\left(k\right){\tilde{u}}_L\left(k\right).$$

where SÔC (k) represents an estimated value of SOC (k), SÔC ^. (k) represents a prediction of SOC (k), g (k) represents a Kalman gain, and ũ _L (k) represents a predicted error of u (k).

wobei C_hatt eine volle Ladekapazität der Sekundärbatterien 31 darstellt.As shown in the following expression 14, a pre-estimate of an SOC value (the first term on the right side of the expression 13) is added by adding a change caused by a current integration method (the second term on the right side of the term) to a previously estimated one SOC value (the first term on the right side of expression 14) is determined. Therefore, the Kalman filter adjusts the SOC estimated by a current integration method by reducing the estimated error. $$S\widehat{O}{C}^{-}\left(k\right)=S\widehat{O}C\left(k-1\right)+\frac{\Delta t}{C_{batt}}\times i\left(k\right).$$

where C _has a full charge capacity of the secondary batteries 31 represents.

The SOC estimation process using a Kalman filter can be schematically shown in a flowchart with three steps S1 to S3, as in FIG 5 shown.

In step S1, the latest internal state (SOC, U1, U2, R0) is determined based on the previously estimated internal state (SOC, U1, U2, R0) and current current I flowing through the current sensor 40 is estimated, estimated. At the same time, the latest internal state error information (SOC, U1, U2, R0) is estimated based on the previously estimated internal state error information (SOC, U1, U2, R0) and current I error information.

In step S2, a terminal voltage U _{L of} the secondary batteries 31 based on the latest internal state (SOC, U1, U2, R0) estimated in step S1. In particular, an OCV will be described with reference to the SOC-OCV correlation characteristics set forth in FIG 2 are calculated to find the OCV corresponding to the estimated SOC, and then the terminal voltage U _{L is} predicted based on the OCV and the total of U0, U1 and U2.

Based on the result of a comparison between the predicted terminal voltage U _L and an observed terminal voltage U _{L detected} by the voltage detection circuit 80 is detected, and the error information of the voltage detection circuit 80 For example, a Kalman gain g (k) for minimizing the error information of the latest internal state (SOC, U1, U2, R0) is estimated in step S1.

In step S3, the latest internal state (SOC, U1, U2, R0) estimated in step S1 and the internal state error information (SOC, U1, U2, R0) are determined using the Kalman G (k) gain. set. The Kalman gain g (k) is used to set the SOC estimated by a current integration method in step S1.

As described above, the Kalman filter is for estimating an internal state (SOC, U1, U2, R0) of the secondary batteries 31 with minimizing an estimated error of a state by repeating the above three steps S1 to S3 while using the data on a terminal voltage U _{L of} the secondary batteries 31 from the voltage detection circuit 80 and the data on a current I from the current sensor 40 are received.

The secondary batteries 31 however, have the ranges of small changes L1 and L2 in the SOC-OCV correlation characteristics. Since an OCV change amount relative to an SOC change amount is small in the regions of small changes L1 and L2, a small error of an estimated OCV value causes a significant error of an estimated SOC value. Therefore, the effect on the accuracy of the SOC estimate in the areas of small changes L1 and L2 is more significant than in the areas of large changes H1, H2, and H3, which can reduce the accuracy of the SOC estimate in areas L1 and L2. In other words, the SOC estimation accuracy can be significantly reduced if an estimated SOC value is set by using the Kalman gain g (k) generated in the areas of small changes L1 and L2 in step S2.

Limitation of the setting using a Kalman filter

In view of the above circumstances, the adjustment is made using the Kalman filter in the BM 50 only performed when the SOC of the secondary batteries 31 in one of the regions of large changes is H1, H2 and H3. If the SOC of the secondary batteries 31 is not in a range of large changes H1, H2 and H3, that is, the SOC is in a range of small changes L1 and L2, the adjustment using the Kalman filter is not performed.

6 FIG. 12 is a flowchart showing a six-step SOC estimation process S10 to S60. The in 6 shown SOC estimation process is by the controller 60 running, for example, the battery group 30 to monitor if the BM 50 starts.

When the process starts, the controller takes over 60 to the memory and the like to obtain the present SOC value (S10).

The control 60 then performs the SOC estimation by a current integration method (step S20). In other words, the controller 60 calculates a cumulative charge-discharge amount by integrating currents I received from the current sensor 40 be issued. The control 60 then estimates the SOC at the next time by adding an SOC change amount, which is calculated based on the cumulative charge-discharge amount, to the current SOC value obtained from the memory 63 is read. In expression 15, the first term expresses the current SOC value and the second term expresses the SOC change amount with respect to the current value. $$SOC=SO{C}_O+\frac{1}{C_{batt}}\times {\displaystyle \int I\Delta t}$$

The control 60 then determines whether the SOC value estimated by the current integration method is H1, H2 and H3 in one of the regions of large changes (step S30). In particular, the controller performs 60 the determination by comparing the SOC regions in the respective regions of large changes H1, H2 and H3 with the SOC value estimated by the current integration method.

The control 60 performs the adjustment using the Kalman filter to the SOC estimated by the current integration method when the SOC estimated by the current integration method is in a range of large changes H1, H2 and H3 (step S40).

