Battery state detection device and battery state detection method

Abstract

To provide a battery state detection device that can highly accurately detect the state of a secondary battery with small data amount.SOLUTION: A battery state detection device 1 comprises: a data acquisition part 2 acquiring battery information including at least voltage and current of a secondary battery; a storage part 4 storing a battery parameter including resistance function of a positive electrode and a negative electrode prepared for each type of the secondary batteries and the resistance function of the positive electrode and the negative electrode; a parameter calculation part 3 calculating a fitting coefficient for identifying OCV on the basis of battery information acquired by the data acquisition part from charge start to charge end or from discharge start to discharge end at each of some times of charging and discharging the secondary battery; and an OCV identification part 5 identifying SOC-OCV characteristics on the basis of the fitting coefficient calculated by the parameter calculation part and potential function of the positive electrode and the negative electrode stored in the storage part.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a battery state detection device and a battery state detection method for a secondary battery. [Background technology]

[0002] In recent years, the use of secondary batteries has expanded beyond electric vehicles (hereinafter referred to as EVs) to include power storage devices for grid stabilization, power storage devices for railway systems, industrial battery systems, etc. Furthermore, the reuse of used secondary batteries is also being considered.

[0003] For this reason, it is necessary to identify the relationship between the steady-state OCV (Open Circuit Voltage) and SOC (State Of Charge) of the secondary battery, and the relationship between the battery resistance and SOC on the charge and discharge sides.

[0004] For example, Patent Document 1 discloses a technique for estimating the relationship between OCV and SOC and the relationship between battery resistance and SOC from continuous voltage data in a BMS (battery management system) of a secondary battery. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Publication No. 2020-134279 Summary of the Invention [Problem to be solved by the invention]

[0006] The technology of Patent Document 1 requires high calculation performance because continuous voltage data is used to analyze the discharge curve and identify the OCV and resistance.

[0007] An object of the present invention is to provide a battery state detection device that detects the state of a secondary battery with high accuracy using a small amount of data. [Means for solving the problem]

[0008] In order to solve the above problems, the battery state detection device of the present invention includes: a data acquisition unit that acquires battery information including at least the voltage and current of a secondary battery; a memory unit that stores battery parameters including a resistance function of a positive electrode and a negative electrode that is prepared in advance for each type of secondary battery; a parameter calculation unit that calculates a fitting coefficient for identifying an OCV based on the battery information acquired by the data acquisition unit from the start of charging to the end of charging or from the start of discharging to the end of discharging in each of a plurality of charge and discharge cycles of the secondary battery; and an OCV identification unit that identifies SOC-OCV characteristics based on the fitting coefficient calculated by the parameter calculation unit and the potential functions of the positive electrode and the negative electrode stored in the memory unit. The parameter calculation unit calculates the battery resistance of the battery at the start of charging or the start of discharging of a battery that is not fully charged from the battery information, calculates the amount of charge by integrating the charging current from the start of charging until the battery is fully charged, or the discharging current from the start of discharging until the battery is completely discharged, and calculates the fitting coefficient by fitting a battery resistance function of the battery parameters to data indicated by the battery resistance and the amount of charge. I made it so that. [Effects of the Invention]

[0009] According to the present invention, the battery state is detected from the state at the start and end of charging multiple times, so that a battery state detection device can be provided that detects the state of a secondary battery with high accuracy using a small amount of data. [Brief explanation of the drawings]

