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

Abstract

An equivalent circuit model generation unit 112 generates an equivalent circuit model that is configured to include an open circuit voltage (OCV) and an n-stage (n is a natural number) RC parallel circuit by fitting measured voltage values at a plurality of points of a secondary battery during a pause period after charging and discharging of the secondary battery. A curve shape determination unit 113 determines whether or not the voltage relaxation curve after the completion of charging and discharging of the secondary battery has a complex shape, on the basis of a degree of matching between a transition of predicted voltage values for the pause period which are predicted from an equivalent circuit model generated by fitting measured voltage values of at least a partial section of the pause period and a transition of the measured voltage values for the pause period. When it has been determined that the voltage relaxation curve has a complex shape, the equivalent circuit model generation unit 112 changes the processing from default processing to special processing.

Description

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

[0001] The present disclosure relates to a battery state estimation system, a battery state estimation method, and a battery state estimation program that estimate the internal state of a secondary battery.

[0002] In order to estimate the internal state of a secondary battery, such as its SOC (State Of Charge), FCC (Full Charge Capacity), and SOH (State Of Health), with high accuracy, it is necessary to estimate the OCV (Open Circuit Voltage) of the secondary battery with high accuracy as a prerequisite.

[0003] Secondary batteries are electrochemical products, and when a charging current flows through a secondary battery, the measured voltage rises nonlinearly, and when a discharging current flows through a secondary battery, the measured voltage drops nonlinearly. The voltage measured when current is flowing through a secondary battery is called the closed circuit voltage (CCV) or operating voltage. After charging or discharging is completed, a secondary battery slowly relaxes (converges) to an OCV that does not contain overvoltage components. The relaxation time to the OCV depends on the cell type, temperature, SOH, etc.

[0004] In order to accurately estimate the OCV when the rest period after the end of charging and discharging is short, it is conceivable to generate an equivalent circuit model by fitting the actually measured voltage value and predict the subsequent voltage relaxation. However, the voltage relaxation curve does not have an exponential shape, and although it matches the actually measured value to some extent during the fitting period, the predicted value after the fitting period may deviate from the actually measured value.

[0005] Patent Document 1 discloses a battery state measuring device. The battery state measuring device detects the discharge current of a secondary battery during continuous high-current discharge and calculates the area of ​​the current waveform obtained from the detected value. The area of ​​the current waveform is calculated, for example, by summing detected values ​​sampled within a predetermined period. The device also obtains the differential voltage between the secondary battery's voltage and a reference voltage and calculates the area of ​​the voltage waveform obtained from the obtained value. The area of ​​the voltage waveform is calculated, similar to the area of ​​the current waveform, by summing the values ​​obtained over the same period. Next, the device calculates an area ratio, which is the ratio of the area of ​​the current waveform to the area of ​​the voltage waveform. The difference between the previous area ratio and the current area ratio (Δ area ratio) is calculated for this area ratio. The device then determines whether the secondary battery is good or bad based on the Δ area ratio. For example, a threshold value is set for determining whether the battery is good or bad, and if the Δ area ratio exceeds the threshold value, the battery is determined to be defective. However, the battery state measuring device does not calculate the ratio between the area derived from the transition curve of the predicted voltage value during a pause period of an equivalent circuit model and the area derived from the transition curve of the measured voltage value during the pause period.

[0006] JP 2010-060381 A

[0007] The present disclosure has been made in consideration of these circumstances, and its purpose is to provide a technology for improving the prediction accuracy when predicting a voltage relaxation curve by fitting voltage measurement values ​​after charging and discharging a secondary battery.

[0008] In order to solve the above problem, a battery state estimation system according to an embodiment of the present disclosure includes: an equivalent circuit model generation unit that generates an equivalent circuit model including an OCV and an n-stage RC parallel circuit (n is a natural number) by fitting voltage measurement values ​​at multiple points of the secondary battery during a rest period after charging or discharging the secondary battery; and a curve shape determination unit that determines whether a voltage relaxation curve after charging or discharging of the secondary battery has a complex shape based on a degree of agreement between a transition of a predicted voltage value during the rest period predicted from the equivalent circuit model generated by fitting the voltage measurement values ​​for at least a portion of the rest period and a transition of the voltage measurement values ​​during the rest period. If the equivalent circuit model generation unit determines that the voltage relaxation curve has a complex shape, the equivalent circuit model generation unit switches from default processing to special processing.

[0009] Any combination of the above components, and conversion of the expression of the present disclosure into an apparatus, system, method, computer program, etc., are also valid aspects of the present disclosure.

[0010] According to the present disclosure, it is possible to improve the prediction accuracy when predicting a voltage relaxation curve by fitting voltage measurement values ​​after charging and discharging a secondary battery.

[0011] 9 is a diagram for explaining a battery-equipped device according to an embodiment. It is a diagram showing an example of the configuration of a battery state estimation system according to an embodiment. FIGS. 3(a)-(c) are diagrams showing an example of an equivalent circuit model of each cell constituting an assembled battery in a battery pack. It is a diagram showing an example of a transition curve predicted from a cell equivalent circuit model generated by fitting a transition curve of voltage measurement values ​​at multiple points after the end of cell discharge and voltage measurement values ​​at multiple points during a rest period after discharge. It is a diagram showing an example of a voltage relaxation curve of a general shape after the end of cell discharge. It is a diagram showing an example of a voltage relaxation curve of a complex shape after the end of cell discharge. It is a diagram showing an example of a graph plotting the logarithm of the time derivative of voltage measurement values ​​after the end of cell discharge. It is a diagram showing an example of a transition curve predicted from a cell equivalent circuit model generated by fitting a transition curve of voltage measurement values ​​at multiple points after the end of cell discharge used in FIG. 7 and voltage measurement values ​​at multiple points during a rest period after discharge. It is a flowchart showing the flow of OCV estimation processing by a battery state estimation system according to an embodiment. It is a flowchart showing a subroutine of special processing in step S15 of FIG.

[0012] 1 is a diagram illustrating a battery-equipped device 2 according to an embodiment. The battery-equipped device 2 according to the embodiment is a device equipped with a chargeable and dischargeable battery pack 30. Examples of the battery-equipped device 2 include consumer information devices (e.g., PCs, tablets, and smartphones), home appliances (e.g., cleaning robots), electric cars, electric motorcycles, electric bicycles, electric kick scooters, and multicopters (drones).

[0013] The battery-equipped device 2 includes a control unit 21, a load unit 22, a charging unit 23, a communication unit 24, and a battery pack 30. The control unit 21 controls the entire battery-equipped device 2. The functions of the control unit 21 can be realized by a combination of hardware and software resources, or by hardware resources alone. Hardware resources that can be used include a CPU, ROM, RAM, GPU (Graphics Processing Unit), NPU (Neural Network Processing Unit), ASIC (Application Specific Integrated Circuit), FPGA (Field Programmable Gate Array), and other LSIs. Software resources that can be used include programs such as an operating system and applications.

[0014] The load unit 22 is a general term for components (excluding the control unit 21, the charging unit 23, and the communication unit 24) that consume power in the battery-equipped device 2. The charging unit 23 is connected to the commercial power grid 4, and converts AC power input from the commercial power grid 4 into DC power of a predetermined voltage or current and outputs it.