In particular, the controller generates 60 a Kalman gain g (k) for minimizing the error information of an internal state (SOC, U1, U2, R0) of the secondary batteries 31 generated by the equivalent circuit model M based on a terminal voltage U _{L supplied} by the voltage detection circuit 80 is detected, and a terminal voltage U _L , which is predicted by the equivalent circuit model M, is estimated.

The control 60 then sets the SOC estimated by the current integration method based on the Kalman gain g (k) and the predicted error of the terminal voltage U _L (see Expression 13). The control 60 then takes the adjusted SOC as the SOC at the next time, that is, the latest SOC value.

However, if the current SOC value is not in one of the regions of large changes H1, H2, and H3, that is, if the current SOC value is in one of the regions of small changes L1 and L2, the controller 60 performs the adjustment using the Kalman filters and does not accept the SOC estimated by the current integration method as the SOC at the next time, that is, the latest SOC value.

Whenever the controller 60 performs step S40 or S50, the controller takes control 60 on the memory 63 to update the SOC value. The memory 63 thus always stores the latest SOC value.

After the step S40 or S50, the controller determines 60 whether the monitoring of secondary batteries 31 to be ended (step S60).

When monitoring the secondary batteries 31 is continued, the process goes back to step S20 and the control 60 repeats the SOC estimate through the current integration procedure. In other words, the controller 60 calculates an SOC change amount by the current integration method based on the current SOC value determined in step S40 or S50. The control 60 then estimates the SOC at the next time by adding the SOC change amount to the current SOC value.

The control 60 Then, step S30 is performed to determine whether the SOC estimated by the current integration method is H1, H2 and H3 in one of the regions of large changes. The control 60 performs the adjustment using the Kalman filter on the SOC estimated by the current integration method, and assumes the set SOC as the SOC at the next time (step S40) only when the SOC estimated by the current integration method is in a range of large changes H1, H2 and H3 is.

However, when the SOC estimated by the current integration method is not in a range of large changes H1, H2 and H3, the controller 60 assumes the SOC estimated by the current integration method as the SOC at the next time (step S50).

When monitoring the secondary batteries 31 is continued, the process goes back to step S20 and the same steps are executed again.

As described above, the controller performs 60 The SOC estimate is always passed through the current integration process regardless of the SOC range, while using the secondary batteries 31 supervised. In other words, the controller 60 performs the SOC estimation by the current integration method not only when the SOC is in one of the regions of small changes L1 and L2, but also when the SOC is in one of the regions of large changes H1, H2, and H3 is. The control 60 then performs the adjustment using the Kalman filter on the SOC estimated by the current integration method only when the SOC estimated by the current integration method is in a range of large changes H1, H2 and H3. However, if the estimated SOC is not in a range of large changes H1, H2, and H3, that is, the estimated SOC is in one of the ranges of small changes L1 and L2, the controller performs 60 setting using the Kalman filter is not enough.

If the secondary batteries 31 for example, from the point P to the in 2 shown full charge state, the controller performs 60 the setting using the Kalman filter on the state of charge estimated by the current integration method after the start of charging by as long as the estimated SOC is less than 31% (the SOC is in the range of large changes H2) and takes the set SOC as the latest SOC Value. As long as the estimated SOC is in the range of 31% to 62% (the SOC is in the range of small changes L1), the controller performs 60 does not complete the adjustment using the Kalman filter and adopts the SOC estimated by the current integration method as the latest SOC value.

As long as the estimated SOC is more than 62% and less than 68% (the SOC is in the range of large changes H1), the controller performs 60 adjusting by using the Kalman filter to the SOC estimated by the current integration method, and taking the set SOC as the latest SOC value.

As long as the estimated SOC is in the range of 68% to 97% (the SOC is in the range of small changes L2), the controller performs 60 does not complete the adjustment using the Kalman filter and adopts the SOC estimated by the current integration method as the latest SOC value. As long as the estimated SOC is more than 97% (the SOC is in the range of large changes H3), the controller performs 60 setting using the Kalman filter to the SOC estimated by the current integration method, and assuming the set SOC as the latest SOC value.

As described above, in this embodiment, the adjustment using the Kalman filter is performed only in the regions of large changes H1 to H3 in which the SOC estimation accuracy is expected to be improved by the adjustment. This can improve the SOC estimation accuracy.

In other words, the adjustment using the Kalman filter is not performed in the areas of small changes L1 and L2 in which the SOC estimation accuracy can be reduced by the adjustment. This can improve the SOC estimation accuracy.

In addition, the battery group became 30 charged and discharged at a battery temperature T of 25 ° C in experiments. Then, a comparison was made between the case where the adjustment using the Kalman filter to the SOC estimated by the current integration method was performed in all ranges of SOC values (0% to 100%) and the case where the adjustment only in the areas of large changes H1, H2 and H3 was done to find the difference between the effects in these cases.

In the 7A and 8th the dot-dash lines show the true SOC values. The dashed lines show the SOC values estimated by the current integration method. The estimated SOC values are determined by measuring currents through the battery group 30 determined with the current sensor 40 and integrating the measured values in the experiments. In this example, the current sensor points 40 a current measurement error of about 20 mA on the discharge side. Estimated errors in SOC values caused by current integration are accumulated over time.

