Battery analysis system, battery analysis method, and battery analysis program

Abstract

A Q-OCV curve generating unit (112) extracts voltage data that can be regarded as an open circuit voltage (OCV), on the basis of battery data for a fixed period, calculates a capacity Q when the extracted voltage data were measured, and generates a Q-OCV curve. An inflection point extracting unit (113) extracts an inflection point of the Q-OCV curve. A deterioration estimating unit (115) estimates deterioration of a secondary battery on the basis of a plurality of the inflection points extracted respectively from a plurality of the Q-OCV curves generated on the basis of the battery data for each of a plurality of fixed periods.

Description

Battery analysis system, battery analysis method, and battery analysis program

[0001] The present disclosure relates to a battery analysis system, a battery analysis method, and a battery analysis program for estimating the internal state of a secondary battery.

[0002] Assembled battery systems, in which battery cells are connected in multiple parallel and multiple series, are used in electric vehicles (EVs), etc., and there is a need to estimate the state of health (SOH) of the assembled battery system. One possible way to calculate the SOH of an assembled battery system is to use the dV / dQ characteristics obtained from battery data (voltage, current, temperature) during low-rate charging and discharging. However, in EV operation, many users prefer rapid charging, making it difficult to collect battery data during low-rate charging. Furthermore, in EV operation, the period of low-speed driving is short, making it difficult to collect battery data during low-rate discharging.

[0003] Patent Document 1 discloses a method for evaluating the degradation of a secondary battery by extracting feature quantities from the dV / dQ curve of the secondary battery. However, because the dV / dQ curve is estimated without excluding measurement data during high-rate charging and discharging, it is affected by internal resistance, which may result in a large error from the original dV / dQ measurement value. For example, during high-rate charging and discharging of an LFP (LiFePO4) battery, the voltage fluctuation due to internal resistance changes is more affected than the voltage fluctuation due to negative electrode stage changes, making it difficult to extract feature quantities due to negative electrode stage changes from the dV / dQ curve. Furthermore, there is a possibility that the material-specific dV / dQ peak may not be visible.

[0004] JP 2016-126891 A

[0005] The present disclosure has been made in light of these circumstances, and its purpose is to provide a technique for estimating the deterioration of a secondary battery with high accuracy.

[0006] In order to solve the above problems, a battery analysis system according to one aspect of the present disclosure includes a data acquisition unit that acquires time-series battery data including the voltage and current of a secondary battery; a Q-OCV curve generation unit that extracts voltage data that can be considered as OCV based on the battery data for a certain period of time, calculates the capacity Q when the extracted voltage data was measured, and generates a Q-OCV curve; an inflection point extraction unit that extracts an inflection point of the Q-OCV curve; and a degradation estimation unit that estimates degradation of the secondary battery based on a plurality of inflection points extracted from a plurality of Q-OCV curves that are generated based on the battery data for a plurality of certain periods.

[0007] 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.

[0008] According to the present disclosure, the deterioration of a secondary battery can be estimated with high accuracy.

[0009] FIG. 1 is a diagram for explaining a battery analysis system according to an embodiment; FIG. 2 is a diagram showing an example of the configuration of a battery pack system; FIG. 3 is a diagram showing an example of the configuration of a battery pack system; and FIG. 4 is a diagram showing an example of an equivalent circuit model of a cell E1. 4A shows a dV / dQ curve obtained by differentiating the Q-OCV approximate curve of FIG. 4A; FIG. 5A shows an example of a Q-OCV plot and approximate curve for the same month one year later; FIG. 5A shows a dV / dQ curve obtained by differentiating the Q-OCV approximate curve of FIG. 5A; FIG. 5B shows an image of approximating the Q-OCV curve of an LFP battery with a composite function of a logarithmic function, a sigmoid function, and an exponential function; FIG. 5C shows an example of a monthly plot of capacity from the peak of the dV / dQ curve to full charge, and a regression line of multiple capacity plots;

[0010] 1 is a diagram illustrating a battery analysis system 10 according to an embodiment. The battery analysis system 10 is a system that estimates the deterioration of an assembled battery system 21 included in a battery pack mounted on an electric vehicle 20. The battery analysis system 10 can also predict the future deterioration progression of the assembled battery system 21, and can also provide the user with an indication of when to replace the battery pack including the assembled battery system 21.

[0011] The battery analysis system 10 may be constructed, for example, on an in-house server installed in the in-house facility or data center of a business that provides an analysis service for battery packs mounted on electric vehicles 20. The battery analysis system 10 may also be constructed on a cloud server used based on a cloud service. The battery analysis system 10 may also be constructed on multiple servers that are distributed and installed at multiple locations (data centers, in-house facilities). The multiple servers may be a combination of multiple in-house servers, a combination of multiple cloud servers, or a combination of an in-house server and a cloud server.

[0012] An assembled battery system 21 included in a battery pack mounted on the electric vehicle 20 supplies power to a drive motor (not shown). The assembled battery system 21 includes a plurality of unit cells or a plurality of cell blocks connected in series.

[0013] 2A and 2B are diagrams showing an example configuration of a battery pack system 21. The battery pack system 21 shown in Fig. 2A includes a plurality of unit cells E1-Em connected in series. The battery pack system 21 shown in Fig. 2B includes a plurality of cell blocks Eb1-Ebm connected in series. Each cell block Eb1-Ebm includes a plurality of cells E1a-E1n-Ema-Emn connected in parallel.

[0014] The cells can be lithium-ion batteries, nickel-metal hydride batteries, lead-acid batteries, etc. In the following description, we will assume that lithium-ion batteries are used. The number of unit cells or cell blocks connected in series is determined according to the voltage of the drive motor.

[0015] The voltage sensor 22 detects the voltage across each of the series-connected single cells or cell blocks. A shunt resistor is connected in series with the multiple series-connected single cells or cell blocks. The current sensor 23 detects the current flowing through the series-connected single cells or cell blocks based on the voltage across the shunt resistor. Note that a Hall element may be used instead of the shunt resistor. Multiple temperature sensors 24 are installed in the battery pack including the battery pack system 21. For example, a thermistor may be used as the temperature sensor 24. For example, one temperature sensor 24 may be installed for every 6 to 8 single cells or cell blocks.