[0015] Charging unit 23 includes an AC / DC converter and a DC / DC converter. The DC / DC converter (e.g., a switching regulator) controls the voltage or current of the DC power supplied from the AC / DC converter in accordance with a voltage command value or a current command value supplied from control unit 21, and outputs the voltage or current to at least one of load unit 22 and battery pack 30. When charging is performed from an externally installed quick charger, charging is performed using DC power.

[0016] The battery pack 30 includes a battery assembly 31 and a battery management device 32. The battery assembly 31 includes multiple cells E1-En connected in series. The number of cells connected in series is determined by the specifications of the load unit 22. The cells may be lithium-ion battery cells, nickel-metal hydride battery cells, lead battery cells, or the like. In the following description, an example using lithium-ion battery cells is assumed. Note that, in each series stage of cells, multiple cells may be connected in parallel to increase capacity.

[0017] A switch SW1 that switches between electrical continuity with the load unit 22 or the charging unit 23 is inserted in a power line connecting the battery pack 31 to the load unit 22 or the charging unit 23. A semiconductor switch or a relay can be used as the switch SW1.

[0018] The battery management device 32 includes a measurement unit 33 and a control unit 34. The measurement unit 33 is configured with an AFE (Analog Front End) IC or an ASIC (Application Specific Integrated Circuit), and the control unit 34 is configured with a microcontroller.

[0019] The measurement unit 33 is connected to each node of the multiple cells E1-En connected in series by multiple voltage measurement lines, and measures the voltage of each cell E1-En by measuring the voltage between each two adjacent voltage measurement lines.

[0020] The measurement unit 33 includes a multiplexer and an A / D converter. The multiplexer outputs the voltages of the multiple cells E1-En to the A / D converter in a predetermined order. The A / D converter converts the analog voltages input from the multiplexer into digital values. The measurement unit 33 transmits the voltage values ​​of the cells E1-En, converted into digital values, to the control unit 34 via a serial communication interface.

[0021] The measurement unit 33 measures the current flowing through the battery pack 31. A shunt resistor Rs is connected to a power line connecting the battery pack 31 to the load unit 22 or the charging unit 23. A differential amplifier (not shown) amplifies the voltage across the shunt resistor Rs and outputs it to an A / D converter in the measurement unit 33. The A / D converter converts the analog voltage indicating the current flowing through the battery pack 31, which is input from the differential amplifier, into a digital value. The measurement unit 33 transmits the current value converted into a digital value to the control unit 34 via a serial communication interface.

[0022] A temperature sensor T1 (e.g., a thermistor) is installed on the surface of the battery pack 31. A divided voltage between the temperature sensor T1 and a voltage dividing resistor (not shown) is input to a measurement unit 33. An A / D converter in the measurement unit 33 converts the input analog voltage indicating the temperature into a digital value. The measurement unit 33 transmits the converted digital temperature value to the control unit 34 via a serial communication interface.

[0023] The control unit 34 manages the states of the cells E1-En based on the voltage values ​​of the cells E1-En, the current values ​​flowing through the battery pack 31, and the temperature values ​​of the battery pack 31 received from the measurement unit 33. When the control unit 34 detects overcharge, overdischarge, overcurrent, abnormally high temperature, or abnormally low temperature, it sends a shutoff signal for the switch SW1 to the measurement unit 33 to turn off the switch SW1.

[0024] The control unit 34 executes programs such as firmware within the microcontroller to achieve the following functions. The control unit 34 estimates the SOC by combining the OCV method and the current integration method. The OCV method is a method for estimating the SOC based on the measured cell OCV and the cell's SOC-OCV curve. The cell's SOC-OCV curve is created in advance by the battery manufacturer based on characteristic tests and is registered in the control unit 34 at the time of shipment.

[0025] The current integration method is a method for estimating the SOC based on the OCV at the start of cell charging and discharging and the integrated value of the measured current. With the current integration method, current measurement errors accumulate as the charging and discharging time increases. Therefore, it is preferable to use a weighted average of the SOC estimated by the current integration method and the SOC estimated by the OCV method.

[0026] The control unit 34 transmits battery data including the voltage, current, temperature, and SOC of each cell of the battery pack 30 to the control unit 21 at a predetermined interval (for example, every 10 seconds) via a network inside the device. Note that if the number of cells E1-En connected in series is large, only the maximum and minimum voltages of the voltages of the cells E1-En may be transmitted.

[0027] The communication unit 24 is an external communication interface (for example, a network interface card (NIC)) for connecting to an external network 5 via a wired or wireless connection. The control unit 21 accesses the network 5 via the communication unit 24 and transmits the battery data received from the control unit 34 of the battery pack 30 to the battery state estimation system 1 at a predetermined transmission period (for example, every 10 seconds). The control unit 21 may store the received battery data in an internal memory and transmit the battery data stored in the memory all at once at a predetermined timing.

[0028] The network 5 is a general term for communication paths such as the Internet, a dedicated line, and a Virtual Private Network (VPN), and the communication medium and protocol are not important. Examples of communication media that can be used include a wired LAN, a wireless LAN, a mobile phone network, an optical fiber network, an ADSL network, and a CATV network. Examples of communication protocols that can be used include TCP (Transmission Control Protocol) / IP (Internet Protocol), UDP (User Datagram Protocol) / IP, and Ethernet (registered trademark).

[0029] The battery state estimation system 1 is a system for estimating and analyzing the state of a battery pack 30 mounted in a battery-equipped device 2. In this embodiment, the battery state estimation system 1 is built on a cloud server installed in a data center managed by a cloud service provider. A battery analysis service provider that provides an analysis service for the battery pack 30 uses the cloud server by entering into a contract with the cloud service provider. Note that the battery state estimation system 1 may also be built on the battery analysis service provider's own server installed in its own facility or data center.

[0030] 2 is a diagram showing an example of the configuration of a battery state estimation system 1 according to an embodiment. The battery state estimation system 1 includes a control unit 11, a storage unit 12, and a communication unit 13. The communication unit 13 is an external communication interface for connecting to a network 5 via a wired or wireless connection.

[0031] The control unit 11 includes a data acquisition unit 111, an equivalent circuit model generation unit 112, a curve shape determination unit 113, a prediction unit 114, and a battery state estimation unit 115. The functions of the control unit 11 can be realized by a combination of hardware resources and software resources, or by hardware resources alone. Examples of hardware resources that can be used include a CPU, ROM, RAM, GPU, NPU, ASIC, FPGA, and other LSIs. Examples of software resources that can be used include programs such as an operating system and applications.

[0032] The storage unit 12 includes a non-volatile recording medium such as an HDD or SSD, and stores various data. The storage unit 12 includes a battery data holding unit 121. The data acquisition unit 111 acquires battery data of the battery pack 30 installed in the battery-equipped device 2 from the battery-equipped device 2 via the network 5. The data acquisition unit 111 saves the acquired battery data in the battery data holding unit 121.