In 7A The solid line shows the SOC values in the case where the adjustment is performed using the Kalman filter on the SOC values estimated by the current integration method in all ranges from SOC values (0% to 100%) ,

As indicated by the solid line in 7A In the regions A1 to A3 (regions of large changes), the SOC values set using the Kalman filter are closer to the true values than the SOC values estimated by the current integration method. This shows that the SOC estimation accuracy is higher in the regions of large changes in the case where the adjustment using the Kalman filter is used than in the case where only the current integration method is used. 7B provides an enlarged view of the range of large changes A3 in FIG 7A represents.

However, in the areas B1 to B4 (small change area), the SOC values set using the Kalman filter are more remote from the true values than the SOC values estimated by the current integration method. This shows that the SOC estimation accuracy is lowered when the adjustment using the Kalman filter is performed on the SOC values estimated by the current integration method in the regions of small change L1 and L2.

In 8th The solid line shows the SOC values in the case where the adjustment is performed using the Kalman filter on the SOC values estimated by the current integration method only in the regions of large changes H1 to H3, and the setting under Use of the Kalman filter is not performed in the areas of small changes L1 and L2.

As indicated by the solid line in 8th In the case where the adjustment using the Kalman filter is performed only in the regions of large changes H1 to H3, the SOC values are shown closer to the true values in general to all regions than the SOC values obtained by the current integration method can be estimated. This shows that the SOC estimation accuracy is improved.

9 represents a table showing the root mean square (RMS) of SOC values in terms of the true values in the respective methods. If only the current integration method is used, the RMS of the estimated SOC values with respect to the true values is 4.738. When the adjustment is made using the Kalman filter in all SOC ranges, the RMS of the set SOC values with respect to the true values is 6.619. When the adjustment using the Kalman filter is performed only in the areas of large changes H1 to H3, the RMS of the set SOC values with respect to the true values is 4.406. The estimated SOC values represent the SOC values estimated by the current integration method and the dashed lines in FIG 7 and 8th are shown. The set SOC values represent the SOC values set using the Kalman filter and the solid lines in the 7 and 8th are shown.

According to the table, it is clear that the RMS is the smallest when the adjustment using the Kalman filter is performed only in the areas of large changes H1 to H3, and the SOC estimation accuracy is improved.

<Second Embodiment>

A second embodiment 2 will now be described with reference to FIGS 10 to 15 described. A battery pack 20 The second embodiment includes a battery pack 30 , a current sensor 40 and a battery manager (BM) 50 for managing the battery group 30 as in the battery pack 20 the first embodiment.

In the second embodiment, the battery group became 30 as in the experiments in the first embodiment at different battery temperatures T in experiments loaded and unloaded to find the relationship between the effect of adjustment using the Kalman filter and the different battery temperatures T.

10 FIG. 12 is a graph showing SOC transitions with time at a battery temperature T of 40 ° C. FIG. 11 FIG. 12 is a graph showing SOC transitions with time at a battery temperature T of 10 ° C. FIG. 12 FIG. 12 illustrates a graph showing SOC transitions with time at a battery temperature T of 0 ° C. FIG. In the 10 to 12 The solid lines show the SOC values in the case where the adjustment is performed using the Kalman filter on the SOC values estimated by the current integration method only in the regions of large changes H1, H2 and H3.

If a comparison between the SOC transition with time at the battery temperature T of 40 ° C (the solid line in 10 ) and the SOC transition with time at the battery temperature T of 25 ° C (the solid line in FIG 8th ) is performed in the case where the adjustment using the Kalman filter is performed only in the areas of large changes H1, H2 and H3; It is understood that the set SOC values are closer to the true values and that the effect of adjustment using the Kalman filter is higher at the battery temperature of 40 ° C than at the battery temperature of 25 ° C.

Similarly, if a comparison between the SOC transition with time at the battery temperature T is 25 ° C (the solid line in FIG 8th ) and the SOC transition with time at the battery temperature T of 10 ° C (the solid line in FIG 11 ), in the case where the adjustment using the Kalman filter is performed only in the areas of large changes H1, H2 and H3, it is understood that the set SOC values are more remote from the true values, and the Effect of setting using the Kalman filter is lower at the battery temperature of 10 ° C than at the battery temperature of 25 ° C.

If a comparison between the SOC transition with time at the battery temperature T of 10 ° C (the solid line in 11 ) and the SOC transition with time at the battery temperature T of 0 ° C (the solid line in FIG 12 ) is performed in the case where the adjustment using the Kalman filter is performed only in the areas of large changes H1, H2 and H3; it is understood that the set SOC values are more remote from the true values and the setting using the Kalman filter is lower at the battery temperature T of 0 ° C than at the battery temperature T of 10 ° C.

13 Fig. 12 is a table showing the root mean square (RMS) values of the estimated SOC values with respect to the true values and the root-mean-squared values (RMS) of the set SOC values with respect to the true values at the respective battery temperatures T. The estimated SOC values represent the SOC values estimated by the current integration method and those indicated by the dashed lines in FIG 8th . 10 . 11 and 12 are shown. The set SOC values represent the SOC values set using the Kalman filter and the solid lines in the 8th . 10 . 11 and 12 are shown.

The RMS of the set SOC values with respect to the true values is 3, 879 at the battery temperature T of 40 ° C and 4.406 at the battery temperature T of 25 ° C. The RMS is 9.804 at the battery temperature T of 10 ° C and 12.038 at the battery temperature T of 0 ° C.