[0010] [Figure 1] 1 is a diagram illustrating a configuration of a secondary battery state diagnosis device according to an embodiment; [Figure 2] FIG. 2 is a diagram showing an equivalent circuit model of a battery. [Figure 3] FIG. 10 is a diagram showing an example of the voltage behavior of a battery when a square wave current is applied to the battery. [Figure 4A] FIG. 1 is a characteristic diagram of the negative electrode potential versus discharge amount of a battery using graphite as the negative electrode active material. [Figure 4B] FIG. 1 is a characteristic diagram of the negative electrode resistance versus discharge amount of a battery using graphite as the negative electrode active material. [Figure 5A] FIG. 1 is a characteristic diagram of the positive electrode potential versus discharge amount of a battery using a ternary positive electrode material (NMC) as the positive electrode active material. [Figure 5B]FIG. 1 is a characteristic diagram of the negative electrode resistance versus discharge amount of a battery using a ternary positive electrode material (NMC) as the positive electrode active material. [Figure 6A] FIG. 1 is a characteristic diagram of the positive electrode potential versus discharge amount of a battery using lithium iron phosphate (LFP) as the positive electrode active material. [Figure 6B] FIG. 1 is a characteristic diagram of the negative electrode resistance versus discharge amount of a battery using lithium iron phosphate (LFP) as the positive electrode active material. [Figure 7] FIG. 10 is a diagram illustrating an outline of processing by a parameter calculation unit. [Figure 8] FIG. 1 is a diagram illustrating an outline of a process for identifying an OCV relative to an SOC of a battery. [Figure 9] FIG. 10 is a diagram illustrating processing by a parameter calculation unit. [Figure 10A] FIG. 10 is a diagram showing an element φ1(k) of a matrix φ(k) of resistance sensitivity and voltage sensitivity. [Figure 10B] FIG. 10 is a diagram showing an element φ2(k) of a matrix φ(k) of resistance sensitivity and voltage sensitivity. [Figure 10C] FIG. 10 is a diagram showing an element φ3(k) of a matrix φ(k) of resistance sensitivity and voltage sensitivity. [Figure 10D] FIG. 10 is a diagram showing an element φ4(k) of a matrix φ(k) of resistance sensitivity and voltage sensitivity. [Figure 11] FIG. 10 is a diagram of a formula showing P(k) in the case of recursive least squares. [Figure 12] FIG. 10 is a diagram of a mathematical formula showing P(k) in the case of a nonlinear Kalman filter. [Figure 13] FIG. 10 is a diagram showing a formula for calculation by recursive least squares. [Figure 14] FIG. 10 is a diagram showing a formula for calculation by recursive least squares. DETAILED DESCRIPTION OF THE INVENTION

[0011] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In the drawings, functionally identical elements may be denoted by the same numerals. Note that the drawings show embodiments in accordance with the principles of the present disclosure, but these are for understanding the present disclosure and are not to be used to interpret the present disclosure in a limiting manner. The description in this specification is merely a typical example and does not limit the scope of the invention according to the present disclosure.

[0012] Furthermore, although the embodiments are described in detail to enable those skilled in the art to practice the contents of the present disclosure, other embodiments are possible, and changes in configuration and structure and substitutions of various elements are possible without departing from the scope and spirit of the technical ideas of the present disclosure.

[0013] FIG. 1 is a diagram showing the configuration of a secondary battery state diagnosis device according to an embodiment. Below, we will explain how the secondary battery status diagnosis device of the embodiment diagnoses the status of a secondary battery installed in an EV (Electric Vehicle), but the device can also be applied to secondary battery status diagnosis in devices other than EVs, such as power storage devices for grid interconnection stabilization, power storage devices for railway systems, and industrial battery systems.

[0014] The state detection device 1 of the embodiment is provided as a server on a network to which multiple EVs are connected, and estimates the state of each battery of each EV from a remote location, and performs remote diagnosis of the battery based on the estimated state.

[0015] The state detection device 1 of the embodiment is composed of a data acquisition unit 2, a parameter calculation unit 3, a battery parameter table 4, an OCV identification unit 5, a deterioration diagnosis / remaining life determination unit 6, and a battery parameter BMS transfer unit 8, and calculates the battery parameters of a battery pack 7 and performs deterioration diagnosis / remaining life determination.