[0033] When it is desired to estimate the OCV before the end of polarization relaxation during a rest period after charging or discharging the battery pack 30, the equivalent circuit model generation unit 112 acquires voltage measurement values ​​at multiple points of each of the cells E1-En during the rest period after charging or discharging the battery pack 30. The voltage measurement value of each of the cells E1-En indicates the voltage across each of the cells E1-En. In the following description, an example is assumed in which an equivalent circuit model is generated for each of the cells E1-En, but it is also possible to generate only equivalent circuit models for the maximum voltage cell and the minimum voltage cell, or to generate only an equivalent circuit model for the minimum voltage cell.

[0034] The equivalent circuit model generation unit 112 generates an equivalent circuit model for each of the cells E1-En by fitting voltage measurement values ​​at multiple points for each of the cells E1-En. In this embodiment, the equivalent circuit model for each of the cells E1-En is generated using only the terminal voltages measured during the rest period, without using the terminal voltages measured during the charge / discharge period.

[0035] 3(a)-(c) are diagrams showing examples of equivalent circuit models of the cells E1-En constituting the assembled battery 31 in the battery pack 30. In this embodiment, the transition of voltage relaxation during a rest period after charging and discharging of the cells is predicted using an equivalent circuit model. As shown in FIG. 3(a), the equivalent circuit model of a cell during a rest period after charging and discharging, when no current is flowing, is composed of an OCV and an n-stage (n is a natural number) RC parallel circuit connected in series. In the equivalent circuit model shown in FIG. 3(a), the terminal voltage V during the rest period can be described by the following (Equation 1):

[0036] V = OCV + U １ ・exp(-t / τ １ ) + U ２ ・exp(-t / τ ２ ) + ... + U ｎ ・exp(-t / τ ｎ ) ... (Formula 1) τ ｊ =R ｊ C ｊ

[0037] The equivalent circuit model shown in Fig. 3(b) is a third-order equivalent circuit model in which three stages of RC parallel circuits are connected in series, while the equivalent circuit model shown in Fig. 3(c) is a first-order equivalent circuit model in which a single stage of RC parallel circuit is used.

[0038] 4 shows an example of a transition curve predicted from an equivalent circuit model of a cell generated by fitting a transition curve of voltage measurement values ​​at multiple points after the end of discharge of the cell to a transition curve of voltage measurement values ​​at multiple points during a rest period after discharge. In the example shown in FIG. 4, the equivalent circuit model is generated by fitting voltage measurement values ​​during a rest period of 3 to 10 minutes from the end of discharge.

[0039] Specifically, as shown in the following (Equation 2), a third-order equivalent circuit model in which the degree of freedom of the time constant τ is simply set to one is fitted by the least squares method. In the following (Equation 2), the second-order time constant τ is set to 1 / 9 times the first-order time constant τ, and the third-order time constant τ is set to 1 / 25 times the first-order time constant τ.

[0040] V = OCV + U １ ・exp(-t / τ)+U ２・exp(-9t / τ)+U3・exp(-25t / τ)...(Formula 2)

[0041] In the example shown in Fig. 4, during the rest period (3 to 10 minutes after the end of discharge), the transition of the predicted voltage value predicted by the equivalent circuit model and the transition of the actual measured voltage value nearly match, but after 10 minutes from the end of discharge, the two diverge. In the example shown in Fig. 4, the measured voltage value after the end of discharge describes a voltage relaxation curve with a complex shape.

[0042] For example, in a lithium-ion battery cell using a mixed material of carbon (C) and silicon (Si) for the negative electrode, two-stage voltage relaxation may occur after the end of charging and discharging. Furthermore, the voltage may relax exponentially for a certain period of time, and then change in a nearly linear pattern. In this embodiment, in the case of complex voltage relaxation, the equivalent circuit model generation unit 112 performs special processing.

[0043] The equivalent circuit model generation unit 112 generates an equivalent circuit model by fitting voltage measurement values ​​for at least a portion of a rest period after the end of charging and discharging of each of the cells E1-En. The curve shape determination unit 113 determines whether the voltage relaxation curve after charging and discharging of each of the cells E1-En has a complex shape based on the degree of agreement between the transition of the predicted voltage value during the rest period predicted from the generated equivalent circuit model and the transition of the measured voltage value during the rest period.

[0044] For example, the equivalent circuit model generation unit 112 generates an equivalent circuit model by fitting voltage measurement values ​​for the start and end sections of a rest period after charging and discharging of each of the cells E1-En. The curve shape determination unit 113 determines whether the voltage relaxation curve after charging and discharging of each of the cells E1-En has a complex shape based on the ratio between the area derived from the transition curve of the predicted voltage values ​​for the rest period of the generated equivalent circuit model and the area derived from the transition curve of the measured voltage values ​​for the rest period.

[0045] In the case of a rest period after discharge of each of the cells E1-En, the equivalent circuit model generation unit 112 generates an equivalent circuit model by fitting voltage measurement values ​​for the start and end sections of the rest period after discharge of each of the cells E1-En. The curve shape determination unit 113 determines that the voltage relaxation curve of each of the cells E1-En has a complex shape if the cumulative value of the upward deviation of the transition curve of the predicted voltage values ​​during the rest period of the generated equivalent circuit model from the transition curve of the measured voltage values ​​is greater than a threshold value.

[0046] In the case of a rest period after charging of each of the cells E1-En, the equivalent circuit model generation unit 112 generates an equivalent circuit model by fitting voltage measurement values ​​for the start and end sections of the rest period after charging of each of the cells E1-En. The curve shape determination unit 113 determines that the voltage relaxation curve of each of the cells E1-En has a complex shape if the cumulative value of downward deviation of the transition curve of the predicted voltage values ​​during the rest period of the generated equivalent circuit model from the transition curve of the measured voltage values ​​is greater than a threshold value.

[0047] Fig. 5 shows an example of a voltage relaxation curve with a general shape after the end of cell discharge. Fig. 6 shows an example of a voltage relaxation curve with a complex shape after the end of cell discharge. The upper graphs in Fig. 5 and Fig. 6 respectively show the transition of predicted voltage values ​​for one hour after the end of discharge of the battery pack 30. The lower graphs are enlarged graphs of the upper graphs for the period from the end of discharge to 10 minutes later.

[0048] In the examples shown in FIGS. 5 and 6, voltage measurement values ​​at both ends (0 to 1 minute, 9 to 10 minutes) of the target pause period (10 minutes from the end of discharge) are fitted to generate a first-order equivalent circuit model shown in Equation 3 below.

[0049] V=OCV+U・exp(-t / τ)...(Formula 3)

[0050] The curve shape determination unit 113 calculates the area ratio shown in the following (Equation 4).

[0051] Area ratio=(Σrest period (predicted voltage value−measured voltage value)) / (Σrest period (predicted voltage value−minimum predicted voltage value)) (Equation 4) If the numerator (predicted voltage value−measured voltage value) is a negative value, it is set to 0.

[0052] The predicted voltage values ​​are derived from an equivalent circuit model generated by fitting voltage measurement values ​​at both ends (0-1 minute, 9-10 minutes) of the above-mentioned rest period (0-10 minutes). When voltage measurement values ​​are obtained at 10-second intervals, the rest period is 10 minutes, so predicted voltage values ​​at 60 points are derived.

[0053] When the transition of the predicted voltage value and the transition of the measured voltage value during the pause period ideally match, the area ratio is 0. The greater the deviation between the transition of the predicted voltage value and the transition of the measured voltage value, the larger the area ratio becomes.