As shown in the table, as the battery temperature T increases, the RMS of the set SOC values decreases with respect to the true values. Accordingly, as the battery temperature T increases, the effect of adjustment using the Kalman filter increases. Specifically, the RMS of the set SOC values with respect to the true values is 4.406 at the battery temperature T of 25 ° C and 3.879 at the battery temperature T of 40 ° C. These RMS values are smaller than the respective RMS values of the SOC values estimated by the current integration method, that is, 4,738 and 4,515.

This shows that the SOC estimation accuracy can be improved when the adjustment using the Kalman filter is performed on the SOC values estimated by the current integration method at a battery temperature T of only 25 ° C or more.

The increase of the SOC estimation accuracy with the increase of the battery temperature T can be explained by the following factors. In general, the impedance of the secondary batteries is decreasing 31 when the battery temperature T increases. The reduced impedance reduces a voltage change at the time of current flow. This improves the accuracy in the estimation of the terminal voltage U _L by the estimation model M, which is considered to improve the SOC estimation accuracy. On the other hand, a high impedance at a low battery temperature T increases a voltage change at the time of current flow. This reduces the accuracy in the estimation of the terminal voltage U _L by the estimation model M, which is assumed to reduce the SOC estimation accuracy.

14 FIG. 10 illustrates a flowchart with a term related to the battery temperature T added to the terms for determining whether to perform the adjustment using the Kalman filter to an SOC estimated by the current integration method. This process includes an additional step S37A for the in 6 shown SOC estimation process. In particular, this process includes the additional step of having the controller 60 determines whether the battery temperature T detected by the temperature sensor 95 is a first temperature (25 ° C in this example) or more.

In the in 14 The SOC estimation process shown performs the control 60 setting by using the Kalman filter on an SOC estimated by the current integration method by (step S40) when the current SOC value in one of the large change regions is H1, H2, and H3, and the battery temperature T is 25 ° C or more (Step S30: YES, Step S37A: YES). The control 60 then takes the adjusted SOC as the SOC at the next time, that is, the latest SOC value.

However, if the current SOC value is not in any of the regions of large changes H1, H2, and H3 (step S30: NO) or the battery temperature T is less than 25 ° C (step S37A: NO), the controller 60 defaults the setting Use of the Kalman filter is not accepted, and assumes the SOC estimated by the current integration method as the SOC at the next time, that is, the latest SOC value.

As in 13 1, the RMS of the set SOC values with respect to the true values is 9.804 at the battery temperature of 10 ° C and 12.038 at the battery temperature T of 0 ° C. These RMS values are greater than the respective RMS values of the SOC values estimated by the current integration method, that is, 4,440 and 4,427.

This shows that the SOC estimation accuracy tends to be reduced by making the adjustment using the Kalman filter at the battery temperature T of 10 ° C or less. Therefore, the reduction of the SOC estimation accuracy can be prevented by not performing the adjustment using the Kalman filter to an SOC estimated by the current integration method at the battery temperature T of 10 ° C. or less.

15 FIG. 10 illustrates a flowchart with a term related to the battery temperature T added to the terms for determining whether to perform the adjustment using the Kalman filter to an SOC estimated by the current integration method. This process includes an additional step S37B for the in 6 shown SOC estimation process. In particular, this process includes the additional step of having the controller 60 determines whether the battery temperature T detected by the temperature sensor 95 is a second temperature (10 ° C. in this example) or less.

In the in 15 The SOC estimation process shown performs the control 60 setting by using the Kalman filter on an SOC estimated by the current integration method (step S40) when the current SOC value in one of the large change regions is H1, H2 and H3 and the battery temperature T is more than 10 ° C ( Step S30: YES, step S37B: NO). The control 60 then takes the adjusted SOC as the SOC at the next time, that is, the latest SOC value.

However, if the current SOC value is not in one of the regions of large changes H1, H2, and H3 (step S30: NO) or the battery temperature T is 10 ° C or less (step S37B: YES), the controller 60 defaults the setting Use of the Kalman filter is not accepted, and assumes the SOC estimated by the current integration method as the SOC at the next time, that is, the latest SOC value.

<Third Embodiment>

A third embodiment will now be described with reference to FIGS 16 and 17 described. A battery pack 20 In the third embodiment, a battery pack 30 includes a current sensor 40 and a battery manager (BM) 50 for managing the battery group 30 as in the battery pack 20 the first embodiment.

In the third embodiment, the battery group became 30 charged and discharged at a battery temperature T of 25 ° C. Then, a term relating to a minimum current flow state is added to the terms for determining whether to make the adjustment using the Kalman filter, the difference between the effects in the case with the additional term and the case without the term find additional term. The minimum current flow state is defined as a state in which the current through the battery group 30 is a predetermined value (for example, 100 mA) or less.

In 16 the dot-dash line shows the true SOC values. The dashed line shows the SOC values estimated by the current integration method. The estimated SOC values are determined by measuring currents through the battery group 30 determined with the current sensor 40 and integrating the measured values in the experiments.

The solid line shows the SOC values in the case where the adjustment is performed using the Kalman filter at SOC values estimated by the current integration method when the SOC value in one of the large change regions H1 to H3 and the battery pack 30 is in the minimum current flow state.

The SOC values are closer to the true values in the case where the process includes the additional term with respect to the minimum current flow state to determine whether the adjustment using the Kalman filter (the solid line in FIG 16 ) than in the case where the process does not include the additional term (the solid line in FIG 8th ). This shows that the SOC estimation accuracy is improved.

