Battery state estimation device, battery state estimation method, and battery state estimation program

Abstract

An electrochemical model computation unit 225 calculates a battery voltage between terminals of a battery as a state estimation value by using an electrochemical model of the battery including, as state variables, an open circuit voltage, an activation overvoltage, an ohmic overvoltage, and a concentration overvoltage that are each estimated on the basis of current pattern data of a flow that should be present in the battery. A temperature model computation unit 226 calculates a battery temperature of the battery as a state estimation value by using a battery temperature model including, as state variables, an open circuit voltage, the sum of an activation overvoltage and an ohmic overvoltage, and a current based on the current pattern data that are each estimated on the basis of the specifications of the battery and the current pattern data. The electrochemical model computation unit 225 further calculates, as a state estimation value, a battery voltage difference between a battery voltage at a current time step and a battery voltage at a previous time step.

Description

Battery state estimation device, battery state estimation method, and battery state estimation program

[0001] This disclosure relates to a battery state estimation device, a battery state estimation method, and a battery state estimation program for estimating battery conditions such as temperature.

[0002] With the increasing adoption of rechargeable batteries, their applications have expanded from consumer electronics (PCs, other electronic devices, etc.) to EVs (electric vehicles), PHEVs (plug-in hybrid vehicles), automotive starter batteries, electric assist bicycles, and industrial backup power supplies. In the future, their applications are expected to expand to power supplies for data centers used in generational AI, agricultural and construction machinery, electric ships, and electric aircraft. In these anticipated applications, there will likely be many situations requiring irregular and high loads. High loads necessitate the flow of large currents, which can easily lead to high temperatures due to Joule heating and other factors. Furthermore, due to the characteristics of the product, it is difficult to implement comprehensive cooling mechanisms. Therefore, battery temperature management and temperature estimation technologies will become increasingly important in the future.

[0003] Patent Document 1 discloses a method for maintaining a map describing the relationship between duration, SOC (State of Charge), temperature, and internal resistance, and for appropriately calibrating the system when there is a difference of more than a specified value between the actual temperature and the estimated temperature. In this method, the internal resistance is a function of the duration, which is the energizing time, but in actual battery usage, it is rare for a constant current to flow continuously. With this map-based method, it is difficult to predict the temperature in accordance with the behavior of the actual device, where the current is constantly fluctuating. Furthermore, since the concentration overvoltage included in the internal resistance does not contribute to heat generation, the estimation accuracy of this method is limited. In particular, the estimation accuracy decreases when the current is large or in the low SOC region.

[0004] Patent Document 2 discloses a method for estimating the internal state of a battery using a Kalman filter based on three sensor measurements: battery voltage, current flowing through the battery, and battery temperature, employing an electrochemical model and a temperature model. In the electrochemical model, there are many unknown state quantities, such as the active material concentration distribution within the particles of the positive and negative electrodes, as well as the open-circuit potential and activation overpotential, and these quantities exhibit nonlinearity. Accurately estimating the internal state of a battery from three sensor measurements is difficult.

[0005] Japanese Patent Publication No. 2013-101884 Japanese Patent Publication No. 2022-54018

[0006] This disclosure is made in light of these circumstances, and its purpose is to provide a technology that improves the accuracy of battery temperature estimation, even for high currents.

[0007] To solve the above problems, a battery state estimation device according to one embodiment of the present disclosure includes: an electrochemical model calculation unit that calculates the battery voltage between the terminals of the battery as a state estimate using an electrochemical model of the battery in which the open-circuit voltage, activation overvoltage, ohm overvoltage, and concentration overvoltage are estimated based on current pattern data that should flow through the battery, respectively; and a temperature model calculation unit that calculates the battery temperature of the battery as a state estimate using a temperature model of the battery in which the open-circuit voltage, the sum of activation overvoltage and ohm overvoltage, and the current based on the current pattern data are estimated based on the specifications of the battery and the current pattern data, respectively, as state variables. The electrochemical model calculation unit further calculates the difference in battery voltage between the battery voltage at the current time step and the battery voltage at the previous time step as a state estimate.

[0008] Furthermore, any combination of the above components, as well as any conversion of the expressions of this disclosure between devices, systems, methods, computer programs, etc., are also valid forms of this disclosure.

[0009] According to this disclosure, it is possible to improve the accuracy of battery temperature estimation, even for high currents.

[0010] This is a diagram illustrating a battery pack according to an embodiment. This is a diagram illustrating the relationship between OCV and battery voltage. This is a diagram showing an example of an SOC-OCV curve. This is a diagram summarizing the types of battery overvoltages. This is a diagram illustrating overvoltage estimation using a single-particle model. This is a diagram showing an example of a Cp-OCP curve.

[0011] Figure 1 is a diagram illustrating a battery pack 1 according to an embodiment. The battery pack 1 according to the embodiment includes a battery pack 10 and a battery management device 20. The battery pack 1 can supply power to a load 2. For example, if the load 2 is a server or storage in a data center, the battery pack 1 plays the role of a backup power source for the server or storage. When the battery pack 1 is installed in a mobility device such as an EV, the main load 2 is an inverter and a motor.

[0012] The charger 4 is connected to the commercial power grid 3 and converts the AC power input from the commercial power grid 3 into DC power of a predetermined voltage or current and supplies it to the battery pack 1. The charger 4 may also be built into the battery pack 1.

[0013] The battery pack 10 includes multiple cells E1-En connected in series. The number of cells in series is determined by the specifications of the load 2. Lithium-ion battery cells, nickel-metal hydride battery cells, lead-acid battery cells, etc., can be used as cells. Hereinafter, this specification assumes the use of lithium-ion battery cells (nominal voltage: 3.6-3.7V). In addition, multiple cells may be connected in parallel in the series stage of each cell to increase the capacity.

[0014] A switch SW1 is inserted into the power line connecting the battery pack 10 to the load 2 or charger 4 to switch between continuity and non-continuity with the load 2 or charger 4. A semiconductor switch or relay can be used for switch SW1.