[0054] Because the numerator of Equation 4 is a value obtained by accumulating the difference between the predicted voltage value and the measured voltage value when the predicted voltage value is greater than the measured voltage value, only the deviation when the predicted voltage value is greater than the measured voltage value is reflected in the area ratio. In other words, the deviation when the predicted voltage value is greater than the measured voltage value is not offset by the deviation when the predicted voltage value is smaller than the measured voltage value, and it is possible to determine with high accuracy whether the transition curve of the measured voltage values ​​includes a section that is concave downward relative to the transition curve of the predicted voltage values.

[0055] The voltage relaxation curve shown in Figure 5 is an example where the area ratio defined in (Equation 4) is small, while the voltage relaxation curve shown in Figure 6 is an example where the area ratio defined in (Equation 4) is large. The voltage relaxation curve shown in Figure 6 is a voltage relaxation curve for a lithium-ion battery cell using a composite material of carbon (C) and silicon (Si) for the anode, and it has two relaxation stages. The voltage relaxation curve between the first and second voltage relaxation stages after the end of discharge exhibits a concave-downward shape relative to the exponential curve. When generating the equivalent circuit model, voltage measurement values ​​from both end sections (0-1 minute, 9-10 minutes) of the rest period (0-10 minutes) are used to exclude voltage measurement values ​​during the period where the exponential curve exhibits a concave-downward shape. This prevents the exponential curve after the end of discharge generated by fitting from being pulled downward overall, making it easier to extract the difference from a voltage relaxation curve that includes a two-stage relaxation.

[0056] The curve shape determination unit 113 compares the area ratio defined by Equation 4, which indicates a normalized cumulative value of upward deviation of the transition curve of the predicted voltage values ​​during the pause period from the transition curve of the measured voltage values, with a threshold value. If the area ratio is equal to or less than the threshold value, the curve shape determination unit 113 determines that the voltage relaxation curve has a typical shape, and if the area ratio is greater than the threshold value, the curve shape determination unit 113 determines that the voltage relaxation curve has a complex shape.

[0057] Consider the voltage relaxation curve after the end of charging of the battery pack 30. For example, voltage measurement values ​​at both ends (0 to 1 minute, 9 to 10 minutes) of the target pause period (10 minutes from the end of charging) are fitted to generate a first-order equivalent circuit model shown in Equation 3 above.

[0058] The curve shape determination unit 113 calculates the area ratio shown in the following (Equation 5).

[0059] Area ratio=(Σrest period (measured voltage value−predicted voltage value)) / (Σrest period (maximum predicted voltage value−predicted voltage value)) (Equation 5) If the numerator (measured voltage value−predicted voltage value) is a negative value, it is set to 0.

[0060] The predicted voltage values ​​are derived from an equivalent circuit model generated by fitting voltage measurement values ​​at both ends (0-1 minute, 9-10 minutes) of the above-mentioned rest period (0-10 minutes). When voltage measurement values ​​are obtained at 10-second intervals, the rest period is 10 minutes, so predicted voltage values ​​at 60 points are derived.

[0061] When the transition of the predicted voltage value and the transition of the measured voltage value during the pause period ideally match, the area ratio is 0. The greater the deviation between the transition of the predicted voltage value and the transition of the measured voltage value, the larger the area ratio becomes.

[0062] Since the numerator of (Equation 5) is a value obtained by accumulating the difference between the predicted voltage value and the measured voltage value when the predicted voltage value is smaller than the measured voltage value, only the deviation when the predicted voltage value is smaller than the measured voltage value is reflected in the area ratio. In other words, the deviation when the predicted voltage value is smaller than the measured voltage value is not offset by the deviation when the predicted voltage value is larger than the measured voltage value, and it is possible to determine with high accuracy whether the transition curve of the measured voltage value includes a section that is concave upward relative to the transition curve of the predicted voltage value.

[0063] The voltage relaxation curve of a lithium-ion battery cell using a carbon (C) and silicon (Si) composite material for the anode exhibits a concave-upward curve relative to the exponential curve between the first and second voltage relaxation stages after the end of charging. When generating the equivalent circuit model, voltage measurements from both ends of the rest period (0-1 minute, 9-10 minutes) were used to eliminate the concave-upward curve relative to the exponential curve. This prevents the exponential curve after the end of charging generated by fitting from being pulled upward overall, making it easier to extract the difference from the voltage relaxation curve that includes the two-stage relaxation.

[0064] The curve shape determination unit 113 compares an area ratio defined by Equation 5, which indicates a normalized cumulative value of downward deviation of the transition curve of the predicted voltage values ​​during the pause period from the transition curve of the measured voltage values, with a threshold value. If the area ratio is equal to or less than the threshold value, the curve shape determination unit 113 determines that the voltage relaxation curve has a typical curved shape, and if the area ratio is greater than the threshold value, the curve shape determination unit 113 determines that the voltage relaxation curve has a complex shape.

[0065] The curve shape determination unit 113 may change the threshold value to be compared with the area ratio of each of the cells E1-En depending on the SOC of each of the cells E1-En. For example, the curve shape determination unit 113 sets the threshold value to 7% when the SOC is in the low to medium range (0 to 70%), and sets the threshold value to 10% when the SOC is in the high range (70 to 100%).

[0066] In experiments using lithium-ion battery cells with mixed negative electrodes, voltage relaxation curves including two-stage relaxation, as shown in Figure 6, frequently occurred in the mid-SOC range (30-70%). For high SOC ranges (70-100%), the voltage relaxation tended to be approximately linear and then flatten out midway through. Therefore, the area ratio defined by (Equation 4) or (Equation 5) for high SOC ranges (70-100%) tended to be smaller than for low-to-mid SOC ranges (0-70%). Therefore, the threshold for high SOC ranges (70-100%) was set higher than the threshold for low-to-mid SOC ranges (0-70%) to reduce the risk of misjudging a voltage relaxation curve with a typical curve shape as a voltage relaxation curve with a complex curve shape when the SOC range is high (70-100%).

[0067] In the above example, the threshold is set to 70%, but 70% is just an example, and the optimum value is set as appropriate depending on the content of the mixed material (silicon in the above example). The threshold may be set in three or more stages depending on the cell's SOC. The shape of the cell's voltage relaxation curve is also affected by temperature. The threshold may be set more finely depending on the combination of the cell's SOC and temperature. A threshold that can optimally determine whether the voltage relaxation curve has a complex shape for each combination of the cell's SOC and temperature may be determined based on the results of experiments or simulations, and the curve shape determination unit 113 may store the threshold as a map.

[0068] If the voltage relaxation curve of each cell E1-En is determined to have a complex shape, the equivalent circuit model generation unit 112 changes from default processing to special processing. In the default processing, the equivalent circuit model generation unit 112 fits an equivalent circuit model within a predetermined range of rest periods and parameter search ranges. In this embodiment, in the default processing, the third-order equivalent circuit model shown in Equation 2 above is fitted using voltage measurement values ​​during a rest period of 3 to 10 minutes from the end of discharge or charge.