When the adjustment using the Kalman filter is performed only in the regions of large changes H1 to H3 and in the minimum current flow state, the root mean square (RMS) of the set SOC values with respect to the true values is 2.563. This RMS is smaller than 4.406, which corresponds to the RMS of the set SOC values with respect to the true values in the case where the adjustment is made also in the minimum current flow state (the case in the first embodiment). The set SOC values represent the SOC values set using the Kalman filter.

This shows that the adjustment using the Kalman filter is performed more efficiently in the minimum current flow state. In the third embodiment, the Kalman filter is used when the following two conditions (a) and (b) are satisfied.

1. (a) An SOC is in one of the regions of large changes H1, H2 and H3.
2. (b) The secondary batteries are in the minimum current flow state.

17 FIG. 12 is a flowchart of the SOC estimation process used in the third embodiment. This process includes an additional step S35 for the SOC estimation process used in the first embodiment (6). In step S35, the controller determines 60 by comparing one by the current sensor 40 measured value with a threshold, whether the secondary batteries 31 are in the minimum current flow state. In this example, the controller determines 60 in that the secondary batteries 31 are in the minimum current flow state when the current is a predetermined value or less.

In the in 17 The SOC estimation process shown determines the control 60 in step S30, whether the condition (a) is satisfied and in step S35 whether the condition (b) is satisfied. When the two conditions are satisfied, the controller 60 performs the adjustment using the Kalman filter to an SOC estimated by the current integration method (step S40). The control 60 then takes the SOC set using the Kalman filter as the SOC at the next time, that is, the latest SOC value. If at least one of the conditions is not met, the controller uses 60 does not accept the Kalman filter and adopts the SOC estimated by the current integration method as the SOC at the next time, that is, the latest SOC value (step S50).

As described above, in the third embodiment, the adjustment using the Kalman filter is performed only when an SOC is in a range of large changes H1, H2 and H3 and the battery pack 30 is in the minimum current flow state. Therefore, the SOC estimation accuracy can be improved as compared with the first embodiment.

<Fourth Embodiment>

A fourth embodiment will now be described with reference to FIGS 18 and 23 described. A battery pack 20 The fourth embodiment includes a battery pack 30 , a current sensor 40 and a battery manager (BM) 50 for managing the battery group 30 as in the battery pack 20 the first embodiment.

In the fourth embodiment, the battery group became 30 as in the experiments in the third embodiment at different battery temperatures T in experiments charged and discharged to find the relationship between the effect of the adjustment using the Kalman filter and the different battery temperatures T.

18 FIG. 12 is a graph showing SOC transitions with time at a battery temperature T of 40 ° C. FIG. 19 FIG. 12 is a graph showing SOC transitions with time at a battery temperature T of 10 ° C. FIG. 20 FIG. 12 illustrates a graph showing SOC transitions with time at a battery temperature T of 0 ° C. FIG. In the 18 to 20 The solid lines show the SOC values in the case where the adjustment is performed using the Kalman filter on the SOC values estimated by the current integration method only in the regions of large changes H1, H2 and H3, and the are in the minimum current flow state.

If a comparison between the SOC transition with time at the battery temperature T of 40 ° C (the solid line in 18 ) and the SOC transition with time at the battery temperature T of 25 ° C (the solid line in FIG 16 ) is performed in the case where the adjustment is performed using the Kalman filter; It is understood that the set SOC values are closer to the true values and that the effect of adjustment using the Kalman filter is higher at the battery temperature of 40 ° C than at the battery temperature of 25 ° C.

Similarly, if a comparison between the SOC transition with time at the battery temperature T is 25 ° C (the solid line in FIG 16 ) and the SOC transition with time at the battery temperature T of 10 ° C (the solid line in FIG 19 ) is performed in the case where the adjustment is performed using the Kalman filter; It is understood that the set SOC values are more remote from the true values and the effect of adjustment using the Kalman filter is lower at the battery temperature of 10 ° C than at the battery temperature of 25 ° C.

In addition, when a comparison between the SOC transition with time at the battery temperature T of 10 ° C (the solid line in FIG 19 ) and the SOC transition with time at the battery temperature T of 0 ° C (the solid line in FIG. 20), in the case where the adjustment is performed using the Kalman filter; it is understood that the set SOC values are more remote from the true values and the effect of the adjustment using the Kalman filter is lower at the battery temperature T of 0 ° C than at the battery temperature T of 10 ° C.

21 Fig. 12 is a table showing the root mean square (RMS) values of the estimated SOC values with respect to the true values and the root-mean-squared values (RMS) of the set SOC values with respect to the true values at the respective battery temperatures T. The estimated SOC values represent the SOC values estimated by the current integration method and the dashed lines in FIG 16 . 18 . 19 and 20 are shown. The set SOC values represent the SOC Values set using the Kalman filter and the solid lines in the 16 . 18 . 19 and 20 are shown.

The RMS of the set SOC values with respect to the true values is 2.598 at the battery temperature T of 40 ° C and 2.621 at the battery temperature T of 25 ° C. The RMS is 6.461 at the battery temperature T of 10 ° C and 8.522 at the battery temperature T of 0 ° C.

As shown in the table, as the battery temperature T increases, the RMS of the set SOC values decreases with respect to the true values. Accordingly, as the battery temperature T increases, the effect of adjustment using the Kalman filter increases. Specifically, the RMS of the set SOC values with respect to the true values is 2.621 at the battery temperature T of 25 ° C and 2.598 at the battery temperature T of 40 ° C. These RMS values are smaller than the respective RMS values of the SOC values estimated by the current integration method, that is, 4,738 and 4,515.