[0015] The battery management device 20 includes a measurement unit 21 and a control unit 22. The measurement unit 21 is composed of an AFE (Analog Front End) IC or an ASIC (Application Specific Integrated Circuit). The control unit 22 is composed of a microcomputer. The microcomputer includes a CPU, RAM, ROM, and I / O. The control unit 22 may also be composed of a SoC (System on a Chip) that further includes any IC such as a GPU (Graphics Processing Unit), NPU (Neural Network Processing Unit), ASIC (Application Specific Integrated Circuit), or FPGA (Field Programmable Gate Array). In this embodiment, the control unit 22 implements the functions of a battery state estimation device.

[0016] The measurement unit 21 is connected to each node of the multiple series-connected cells E1-En by multiple voltage measurement lines, and measures the voltage of each cell E1-En by measuring the voltage between two adjacent voltage measurement lines.

[0017] The measurement unit 21 includes a multiplexer and an A / D converter. The multiplexer outputs the voltages of multiple cells E1-En in a predetermined order to the A / D converter. The A / D converter converts the analog voltages input from the multiplexer into digital values. The measurement unit 21 transmits the voltage values ​​of each cell E1-En, which have been converted into digital values, to the control unit 22 via a serial communication interface.

[0018] The measurement unit 21 measures the current flowing through the battery pack 10. A shunt resistor Rs is connected to the power line connecting the battery pack 10 to the load 2 or charger 4. A differential amplifier (not shown) amplifies the voltage across the shunt resistor Rs and outputs it to the A / D converter in the measurement unit 21. The A / D converter converts the analog voltage indicating the current flowing through the battery pack 10, which is input from the differential amplifier, into a digital value. The measurement unit 21 transmits the digitally converted current value to the control unit 22 via a serial communication interface.

[0019] A temperature sensor T1 (for example, a thermistor) is installed on the surface of the battery pack 10. The divided voltage of the temperature sensor T1 and a voltage divider resistor (not shown) is input to the measurement unit 21. The A / D converter in the measurement unit 21 converts the input analog voltage representing the temperature into a digital value. The measurement unit 21 transmits the converted digital temperature value to the control unit 22 via a serial communication interface.

[0020] The control unit 22 includes a measurement value acquisition unit 221, a measurement voltage difference calculation unit 222, a current pattern data holding unit 223, an SOC / OCV (Open Circuit Voltage) estimation unit 224, an electrochemical model calculation unit 225, a temperature model calculation unit 226, and a calibration unit 227.

[0021] The measurement value acquisition unit 221 acquires the voltage value of each cell E1-En, the current value flowing through the battery pack 10, and the temperature value of the battery pack 10, which are received from the measurement unit 21, as observed values ​​of the battery state. The measurement voltage difference calculation unit 222 calculates the cell voltage difference between each cell E1-En observed at the current time step and each cell E1-En observed at the previous time step. The unit time step width depends on the type of load 2 and the performance of the control unit 22, but it is desirable to set it to 1 second or less. The measurement voltage difference calculation unit 222 outputs, for example, the smallest cell voltage difference, the largest cell voltage difference, the average value of the multiple cell voltage differences, or the median value of the multiple cell voltage differences as the observed value of the battery voltage difference. Alternatively, a battery voltage difference processed using a statistical method may be used as the observed value.

[0022] The current pattern data holding unit 223 holds current pattern data that has been prepared in advance by the designer. The current pattern data is used as basic data for estimating the battery state by simulation and is created according to the type of load 2. The current pattern data may be a periodic sine wave, square wave, triangular wave, sawtooth wave, or an irregular current pattern generated randomly. Alternatively, it may be a current pattern generated by the designer based on measured data of the load 2. Furthermore, the current pattern does not have to be one prepared in advance; the current value actually applied to the load 2 during operation may be used. In this way, it is also possible to perform calculations during operation.

[0023] When battery pack 1 is used in an EV, driving pattern data used for testing the EV's battery may be used. Alternatively, multiple current pattern data may be prepared according to the usage scenario of load 2. For example, when battery pack 1 is used in an EV, current patterns may be prepared for acceleration, deceleration, low-speed driving, medium-speed driving, high-speed driving, and charging. In this case, the current pattern switching unit (not shown) estimates the vehicle state according to the current value flowing through the battery pack 10 measured by the measurement unit 21, and switches the current pattern as appropriate according to the vehicle state. Furthermore, the current pattern does not have to be one of a set, but may be the actual current value of the EV while it is running. In this way, it is also possible to perform calculations while driving. Not only EVs, but also PHEVs, electric assist bicycles, agricultural and construction machinery, electric ships, electric aircraft, etc., can be estimated using the same procedure.

[0024] The SOC / OCV estimation unit 224, the electrochemical model calculation unit 225, and the temperature model calculation unit 226 function as state estimators that estimate the battery state based on current pattern data. The SOC / OCV estimation unit 224 estimates the state of charge (SOC) of a cell based on the current pattern data that should flow through the cell, and estimates the OCV corresponding to the estimated SOC by referring to the cell's SOC-OCV curve.

[0025] The electrochemical model calculation unit 225 uses an electrochemical model of the cell that includes open-circuit voltage (OCV), activation overpotential, ohmic overpotential, and concentration overpotential, which are estimated based on current pattern data, as state variables to calculate the battery voltage between the cell terminals as a state estimate.

[0026] In this embodiment, the electrochemical model calculation unit 225 further calculates the battery voltage difference between the battery voltage at the current time step and the battery voltage at the previous time step as a state estimate. At that time, the electrochemical model calculation unit 225 calculates the battery voltage difference by assuming that the open-circuit voltage at the current time step and the open-circuit voltage at the previous time step, and the concentration overvoltage at the current time step and the concentration overvoltage at the previous time step are equal values. That is, the electrochemical model calculation unit 225 calculates the battery voltage difference as the difference between the sum of the activation overvoltage and ohm overvoltage at the current time step and the sum of the activation overvoltage and ohm overvoltage at the previous time step.