[0016] The control unit 25 is composed of a Battery Management Unit (BMU) and an Electronic Control Unit (ECU) working together. The BMU estimates the State Of Charge (SOC) by combining the Open Circuit Voltage (OCV) method and the current integration method. The OCV method estimates 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 BMU at the time of shipment.

[0017] The current integration method is a method for estimating the SOC based on the OCV at the start of charging and discharging the cell and the integrated value of the measured current. In 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.

[0018] The BMU periodically (e.g., every 10 seconds) transmits battery data including the voltage, current, temperature, and SOC of multiple single cells or cell blocks (hereinafter, both will be collectively referred to as cells as appropriate) to the ECU via an in-vehicle network, and the ECU samples the battery data in chronological order. A controller area network (CAN) or a local interconnect network (LIN) can be used as the in-vehicle network, for example.

[0019] The cell voltages transmitted from the BMU to the ECU may be the voltages of all the series-connected cells or only the maximum and minimum cell voltages, and the temperatures transmitted to the ECU may be the temperatures at multiple observation points in the battery pack or only the maximum and minimum temperatures.

[0020] The communication unit 26 has a function of performing communication signal processing with the communication unit 33 of the charging stand 30 and a function of performing wireless signal processing for connecting to the network 5. The communication unit 26 can access the network 5 using, for example, a mobile phone network (cellular network), a wireless LAN, V2I (Vehicle to Infrastructure), V2V (Vehicle to Vehicle), an ETC system (Electronic Toll Collection System), or DSRC (Dedicated Short Range Communications).

[0021] The network 5 is a general term for communication paths such as the Internet, dedicated lines, and VPNs (Virtual Private Networks), and the communication media and protocols are not important. Examples of communication media that can be used include a mobile phone network, a wireless LAN, a wired LAN, 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).

[0022] The ECU may transmit the sampled battery data to the battery analysis system 10 each time, or may store the sampled battery data in an internal memory and transmit the battery data stored in the memory at a predetermined timing all at once to the battery analysis system 10. Note that when the electric vehicle 20 and the charging stand 30 are connected by a charging cable, the ECU may transmit the battery data stored in the memory to the battery analysis system 10 via the charging stand 30.

[0023] By connecting the electric vehicle 20 to a charging stand 30 via a charging cable, the assembled battery system 21 inside the electric vehicle 20 can be charged externally. The charging stand 30 is connected to a commercial power system 2 and charges the assembled battery system 21.

[0024] Generally, charging is performed with alternating current (AC) for normal charging and with direct current (DC) for rapid charging. When charging with AC (e.g., single-phase 100 / 200 V), the charging voltage or charging current is controlled by a charger (not shown) inside the electric vehicle 20. When charging with DC, the charging voltage or charging current is controlled by a power supply unit 31 in the charging stand 30. The power supply unit 31 includes a rectifier circuit, a filter, and a DC / DC converter, and generates DC power by full-wave rectifying AC power supplied from the commercial power system 2 using the rectifier circuit and smoothing it using a filter. The DC / DC converter controls the voltage or current of the generated DC power.

[0025] For example, CHAdeMO (registered trademark), ChaoJi, GB / T, and Combo (Combined Charging System) can be used as fast charging standards. CHAdeMO, ChaoJi, and GB / T use CAN as the communication method. Combo uses PLC (Power Line Communication) as the communication method.

[0026] A charging cable that employs the CAN system includes a communication line in addition to a power line. When the charging cable connects the electric vehicle 20 and the charging stand 30, the control unit 25 of the electric vehicle 20 establishes a communication channel with the control unit 32 of the charging stand 30. Note that in a charging cable that employs the PLC system, communication signals are transmitted superimposed on the power line.

[0027] The communication unit 33 of the charging stand 30 has a function of executing communication signal processing with the communication unit 26 of the electric vehicle 20 and a function of executing signal processing for connecting to the network 5. The communication unit 33 can access the network 5 using, for example, a wired LAN, a wireless LAN, or a mobile phone network.

[0028] The battery analysis system 10 includes a control unit 11, a storage unit 12, and a communication unit 13. The communication unit 13 is a communication interface (for example, a network interface card (NIC)) for connecting to the network 5 via a wired or wireless connection.

[0029] The control unit 11 includes a data acquisition unit 111, a Q-OCV curve generation unit 112, an inflection point extraction unit 113, a regression equation generation unit 114, and a deterioration 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 (Graphics Processing Unit), NPU (Neural Network Processing Unit), ASIC (Application Specific Integrated Circuit), FPGA (Field Programmable Gate Array), and other LSIs. Examples of software resources that can be used include an operating system, an application, and other programs.

[0030] 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 time-series battery data including the voltage, current, temperature, and SOC of the battery pack system 21 from the electric vehicle 20 or the charging stand 30, and stores the acquired battery data in the battery data holding unit 121.

[0031] The Q-OCV curve generator 112 reads battery data from the battery data storage unit 121 at regular intervals (for example, once a month). Based on the read battery data for the regular interval, the Q-OCV curve generator 112 extracts voltage data that can be regarded as OCV. Based on time-series current data, the Q-OCV curve generator 112 calculates the capacity Q at the time the extracted voltage data was measured. The Q-OCV curve generator 112 generates a Q-OCV approximate curve based on multiple combinations of capacity Q and OCV (hereinafter referred to as Q-OCV plots).

[0032] In this embodiment, the capacity Q used to generate the Q-OCV approximate curve is the current integrated value Q when the battery is operated from a fully charged state. In this specification, the fully charged state is defined as (0 [Ah]), and the capacity Q is defined as the net discharge integrated amount (-X [Ah]).