[0016] The battery pack 7 is installed in an EV and is composed of multiple LFP (lithium iron phosphate) or MNC (nickel-cobalt-manganese ternary cathode lithium-ion) battery cells (hereinafter referred to as batteries) connected in series and parallel. In addition to LFP and MNC batteries, the cathode active material may be LCO (lithium cobalt dioxide), LMO (manganese-based), or NCA (nickel-cobalt-aluminum oxide), and the anode active material may be graphite, soft carbon, or LTO (lithium titanate).

[0017] BMS71 is a battery management system implemented in the battery pack 7. BMS71 measures the voltage, charge / discharge current, and temperature of multiple batteries connected in parallel, calculates the remaining battery capacity SOC, and determines whether there is a battery abnormality.

[0018] The data acquisition unit 2 of the state detection device 1 acquires the vehicle or battery pack identification information, charging date and time, battery voltage, charging / discharging current, and temperature as battery information from the BMS 71 of each battery pack 7 via road-to-vehicle communication of the EV or a mobile communication network such as a mobile phone, and stores the information in EV data 9 for each EV or battery pack identification information.

[0019] Specifically, the data acquisition unit 2 acquires the voltage, charging current, and temperature of each battery during charging. The data acquisition unit 2 acquires at least any of the following data (1) to (4). (1) Three pieces of data: voltage at the start of charging, voltage at the end of charging, and charge amount during charging. (2) Three pieces of data when charging to full charge: resistance at the start of charging (calculated by subtracting the voltage immediately after the start of charging / discharging from the voltage at x seconds, then dividing by the current), the amount of charge during charging, and the temperature at the start of charging (3) Five pieces of data: resistance at the start of charging, resistance at the end of charging, charge amount during charging, and temperature at the start and end of charging. (4) Seven data points: resistance at the start and end of charging, voltage at the start and end of charging, charge amount during charging, and temperature at the start and end of charging.

[0020] The above describes the case where the data acquisition unit 2 acquires data on the battery during charging, but when acquiring data on the battery during discharging, it acquires any of the following data (5) to (8). (5) Three pieces of data: voltage at the start of discharge, voltage at the end of discharge, and amount of charge during discharge. (6) When discharging to the end of discharge, the resistance at the start of discharge (if the initial discharge current is constant from the start of discharge, calculate by subtracting the voltage at x seconds from the voltage immediately after discharge starts and then dividing by the current), the amount of charge during discharge, and the temperature at the start of discharge. (7) Five pieces of data: resistance at the start of discharge, resistance at the end of discharge, charge amount during discharge, and temperature at the start and end of discharge. (8) Seven data items: resistance at the start and end of discharge, voltage at the start and end of discharge, charge amount during discharge, and temperature at the start and end of discharge.

[0021] The data acquisition unit 2 acquires the above data, for example, when the vehicle is charging at a charging station or when the vehicle is charging while parked overnight, or from the start of the vehicle's journey until just before the end of the journey.

[0022] Here, a method for calculating the polarization resistance of a battery during charging will be described. The polarization resistance of the battery is determined using the equivalent circuit model of the battery shown in Figure 2. In Figure 2, R0 represents the direct current resistance (hereinafter sometimes referred to as DCR), Rpr represents the polarization resistance, and Cp represents the polarization capacity. The open circuit voltage of the battery is indicated by OCV.

[0023] 3 is a diagram showing an example of the voltage behavior of a battery when a square wave current I is applied to the battery, where the horizontal axis represents elapsed time. When a current I is applied, the battery voltage V changes as shown in Figure 4. This change in voltage V can be broadly divided into three components: a DC voltage component I×R0, a polarization voltage component Vpr, and an OCV fluctuation component ΔOCV.