[0027] The electrochemical model calculation unit 225 may skip calculating the battery voltage difference if the current value at the current step matches the current value at the previous step, based on the current pattern data, in order to reduce the number of calculations.

[0028] The temperature model calculation unit 226 calculates the cell's battery temperature as a state estimate using a cell temperature model that includes the cell specifications, the open-circuit voltage estimated by the SOC / OCV estimation unit 224, the sum of the activation overvoltage and ohm overvoltage estimated based on the current pattern data, and the current based on the current pattern data as state variables.

[0029] The temperature model calculation unit 226 estimates the amount of heat generated between the previous time step and the current time step using the difference between the open-circuit voltage at the current time step and the open-circuit voltage at the previous time step, the sum of the activation overvoltage and the ohmic overvoltage at the current time step, and the current at the current time step.

[0030] The calibration unit 227 corrects the state estimate, which includes battery voltage, battery temperature, and battery voltage difference, using a Kalman filter or the like with the observed values, which also include battery voltage, battery temperature, and battery voltage difference. The calibration unit 227 may skip correcting the state estimate if the current value at the current step matches the current value at the previous step, based on the current pattern data.

[0031] The temperature protection unit (not shown) protects the battery pack 10 by controlling the switch SW1 to turn off if the estimated battery temperature, corrected by the Kalman filter, exceeds a preset temperature threshold. This will be explained in detail below.

[0032] The cell voltage measured by the measurement unit 21 is the sum of the open-circuit voltage and the overvoltage. There are three types of overvoltage: ohmic overvoltage and activation overvoltage, which occur only when current is flowing, and concentration overvoltage, which occurs even when no current is flowing. The overvoltage is the sum of these three types of overvoltage. Concentration overvoltage should not be included in the overvoltage of the heat generation formula (heat generation [W] = overvoltage [V] × current [A]).

[0033] Figure 2 illustrates the relationship between OCV and battery voltage. Since OCV cannot be measured directly, it is estimated using SOC estimation by current integration and an SOC-OCV map. Battery voltage is the terminal voltage between the positive and negative terminals of a working battery, and corresponds to the cell voltage described above.

[0034] When the current is stopped after the current has been turned on, the voltage rises instantaneously (see difference b in Figure 2), and this rise is the sum of the activation overvoltage and the ohmic overvoltage (= ΔV). ＩＲ ) The battery voltage at the moment the current stops has a difference from the OCV (see difference a in Figure 2), and this is the concentration overvoltage. When the current is flowing, only the sum of difference b and difference a is known. When the current stops, the individual values ​​of difference b and difference a can be determined. In order to accurately estimate only the overvoltage that contributes to heat generation, in this embodiment, the battery state estimation is performed while comparing whether the magnitude of difference b matches the actual measurement and the model, and calibrating the results.

[0035] Figure 3 shows an example of an SOC-OCV curve. SOC = 100% (right end of the horizontal axis) represents a fully charged state, and SOC = 0% (left end of the horizontal axis) represents a completely discharged state. The SOC-OCV curve of the cell is created in advance based on characteristic tests conducted by the battery manufacturer and is registered in the ROM in the control unit 22 at the time of shipment.

[0036] The SOC / OCV estimation unit 224 estimates the State of Charge (SOC) and OCV of a battery in operation. Specifically, in the first stage, the SOC / OCV estimation unit 224 estimates the SOC using the current integration method. In the second stage, the SOC / OCV estimation unit 224 estimates the OCV based on the estimated SOC by referring to a pre-prepared SOC-OCV curve. The SOC estimation using the current integration method can be performed using equation (1) or equation (2). Equation (2) is the time-discretized form of equation (1).

[0037] SOC ｋ＋１ = SOC ｋ +I / FCCΔt (2) SOC(t) is the SOC at time t, SOC ｉｎｉｔ represents the State of Charge (SOC) at the start of power application. FCC (Full Charge Capacity) is the full charge capacity and is a quantity with the dimensions of electric charge. I is the current. The subscript k is a natural number representing the time step due to time discretization. Δt represents the time step size due to time discretization.

[0038] The SOC / OCV estimation unit 224 calculates the OCV corresponding to the estimated SOC using equation (3) or equation (4). Equation (4) is the time-discretized form of equation (3).

[0039] OCV=OCV(SOC) (3) OCV ｋ =OCV(SOC) ｋ ) (4)

[0040] Figure 4 is a diagram summarizing the types of overvoltage of a battery. The total overvoltage of a battery is classified into ohmic overvoltage, activation overvoltage, and concentration overvoltage. The ohmic overvoltage includes the electronic conduction overvoltage, ion conduction overvoltage, ion diffusion overvoltage of the positive electrode, the ion conduction overvoltage and ion diffusion overvoltage of the separator, and the electronic conduction overvoltage, ion conduction overvoltage, and ion diffusion overvoltage of the negative electrode. The activation overvoltage occurs at the positive and negative electrodes, and the concentration overvoltage also occurs at the positive and negative electrodes. Thus, there are 12 types of overvoltage in a battery, and the total overvoltage of the battery is defined as the sum of the 12 types of overvoltage.

[0041] As described above, the overvoltage of a battery is the difference between the OCV and the battery voltage, and is also referred to as a voltage drop. Also, the relationship overvoltage [V] = internal resistance [Ω] × current [A] holds. In this embodiment, the concentration overvoltage is excluded from the overvoltage of the heat generation type (heat generation amount [W] = overvoltage [V] × current [A]).

[0042] In electrochemistry, the following (1)-(3) are solved. (1) Diffusion equation of Li ions inside the active materials of the positive and negative electrodes respectively ＋ (2) Activation overvoltage at the interface between the active materials and the electrolyte of the positive and negative electrodes respectively (3) Ohmic overvoltage in the positive electrode, negative electrode, and separator

[0043] In the following description, a discretized equation with a small computational load is used so that the simulation model can be mounted on a real machine and calculated with a practical calculation time using an inexpensive microcomputer. If there is a margin in the computational load, the inside of the battery may be further finely meshed and calculations of electrochemistry and temperature may be performed.