[0033] The Q-OCV curve generator 112 extracts voltage data when the battery is in a resting state for a first predetermined time or more and less than a second predetermined time as voltage data that can be considered as OCV. A secondary battery is an electrochemical product, and the measured voltage increases nonlinearly when a charging current flows through the secondary battery, and decreases nonlinearly when a discharging current flows through the secondary battery. The voltage measured when current is flowing through the secondary battery is called the closed circuit voltage (CCV) or operating voltage. After charging and discharging are completed, the secondary battery gradually converges to an OCV that does not include overvoltage components. The convergence time to the OCV depends on the cell type, temperature, SOH, etc.

[0034] The reason for setting an upper limit to a sufficient rest time is that in the case of hysteresis materials such as silicon (Si), if the rest time is too long, the voltage may change. For example, at room temperature (25°C), the first predetermined time may be set to 1 hour and the second predetermined time may be set to 10 hours. Note that in low-temperature environments, the chemical reaction rate within the cell slows, and it takes longer to converge to the OCV. Therefore, the first predetermined time and the second predetermined time may be set to be longer as the temperature decreases.

[0035] The Q-OCV curve generator 112 may generate voltage data that can be regarded as the OCV based on an equivalent circuit model of the cell. For example, the Q-OCV curve generator 112 generates voltage data that can be regarded as the OCV based on voltage data in which the downtime is less than the first predetermined time and the equivalent circuit model.

[0036] Figure 3 shows an example of an equivalent circuit model for cell E1. The equivalent circuit model shown in Figure 3 is an equivalent circuit model in which the diffusion resistance components of each material are simply arranged in series. As an example, it shows an equivalent circuit model of a lithium-ion battery cell that uses lithium cobalt oxide (LiCoO2) for the positive electrode and a mixed material of graphite (C) and silicon (Si) for the negative electrode. The equivalent circuit model shown in Figure 3 can be defined by the following (Equation 1).

[0037] OCV = CCV - I * Rohm - Vpp - Vpc - Vpsi (Equation 1) CCV is the measured observed voltage of cell E1. I is the measured observed current of cell E1. Rohm is a resistance component that collectively represents the reaction resistance, conduction resistance, current collector resistance, etc. of lithium (Li). Zwp represents the diffusion resistance component of lithium (Li) in the positive electrode active material, Zwc represents the diffusion resistance component of lithium (Li) in the negative electrode active material (graphite (C)), and Zwsi represents the diffusion resistance component of lithium (Li) in the negative electrode active material (silicon (Si)). Vpp represents the polarization voltage Vp applied to Zwp, Vpc represents the polarization voltage Vp applied to Zwc, and Vpsi represents the polarization voltage Vp applied to Zwsi.

[0038] In the following description of the equivalent circuit model, it is assumed that the parameters R, C, and Zw have been tuned to values ​​that take into account temperature dependency in accordance with the usage environment.

[0039] The diffusion resistance component is expressed by the Warburg impedance, and it is known that it can be approximated by the sum of an infinite series of RC parallel circuits, as shown in the following (Equation 2).

[0040] Zw(s) = Σ ｎ＝１～∞ (Rn / (sCnRn+1)) ... (Formula 2) Cn=Cd / 2, Rn=8Rd / ((2n-1) ２ π ２ ) The time constant is (2n-1) ２ When used for calculation processing on the order of seconds, a third to fifth order is generally sufficient.

[0041] Each diffusion resistance component can be approximated by a Foster-type circuit as shown in the above (Equation 2). FIG. 3 shows an example of approximation by a third-order Foster-type circuit. In each RC parallel circuit of the Foster-type circuit, the relationship shown in the following (Equation 3) holds. The following (Equation 3) shows the RC parallel circuit of the first stage of the Foster-type circuit in FIG. 3.

[0042] (dVp1 / dt) = (I / Cp1) - (Vp1 / tau1) ... (Equation 3) iCp1 = Cp1 (dVp1 / dt), iRp1 = Vp1 / Rp1, I = iCp1 + iRp1, tau1 = Cp1 * Rp1 Once the Rp1 and Cp1 parameters of the first stage in Equation 3 above are determined, the parameters of the subsequent stages (Rp2, Cp2, Rp3, Cp3) ​​are also automatically determined. The same relationship as Equation 3 for the first stage RC parallel circuit holds for the second and third stages. That is, from Equation 2 above, Cp1 = Cp2 = Cp3, Rp2 = 1 / 9 * Rp1, and Rp3 = 1 / 25 * Rp.

[0043] Based on the above (Equation 2) and (Equation 3), the polarization voltage Vpp applied to Zwp, the polarization voltage Vpc applied to Zwc, and the polarization voltage Vpsi applied to Zwsi can be calculated using the observed current I as an input. The OCV can be estimated by substituting the calculated Vpp, Vpc, and Vpsi into the above (Equation 1). To estimate the OCV after charging and discharging have stopped, simply substitute 0 for I.

[0044] Contrary to the case where the pause time is less than the first predetermined time, when the pause time exceeds the second predetermined time, the Q-OCV curve generating unit 112 may replace the voltage data in the state where the pause time exceeds the second predetermined time with the voltage data at the time when the second predetermined time has elapsed.

[0045] When extracting voltage data in a resting state, the Q-OCV curve generating unit 112 may limit the voltage data to be extracted to voltage data when the resting state is entered after low-rate charging / discharging.

[0046] The Q-OCV curve generating unit 112 may extract voltage data when the charge / discharge rate is a low rate of a predetermined value (for example, 0.025 C) or less as voltage data that can be considered as OCV.

[0047] When the Q-OCV curve generator 112 is able to extract a certain number (e.g., about 25 to 40) or more of Q-OCV plots from battery data for a certain period of time, it generates a Q-OCV approximate curve based on the extracted Q-OCV plots.When the Q-OCV curve generator 112 is able to extract a certain number or more of Q-OCV plots based on voltage data when the battery is in a resting state for a first predetermined time or more and less than a second predetermined time (hereinafter referred to as a specified resting time), it generates a Q-OCV approximate curve based only on the extracted Q-OCV plots for the specified resting time.

[0048] If a certain number or more of Q-OCV plots for the specified rest time cannot be extracted, the Q-OCV curve generator 112 extracts voltage data during a low-rate charge / discharge state to increase the number of Q-OCV plots, or the Q-OCV curve generator 112 estimates the OCV based on an equivalent circuit model to increase the number of Q-OCV plots, or performs both of these.