[0024] The DC voltage component I×R0 responds instantaneously to changes in the current I. That is, it rises instantaneously in response to the rise of the current I, remains at a constant level, and then disappears as the current I falls. The polarization voltage component Vpr fluctuates with a delay relative to the change in the current I. That is, it gradually increases after the current I rises and gradually decreases after the current I falls. The OCV fluctuation component ΔOCV represents the change in the OCV of the battery and corresponds to the difference between OCV1, the OCV value before the start of charging, and OCV2, the OCV value after the start of charging. This OCV fluctuation component ΔOCV corresponds to the amount of change in the battery's state of charge according to the charge amount.

[0025] 3, the voltage V at the time of the rise of the current I is expressed as the current I × DC resistance R0. Then, if the OCV fluctuation component ΔOCV is minute for X time, the voltage Vx X seconds after the rise of the current I is approximately equal to the sum of the DC voltage component I × R0 and the polarization voltage component Vpr, and Vx = I × R0 + Vpr holds.

[0026] From the above, the parameter calculation unit 3 periodically acquires the measured values of the battery voltage and charging current from the state when the battery is not fully charged until it is fully charged. If Vs is the voltage data immediately after the current starts to rise, Vx is the voltage data X seconds after that time, and Ix is the current at that time, then Vs = Ix × R0, and therefore the polarization resistance Rpr can be calculated using equation (1).

[0027]

number

[0028] The above X seconds may be, for example, 30 seconds or a value greater than that, but is a fixed value. Furthermore, the resistance may be a DC resistance, or the sum of the polarization resistance and the DC resistance. During discharge, if the initial discharge current is constant from the start of discharge, it can be calculated by subtracting the voltage at x seconds from the voltage immediately after discharge starts and then dividing the result by the current. The above is the resistance obtained from the measurement values collected by the data acquisition unit 2.

[0029] Returning to Figure 1, the EV data 9 is a memory unit that stores, for each battery pack identified by the identification information acquired by the data acquisition unit 2, the time series of battery voltage, charging current, and temperature at predetermined time intervals until full charge.

[0030] Specifically, the EV data 9 stores in advance a negative electrode potential function, a positive electrode potential function, a negative electrode resistance function, and a positive electrode resistance function prepared for each battery type. Here, the negative electrode potential function represents the relationship between the state of charge, the discharge amount or charge amount, and the negative electrode potential, and the positive electrode potential function represents the relationship between the state of charge, the discharge amount or charge amount, and the positive electrode potential. Also, the negative electrode resistance function represents the relationship between the state of charge, the discharge amount or charge amount, and the negative electrode resistance, and the positive electrode resistance function represents the relationship between the state of charge, the discharge amount or charge amount, and the positive electrode resistance.

[0031] The parameter calculation unit 3 calculates parameters (also called fitting coefficients) for identifying the OCV from any of the data (1) to (8) acquired by the data acquisition unit 2 and the negative electrode potential function and positive electrode potential function or the negative electrode resistance function and positive electrode resistance function stored in the EV data 9.

[0032] The OCV identification unit 5 calculates an OCV table, an SOC table, and a resistance table based on the fitting coefficients obtained by the parameter calculation unit 3 using a calculation method described in detail below, identifies the relationship between the OCV and the SOC of the battery (SOC-OCV characteristics), and stores them in the battery parameter table 4.

[0033] The battery parameter table 4 is a storage unit that stores parameters of the negative electrode potential function, positive electrode potential function, negative electrode resistance function, and positive electrode resistance function for each battery, as well as an OCV table, an SOC table, and a resistance table.

[0034] Here, an example of the characteristics of the positive electrode resistance, negative electrode resistance, negative electrode potential, and positive electrode potential of the battery with respect to the discharge amount is shown in FIGS. 4A to 6B. FIG. 4A is a graph showing the characteristics of the negative electrode potential versus the discharge amount of a battery using graphite as the negative electrode active material, and FIG. 4B is a graph showing the characteristics of the negative electrode resistance versus the discharge amount. FIG. 5A is a graph showing the characteristics of the positive electrode potential versus the discharge amount of a battery using a ternary positive electrode material (NMC) as the positive electrode active material, and FIG. 5B is a graph showing the characteristics of the negative electrode resistance versus the discharge amount. FIG. 6A is a graph showing the characteristics of the positive electrode potential versus the discharge amount of a battery using lithium iron phosphate (LFP) as the positive electrode active material, and FIG. 6B is a graph showing the characteristics of the negative electrode resistance versus the discharge amount.