[0044] Figure 5 is a diagram for explaining overvoltage estimation using a single-particle model. For discretization, the sphere is divided into n ｐ computational elements. Here, an explanation is given with n ｐ = 5. With the coordinate of the center of the sphere as r １ and the coordinate of the outer periphery of the sphere as r ５ , r １ to r ５ are defined as shown in Figure 5. Δr ｐ is the interval between computational elements, and Δr ｐ = r ｐ / (n ｐ-1) The intervals between each calculation element do not need to be equal; for example, they may change in a geometric progression.

[0045] Lithium ion concentration C of the active material particles in the positive electrode ｐ The distribution of can be expressed by equation (5). For i ≠ 1 and 5, discretization results in equation (6). Here, discretization is performed using central differences in the position direction and forward differences in time. Here, the subscript i represents the position (i = 1 to 5), and the subscript k is a natural number representing the time step due to the discretization of time.

[0046] D ｓ，ｐ This is the diffusion coefficient of lithium ions in the positive electrode active material particles.

[0047] Since i=1 is the center of the sphere, it exhibits symmetry, and discretization yields equation (7).

[0048] At i=5, since it is the outer periphery of the active material particle, we consider that a charge transfer reaction occurs, resulting in equation (8).

[0049] I is the total current flowing through the battery, S ｐ is the total reaction area of ​​the positive electrode active material particles, z is the valence of the reaction (for a typical lithium-ion battery, z = 1), and F is the Faraday constant.

[0050] When the equations shown in equations (6), (7), and (8) are combined into a matrix, we obtain equation (9) or equation (10). Equation (10) is a simplified notation of equation (9). Hereafter, u ｋ Let I = (current I at time step k). Current I is positive in the case of discharge and negative in the case of charge.

[0051]

[0052] For simplicity, if we rearrange only the case where i = 3 in equation (6), we get equation (11).

[0053] Compared with the coefficient of the third column of the matrix, D ｐ，３１ = 0, D ｐ，３２ This is expressed by equation (12), Dｐ，３３ This is expressed by equation (13), D ｐ，３４ This is expressed by equation (14), D ｐ，３５ = 0, b ｐ３ = 0.

[0054]

[0055] column vector b ｐｉ The components are, ｐ５ It is 0 except for b. ｐ５ b is expressed by equation (15). ｐ5 = zFΔt / S ｐ (15)

[0056] Equation (9) represents the diffusion phenomenon and concentration distribution of lithium ions in the positive electrode active material particles, but the diffusion phenomenon and concentration distribution of lithium ions in the negative electrode active material particles can be derived in the same way as equations (5) to (10). Note that in the negative electrode, the subscript p is replaced with the subscript n.

[0057]

[0058] column vector b ｎｉ The components are, ｎ５ It is 0 except for b. ｎ５ b is expressed by equation (18). ｎ5 = -zFΔt / S ｎ (18)

[0059] Of equations (9) and (16), the lithium ion concentration C at the outermost periphery of the active material particle, i.e., at the interface between the active material particle and the electrolyte. ｐ，５，ｋ , C ｎ，５，ｋ These values ​​are related to the activation overpotential and concentration overpotential, which will be discussed later.

[0060] The activation overpotential of the positive electrode is determined by the Butler-Volmer equation, which is shown in equation (19).

[0061] α is the transfer coefficient, R is the gas constant, and T is the temperature. Temperature is a variable that changes during the calculation process, as will be explained later. ｐ０This is the exchange current density of the positive electrode and is a parameter that represents the activity of the electrode. The lithium ion concentration C is at the outermost periphery of the active material particles, i.e., at the interface between the active material particles and the electrolyte. ｐ，5，ｋ It is often expressed as a function of .

[0062] By solving equation (19) using, for example, Newton's method, the activation overpotential η of the positive electrode at current I can be obtained. ａｃｔ，ｐ We find the activation overpotential η of the positive electrode at current I by solving equation (19). ａｃｔ，ｐ For convenience, we will describe the process of finding this as shown in equation (20).

[0063]

[0064] The same equations (19) and (20) hold at the negative electrode. At the negative electrode, the subscript p is replaced with the subscript n.

[0065]

[0066] The overall activation overvoltage of the battery is equal to the activation overvoltage η of the positive electrode. ａｃｔ，ｐ and the activation overpotential η of the negative electrode ａｃｔ，ｎ Since it is the sum of the two, the activation overvoltage η of the entire battery ａｃｔ This is given by equation (22).

[0067]

[0068] The positive electrode, negative electrode, and separator all have the following resistance components: Positive electrode electron conduction resistance R ｐｓ , positive electrode ion conduction resistance R ｐＬ , separator ion conduction resistance R ｓＬ , negative electrode electron conduction resistance R ｎｓ , negative electrode ion conduction resistance R ｎＬ It is. Subscript ｐｓ This consists of a positive electrode and a solid electron conductor. ｐＬ The positive electrode and the liquid ion conductor are ｓＬ It consists of a separator and an ion conductor liquid. ｎｓ The negative electrode and the solid electron conductor are... ｎＬ These are derived from the negative electrode and the liquid ion conductor, respectively.

[0069] In addition to these, there are many other types of resistors, such as ion diffusion resistance, current collector foil electronic resistance, interface resistance, and tab resistance, but since they are treated the same way, this specification will limit it to the five types of resistors listed above.

[0070] Since linearity holds for ohm resistance, the ohm resistance can be calculated by taking the sum of the five types of resistances mentioned above. Sum of ohm resistances R ｏｈｍ Equation (23) is obtained.