[0069] In this way, in the default processing, an equivalent circuit model is generated by fitting voltage measurement values ​​obtained by excluding voltage measurement values ​​for a predetermined period (0 to 3 minutes in this embodiment) from the voltage measurement values ​​at multiple points during the rest period (0 to 10 minutes in this embodiment) from the start of the rest period. Since the voltage may behave unstable immediately after the start of the rest period, the default processing excludes voltage measurement values ​​at the beginning of the rest period.

[0070] As a special process, the equivalent circuit model generation unit 112 changes the range of voltage measurement values ​​used for fitting among the voltage measurement values ​​at multiple points during the pause period, or changes the search range for the parameters of the equivalent circuit model. This will be described in detail below.

[0071] When the SOC of each cell E1-En is in the high range, the equivalent circuit model generation unit 112 performs special processing by using, for fitting, voltage measurement values ​​from multiple points during the rest period that are in a wider range than the range of voltage measurement values ​​used for fitting in the default processing.

[0072] The equivalent circuit model generation unit 112 determines that the SOC of each cell E1-En is in the high range when the SOC of each cell E1-En is higher than a set value (e.g., 70%). When the SOC of each cell E1-En is in the high range, the equivalent circuit model generation unit 112 uses, for example, voltage measurement values ​​for the entire interval of the rest period (0 to 10 minutes in this embodiment) for fitting. For the equivalent circuit model, a third-order equivalent circuit model similar to the default processing is used.

[0073] As described above, when a lithium-ion battery cell with a mixed negative electrode was used, the voltage tended to relax linearly halfway through when the SOC was in the high range (70 to 100%). When the SOC was in the high range (70 to 100%), the fitting accuracy was improved by using the voltage measurement values ​​for the entire interval (0 to 10 minutes in this embodiment) of the rest period, rather than excluding the voltage measurement values ​​for the first 3 minutes of the rest period.

[0074] When the SOC of each cell E1-En is in the low-middle range, the equivalent circuit model generation unit 112 performs special processing by time-differentiating voltage measurement values ​​at multiple points during the pause period after a set reference time point, and performing linear regression on the logarithms of the obtained differential values ​​to generate a regression line. When the SOC of each cell E1-En is equal to or lower than a set value (e.g., 70%), the equivalent circuit model generation unit 112 determines that the SOC of each cell E1-En is in the low-middle range. During the pause period, the reference time point is set within the interval from the end of the first stage of voltage relaxation to the start of the second stage of voltage relaxation.

[0075] The set value may be adjusted according to the cell temperature. For each combination of the cell SOC and temperature, a set value that can optimally determine whether the voltage relaxation curve has a shape including two-stage voltage relaxation may be determined according to the results of an experiment or a simulation, and the equivalent circuit model generation unit 112 may store the set value as a map.

[0076] The equivalent circuit model generation unit 112 estimates the time constant τ of voltage relaxation after the reference time point from the slope of the generated regression line. If the estimated time constant τ falls outside a preset reference range (e.g., 500 seconds≦τ≦1500 seconds), the equivalent circuit model generation unit 112 skips the generation of an equivalent circuit model for each cell E1-En. The reference range is set for each type of cell based on the results of experiments and simulations. If the time constant τ estimated from the slope of the regression line does not fall within the reference range, the voltage behavior during the rest period exceeds the expected realistic range, and therefore OCV estimation using the equivalent circuit model is not performed.

[0077] In rare cases, the slope of the generated regression line may be positive. In such cases, the time constant τ becomes negative, and it can be estimated that the voltage does not relax and the overall voltage change is accelerating. In such cases, the OCV estimation using the equivalent circuit model is not performed.

[0078] If the estimated time constant τ falls within a reference range, the equivalent circuit model generation unit 112 generates a first-order equivalent circuit model. If the estimated time constant τ falls within the reference range, the equivalent circuit model generation unit 112 changes the search range of the time constant τ in the equivalent circuit model to a predetermined range centered on the estimated time constant τ. For example, the search range of the time constant τ in default processing may be set to the reference range, and in special processing, if the time constant τ estimated from the slope of the regression line falls within the reference range, the search range of the time constant τ may be changed to a range of ±50% of the time constant τ estimated from the slope of the regression line.

[0079] Fig. 7 shows an example of a graph plotting the logarithm of the time derivative of the voltage measured after the end of cell discharge. The graph shown in Fig. 7 shows data from a 10-minute rest period from the end of discharge. The data shown in Fig. 7 was obtained under conditions where the cell's SOC was 40% and the temperature was 45°C.

[0080] The logarithm of the time differential value of the voltage measurement value, log(dv / dt), can be regressed by a linear function with a slope of −1 / τ and an intercept of b, as shown in the following (Equation 6).

[0081] log(dv / dt)=-1 / τ・t+b...(Formula 6)

[0082] The equivalent circuit model generation unit 112 generates a regression line by linearly regressing the logarithmic values ​​log(dv / dt) of the voltage measurement values ​​at multiple points during the rest period (0 to 10 minutes in FIG. 7 ) after the reference point (5 minutes in FIG. 7 ) (5 to 10 minutes in FIG. 7 ). The equivalent circuit model generation unit 112 can roughly estimate the time constant τ of the second-stage voltage relaxation in advance from the slope of the generated regression line. In the example shown in FIG. 7 , the time constant τ was estimated to be 290 seconds from the slope of the regression line.

[0083] 8 shows an example of a transition curve predicted from an equivalent circuit model of a cell generated by fitting the transition curve of voltage measurement values ​​at multiple points after the end of discharge of the cell used in FIG. 7 and the transition curve of voltage measurement values ​​at multiple points during the rest period after discharge. In the example shown in FIG. 8, a first-order equivalent circuit model is generated by fitting the voltage measurement values ​​during the rest period of 3 to 50 minutes from the end of discharge. When modeling only the second stage of voltage relaxation, the first low-frequency RC parallel circuit can be sufficiently approximated, and the second and subsequent high-frequency RC parallel circuits are not required.

[0084] The time constant τ was estimated to be 261 seconds from the first-order equivalent circuit model shown in Fig. 8. A value close to the time constant τ = 290 seconds estimated in advance from the regression line shown in Fig. 7 was obtained.

[0085] In the above example, when the cell's SOC is higher than the set value, the equivalent circuit model generation unit 112 performs fitting including the voltage measurement value at the beginning of the idle period. When the cell's SOC is equal to or lower than the set value, the equivalent circuit model generation unit 112 estimates the time constant of the second-stage voltage relaxation by generating a regression line of the logarithmic value of the voltage differential value. In special processing, the above-described case distinction may be performed based on a criterion other than the cell's SOC.

[0086] For example, the equivalent circuit model generation unit 112 may perform linear regression of voltage measurement values ​​at multiple points during a rest period after the end of cell discharge using a linear function of time, and determine the content of the special processing based on the prediction accuracy of the generated regression line. Specifically, the equivalent circuit model generation unit 112 may determine the content of the special processing based on the coefficient of determination R ２is higher than the set value, it is estimated that the voltage is relaxing linearly, and fitting is performed including the voltage measurement value at the beginning of the rest period. ２ If R is equal to or less than the set value, it is assumed that the voltage is not relaxed linearly, and the time constant of the second stage of voltage relaxation is estimated by generating a regression line of the logarithmic value of the voltage differential value. ２ The smaller the value, the greater the variation in voltage measurement values ​​at multiple points during the rest period, and it is estimated that the voltage relaxation curve has a complex shape.