This shows that the SOC estimation accuracy can be improved by performing the adjustment using the Kalman filter on the SOC values estimated by the current integration method at a battery temperature T of only 25 ° C or more.

22 FIG. 10 illustrates a flowchart with a term related to the battery temperature T added to the terms for determining whether to perform the adjustment using the Kalman filter to an SOC estimated by the current integration method. This process includes an additional step S37A for the in 17 shown SOC estimation process. In particular, this process includes the additional step of having the controller 60 determines whether the battery temperature T detected by the temperature sensor 95 is the first temperature (25 ° C in this example) or more.

In the in 22 The SOC estimation process shown performs the control 60 setting by using the Kalman filter on an SOC estimated by the current integration method (step S40) when the current SOC value in one of the large change regions is H1, H2 and H3 and the battery group 30 is in the minimum current flow state and the battery temperature T is 25 ° C or more (step S30: YES, step S35: YES, step S37A: YES). The control 60 then takes the adjusted SOC as the SOC at the next time, that is, the latest SOC value.

However, if the current SOC value is not in one of the regions of large changes H1, H2 and H3 (step S30: NO) or the battery group 30 is not in the minimum current flow state (S35: NO) or the battery temperature T is lower than 25 ° C is (step S37A: NO), performs the control 60 does not complete the adjustment using the Kalman filter, and adopts the SOC estimated by the current integration method at the next time, that is, the latest SOC value.

As in 21 1, the RMS of the set SOC values with respect to the true values is 6.461 at the battery temperature T of 10 ° C. and 8.522 at the battery temperature T of 0 ° C. These RMS values are greater than the respective RMS values of the SOC values estimated by the current integration method, that is, 4,440 and 4,427.

This shows that the SOC estimation accuracy tends to be reduced by making the adjustment using the Kalman filter at the battery temperature T of 10 ° C or less. Therefore, the reduction of the SOC estimation accuracy can be prevented by not performing the adjustment using the Kalman filter to an SOC estimated by the current integration method at the battery temperature T of 10 ° C. or less.

23 FIG. 10 illustrates a flowchart with a term related to the battery temperature T added to the terms for determining whether to perform the adjustment using the Kalman filter to an SOC estimated by the current integration method. This process includes an additional step S37B for the in 17 shown SOC estimation process. In particular, this process includes the additional step of having the controller 60 determines whether the battery temperature T detected by the temperature sensor 95 is a second temperature (10 ° C. in this example) or less.

In the in 23 The SOC estimation process shown performs the control 60 setting by using the Kalman filter on an SOC estimated by the current integration method (step S40) when the current SOC value in one of the large change regions is H1, H2 and H3 and the battery group 30 is in the minimum current flow state and the battery temperature T is more than 10 ° C (step S30: YES, step S35: YES, step S37B: NO). The control 60 then takes the adjusted SOC as the SOC at the next time, that is, the latest SOC value.

However, if the present SOC value is not in one of the regions of large changes H1, H2 and H3 (step S30: NO) or the secondary batteries 31 are in the minimum current flow state (step S35: YES) or the battery temperature T is 10 ° C or less is (step S37B: YES), the controller performs 60 does not complete the adjustment using the Kalman filter, and adopts the SOC estimated by the current integration method as the SOC at the next time, that is, the latest SOC value.