[0071] R ｏｈｍ = R ｐｓ +R ｐＬ +R ｓＬ +R ｎｓ +R ｎＬ (23)

[0072] The ohm overvoltage η of the entire battery ｏｈｍ R is the sum of the ohm resistances. ｏｈｍ It is the product of and the current I, resulting in equation (24).

[0073] η ｏｈｍ = R ｏｈｍ I (24)

[0074] The activation overvoltage η of the entire battery shown in equation (22) ａｃｔ And the ohm overvoltage η of the entire battery shown in equation (24) ｏｈｍ When these are discretized in time, we obtain equations (25) and (26), respectively. η ｏｈｍ，ｋ = R ｏｈｍ u ｋ (26) As explained in equation (9), u ｋ = I (current).

[0075] Activation overvoltage and Ohmian overvoltage are overvoltages that arise when current flows; in other words, if no current is flowing, both activation overvoltage and Ohmian overvoltage are 0.

[0076] The concentration overpotential in positive electrode active material particles is due to the distribution of lithium ion concentration within the active material particles. The average lithium ion concentration within the positive electrode active material particles is defined by equation (27) (continuous equation) or equation (28) (discrete equation).

[0077]

[0078] The relationship between the open circuit potential (OCP: Open Circuit Potential) of the positive electrode and the lithium ion concentration C within the positive electrode active material particles will be described below. ｐ of will be described.

[0079] Fig. 6 is a diagram showing an example of a C ｐ - OCP curve. The relationship between the open circuit potential and the lithium ion concentration of the active material particles of the positive and negative electrodes generally forms a downward-sloping curve as shown in Fig. 6. When there is a difference between the lithium ion concentration C ｐ，５，ｋ at the interface between the positive electrode active material particles and the electrolyte and the average lithium ion concentration C ｐ，ａｖｅ，ｋ within the positive electrode active material particles, the difference in the open circuit potential corresponding to the difference in the lithium ion concentrations of the two is the concentration overvoltage η ｐ，Ｄｉｆｆ，ｋ .

[0080] The concentration overvoltage η ｐ，Ｄｉｆｆ，ｋ of the positive electrode is given by Equation (29), the concentration overvoltage η ｎ，Ｄｉｆｆ，ｋ of the negative electrode is given by Equation (30), and the concentration overvoltage η Ｄｉｆｆ，ｋ of the entire battery is given by Equation (31).

[0081] η ｐ，Ｄｉｆｆ，ｋ = OCP ｐ (C ｐ，ａｖｅ，ｋ ) - OCP ｐ (C ｐ，５，ｋ ) (29) η ｎ，Ｄｉｆｆ，ｋ = OCP ｎ (C ｎ，ａｖｅ，ｋ ) - OCP ｎ (C ｎ，５，ｋ ) (30) η Ｄｉｆｆ，ｋ = η ｐ，Ｄｉｆｆ (C ｐ，ａｖｅ，ｋ , C ｐ，５，ｋ ) - η n，Ｄｉｆｆ (C n，ａｖｅ，ｋ , C n，５，ｋ ) (31) OCP ｐ (C ｐ ) is the open circuit potential of the positive electrode, and OCP ｎ (C ｎ ) is the open circuit potential of the negative electrode.

[0082] As described above, the activation overvoltage η ａｃｔ，ｋ , the ohmic overvoltage η ｏｈｍ，ｋ , the concentration overvoltage ηＤｉｆｆ，ｋ The three overvoltages were determined. Therefore, the battery voltage Vk at time step k is given by equation (32).

[0083] V ｋ =OCV(SOC) ｋ ) - η ａｃｔ，ｋ -η ｏｈｍ，ｋ -η Ｄｉｆｆ，ｋ (32)

[0084] The following explains how to calculate the temperature model. The common equations in thermodynamics for calculating the temperature of a battery are equation (33) or equation (34). Equations (33) and (34) are equivalent, and either can be used. Equation (33) will be used in the following explanation.

[0085] ρ: Density of the battery [kg / m³] ３ ], Vol: Battery volume [m ３ ], Cp: Specific heat of the battery [J / kg / K], I: Current [A], T: Temperature of the battery [K], OCV: Open circuit voltage [V], h: Heat transfer coefficient [W / m] ２ / K], S: Surface area of ​​the battery [m ２ ], T ０ : Outside temperature [K], η ｏｈｍ : Ohm overvoltage [V], η ａｃｔ : Activation overpotential [V], η Ｄｉｆｆ : Concentration overvoltage [V].

[0086] The terms on the left side of equations (33) and (34) represent the temperature rise. The first term on the right side of equation (33) or equation (34) represents the amount of heat generated by the flow of current. The second term on the right side of equation (33) or equation (34) represents the entropy heat generated by the flow of current. The third term on the right side of equation (33) or equation (34) represents the heat dissipation to the outside by heat transfer. Although not shown in equations (33) or (34), radiative heat transfer may also be considered.

[0087] Outside temperature T ０ If the battery pack 1 is equipped with an ambient temperature sensor for measuring ambient temperature, the value obtained from the ambient temperature sensor can be used. If an ambient temperature sensor is not installed, for example, the average temperature of the area where the battery pack 1 is used will be used.

[0088] Equation (35) is obtained by discretizing equation (33) over time.

[0089] Temperature T ｋ The initial value is set to, for example, room temperature. ｋ Since this is updated by a Kalman filter, etc., it can be set to any value (for example, 300K at room temperature). When the current I changes, the ohmic overpotential and activation overpotential change and affect the amount of heat generated, but the concentration overpotential does not affect the amount of heat generated.

[0090] Up to this point, we have explained the calculation method for estimating the battery state using electrochemical and temperature models. Below, we will explain the observation of the battery state in an actual operating device. In this embodiment, the quantities to be observed are battery temperature, battery voltage, and the battery voltage difference value ΔV during current fluctuations. ＩＲ These are the three.