[0087] Furthermore, for example, the equivalent circuit model generation unit 112 may determine the content of the special processing based on the second-order differential value (change acceleration) of voltage measurement values ​​at multiple points during the rest period after the cell has finished discharging. For example, in a typical voltage relaxation, the voltage relaxation curve is always convex upward during the rest period after discharging and always convex downward during the rest period after charging, so the second-order differential value is always negative during the rest period after discharging and always positive during the rest period after charging. On the other hand, in a voltage relaxation including two-stage relaxation as shown in FIG. 6, there is a section where the curve's convex and concave sides are reversed.

[0088] Specifically, if the number of times the positive and negative signs of the second-order differential values ​​of the voltage measurement values ​​during the pause period are reversed is equal to or less than a set number, the equivalent circuit model generation unit 112 estimates that two-stage voltage relaxation has not occurred, and performs fitting including the voltage measurement value at the beginning of the pause period.If the number of times the positive and negative signs of the second-order differential values ​​of the voltage measurement values ​​during the pause period are reversed exceeds a set number, the equivalent circuit model generation unit 112 estimates that two-stage voltage relaxation has occurred, and estimates the time constant of the second-stage voltage relaxation by generating a regression line of the logarithmic values ​​of the voltage differential values.

[0089] Returning to Fig. 2, the prediction unit 114 can use the equivalent circuit model generated by the equivalent circuit model generation unit 112 to predict the terminal voltage and OCV of each of the cells E1-En at a set time ahead.

[0090] The battery state estimation unit 115 estimates the SOC of each cell E1-En based on the predicted OCV of each cell E1-En and the SOC-OCV curve of the cell. The SOC-OCV curve of the cell may be acquired from the battery-equipped device 2, or may be estimated by acquiring the OCV and SOC of multiple battery packs 30 of the same model.

[0091] The battery state estimation unit 115 estimates the FCC based on the depth of discharge DOD and the current integrated value ΣI as shown in the following (Equation 7). The depth of discharge DOD is calculated as the difference between the SOC at the start of charging / discharging and the SOC at the end of charging / discharging. The current integrated value ΣI is calculated by integrating the measured values ​​of the current during the charging / discharging period. FCC=ΣI / DOD (Equation 7)

[0092] The battery state estimation unit 115 estimates the SOH based on the estimated FCC and the initial FCC, as shown in the following (Equation 8). The SOH is defined as the ratio of the current FCC to the initial FCC, and the lower the value (closer to 0%), the more advanced the deterioration. SOH = current FCC / initial FCC (Equation 8)

[0093] 9 is a flowchart illustrating the flow of the OCV estimation process performed by the battery state estimation system 1 according to the embodiment. The equivalent circuit model generation unit 112 acquires, from the battery data storage unit 121, voltage measurement values ​​between the terminals of each of the cells E1-En during a rest period after charging / discharging of the target battery pack 30 (S10). The equivalent circuit model generation unit 112 fits the voltage measurement values ​​during the start and end periods of the rest period after charging / discharging of each of the cells E1-En to generate an equivalent circuit model (S11).

[0094] The curve shape determination unit 113 estimates the shape of the voltage relaxation curve after charging and discharging of each cell E1-En based on the transition of the predicted voltage value during the pause period predicted from the generated equivalent circuit model (S12). If the estimated voltage relaxation curve shape is typical (N in S13), the equivalent circuit model generation unit 112 generates a third-order equivalent circuit model by fitting voltage measurement values ​​obtained by subtracting the voltage measurement value at the beginning of the pause period from the voltage measurement value during the pause period as default processing (S14). If the estimated voltage relaxation curve shape is complex (Y in S13), the equivalent circuit model generation unit 112 generates an equivalent circuit model using special processing (S15).

[0095] Fig. 10 is a flowchart showing a subroutine of the special processing in step S15 of Fig. 9. If the cell SOC is higher than a set value (e.g., 70%) (Y in S151), the equivalent circuit model generation unit 112 uses the voltage measurement values ​​for all sections of the pause period for fitting to generate a third-order equivalent circuit model (S152).

[0096] If the cell SOC is equal to or lower than the set value (N in S151), the equivalent circuit model generation unit 112 time-differentiates the voltage measurement values ​​at multiple points after the reference point in the idle period and calculates the logarithm of the obtained multiple differential values ​​(S153). The equivalent circuit model generation unit 112 performs linear regression on the logarithm of the multiple differential values ​​to generate a regression line, and estimates the time constant of voltage relaxation after the reference point from the slope of the generated regression line (S154).

[0097] The equivalent circuit model generation unit 112 determines whether the estimated time constant falls within a reference range (S155). If the estimated time constant falls within the reference range (Y in S155), the equivalent circuit model generation unit 112 generates a first-order equivalent circuit model by fitting voltage measurement values ​​obtained by subtracting the voltage measurement value at the beginning of the pause period from the voltage measurement values ​​during the pause period (S156). Note that the equivalent circuit model generation unit 112 may also generate a first-order equivalent circuit model by fitting voltage measurement values ​​after the reference point in the pause period.

[0098] If the estimated time constant does not fall within the reference range (N in S155), the process of step S156 is skipped. Return to Fig. 9. The prediction unit 114 predicts the terminal voltage and OCV of each of the cells E1-En at a set time ahead using the equivalent circuit model generated by the equivalent circuit model generation unit 112 (S16).

[0099] As described above, according to this embodiment, an equivalent circuit model is generated by fitting voltage measurement values ​​of the start and end sections of a rest period after charging and discharging cells E1-En, and the content of the process for generating the equivalent circuit model is changed depending on whether the voltage relaxation curve predicted from the equivalent circuit model has a complex shape. This makes it possible to improve the prediction accuracy when generating an equivalent circuit model and predicting a voltage relaxation curve.

[0100] When determining whether a voltage relaxation curve has a complex shape, generating an equivalent circuit model by fitting the voltage measurement values ​​at the start and end of the rest period makes it possible to accurately capture voltage transitions, including two-stage voltage relaxation. Two-stage voltage relaxation occurs in lithium-ion battery cells that use composite materials for the anode.

[0101] The OCV estimation method according to this embodiment is particularly effective for battery-equipped devices 2 in which the rest period is short and new charging / discharging is often started before the terminal voltage has completely converged.

[0102] The present disclosure has been described above based on the embodiments. The embodiments are merely examples, and it will be understood by those skilled in the art that various modifications are possible in the combination of the respective components and processing steps, and that such modifications are also within the scope of the present disclosure.

[0103] In the above-described embodiment, an example has been described in which a threshold value is set according to the SOC of each of the cells E1-En to be compared with the area ratio of each of the cells E1-En defined in the above (Equation 4) or (Equation 5). In this regard, the threshold value may be set using a criterion other than the SOC of each of the cells E1-En.

[0104] For example, the curve shape determination unit 113 may perform linear regression of voltage measurement values ​​at multiple points during a rest period after the end of charging and discharging of each of the cells E1 to En using a linear function of time, and set a threshold value to be compared with the area ratio of each of the cells E1 to En based on the prediction accuracy of the generated regression line. Specifically, the curve shape determination unit 113 may set a coefficient of determination R ２ is higher than the set value, it is estimated that the voltage is relaxed linearly, and the threshold value is set to 10%. ２ If is less than or equal to a set value, it is assumed that the voltage is not relaxing linearly and the threshold is set to 7%.