Other Embodiments

• (1) In der ersten bis vierten Ausführungsform werden die Lithiumionensekundärbatterien 31 als ein Beispiel der Energiespeicher-Vorrichtung genommen. Alternativ kann die Energiespeicher-Vorrichtung eine andere elektrochemische Zelle als eine Lithiumionenbatterie darstellen, solange die alternative elektrochemische Zelle einen Bereich kleiner Änderungen, in dem das Verhältnis einer OCV-Änderung zu einer SOC-Änderung relativ klein ist und einen Bereich großer Änderungen, in dem das Verhältnis einer OCV-Änderung zu einer SOC-Änderung relativ groß ist, aufweist.
• (2) In der ersten bis vierten Ausführungsform bestimmt die Steuerung 60 basierend auf einem geschätzten SOC-Wert, ob die Sekundärbatterien 31 in einem der Bereiche großer Änderungen H1 bis H3 sind. Alternativ kann die Steuerung 60 den Bereich durch Vergleichen des Verhältnisses einer Spannungsschwankung mit einer Ladungs-Entladungs-Kapazitätsschwankung der Sekundärbatterien 31 (ΔV/ΔAh) mit einem Schwellenwert bestimmen. Mit anderen Worten, die Steuerung 60 kann den Bereich durch Berechnen einer Lade-Entlade-Kapazitätsänderung ΔAh in einer Zeiteinheit und einer Spannungsänderung ΔV in der Zeiteinheit basierend auf der Ausgabe von dem Stromsensor 40 und der Ausgabe von der Spannungserfassungsschaltung 80 bestimmen und ihr Verhältniss (ΔV/ΔAh) mit einem Schwellenwert vergleichen.
• (3) Wenn die Einstellung unter Verwendung des Kalman-Filters durchgeführt wird, hängt die SOC-Schätzgenauigkeit von der Genauigkeit der Schätzung des internen Zustands des Ersatzschaltungsmodells M der Sekundärbatterien 31 ab. Wenn die Stromwerte groß sind, neigt zum Beispiel die Genauigkeit der Schätzung des internen Zustands des Ersatzschaltungsmodells M dazu, reduziert zu werden. Die erste Ausführungsform zeigt das Beispiel, in dem die Einstellung unter Verwendung des Kalman-Filters an einen SOC durchgeführt wird, der durch das Stromintegrationsverfahren geschätzt wird, wenn die Sekundärbatterien 31 nur in einem der Bereiche großer Änderungen H1 bis H3 sind. Wenn die Genauigkeit der Schätzung des internen Zustands zum Beispiel durch große Stromwerte verringert wird, kann die Einstellung unter Verwendung des Kalman-Filters selbst dann nicht durchgeführt werden, wenn die Sekundärbatterien 31 in einem der Bereiche großer Änderungen H1 bis H3 sind.
• 4) In der ersten Ausführungsform wird das Ersatzschaltungsmodell M als ein Beispiel des Schätzmodells genommen, das einen internen Zustand (SOC, U1, U2, R0) der Sekundärbatterien 31 schätzt. Alternativ kann das Ersatzschaltungsmodell M ein anderes Modell als ein Ersatzschaltungsmodell darstellen, solange das alternative Modell ein Modell zum Schätzen eines internen Zustands (SOC, U1, U2, R0) der Sekundärbatterien 31 darstellt. Selbst wenn ein Ersatzschaltungsmodell verwendet wird, kann ein anderes Modell verwendet werden, das sich von dem in 1 gezeigten Ersatzschaltungsmodell M unterscheidet. Zum Beispiel kann ein solches Modell eine unterschiedliche Term-Anzahl (nicht zwei) von Impedanzen Z zum Simulieren von Polarisationsspannungen der Sekundärbatterien 31 aufweisen.
• (5) Die erste Ausführungsform zeigt das Beispiel, in dem ein durch das Stromintegrationsverfahren geschätzter SOC unter Verwendung des Kalman-Filters eingestellt wird. Alternativ kann ein geschätzter SOC unter Verwendung eines adaptiven digitalen Filters eingestellt werden. Mit anderen Worten, ein anderer Filter als ein Kalman-Filter, wie etwa ein adaptives digitales Filter, kann für die Einstellung verwendet werden, solange ein durch das Stromintegrationsverfahren geschätzter SOC auf der Grundlage eines beobachteten Werts einer Anschlussspannung einer Energiespeicher-Vorrichtung und einer Anschlussspannung, die von einem Schätzmodell vorhergesagt wird, das einen internen Zustand der Energiespeichervorrichtung schätzt, eingestellt wird.
• (6) In der ersten Ausführungsform wird der Kalman-Filter auf die SOC-Schätzung einer in einem Fahrzeug montierten Energiespeicher-Vorrichtung angewendet. Alternativ kann die vorliegende Erfindung auf die SOC-Schätzung einer Energiespeicher-Vorrichtung angewendet werden, die in einem Zweiradfahrzeug, einem Schienenfahrzeug, einem unterbrechungsfreien Stromversorgungssystem, einer regenerierten Energieempfangs-Vorrichtung, einer Energiespeicher-Vorrichtung zum Erzeugen natürlicher Energie, und dergleichen montiert ist. Ein Teil oder alle Funktionen der Zustandsschätz-Vorrichtung können entfernt installiert und über ein Netzwerk mit der Energiespeicher-Vorrichtung oder dem Batteriepack (Energiespeicher-Vorrichtung) verbunden sein. Die Zustandsschätz-Vorrichtung kann als ein Server in einem Netzwerk installiert sein.

The present invention is not limited to the embodiments described above with reference to the drawings. For example, the following embodiments are also included in the technical scope of the present invention.