[0091] Battery temperature (T temperature model) ｋ The battery voltage (corresponding to the electrochemical model V) is observed using the measurement value of the temperature sensor T1. ｋ (This corresponds to) which is observed using the voltage sensor's measurement value. Battery voltage difference value ΔV during current fluctuation ＩＲ，ｋ As shown in equation (36), the battery voltage V observed at the current time step k is expressed as follows: ｋ And the battery voltage V observed in the previous time step (k-1) ｋ－１ That is the difference.

[0092] ΔV ＩＲ，ｋ = V ｋ -V ｋ－１ (36)

[0093] Therefore, in the battery state estimation device according to this embodiment, the observed value vector y ｏｂｓ ｋ This is given by equation (37). Hereafter, vectors will be shown in bold except within the equations.

[0094]

[0095] Battery voltage difference value ΔV ＩＲ，ｋ This can be rewritten from equations (32) and (36) as equation (38).

[0096] ΔV ＩＲ，ｋ = V ｋ -V ｋ－１ =OCV(SOC) ｋ )-OCV(SOC ｋ－１ ) - (η ａｃｔ，ｋ -η ａｃｔ，ｋ－１ ) - (η ｏｈｍ，ｋ -η ｏｈｍ，ｋ－１ ) - (η Ｄｉｆｆ，ｋ -η Ｄｉｆｆ，ｋ－１ ) (38)

[0097] Here, if Δt is sufficiently small, more specifically, if the change in current I at Δt is sufficiently small with respect to FCC, then OCV (SOC ｋ ) and OCV (SOC ｋ－１ ) can be considered to have the same value, η Ｄｉｆｆ，ｋ and η Ｄｉｆｆ，ｋ－１ These can also be considered to have the same value. Therefore, equation (38) can be simplified to equation (39).

[0098] ΔV ＩＲ，ｋ = V ｋ -V ｋ－１ = -(η) ａｃｔ，ｋ -η ａｃｔ，ｋ－１ ) - (η ｏｈｍ，ｋ -η ｏｈｍ，ｋ－１ ) (39)

[0099] During current fluctuations, the activation overpotential η ａｃｔ，ｋ and Ohm overvoltage η ｏｈｍ,k It fluctuates without time delay or wasted time, and the battery voltage difference value ΔV ＩＲ，ｋ Because it can measure quickly and directly with a voltage sensor, it is possible to accurately determine the difference between the estimated value and the observed value.

[0100] The observed vector y derived from the estimation calculation ｋ This is given by equation (40).

[0101] V ｋ Equation (32), ΔV ＩＲ，ｋ Equation (39), T ｋ These can be derived from equation (35).

[0102] By adjusting the values ​​in the electrochemical and temperature models according to the difference between equations (37) and (40), highly accurate battery temperature estimation becomes possible. A Kalman filter can be used as a method of adjustment. There are several methods, such as nonlinear Kalman filters, ensemble Kalman filters, and particle filters, but since the observation equations are nonlinear, using a particle filter provides the highest accuracy.

[0103] The following describes how to modify the battery state estimate using a particle filter. Note that nonlinear Kalman filters, ensemble Kalman filters, and fragrance-free Kalman filters are described in detail, for example, in Chapter 7 of "Fundamentals of Kalman Filters," by Shuichi Adachi and Ichiro Maruta, first edition October 1, 2012, Tokyo Denki University Press, ISBN: 9784501328900, and the battery state estimate can be modified using a procedure similar to that for particle filters.

[0104] For example, equations (2), (4), (10), (17), (25), (26), (31), (32), (35), and (39) mentioned above can be used as equations to construct the particle filter. For example, the state vector x ｋ Determine this as shown in equation (41).

[0105]

[0106] Regarding the time evolution of the state vector in equation (40), the system noise v ｋ The state equation when is added is described using a general function F as shown in equation (42). The system noise may be normal noise or irregular noise, but the noise distribution is known or can be assumed. Function F is a symbol that collectively describes equations (2), (4), (10), (17), (25), (26), (31), (32), (35), and (39).

[0107]

[0108] For the system model in equation (42), the observation model y ｋ This is written as in equation (43). In the observation model, the observation noise w ｋIt is assumed that this is added to the observed quantity. In this embodiment, since all observed quantities are simply the values ​​of the state vector, the observation model is linear, but for generality, it is described using the function H.

[0109]

[0110] Also, the state vector x ｋ Given the observed value y ｏｂｓ ｋ Let P be the probability of obtaining the system noise v. ｋ The probability P can serve as a tuning parameter to ensure stable computation.

[0111] The following describes the procedure for implementing particle filtering.

[0112] In procedure (1), N particles are generated, and the state vector x is determined for each of them. ｎ ｋ This provides the initial values. The superscript n represents the particle's serial number (1 ≤ n ≤ N). In this embodiment, the state vector at time step k also requires the values ​​at time step (k-1) for some variables. Initially, the values ​​at time step k and the values ​​at time step (k-1) should be the same.

[0113] In step (2), random numbers are used to determine the system noise v of each of the N particles. ｎ ｋ Generates.

[0114] In step (3), using the values ​​generated in steps (1) and (2), the state vector x at time step (k-1) is obtained by equation (42). ｎ ｋ－１ and system noise v ｎ ｋ From this, the state vector x of each of the N particles at the next time step ｎ ｋ The following is calculated. At this stage, the state vector is a provisional value before filtering.

[0115] In step (4), the weight of the nth particle is expressed by equation (44).

[0116] The right-hand side is the state vector x ｎ ｋ If you obtain the observed value y ｏｂｓ ｋ This represents the probability P of obtaining [the desired result].

[0117] In step (5), the weights of all particles obtained in step (4) are normalized. The normalized particle weight p ｎ ｋ This is expressed by equation (45).

[0118]

[0119] In step (6), the state vectors of the n particles are averaged by the particle weights, and the state vectors are updated as shown in equation (46).

[0120] The state vector on the left side has a different value from the state vector in equation (42), and is a value filtered using the observed values.