[0105] For example, the curve shape determination unit 113 may set a threshold value to be compared with the area ratio of each cell E1-En based on the second-order differential value (change acceleration) of the voltage measurement values ​​at multiple points during the pause period after the discharge of each cell E1-En is completed. Specifically, if the number of times the positive and negative signs of the second-order differential value of the voltage measurement values ​​during the pause period is equal to or less than a set number, the curve shape determination unit 113 assumes that two-stage voltage relaxation has not occurred and sets the threshold to 10%. If the number of times the positive and negative signs of the second-order differential value of the voltage measurement values ​​during the pause period is equal to or less than a set number, the curve shape determination unit 113 assumes that two-stage voltage relaxation has occurred and sets the threshold to 7%.

[0106] In the above embodiment, an example has been described in which the OCV is estimated by the battery state estimation system 1 built on a cloud server. In this regard, the functions of the battery state estimation system may be implemented on the edge side. For example, the battery state estimation system may be implemented in the battery management device 32 in the battery pack 30, or in the control unit 21 of the battery-equipped device 2.

[0107] The embodiment may be specified by the following items.

[0108] [Item 1] A battery state estimation system (1) comprising: an equivalent circuit model generation unit (112) that generates an equivalent circuit model including an OCV and an n-stage RC parallel circuit (n is a natural number) by fitting voltage measurement values ​​at multiple points of the secondary batteries (E1-En) during a rest period after charging or discharging the secondary batteries (E1-En), and a curve shape determination unit (113) that determines whether a voltage relaxation curve of the secondary batteries (E1-En) after charging or discharging has a complex shape based on a degree of agreement between a transition of a predicted voltage value during the rest period predicted from the equivalent circuit model generated by fitting the voltage measurement values ​​for at least a portion of the rest period and a transition of the voltage measurement values ​​during the rest period, wherein the equivalent circuit model generation unit (112) switches from default processing to special processing when it is determined that the voltage relaxation curve has a complex shape. This can improve the prediction accuracy when generating an equivalent circuit model and predicting a voltage relaxation curve. [Item 2] The battery state estimation system (1) according to Item 1, wherein the equivalent circuit model generation unit (112) generates the equivalent circuit model by fitting voltage measurement values ​​in a start section and an end section of the pause period, and the curve shape determination unit (113) determines whether the voltage relaxation curve has a complex shape based on a ratio between an area derived from a transition curve of predicted voltage values ​​during the pause period of the generated equivalent circuit model and an area derived from a transition curve of measured voltage values ​​during the pause period. This makes it possible to accurately capture a transition of voltage including two-stage voltage relaxation. [Item 3] The battery state estimation system (1) according to Item 1, wherein the equivalent circuit model generation unit (112) generates the equivalent circuit model by fitting voltage measurement values ​​of a start section and an end section of a rest period after discharge of the secondary batteries (E1-En), and the curve shape determination unit (113) determines that the voltage relaxation curve has a complex shape when a cumulative value of upward deviation of a transition curve of predicted voltage values ​​during the rest period of the generated equivalent circuit model from a transition curve of measured voltage values ​​is greater than a threshold value. This makes it possible to accurately capture a transition in voltage, including two-stage voltage relaxation, after discharge is completed.[Item 4] The battery state estimation system (1) according to Item 1, wherein the equivalent circuit model generation unit (112) generates the equivalent circuit model by fitting voltage measurement values ​​for a start section and an end section of a rest period after charging of the secondary batteries (E1-En), and the curve shape determination unit (113) determines that the voltage relaxation curve has a complex shape when a cumulative value of downward deviation of a transition curve of predicted voltage values ​​from a transition curve of measured voltage values ​​during the rest period of the generated equivalent circuit model is greater than a threshold. This allows for accurate capture of a voltage transition including a two-stage voltage relaxation after charging is completed. [Item 5] The battery state estimation system (1) according to Item 3 or 4, wherein the curve shape determination unit (113) changes the threshold value according to the SOC of the secondary batteries (E1-En). This allows for more accurate capture of a voltage transition including a two-stage voltage relaxation. [Item 6] The battery state estimation system (1) according to Item 1, wherein the equivalent circuit model generation unit (112) performs the special processing by changing a range of voltage measurement values ​​used for fitting among the voltage measurement values ​​taken at multiple points during the pause period, or by changing a search range for parameters of the equivalent circuit model. This allows for improved prediction accuracy of the voltage relaxation curve when the voltage relaxation curve has a complex shape by changing the range of voltage measurement values ​​used for fitting or by changing the search range for parameters of the equivalent circuit model. [Item 7] The equivalent circuit model generation unit (112) performs the default processing by fitting voltage measurement values ​​obtained by excluding voltage measurement values ​​for a predetermined period from the start of the pause period from the voltage measurement values ​​taken at multiple points during the pause period, to generate the equivalent circuit model. When the SOC of the secondary batteries (E1-En) is higher than a set value, the special processing is performed by using voltage measurement values ​​from a range of voltage measurement values ​​taken at multiple points during the pause period that is wider than the range of voltage measurement values ​​used for fitting in the default processing. This makes it possible to improve the prediction accuracy of the voltage relaxation curve when the SOC is in the high range.[Item 8] The battery state estimation system (1) according to Item 1, wherein the equivalent circuit model generation unit (112) time-differentiates voltage measurement values ​​at multiple points after a set reference point among voltage measurement values ​​at multiple points during the pause period when the SOC of the secondary batteries (E1-En) is equal to or lower than a set value, performs linear regression on the logarithm of the obtained differential value, estimates a time constant of voltage relaxation after the reference point, and skips generation of the equivalent circuit model when the estimated time constant falls outside a preset reference range. This makes it possible to avoid low-accuracy estimation of an OCV from an unreliable equivalent circuit model when the SOC is in the low-to-mid range. [Item 9] The battery state estimation system (1) according to Item 8, wherein the equivalent circuit model generation unit (112) generates an equivalent circuit model including an OCV and a single-stage RC parallel circuit when the estimated time constant is within the reference range. This makes it possible to generate an equivalent circuit model that fits better with the second-stage voltage relaxation. [Item 10] The battery state estimation system (1) according to Item 8, wherein the equivalent circuit model generation unit (112) changes a search range for the time constant in the equivalent circuit model to a predetermined range centered on the estimated time constant when the estimated time constant falls within the reference range. This makes it possible to reduce the amount of calculation required to generate the equivalent circuit model. [Item 11] A battery state estimation method comprising: generating an equivalent circuit model including an OCV and an n-stage RC parallel circuit (n is a natural number) by fitting voltage measurement values ​​at multiple points of the secondary batteries (E1-En) during a rest period after charging and discharging the secondary batteries (E1-En); determining whether a voltage relaxation curve after charging and discharging of the secondary batteries (E1-En) has a complex shape based on a degree of agreement between a transition of a predicted voltage value during the rest period predicted from the equivalent circuit model generated by fitting the voltage measurement values ​​for at least a portion of the rest period and a transition of the voltage measurement values ​​during the rest period; and switching from default processing to special processing when it is determined that the voltage relaxation curve has a complex shape. This method can improve the prediction accuracy when generating an equivalent circuit model and predicting a voltage relaxation curve.[Item 12] A battery state estimation program that causes a computer to execute the following steps: generating an equivalent circuit model including an OCV and an n-stage RC parallel circuit (n is a natural number) by fitting voltage measurement values ​​at multiple points of the secondary batteries (E1-En) during a rest period after charging and discharging the secondary batteries (E1-En); determining whether a voltage relaxation curve after charging and discharging of the secondary batteries (E1-En) has a complex shape based on a degree of agreement between a transition of a predicted voltage value during the rest period predicted from the equivalent circuit model generated by fitting the voltage measurement values ​​for at least a portion of the rest period and a transition of the voltage measurement values ​​during the rest period; and switching from default processing to special processing when it is determined that the voltage relaxation curve has a complex shape. This can improve the prediction accuracy when generating an equivalent circuit model and predicting a voltage relaxation curve.