• (1) In the first to fourth embodiments, the lithium-ion secondary batteries 31 are taken as an example of the energy storage device. Alternatively, the energy storage device may constitute another electrochemical cell than a lithium ion battery as long as the alternative electrochemical cell has a range of small changes in which the ratio of OCV change to SOC change is relatively small and a range of large changes in which Ratio of an OCV change to a SOC change is relatively large.
• (2) In the first to fourth embodiments, the controller determines 60 based on an estimated SOC value, whether the secondary batteries 31 in one of the areas of large changes H1 to H3. Alternatively, the controller 60 may determine the range by comparing the ratio of a voltage fluctuation with a charge-discharge capacity fluctuation of the secondary batteries 31 (ΔV / ΔAh) with a threshold. In other words, the controller 60 may determine the range by calculating a charge-discharge capacity change ΔAh in a unit time and a voltage change ΔV in the unit time based on the output from the current sensor 40 and the output from the voltage detection circuit 80 and compare their ratio (ΔV / ΔAh) with a threshold.
• (3) When the adjustment is performed using the Kalman filter, the SOC estimation accuracy depends on the accuracy of the estimation of the internal state of the equivalent circuit model M of the secondary batteries 31. For example, when the current values are large, the accuracy of estimating the internal state of the equivalent circuit model M tends to be reduced. The first embodiment shows the example in which the adjustment is performed using the Kalman filter to an SOC estimated by the current integration method when the secondary batteries 31 only in one of the areas of major changes are H1 to H3. For example, when the accuracy of the estimation of the internal state is lowered by large current values, the adjustment using the Kalman filter can not be performed even if the secondary batteries 31 in one of the areas of large changes H1 to H3.
• 4) In the first embodiment, the equivalent circuit model M is taken as an example of the estimation model having an internal state (SOC, U1, U2, R0) of the secondary batteries 31 underestimated. Alternatively, the spare circuit model M may represent a model other than an equivalent circuit model as long as the alternative model is a model for estimating an internal state (SOC, U1, U2, R0) of the secondary batteries 31 represents. Even if an equivalent circuit model is used, another model different from the one in FIG 1 shown equivalent circuit model M differentiates. For example, such a model may have a different term number (not two) of impedances Z for simulating polarization voltages of the secondary batteries 31 exhibit.
• (5) The first embodiment shows the example in which an SOC estimated by the current integration method is adjusted by using the Kalman filter. Alternatively, an estimated SOC may be adjusted using an adaptive digital filter. In other words, a filter other than a Kalman filter such as an adaptive digital filter may be used for the adjustment as long as an SOC estimated by the current integration method based on an observed value of a terminal voltage of an energy storage device and a terminal voltage, which is predicted by an estimation model estimating an internal state of the energy storage device.
• (6) In the first embodiment, the Kalman filter is applied to the SOC estimation of a vehicle-mounted energy storage device. Alternatively, the present invention may be applied to the SOC estimate of an energy storage device that is included in a A two-wheeled vehicle, a rail vehicle, an uninterruptible power supply system, a regenerated power receiving device, an energy storage device for generating natural energy, and the like is mounted. Some or all functions of the state estimator may be remotely installed and connected via a network to the energy storage device or the battery pack (energy storage device). The state estimator may be installed as a server in a network.

LIST OF REFERENCE NUMBERS

20:

battery Pack

30:

battery group

31:

Secondary Battery (Energy Storage Device of the Present Invention)

40:

current sensor

50:

Battery Manager (State Estimator of the Present Invention)

60:

Control (Range Determination Unit and SOC Estimation Unit of the Present Invention)

61:

CPU

63:

Storage

80:

Voltage detection circuit

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• JP 5074830 B2 [0003]

Claims (10)

A state estimation device estimating a state of an energy storage device having a range of small changes and a range of large changes, wherein an OCV change amount relative to an SOC change amount in the range of small changes is relatively small and that in the range is large Changes is relatively large, the state estimator includes: a region determining unit that determines whether the energy storage device is in the range of large changes; and an SOC estimator that estimates an SOC representing one of internal states of the energy storage device, where the SOC estimator unit executes: estimating the SOC of the energy storage device by a current integration method that integrates a current to / from the energy storage device; and an adjustment processing that sets the SOC estimated by the current integration method based on a measured terminal voltage value of the energy storage device and a terminal voltage predicted by an estimation model when the energy storage device is in the large change region, the estimation model determining the internal state of the energy storage device underestimated. State estimator after Claim 1 wherein the range of small changes includes a plateau range in which the OCV change amount relative to the SOC change amount is smaller than a predetermined value, and wherein the adjustment processing is not performed in the plateau region. State estimator after Claim 1 or 2 wherein the adjusting comprises a Kalman filter that reduces an estimation error of the internal state of the energy storage device estimated by the estimation model based on the measured terminal voltage value of the energy storage device and the terminal voltage predicted by the estimation model. State estimation device according to one of the Claims 1 to 3 wherein the estimation model comprises an equivalent circuit model of the energy storage device comprising an OCV, a conductor resistance, a first impedance simulating a short period polarization of the energy storage device, and a second impedance simulating a long period polarization of the energy storage device. State estimation device according to one of the Claims 1 to 4 wherein the SOC estimating unit executes the setting processing when the energy storage device is in the range of large changes and the energy storage device is in a minimum current flow state. State estimation device according to one of the Claims 1 to 5 a temperature detecting unit that detects a temperature of the energy storage device, wherein the SOC estimating unit performs the setting processing when the temperature of the energy storage device is a first temperature or more. State estimation device according to one of the Claims 1 to 5 A temperature sensing unit that detects a temperature of the energy storage device, wherein the SOC estimation unit does not perform the adjustment processing when the temperature of the energy storage device is a second temperature or less. State estimation device according to one of the Claims 1 to 7 wherein the range determining unit determines whether the energy storage device is in the range of large changes based on an SOC value estimated by the SOC estimating unit. State estimation device according to one of the Claims 1 to 7 wherein the area determining unit determines whether the energy storage device is in the range of large changes by comparing a voltage swing relative to a charge-discharge capacity change of the energy storage device with a threshold. A state estimation method that estimates a state of an energy storage device having a range of small changes and a range of large changes, wherein an OCV change amount relative to an SOC change amount in the range of small changes is relatively small and that in the range large Changes is relatively large, the state estimation method comprising: a range determining step that determines whether the energy storage device is in the range of large changes; and an SOC estimation step that estimates an SOC representing one of internal states of the energy storage device, the SOC estimation step comprising: estimating the SOC of the energy storage device by a current integration method that generates a current to / integrated by the energy storage device; and adjusting setting the SOC estimated by the current integration method based on a measured terminal voltage value of the energy storage device and a terminal voltage predicted by an estimation model when the energy storage device is in the range of large changes, wherein the estimation model determines the internal state of the energy storage device. Device estimates.

Images