[0121] In step (7), for example, the value of the right-hand side of equation (40) may be estimated from the observation vector shown in equation (43) based on equation (46). Alternatively, (ΔV in the state vector of equation (46) ＩＲ，ｋ ,V ｋ ,T ｋ ) can be used as the value on the right side of equation (40).

[0122] The above describes the difference in battery voltage ΔV during current fluctuations. ＩＲ The calculation and the correction process for the estimated battery state were explained. In this regard, the correction process for the estimated battery state may be performed only when the current value changes. That is, u ｋ ≠u ｋ－１ The battery state estimate may be corrected only in this case.

[0123] Also, when the current is interrupted, u ｋ = 0 and u ｋ－１ Since ≠ 0, in this case the activation overvoltage and ohm overvoltage at time step (k-1) are both 0. Therefore, equation (39) can be rewritten as equation (47) and further simplified. In this case, more accurate battery state estimation becomes possible.

[0124] ΔVＩＲ，ｋ = V ｋ -V ｋ－１ = V ｋ = -η ａｃｔ，ｋ -η ｏｈｍ，ｋ (47)

[0125] Although not mentioned in the above explanation, a certain time delay may occur in the activation overpotential response due to the capacitive component at the solid-liquid interface. Therefore, it may be preferable to set the time step size used for battery state estimation in the embodiment to be larger than the time constant determined by the activation overpotential and the capacitance at the solid-liquid interface.

[0126] As described above, according to this embodiment, the battery voltage difference value ΔV during current fluctuation ＩＲ By adding this to the state variables, the accuracy of temperature estimation for batteries, even those with high currents, can be improved. Because temperature estimation is performed using an electrochemical model, it can handle constant fluctuations in current. Furthermore, by using a temperature model that eliminates concentration overpotentials, highly accurate heat generation simulations can be performed.

[0127] The present disclosure has been described above based on embodiments. The embodiments are illustrative, and it will be understood by those skilled in the art that various modifications are possible in combinations of their components and processing processes, and that such modifications are also within the scope of the present disclosure.

[0128] In the embodiment described above, an example was explained in which the battery state estimation device is implemented in the control unit 22 within the battery pack 1. However, the battery state estimation device may also be implemented in a cloud server installed in a data center or in a company's own server installed in its own facility. In that case, the control unit 22 within the battery pack 1 transmits the voltage values ​​of each cell E1-En, the current values ​​flowing through the battery pack 10, and the temperature values ​​of the battery pack 10, received from the measurement unit 21, to the cloud server or the company's own server via wireless or wired communication. In this case, more precise electrochemical and temperature models can be used.

[0129] The embodiments may be specified by the following items.

[0130] [Item 1] A battery state estimation device (22) comprising: an electrochemical model calculation unit (225) that calculates the battery voltage between the terminals of the battery (E1) as a state estimate value using an electrochemical model of the battery (E1) that includes open-circuit voltage, activation overvoltage, ohm overvoltage, and concentration overvoltage as state variables, which are estimated based on current pattern data that should flow through the battery (E1); and a temperature model calculation unit (226) that calculates the battery temperature of the battery (E1) as a state estimate value using a temperature model of the battery (E1) that includes open-circuit voltage, the sum of activation overvoltage and ohm overvoltage, and current based on the current pattern data as state variables, which are estimated based on the specifications of the battery (E1) and the current pattern data, wherein the electrochemical model calculation unit (225) further calculates the difference in battery voltage between the battery voltage at the current time step and the battery voltage at the previous time step as a state estimate value. According to this, the battery state can be described with high accuracy with a relatively small number of parameters. [Item 2] The battery state estimation device (22) described in Item 1, further comprising: a measurement value acquisition unit (221) that acquires the battery voltage measured by a voltage sensor (21) between the terminals of the battery (E1) and the temperature measured by a temperature sensor as observed values; and a measurement voltage difference calculation unit (222) that calculates the difference between the battery voltage observed at the current time step and the battery voltage observed at the previous time step as an observed value. According to this, by calculating the battery voltage difference using the observed battery voltage, a comparison target with the state estimation value including the battery voltage difference can be generated. [Item 3] The battery state estimation device (22) described in Item 2, further comprising: a calibration unit (227) that corrects the state estimation value including the battery voltage, the battery temperature, and the battery voltage difference using observed values ​​including the battery voltage, the battery temperature, and the battery voltage difference. According to this, the battery state can be estimated with high accuracy by comparing the state estimation value with the battery voltage difference included in the observed value.[Item 4] The battery state estimation device (22) described in Item 3, wherein the calibration unit (227) corrects the estimated state value, including the battery temperature and the battery voltage difference, using one of a Kalman filter, a nonlinear Kalman filter, an unscented Kalman filter, an ensemble Kalman filter, or a particle filter. According to this, the estimated state value can be corrected with high accuracy by using a Kalman filter. [Item 5] If the current value at the current time step and the current value at the previous time step match, based on the current pattern data, the electrochemical model calculation unit (225) skips calculating the battery voltage difference, and the calibration unit (227) skips correcting the estimated state value. According to this, the amount of calculation can be reduced. [Item 6] The battery state estimation device (22) described in Item 1, wherein the electrochemical model calculation unit (225) calculates the difference between the sum of the activation overvoltage and the ohm overvoltage at the current time step and the sum of the activation overvoltage and the ohm overvoltage at the previous time step as the battery voltage difference. This reduces the amount of computation. [Item 7] The temperature model calculation unit (226) estimates the amount of heat generated between the previous time step and the current time step using the difference between the open-circuit voltage at the current time step and the open-circuit voltage at the previous time step, the sum of the activation overvoltage and the ohmic overvoltage at the current time step, and the current at the current time step, as described in Item 1, the battery state estimation device (22). This makes it possible to estimate the increase in heat generated from the previous time step with high accuracy by eliminating concentration overvoltage.[Item 8] A battery state estimation method comprising: a step of calculating the battery voltage between the terminals of the battery (E1) as a state estimate using an electrochemical model of the battery (E1) in which the open-circuit voltage, activation overvoltage, ohm overvoltage, and concentration overvoltage are estimated based on current pattern data that should flow through the battery (E1); a step of calculating the battery temperature of the battery (E1) as a state estimate using a temperature model of the battery (E1) in which the open-circuit voltage, the sum of activation overvoltage and ohm overvoltage, and the current based on the current pattern data are estimated based on the specifications of the battery (E1) and the current pattern data as state variables; and a step of calculating the difference in battery voltage between the battery voltage at the current time step and the battery voltage at the previous time step as a state estimate. According to this method, the battery state can be described with high accuracy using a relatively small number of parameters. [Item 9] A battery state estimation program that causes a computer to perform the following steps: calculate the battery voltage between the terminals of the battery (E1) as a state estimate using an electrochemical model of the battery (E1) that includes open-circuit voltage, activation overvoltage, ohm overvoltage, and concentration overvoltage as state variables, which are estimated based on current pattern data that should flow through the battery (E1); calculate the battery temperature of the battery (E1) as a state estimate using a temperature model of the battery (E1) that includes open-circuit voltage, the sum of activation overvoltage and ohm overvoltage, and current based on the current pattern data as state variables, which are estimated based on the specifications of the battery (E1) and the current pattern data; and calculate the battery voltage difference between the battery voltage at the current time step and the battery voltage at the previous time step as a state estimate. According to this, the battery state can be described with high accuracy using a relatively small number of parameters.