[0035] 4A to 6B show the characteristics of potential or resistance relative to the amount of discharge, but the curves showing the characteristics of potential or resistance relative to the amount of charge are inverted left and right of the figures.

[0036] The resistance curves of the negative electrode resistance and positive electrode resistance shown in Figures 4B, 6B, and 6B are stored as negative electrode resistance functions and positive electrode resistance functions in the battery parameter table 4. Similarly, the potential curves of the negative electrode potential and positive electrode potential shown in Figures 4A, 5A, and 6A are stored as negative electrode potential functions and positive electrode potential functions in the battery parameter table 4.

[0037] Returning to Figure 1, the deterioration diagnosis and remaining life determination unit 6, described in detail below, diagnoses battery deterioration and determines remaining life based on changes from the initial values of battery parameters. As a result, the deterioration diagnosis and remaining life determination unit 6 recommends battery replacement in the EV, or determines the end of life from the battery deterioration history, predicts the battery replacement time, and notifies the EV.

[0038] When the OCV table and resistance table of the battery parameter table 4 are updated, the battery parameter BMS transfer unit 8 transfers the OCV table and resistance table to the BMS 71 of each EV.

[0039] Specifically, the state detection device 1 of the embodiment is configured by a computer including a CPU that performs arithmetic processing, a memory, a communication unit, an operation unit, a display unit, and a nonvolatile storage medium. The CPU executes a program stored in the nonvolatile storage medium to realize the functions of the parameter calculation unit 3 and the OCV identification unit 5. Furthermore, the data acquisition unit 2 and the battery parameter BMS transfer unit 8 are configured by the communication unit, and the battery parameter table 4 and EV data 9 are configured by the nonvolatile storage medium.

[0040] An outline of the processing of the state detection device 1 of the embodiment will be described below. First, the parameter calculation unit 3 will be described.

[0041] Parameter calculation process 3 first inputs any of the following in the kth charge / discharge, acquired by data acquisition unit 2: initial battery temperature Ts(k), final battery temperature Te(k), initial resistance Rs(k), final resistance Re(k), charge / discharge amount Q(k) (unit: Ah, with charge being +), voltage Vs(k) immediately before the start of charge / discharge, and voltage Ve(k) after sufficient time has passed since the end of charge / discharge. The following describes the case where all information is available.

[0042] Next, let the unknown parameters be an, bn, ap, bp, and Qmax (collectively referred to as the vector θ), and find θ from the input values Ts(k), Te(k), Rs(k), Re(k), and Q(k) (k=1, ...). bn is the position of the negative pole function Vp when SOC=0%, bp is the position of the positive pole function when SOC=0%, an is how much the position of the negative pole function shifts when the SOC changes by 1%, ap is how much the position of the positive pole function shifts when the SOC changes by 1%, and Qmax is the Ah capacity of the battery.

[0043] The function of Vp is expressed as Vp(bp-SOC×ap), and the function of Vn is expressed as Vn(bn-SOC×an). Vp and Vn are stored in the battery parameter table 4 mentioned above. If the SOC at the start of the kth charge / discharge is SOCi(k), the theoretical formula for the OCV before the start of charge / discharge is OCV=Vp(bp-SOCi(k)×ap)-Vn(bn-SOCi(k)×bp) using the positive electrode potential-negative electrode potential, and the OCV after the end of charge / discharge is OCV=Vp(bp-{SOCi(k)+100Q(k) / Qmax}×ap)-Vn(bn-{SOCi(k)+100Q(k) / Qmax}×bp). That is, Vs(k)= Vp(bp-SOCi(k)×ap)-Vn(bn-SOCi(k)×bp), Ve(k)= Vp(bp-{SOCi(k)+100Q(k) / Qmax}×ap)-Vn(bn-{SOCi(k)+100Q(k) / Qmax}×bp).