[0109] The present disclosure can be used to estimate the internal state of a lithium-ion battery.

[0110] 2 Battery-equipped device, 4 Commercial power system, 5 Network, 21 Control unit, 22 Load unit, 23 Charging unit, 24 Communication unit, 30 Battery pack, 31 Assembled battery, 32 Battery management device, 33 Measurement unit, 34 Control unit, E1-En Cell, Rs Shunt resistor, SW1 Switch, 1 Battery state estimation system, 11 Control unit, 111 Data acquisition unit, 112 Equivalent circuit model generation unit, 113 Curve shape determination unit, 114 Prediction unit, 115 Battery state estimation unit, 12 Memory unit, 121 Battery data retention unit, 13 Communication unit.

Claims

1. A battery state estimation system comprising: an equivalent circuit model generation unit that generates an equivalent circuit model including an OCV (Open Circuit Voltage) and n (n is a natural number) stages of RC parallel circuits by fitting voltage measurement values ​​at multiple points of the secondary battery during a rest period after charging or discharging the secondary battery; and a curve shape determination unit that determines whether a voltage relaxation curve after charging or discharging of the secondary battery has a complex shape based on the degree of agreement between a transition of a predicted voltage value during the rest period predicted from the equivalent circuit model generated by fitting voltage measurement values ​​for at least a portion of the rest period and a transition of the voltage measurement values ​​during the rest period, wherein the equivalent circuit model generation unit changes from default processing to special processing when it is determined that the voltage relaxation curve has a complex shape.

2. The battery state estimation system of claim 1, wherein the equivalent circuit model generation unit generates the equivalent circuit model by fitting voltage measurement values ​​for the start and end sections of the pause period, and the curve shape determination unit determines whether the voltage relaxation curve has a complex shape based on the ratio of an area derived from a transition curve of voltage prediction values ​​during the pause period of the generated equivalent circuit model to an area derived from a transition curve of voltage measurement values ​​during the pause period.

3. The battery state estimation system of claim 1, wherein the equivalent circuit model generation unit generates the equivalent circuit model by fitting voltage measurement values ​​for a start section and an end section of a rest period after discharging the secondary battery, and the curve shape determination unit determines that the voltage relaxation curve has a complex shape if the cumulative value of upward deviation of the transition curve of voltage prediction values ​​during the rest period of the generated equivalent circuit model from the transition curve of voltage measurement values ​​is greater than a threshold value.

4. The battery state estimation system of claim 1, wherein the equivalent circuit model generation unit generates the equivalent circuit model by fitting voltage measurement values ​​for a start section and an end section of a rest period after charging of the secondary battery, and the curve shape determination unit determines that the voltage relaxation curve has a complex shape if the cumulative value of downward deviation of the transition curve of voltage prediction values ​​during the rest period of the generated equivalent circuit model from the transition curve of voltage measurement values ​​is greater than a threshold value.

5. The battery state estimation system according to claim 3 or 4, wherein the curve shape determination unit changes the threshold value according to an SOC (State Of Charge) of the secondary battery.

6. The battery state estimation system according to claim 1, wherein the equivalent circuit model generation unit performs the special processing by changing the range of voltage measurement values ​​used for fitting among the voltage measurement values ​​at multiple points during the pause period, or by changing the search range for parameters of the equivalent circuit model.

7. The battery state estimation system according to claim 1, wherein the equivalent circuit model generation unit generates the equivalent circuit model by fitting voltage measurement values ​​obtained by excluding voltage measurement values ​​for a predetermined period from the start of the pause period from voltage measurement values ​​measured at multiple points during the pause period as the default processing, and when the SOC (State Of Charge) of the secondary battery is higher than a set value, uses voltage measurement values ​​from a wider range of voltage measurement values ​​measured at multiple points during the pause period for fitting than the range of voltage measurement values ​​used for fitting in the default processing as the special processing.

8. The battery state estimation system according to claim 1, wherein, when the SOC (State Of Charge) of the secondary battery is equal to or less than a set value, the equivalent circuit model generation unit time-differentiates voltage measurement values ​​at multiple points after a set reference point among the voltage measurement values ​​at multiple points during the rest period, performs linear regression on the logarithm of the obtained differential value, and estimates a time constant for voltage relaxation after the reference point, and skips generation of the equivalent circuit model if the estimated time constant deviates from a predetermined reference range.

9. The battery state estimation system according to claim 8, wherein the equivalent circuit model generation unit generates an equivalent circuit model including an OCV and a single-stage RC parallel circuit when the estimated time constant falls within the reference range.

10. A battery state estimation system as described in claim 8, wherein, when the estimated time constant falls within the reference range, the equivalent circuit model generation unit changes the search range for the time constant in the equivalent circuit model to a predetermined range centered on the estimated time constant.

11. A battery state estimation method comprising: a step of fitting voltage measurement values ​​at multiple points of a secondary battery during a rest period after charging and discharging the secondary battery, to generate an equivalent circuit model including an OCV (Open Circuit Voltage) and an n-stage RC parallel circuit (n is a natural number); a step of determining whether a voltage relaxation curve after charging and discharging of the secondary battery has a complex shape based on the degree of agreement between a trend in a predicted voltage value during the rest period predicted from the equivalent circuit model generated by fitting voltage measurement values ​​for at least a portion of the rest period and a trend in the measured voltage value during the rest period; and a step of changing from default processing to special processing when it is determined that the voltage relaxation curve has a complex shape.

12. A battery state estimation program that causes a computer to execute the following steps: a process of fitting voltage measurement values ​​at multiple points of a secondary battery during a rest period after charging or discharging the secondary battery, and generating an equivalent circuit model including an OCV (Open Circuit Voltage) and an n-stage RC parallel circuit (n is a natural number); a process of determining whether the voltage relaxation curve after charging or discharging of the secondary battery has a complex shape based on the degree of agreement between the trend in the voltage measurement values ​​during the rest period and the trend in the predicted voltage values ​​during the rest period predicted from the equivalent circuit model generated by fitting the voltage measurement values ​​for at least a portion of the rest period; and a process of changing from default processing to special processing when it is determined that the voltage relaxation curve has a complex shape.

Images