[0131] This disclosure can be used to estimate battery conditions such as temperature.

[0132] 1 Battery pack, 2 Load, 3 Commercial power grid, 4 Charger, 10 Battery set, 20 Battery management device, 21 Measurement unit, 22 Control unit, 221 Measurement value acquisition unit, 222 Measurement voltage difference calculation unit, 223 Current pattern data holding unit, 224 SOC / OCV estimation unit, 225 Electrochemical model calculation unit, 226 Temperature model calculation unit, 227 Calibration unit, E1-En Cell, Rs Shunt resistor, SW1 Switch, T1 Temperature sensor.

Claims

1. A battery state estimation device comprising: an electrochemical model calculation unit that calculates the battery voltage between the terminals of the battery as a state estimate using an electrochemical model of the battery in which the open-circuit voltage, activation overvoltage, ohm overvoltage, and concentration overvoltage, each estimated based on current pattern data that should flow through the battery, are state variables; and a temperature model calculation unit that calculates the battery temperature of the battery as a state estimate using a temperature model of the battery in which the open-circuit voltage, the sum of the activation overvoltage and ohm overvoltage, and the current based on the current pattern data, each estimated based on the specifications of the battery and the current pattern data, are state variables, wherein the electrochemical model calculation unit further calculates the difference in battery voltage between the battery voltage at the current time step and the battery voltage at the previous time step as a state estimate.

2. The battery state estimation device according to claim 1, further comprising: a measurement value acquisition unit that acquires the battery voltage measured by a voltage sensor between the terminals of the battery and the temperature measured by a temperature sensor as observed values; and a measurement voltage difference calculation unit that calculates the difference between the battery voltage observed in the current time step and the battery voltage observed in the previous time step as an observed value.

3. The battery state estimation device according to claim 2, further comprising a calibration unit that corrects the state estimation value, which includes the battery voltage, the battery temperature, and the battery voltage difference, using observed values, which include the battery voltage, the battery temperature, and the battery voltage difference.

4. The battery state estimation device according to claim 3, wherein the calibration unit corrects the state estimation value, including the battery temperature and the battery voltage difference, using one of a Kalman filter, a nonlinear Kalman filter, a fragrance-free Kalman filter, an ensemble Kalman filter, or a particle filter.

5. The battery state estimation device according to claim 3, wherein, if the current value at the current step matches the current value at the previous time step based on the current pattern data, the electrochemical model calculation unit skips calculating the battery voltage difference, and the calibration unit skips correcting the state estimation value.

6. The battery state estimation device according to claim 1, wherein the electrochemical model calculation unit calculates the difference between the sum of the activation overvoltage and the ohmic overvoltage at the current time step and the sum of the activation overvoltage and the ohmic overvoltage at the previous time step as the battery voltage difference.

7. The battery state estimation device according to claim 1, wherein the temperature model calculation unit estimates the amount of heat generated between the previous time step and the current time step using the difference between the open-circuit voltage at the current time step and the open-circuit voltage at the previous time step, the sum of the activation overvoltage and the ohmic overvoltage at the current time step, and the current at the current time step.

8. A battery state estimation method comprising: a step of calculating the battery voltage between the terminals of the battery as a state estimate using an electrochemical model of the battery in which the open-circuit voltage, activation overvoltage, ohm overvoltage, and concentration overvoltage, each estimated based on current pattern data that should flow through the battery, are state variables; a step of calculating the battery temperature of the battery as a state estimate using a temperature model of the battery in which the open-circuit voltage, the sum of the activation overvoltage and ohm overvoltage, and the current based on the current pattern data, each estimated based on the specifications of the battery and the current pattern data, are state variables; and a step of calculating the battery voltage difference between the battery voltage at the current time step and the battery voltage at the previous time step as a state estimate.

9. A battery state estimation program that causes a computer to perform the following steps: calculate the battery voltage between the terminals of the battery as a state estimate using an electrochemical model of the battery in which the open-circuit voltage, activation overvoltage, ohmic overvoltage, and concentration overvoltage, each estimated based on current pattern data that should flow through the battery, are state variables; calculate the battery temperature of the battery as a state estimate using a temperature model of the battery in which the open-circuit voltage, the sum of the activation overvoltage and ohmic overvoltage, and the current based on the current pattern data, each estimated based on the specifications of the battery and the current pattern data, are state variables; and calculate the battery voltage difference between the battery voltage at the current time step and the battery voltage at the previous time step as a state estimate.

Images