[0044] Next, we will explain the theoretical formula for the resistance estimation function R. Details of this function will be described in the examples, but it is expressed as R(SOC,Temp; an,bn,ap,bp,Qmax,A,B,C)=A×Rp(SOC,Temp;ap,bp,Qmax)+B×Rn(SOC,Temp;an,bn,Qmax)+C. Rp is the positive electrode resistance stored in the battery parameter table 4 described above, Rn is the negative electrode resistance stored in the battery parameter table 4 described above, A and B are the respective resistance multipliers, and C is an unknown parameter indicating the resistance of the bias component (busbar resistance). Note that if polarization resistance is used for R, C=0. Here, to identify the resistance, it is necessary to identify A, B, and C in addition to θ.

[0045] Here, the process for identifying θ, A, B, and C will be outlined. Fig. 7 shows the relationship between the polarization resistance of a battery (hereinafter referred to as battery resistance) and the charge capacity during charging. Because battery resistance changes due to battery degradation, the measured polarization resistance at a given charge capacity of a battery in use deviates from the initial battery resistance curve of the battery, as shown in Fig. 7.

[0046] The fitting of the battery resistance curve to the measured value of polarization resistance by the parameter calculation unit 3 is nothing but calculating θ, A, B, and C as parameters for OCV identification.

[0047] The parameter calculation unit 3 plots one battery resistance (polarization resistance Rpr) from one battery charge / discharge cycle using the above procedure. Then, after plotting the battery resistance (polarization resistance Rpr) from multiple battery full charge cycles, it fits the battery resistance curve and calculates parameters θ, A, B, and C for OCV identification. That is, it fits a battery resistance function pre-stored in the storage unit to data representing the battery resistance and the charge amount, and calculates values that are scaled and translated during fitting (parameter adjustments for θ, A, B, and C). The following assumes that C is not 0, but if the measured resistance is polarization resistance, C = 0.

[0048] Next, the OCV identification unit 5 determines the OCV. The graph in Figure 8 shows the SOC-OCV characteristics, with the horizontal axis representing SOC (%) and the vertical axis representing OCV (V). The vertical axis in Figure 8 also represents the battery resistance (mΩ).

[0049] The deterioration diagnosis and remaining life determination unit 6 compares Qmax, which indicates the battery's Ah capacity when fully charged, between the time of start of use and the present, and determines that the battery is in a deteriorated state when the capacity drop reaches a predetermined value.

[0050] Furthermore, the degradation diagnosis and remaining life determination unit 6 may determine the degradation state based on the rate of increase in the polarization resistance calculated by the battery parameter calculation unit 3 relative to the time of start of use.

[0051] Furthermore, the deterioration diagnosis and remaining life judgment unit 6 calculates a trend curve for Qmax, which indicates the Ah capacity of the battery when fully charged, against the operating time of the battery, and predicts the number of days until Qmax, the target value for battery replacement, as the remaining life of the battery.

[0052] The deterioration diagnosis / remaining life determination unit 6 may obtain a trend curve of the polarization resistance calculated by the battery parameter calculation unit 3 instead of Qmax, and predict the number of days until the resistance value reaches a target value for battery replacement, as the remaining life of the battery.