[0049] The Q-OCV curve generator 112 may divide the full charge capacity [Ah] of the cell into a plurality of capacity ranges and calculate a representative value of the OCV for each capacity range. For example, the Q-OCV curve generator 112 divides the full charge capacity of the cell into capacity ranges corresponding to 1% SOC and calculates a representative value of the OCV for each capacity range. The Q-OCV curve generator 112 extracts a set number (e.g., five points) or more of Q-OCV plots for each capacity range and calculates a representative value of the OCV for each capacity range. The Q-OCV curve generator 112 calculates the representative value of the OCV, for example, by using the median or weighted average of the OCV.

[0050] When the median is used, the influence of outliers can be eliminated. When the weighted average is used, the Q-OCV curve generator 112 increases the contribution of the OCV that falls within the specified rest time, decreases the contribution of the OCV that does not fall within the specified rest time, and further decreases the contribution of the OCV during low-rate charge / discharge. Furthermore, the Q-OCV curve generator 112 may increase the contribution of an OCV with a lower charge / discharge rate before rest, and decrease the contribution of an OCV with a higher charge / discharge rate before rest. Furthermore, the Q-OCV curve generator 112 may increase the contribution of an OCV with a higher temperature during voltage measurement, and decrease the contribution of an OCV with a lower temperature.

[0051] The Q-OCV curve generator 112 may generate a Q-OCV approximate curve in which the upper limit of Q is the full charge capacity [Ah] and the lower limit of Q is Y [Ah]. For the purpose of extracting the inflection point of the Q-OCV approximate curve, the Q-OCV approximate curve up to a depth close to a fully discharged state corresponding to SOC=0% is not necessary. The Q-OCV curve generator 112 sets the value of Y according to the type of cell.

[0052] For example, in the case of an LFP battery, Y may be set to the capacity Q corresponding to an SOC of 50%. Degradation of an LFP battery is dominated by SEI (Solid Electrolyte Interphase) degradation, which occurs when a film (SEI film) is deposited on the negative electrode. SEI degradation appears early in the degradation process, and as it progresses, the peak approaches the full charge side. Therefore, for degradation of an LFP battery, a Q-OCV approximation curve up to a deep depth of discharge is not necessary. A Q-OCV approximation curve in the range corresponding to an SOC of 50-100% is sufficient.

[0053] In an NCA-based lithium-ion battery using nickel (N), cobalt (C), and aluminum (A) in the positive electrode and graphite (C) in the negative electrode, Y may be set to a capacity Q corresponding to an SOC of 30%. In an NCA-based lithium-ion battery using nickel (N), cobalt (C), and aluminum (A) in the positive electrode and a mixture of graphite (C) and silicon (Si) in the negative electrode, Y may be set to a capacity Q corresponding to an SOC of 5 to 10%.

[0054] Degradation of NCA-based lithium-ion batteries is also significantly affected by degradation due to cracking and isolation of the positive electrode active material. The impact of positive electrode degradation appears as a peak at a deeper depth of discharge. Furthermore, the impact of degradation due to cracking and isolation of graphite (C) appears as a peak at an even deeper depth of discharge. Furthermore, the impact of degradation due to cracking and isolation of silicon (Si) appears as a peak at a deeper depth of discharge than graphite (C). Therefore, NCA-based lithium-ion batteries using a mixed negative electrode of graphite (C) and silicon (Si) require a Q-OCV approximation curve with a deeper depth of discharge.

[0055] The inflection point extraction unit 113 extracts inflection points of the Q-OCV approximate curve generated by the Q-OCV curve generation unit 112. Specifically, the inflection point extraction unit 113 differentiates the Q-OCV approximate curve to generate a dV / dQ curve, and extracts peaks of the dV / dQ curve as inflection points.

[0056] The degradation estimation unit 115 estimates the degradation of the cell based on a plurality of inflection points extracted from a plurality of Q-OCV approximation curves generated based on battery data for a plurality of fixed periods (for example, one month). Specifically, the degradation estimation unit 115 compares the capacity from the peak to full charge of the dV / dQ curve generated based on battery data for a fixed period a with the capacity from the peak to full charge of the dV / dQ curve generated based on battery data for a fixed period b, to estimate the degradation state of the cell.

[0057] The estimation of the cell degradation state described above may be performed for all cells constituting the battery pack system 21, or may be performed for only the maximum voltage cell and the minimum voltage cell. For example, in a battery pack system 21 with a small number of series connections, such as a pedestrian-type battery pack, the degradation state of each cell may be estimated, whereas in a battery pack system 21 with a large number of series connections, such as an EV, the degradation state of only the maximum voltage cell and the minimum voltage cell may be estimated.

[0058] FIG. 4A is a diagram showing an example of a plot and approximation curve of Q-OCV for an LFP battery in month X of year 202X. The horizontal axis represents Q [Ah], and the vertical axis represents OCV [V]. FIG. 4B is a diagram showing a dV / dQ curve obtained by differentiating the Q-OCV approximation curve of FIG. 4A. The horizontal axis represents Q [Ah], and the vertical axis represents dV / dQ [V / Ah]. FIG. 5A is a diagram showing an example of a plot and approximation curve of Q-OCV for the same month one year later. FIG. 5B is a diagram showing a dV / dQ curve obtained by differentiating the Q-OCV approximation curve of FIG. 5A.

[0059] The Q-OCV curve generator 112 generates a Q-OCV approximate curve by regressing multiple Q-OCV plots within a certain period. The Q-OCV curve generator 112 can generate the Q-OCV approximate curve using kernel regression, for example. Kernel regression is defined as the sum of N kernel functions, as shown in the following (Equation 4).

[0060] f(x) = Σα ｉ k(x （ｉ） , x) ... (Equation 4) i = 1 to N Kernel function k(x （ｉ） , x), for example, a Gaussian kernel can be used as shown in the following (Equation 5).

[0061] k(x, x') = exp(-β・√(Σ(x ｊ -x ｊ ') ２ ) ２ ) (Equation 5) j = 1 to n Kernel regression can be applied to any type of cell. When the target cell is an LFP battery, the Q-OCV curve generator 112 can also generate a Q-OCV approximation curve using a composite function of a logarithmic function, a sigmoid function, and an exponential function.

[0062] Figure 6 shows an image of approximating the Q-OCV curve of an LFP battery using a composite function of a logarithmic function, a sigmoid function, and an exponential function. The SOC-OCV curve of an LFP battery can be approximated with high accuracy by approximating the low SOC region of the SOC-OCV curve with a logarithmic function, the high SOC region with an exponential function, and the region between the two with a sigmoid function. The SOC-OCV curve and the Q-OCV curve basically have similar shapes.

[0063] The general formula for the sigmoid function is shown below (Equation 6).

[0064] y = w / (1 + exp (-a * (Q-m))) (Equation 6) w: parameter in the height direction m: Q at which the slope of the sigmoid function peaks a: steepness of the sigmoid function The parameter m in the above (Equation 6) is Q at which the slope peaks, that is, the inflection point unique to the sigmoid function itself, but since the negative electrode stage change point (inflection point) is included between the low SOC side region and the high SOC side region, the inflection point unique to the sigmoid function that approximates that region corresponds to the capacity Q corresponding to the negative electrode stage change point (inflection point). Therefore, when the Q-OCV approximation curve is regressed using a sigmoid function or a composite function including a sigmoid function, the Q-OCV curve generation unit 112 does not need to differentiate the Q-OCV approximation curve to generate a dV / dQ curve, and the inflection point extraction unit 113 can extract the inflection point directly from the peak of the slope of the sigmoid function included in the generated Q-OCV approximation curve. It is also possible to approximate the region between the low SOC region and the high SOC region using, for example, a hyperbolic tangent function, which has its own inflection point. In other words, if the inflection point unique to a function approximating the region including the inflection point of the Q-OCV approximation curve is extracted, it corresponds to the inflection point of the Q-OCV approximation curve, so that it is not necessary to differentiate the Q-OCV approximation curve to generate a dV / dQ curve, and the inflection point of the Q-OCV approximation curve can be extracted. If there are multiple inflection points in the Q-OCV approximation curve, the curve is divided so that each region contains one inflection point, and multiple functions each having their own unique inflection points are associated with the multiple divided regions to synthesize the Q-OCV approximation curve.

[0065] The Q-OCV curve generator 112 may sequentially generate Q-OCV curves for the cell by expanding / contracting or shifting the Q-OCV curves for the positive electrode and the negative electrode of the target cell that have been prepared in advance, and identify the Q-OCV curve that best fits a plurality of Q-OCV plots within a certain period. This method requires default characteristic data for the Q-OCV curves of the positive electrode and the negative electrode of the target cell, and is difficult to apply to cells for which default characteristic data is not available.

[0066] As shown in Figures 4A, 4B, 5A, and 5B, the peaks of the dV / dQ curves corresponding to the inflection points of the Q-OCV approximate curves occur at the stage change points of the negative electrode. Compared to the capacity Q1 from the peak of the dV / dQ curve to full charge based on battery data for month X of year 202X (fixed period a), the capacity Q2 from the peak of the dV / dQ curve to full charge based on battery data for the same month one year later (fixed period b) is smaller.

[0067] The degradation estimation unit 115 uses this change from the capacity Q1 to the capacity Q2 as an index indicating the degradation state of the cell. For example, the degradation estimation unit 115 may regard the difference between the capacity Q1 and the capacity Q2 as the reduced capacity.

[0068] Furthermore, a degradation capacity map is created in advance by experimenting or simulating the capacity from the peak of the dV / dQ curve of the target cell from the initial state to the end of use state, from which the capacity from the peak to full charge is derived. The capacity from the peak to full charge in the initial state may be assigned a degradation level of 100%, and the capacity from the peak to full charge in the end of use state may be assigned a degradation level of 0%, and the capacity from the peak to full charge in each state may be normalized and defined. The degradation estimation unit 115 estimates the degradation level of the target cell by referring to the degradation capacity map based on the capacity from the peak to full charge of the dV / dQ curve based on the battery data for the target period.

[0069] The regression equation generating unit 114 generates a regression equation by regressing the time transition of a plurality of inflection points extracted from a plurality of Q-OCV approximate curves generated based on battery data for a plurality of fixed periods.

[0070] FIG. 7 shows an example of a monthly plot of the capacity Q from the peak of the dV / dQ curve to full charge, and a regression line of the plots of multiple capacity Qs. The horizontal axis represents time [months], and the vertical axis represents capacity Q [Ah]. In FIGS. 4A, 4B, 5A, and 5B, the increase in the cumulative discharge amount from full charge is defined as "-". In FIG. 7, the "-" is omitted to indicate the chargeable amount from the negative electrode peak toward full charge, but the actual capacity Q is the same. While FIG. 7 shows an example of a simple linear regression of multiple capacities Q from the negative electrode peak to full charge with the number of months elapsed, the regression equation generation unit 114 may generate a linear regression equation using the 0.5 power of the number of months elapsed. Depending on the type of cell, a linear regression equation using explanatory variables of other powers, such as the 0.4 power or the 0.6 power of the number of months elapsed, may also be generated.

[0071] The degradation estimation unit 115 predicts the progression of degradation of the target cell based on the regression equation generated by the regression equation generation unit 114. By inputting a future month into the regression equation, the degradation estimation unit 115 can predict the capacity Q from the negative electrode peak to full charge for that month. Based on the capacity Q from the negative electrode peak to full charge, the degradation estimation unit 115 can estimate the degree of degradation of the cell for that month by referring to a degradation capacity map.