[0053] The processing of the state detection device 1 of the embodiment will be described in detail below, taking as an example each case where, for example, only charging is performed during charging and discharging, and the final charge is full charge. When the state detection device 1 of the embodiment detects the state of a battery during discharging, charging can be interpreted as discharging. Specifically, the start of charging should be interpreted as the start of discharging, the end of charging should be interpreted as the end of discharging, and full charge should be interpreted as the end of discharging. [Example]

[0054] In this embodiment, a general calculation method of the parameter calculation unit will be described. Then, as described above, the parameter calculation unit 3 calculates the polarization resistance Rpr shown in equation (1) from the periodic battery voltage, current value, and temperature during one charge from when the battery is not fully charged to when it is fully charged, and performs this process for multiple charges to calculate the polarization resistance Rpr(k) (k is the number of charges). The method will be described.

[0055] First, the method for identifying unknown parameters will be described. Here, to identify the unknown parameters θ, A, B, and C, if the squared error E(k) between the measured value and the theoretical value in the following equation (2) is taken as E(k), then θ, A, B, and C that minimize the following equation (3) are found. Here, γ is a predetermined fixed constant that is the square of the reciprocal of the current, and indicates the weight of the resistance and voltage. Setting γ = 0 means that the parameters are identified using only the resistance information. Variations of γ will be described later. λ is a forgetting factor, and may be set to a predetermined constant such that 0 < λ ≦ 1, for example, 0.99.

[0056]

number

[0057]

number

[0058] Note that equation (2) contains an unknown variable SOCi(k) for each kth charge, and optimizing this variable would increase the amount of calculations. For this reason, SOCi(k) is calculated numerically each time so that the squared error E(k) for the kth charge / discharge is minimized. This is done by numerically calculating SOCi(k) by inputting the current θ, A, B, and C into the equation obtained by partially differentiating E(k) with respect to SOCi(k). Newton's method or bisection method can also be used for this.

[0059] As a numerical calculation method to minimize equation (3), all charging data can be used to find θ, A, B, and C all at once using the quasi-Newton method, or θ, A, B, and C can be found sequentially by extending the nonlinear Kalman filter. When finding them sequentially, initial values for θ, A, B, and C are first set.

[0060] Next, an example of sequentially finding the numerical calculation process that minimizes equation (3) will be described with reference to FIG.

[0061] In step S91, initial values are determined for θ (vectors of an, bn, ap, bp, and Qmax), A, B, and C. Possible values are set as the initial values. For example, bn and bp can be set to the maximum value of the domain of the function or a smaller value. an and ap can be set to the range of the domain of the function divided by 100 or a smaller value. Qmax can be set to the battery catalog value Ah. A, B, and C can be determined from the initially measured resistance and the resistance function values of the positive and negative electrodes, with A:B set to 1:1 and C = 0. The forgetting factor λ can be set to, for example, 0.99 or a value close to 1. Set S(0) = 0, and set P(0) to a constant α (a large value), which is αI. I is the identity matrix.

[0062] In step S92, the k-th charge / discharge data Ts(k), Te(k), Rs(k), Re(k), Q(k), Vs(k), and Ve(k) are acquired from the data acquisition unit 2.

[0063] In step S93, the k-th charge / discharge start SOCSOCi(k) is calculated numerically using the method described above.

[0064] In step S94, the resistance sensitivity and voltage sensitivity (numerical differentiation) matrix φ(k) for the current parameters θ(k-1), A(k-1), B(k-1), and C(k-1) is calculated. φ(k) is (φ1(k),φ2(k),φ3(k),φ4(k))T, where φ1, φ2, φ3, and φ4 are eight-dimensional column vectors shown in Figures 10A, 10B, 10C, and 10D, respectively. The equations in Figures 10A to 10D are partial differentials, and here the partial differentials are calculated numerically.

[0065] Here, we explain the partial differentiation with respect to Qmax. The theoretical resistance function R is expressed by equation (4).

[0066]

number

[0067] The theoretical resistance function R shown in equation (4) does not have Qmax(k) as an argument. However, in reality, when numerically calculating SOCi(k) that minimizes E(k) shown in equation (2), the value of SOCi(k) changes depending on Qmax(k), so SOCi(k) is a function of Qmax(k). Therefore, if SOCi(k) is re-expressed as SOCi(Qmax), and θ(k) = (an(k),bn(k),ap(k),bp(k),Qmax(k)), θ'(k) = (an(k),bn(k),ap(k),bp(k),Qmax(k)+δ), the partial derivative of the theoretical resistance function R with respect to Qmax becomes equation (5).