[0072] The deterioration estimation unit 115 can also estimate the deterioration of each material from the peak of the dV / dQ curve. In a lithium-ion battery, lithium ions move back and forth between the positive electrode active material and the negative electrode active material as the battery is charged and discharged. During discharge, as lithium ions move from the negative electrode to the positive electrode, the negative electrode potential rises and the positive electrode potential falls. The OCV, which is the difference between the two, falls. During charge, as lithium ions move from the positive electrode to the negative electrode, the positive electrode potential rises and the negative electrode potential falls. The OCV, which is the difference between the two, rises. The shape of the Q-OCV curve of a cell varies depending on the type and amount of the positive electrode active material and negative electrode active material. The shape of the Q-OCV also changes with deterioration.

[0073] In general, it is difficult to extract structural changes resulting from the positive electrode active material and the negative electrode active material from the Q-OCV curve of a cell. On the other hand, multiple peaks are observed in the dV / dQ curve of a cell. Whether each peak is due to the structure of the positive electrode active material or the structure of the negative electrode active material can be estimated from the shapes of the dV / dQ curves of the positive electrode and the negative electrode at least in the initial state.

[0074] Each peak corresponds to a maximum or minimum value of the rate of change of the positive electrode potential or negative electrode potential. The positive electrode potential and negative electrode potential are determined by the lithium composition ratio in the active material. The lithium composition ratio in the active material at the time of the maximum or minimum rate of change of potential is constant regardless of the cell capacity. Therefore, the rate of decrease in the distance between the two peaks originating from the positive electrode in the cell's dV / dQ curve corresponds to the rate of decrease in the positive electrode capacity. For example, if the distance between the two peaks originating from the positive electrode in the current dV / dQ curve has decreased to 80% of the distance between the two peaks originating from the positive electrode in the initial dV / dQ curve, it can be estimated that the positive electrode capacity has decreased to 80% of its initial state. The same applies to the negative electrode.

[0075] FIG. 8 shows an example of a dV / dQ curve for a certain cell. In the example shown in FIG. 8, four peaks (maximum values) are extracted from the dV / dQ curve of the cell. The first peak N1 and the third peak N2 are peaks originating from the negative electrode, and the second peak P1 and the fourth peak P2 are peaks originating from the positive electrode. The rate of decrease in the distance between the first peak N1 and the third peak N2 (i.e., the charge amount required to transition from the first peak N1 to the third peak N2) can be used to estimate the rate of decrease in the negative electrode capacity. Similarly, the rate of decrease in the distance between the second peak P1 and the fourth peak P2 can be used to estimate the rate of decrease in the positive electrode capacity.

[0076] The positive electrode potential and the negative electrode potential can be detected by introducing a metallic lithium reference electrode into the cell. In this case, the Q-OCV curve of the positive electrode, the Q-OCV curve of the negative electrode, the dV / dQ curve of the positive electrode, and the dV / dQ curve of the negative electrode can be estimated. This allows the rate of decrease in the distance between two peaks in the dV / dQ curve of the positive electrode and the rate of decrease in the distance between two peaks in the dV / dQ curve of the negative electrode to be directly determined. However, commercially available general cells do not have a reference electrode. Therefore, in this embodiment, the peaks originating from the positive electrode and the peaks originating from the negative electrode are identified among the multiple peaks in the dV / dQ curve of the target cell, and the deterioration of the positive electrode and the negative electrode is estimated.

[0077] In cell design, it is common to add excess capacity to both the positive and negative electrodes. Designers fill the positive electrode with more positive active material than will actually be used for charging and discharging, and fill the negative electrode with more negative active material than will actually be used for charging and discharging. The more positive active material there is, the wider the Q-OCV curve of the positive electrode, and the more negative active material there is, the wider the Q-OCV curve of the negative electrode. The width of the Q-OCV curves of the positive and negative electrodes shrinks due to degradation.

[0078] The area where the Q-OCV curve of the positive electrode and the Q-OCV curve of the negative electrode overlap is the cell capacity, and the designer sets the lower limit of the cell capacity at SOC = 0% (for example, battery voltage = 2.5 V) and the upper limit at SOC = 100% (for example, battery voltage = 4.2 V).

[0079] 9 is a diagram showing an example of the Q-OCV curves of a cell, a positive electrode, and a negative electrode in the initial state of a certain cell. The horizontal axis represents the capacity [Ah], and the vertical axis represents the OCV [V] and the positive electrode potential [V vs. L ＋ / L], negative electrode potential [VvsL ＋ / L]. In the example shown in Figure 9, the lower limit capacity of the positive electrode is set lower than the lower limit capacity of the cell, and further discharge is actually possible from a state of SOC = 0%. The difference between the lower limit capacity of the positive electrode and the lower limit capacity of the cell is the positive electrode excess capacity. The upper limit capacity of the negative electrode is set higher than the cell upper limit capacity, and further charge is actually possible from a state of SOC = 100%. The difference between the upper limit capacity of the negative electrode and the upper limit capacity of the cell is the negative electrode excess capacity.

[0080] FIG. 10 shows examples of contraction of the Q-OCV curves of a cell, positive electrode, and negative electrode. The upper diagram shows a state in which the width of the Q-OCV curve of the positive electrode has contracted by 10% due to degradation such as cracking or isolation of the positive electrode active material. The lower diagram shows a state in which the width of the Q-OCV curve of the negative electrode has contracted by 10% due to degradation such as cracking or isolation of the negative electrode active material. The BMU of the battery pack system 21 sets the range between the lower and upper voltage limits of the cell as the cell capacity. If the positive electrode capacity decreases by 10% as shown in the upper diagram, or if the negative electrode capacity decreases by 10% as shown in the lower diagram, the cell capacity will also decrease by 10%.

[0081] In the case of film degradation caused by a reaction between the lithium on the negative electrode and the electrolyte, the amount of positive and negative electrode active material does not decrease, so the width of the Q-OCV curves for the positive and negative electrodes does not narrow. However, because the amount of lithium ions available for charging and discharging decreases, the Q-OCV curve for the positive electrode shifts to the discharge side, resulting in a decrease in battery capacity.

[0082] 11 is a flowchart showing the flow of a cell degradation state estimation process performed by the battery analysis system 10 according to the embodiment. The Q-OCV curve generator 112 reads battery data for a certain period from the battery data storage unit 121. Based on the read battery data for the certain period, the Q-OCV curve generator 112 extracts voltage data that can be regarded as the OCV (S10). Based on the time-series current data, the Q-OCV curve generator 112 calculates the capacity Q at the time the extracted voltage data was measured (S11). The Q-OCV curve generator 112 generates a Q-OCV approximate curve from multiple Q-OCV plots (S12).

[0083] The inflection point extraction unit 113 differentiates the Q-OCV approximate curve to generate a dV / dQ curve (S13). The degradation estimation unit 115 compares the peak-to-full charge capacity Q1 of the dV / dQ curve generated based on the battery data for a certain period a with the peak-to-full charge capacity Q2 of the dV / dQ curve generated based on the battery data for a certain period b to estimate the degradation state of the cell (S14).

[0084] As described above, according to this embodiment, the degradation state of a secondary battery can be estimated with high accuracy by estimating the degradation state of the secondary battery based on changes in the inflection points of the Q-OCV approximation curve generated by plotting combinations of voltage and capacity Q that can be considered as OCV. A high-quality Q-OCV curve can be obtained even for battery packs that are frequently fast-charged and for which there is almost no battery data for low-rate charging. Therefore, a high-quality Q-OCV curve can be obtained without performing low-rate maintenance charging. Furthermore, this embodiment can extract the inflection points of LFP batteries, which are difficult to capture, with high accuracy.

[0085] 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.

[0086] The functions of the battery analysis system 10 described above may be implemented in the control unit 25 of the electric vehicle 20 or the control unit 32 of the charging station 30 .

[0087] In the above embodiment, the electric vehicle 20 is assumed to be a four-wheeled electric vehicle. In this regard, it may also be an electric motorcycle (electric scooter), an electric bicycle, or an electric kick scooter. Furthermore, electric vehicles include not only full-scale electric vehicles but also low-speed electric vehicles such as golf carts and land cars. Furthermore, the device in which the assembled battery system 21 is installed is not limited to the electric vehicle 20. Devices in which the assembled battery system 21 is installed include electric vehicles such as electric ships, railway vehicles, and multicopters (drones), stationary power storage systems, and consumer electronic devices (smartphones, notebook PCs, etc.).

[0088] The embodiment may be specified by the following items.

[0089] [Item 1] A battery analysis system (10) comprising: a data acquisition unit (111) that acquires time-series battery data including voltage and current of a secondary battery (E1); a Q-OCV curve generation unit (112) that extracts voltage data that can be regarded as an OCV (Open Circuit Voltage) based on the battery data for a certain period, calculates a capacity Q when the extracted voltage data is measured, and generates a Q-OCV curve; an inflection point extraction unit (113) that extracts an inflection point from the Q-OCV curve; and a deterioration estimation unit (115) that estimates deterioration of the secondary battery (E1) based on a plurality of inflection points extracted from a plurality of Q-OCV curves that are generated based on the battery data for a plurality of certain periods.

[0090] This allows the deterioration of the secondary battery (E1) to be estimated with high accuracy.

[0091] [Item 2] The battery analysis system (10) according to Item 1, wherein the inflection point extraction unit (113) differentiates the Q-OCV curve to generate a dV / dQ curve, and the degradation estimation unit (115) compares a capacity from a peak to full charge of the dV / dQ curve generated based on the battery data for a certain period with a capacity from a peak to full charge of a dV / dQ curve generated based on the battery data for another certain period, thereby estimating a degradation state of the secondary battery (E1).

[0092] This allows the deterioration of the secondary battery (E1) to be estimated with high accuracy based on the change in the inflection point over time.

[0093] [Item 3] The battery analysis system (10) according to Item 1, wherein the Q-OCV curve generation unit (112) generates the Q-OCV approximation curve using a composite function including a function having its own specific inflection point that approximates a region including the inflection point, and the degradation estimation unit (115) estimates the degradation state of the secondary battery by comparing a capacity from a peak of a slope of a function having its own specific inflection point, which is generated based on the battery data for a certain period, to a full charge, with a capacity from a peak of a slope of a function having its own specific inflection point, which is generated based on the battery data for another certain period.

[0094] This allows the deterioration of the secondary battery (E1) to be estimated with high accuracy based on the change in the inflection point over time.

[0095] [Item 4] The battery analysis system according to Item 1, further comprising a regression equation generation unit (114) that generates a regression equation by regressing time transitions of a plurality of inflection points extracted from a plurality of Q-OCV curves, the regression equation being generated based on the battery data for a plurality of fixed periods, and the deterioration estimation unit (115) predicts a deterioration transition of the secondary battery (E1) based on the regression equation.

[0096] According to this, by generating a regression equation of the inflection point, it is possible to predict the future deterioration transition of the secondary battery (E1) with high accuracy.

[0097] [Item 5] The battery analysis system according to Item 1, wherein the Q-OCV curve generation unit (112) extracts voltage data when the battery is in a resting state for a first predetermined time or more and less than a second predetermined time.

[0098] This allows for the collection of high quality Q-OCV plots.

[0099] [Item 6] The battery analysis system (10) according to Item 1 or 5, wherein the Q-OCV curve generation unit (112) extracts voltage data when the battery is in a low-rate charge / discharge state where the charge / discharge rate is equal to or less than a predetermined value.

[0100] This allows the collection of Q-OCV plots for battery packs with short rest periods.

[0101] [Item 7] The battery analysis system (10) according to Item 1 or 5, wherein the Q-OCV curve generation unit (112) generates voltage data that can be regarded as the OCV based on an equivalent circuit model of the secondary battery (E1).

[0102] This allows the collection of Q-OCV plots for battery packs with short rest periods.

[0103] [Item 8] The battery analysis system (10) according to Item 1, wherein the Q-OCV curve generation unit (112) divides the full charge capacity of the secondary battery (E1) into a plurality of capacity ranges and calculates a representative value of OCV for each capacity range.

[0104] This can reduce the influence of outliers in the Q-OCV plot.

[0105] [Item 9] The battery analysis system (10) according to Item 1, wherein the Q-OCV curve generation unit (112) generates the Q-OCV curve in which an upper limit value of Q is a full charge capacity [Ah] and a lower limit value of Q is Y [Ah], and sets the value of Y according to the type of the secondary battery (E1).

[0106] This makes it possible to reduce the amount of calculation required to generate the Q-OCV curve.

[0107] [Item 10] A battery analysis method comprising the steps of: acquiring time-series battery data including voltage and current of a secondary battery (E1); extracting voltage data that can be considered as an OCV (Open Circuit Voltage) based on the battery data for a certain period, calculating a capacity Q when the extracted voltage data was measured, and generating a Q-OCV curve; extracting an inflection point from the Q-OCV curve; and estimating deterioration of the secondary battery (E1) based on a plurality of inflection points extracted from a plurality of Q-OCV curves generated based on the battery data for a plurality of certain periods.

[0108] This allows the deterioration of the secondary battery (E1) to be estimated with high accuracy.

[0109] [Item 11] A battery analysis program that causes a computer to execute the following processes: a process of acquiring time-series battery data including voltage and current of a secondary battery (E1); a process of extracting voltage data that can be considered as an OCV (Open Circuit Voltage) based on the battery data for a certain period, calculating a capacity Q when the extracted voltage data was measured, and generating a Q-OCV curve; a process of extracting an inflection point from the Q-OCV curve; and a process of estimating deterioration of the secondary battery (E1) based on a plurality of inflection points extracted from a plurality of Q-OCV curves generated based on the battery data for a plurality of certain periods.

[0110] This allows the deterioration of the secondary battery (E1) to be estimated with high accuracy.

[0111] 2 Commercial power system, 5 Network, 10 Battery analysis system, 11 Control unit, 12 Memory unit, 13 Communication unit, 20 Electric vehicle, 21 Battery pack system, 22 Voltage sensor, 23 Current sensor, 24 Temperature sensor, 25 Control unit, 26 Communication unit, 30 Charging stand, 31 Power supply unit, 32 Control unit, 33 Communication unit, 111 Data acquisition unit, 112 Q-OCV curve generation unit, 113 Inflection point extraction unit, 114 Regression equation generation unit, 115 Deterioration estimation unit, 121 Battery data storage unit.

Claims

1. A battery analysis system comprising: a data acquisition unit that acquires time-series battery data including the voltage and current of a secondary battery; a Q-OCV curve generation unit that extracts voltage data that can be considered as OCV (Open Circuit Voltage) based on the battery data for a certain period of time, calculates the capacity Q when the extracted voltage data was measured, and generates a Q-OCV curve; an inflection point extraction unit that extracts inflection points from the Q-OCV curve; and a deterioration estimation unit that estimates deterioration of the secondary battery based on multiple inflection points extracted from multiple Q-OCV curves that were generated based on the battery data for multiple fixed periods.

2. The battery analysis system of claim 1, wherein the inflection point extraction unit differentiates the Q-OCV curve to generate a dV / dQ curve, and the degradation estimation unit compares the capacity from the peak to full charge of the dV / dQ curve generated based on the battery data for a certain period with the capacity from the peak to full charge of the dV / dQ curve generated based on the battery data for another certain period, thereby estimating the degradation state of the secondary battery.

3. The battery analysis system of claim 1, wherein the Q-OCV curve generation unit generates the Q-OCV approximation curve using a composite function including a function having its own specific inflection point that approximates a region including the inflection point, and the degradation estimation unit estimates the degradation state of the secondary battery by comparing the capacity from the peak of the slope of the function having its own specific inflection point, which is generated based on the battery data for a certain period, to full charge, with the capacity from the peak of the slope of the function having its own specific inflection point, which is generated based on the battery data for another certain period.

4. The battery analysis system according to claim 1, further comprising a regression equation generation unit that generates a regression equation by regressing the time transition of multiple inflection points extracted from multiple Q-OCV curves, the regression equation being generated based on the battery data for multiple fixed periods, and the deterioration estimation unit predicts the deterioration transition of the secondary battery based on the regression equation.

5. The battery analysis system according to claim 1, wherein the Q-OCV curve generation unit extracts voltage data when the battery is in a resting state for a first predetermined time or more and less than a second predetermined time.

6. The battery analysis system according to claim 1 or 5, wherein the Q-OCV curve generation unit extracts voltage data when the battery is in a low-rate charge / discharge state where the charge / discharge rate is equal to or lower than a predetermined value.

7. The battery analysis system according to claim 1 or 5, wherein the Q-OCV curve generation unit generates voltage data that can be regarded as the OCV based on an equivalent circuit model of the secondary battery.

8. The battery analysis system according to claim 1, wherein the Q-OCV curve generation unit divides the full charge capacity of the secondary battery into a plurality of capacity ranges and calculates a representative value of OCV for each capacity range.

9. The battery analysis system according to claim 1, wherein the Q-OCV curve generation unit generates the Q-OCV curve in which the upper limit of Q is the full charge capacity [Ah] and the lower limit of Q is Y [Ah], and sets the value of Y according to the type of the secondary battery.

10. A battery analysis method comprising the steps of: acquiring time-series battery data including the voltage and current of a secondary battery; extracting voltage data that can be considered as OCV (Open Circuit Voltage) based on the battery data for a certain period of time, calculating the capacity Q when the extracted voltage data was measured, and generating a Q-OCV curve; extracting inflection points from the Q-OCV curve; and estimating deterioration of the secondary battery based on multiple inflection points extracted from multiple Q-OCV curves generated based on the battery data for multiple fixed periods.

11. A battery analysis program that causes a computer to execute the following processes: a process of acquiring time-series battery data including the voltage and current of a secondary battery; a process of extracting voltage data that can be considered as OCV (Open Circuit Voltage) based on the battery data for a certain period of time, calculating the capacity Q when the extracted voltage data was measured, and generating a Q-OCV curve; a process of extracting inflection points from the Q-OCV curve; and a process of estimating deterioration of the secondary battery based on multiple inflection points extracted from multiple Q-OCV curves generated based on the battery data for multiple fixed periods of time.

Images