[0068]

number

[0069] δ is a small value (Ah). Similarly, the partial derivative of the theoretical resistance function R with respect to Qmax is given by equation (6).

number

[0070] The theoretical function of OCV is given by equation (7).

number

[0071] Therefore, the partial derivatives of the theoretical function of OCV with respect to Qmax are given by equations (8) and (9).

number

number

[0072] In step S95, the variance-covariance matrix S is updated. Then, P is calculated as the generalized inverse matrix of S. Here, the generalized inverse matrix will be explained.

[0073] Now, given a square matrix D, if D is full rank, then D -1 However, if D is not full rank, then D -1 cannot be defined. In this case, D + DD + =D + , D.D. + Matrix D that satisfies D=D + is the generalized inverse matrix. There are several generalized inverse matrices, but here we use the Moore-Penrose type generalized inverse matrix D + D=DD + shall be adopted.

[0074] The generalized inverse matrix of Moore-Penrose type is uniquely determined. |Dx-b| 2 If D is not full rank, there are countless x's for which the distance from the origin is smallest, but the x's for which the distance from the origin is smallest are uniquely determined, and are expressed as x=D.+ b. If D is full rank, the generalized inverse matrix is the same as the inverse matrix. In other words, the generalized inverse matrix is a method of calculating as if it were an inverse matrix even if D is degenerate.

[0075] This Moore-Penrose type general inverse matrix can be obtained by QR decomposition of the matrix D. That is, D = Qdiag(λ1,...,λn)Q -1 Eigenvalue decomposition is performed as follows: D + = Qdiag(1 / λ1, ..., 1 / λn)Q-1, but if λ1 = 0, then D+ = Qdiag(0, 1 / λ2, ..., 1 / λn)Q-1. This QR decomposition operation can also be used to find P(k) from S(k).

[0076] If we are talking about an inverse matrix rather than a general inverse matrix, P(k) can be found using the inverse matrix lemma. This becomes the usual recursive least squares, resulting in the formula shown in Figure 11. In the case of a nonlinear Kalman filter, λ = 1 and the update of P changes to the formula shown in Figure 12. In the formula in Figure 15, U is a 4x4 observation error matrix, and is a preset matrix such that U = diag(u1, u2, u3, u4). A small positive constant is set for uj (j = 1, 2, 3, 4).

[0077] In step S96, the observed error and theoretical error ε are calculated, and φ, A, B, and C are updated with ε. The updated parameters are passed to the OCV identification unit 5, which identifies the OCV function. Then, the process returns to step S92.

[0078] The above is an overview of the parameter calculation unit 3. However, in the case of NMC, if OCV(0) at SOC 0% and OCV(100) at SOC 100% are specified, ap and bp become functions of an and bn. The processing in this case will be described below. Equation (10) holds true when SOC is 0%.

[0079]

number

[0080] From equation (10), bp = Vp -1 (OCV(0) + Vn(bn)). Here, Vp is the inverse function Vp of Vp, because the positive electrode potential monotonically decreases in the case of NMC. -1 has a unique value. This makes bp a function of bn. Equation (11) holds true under the conditions of SOC 100%.

[0081]

number

[0082] From equation (11), bp-100ap=Vp -1 (OCV(0)+Vn(bn-100an)). bp is a function of bn, so ap={bp- Vp -1 (OCV(0)+Vn(bn-100an))} / 100, which is a function of an and bn. Therefore, θ becomes a vector of three elements (an, bn, Qmax), and there are no terms for elements that are partially differentiated with respect to ap and bp in the equations of Figures 10A to 10D.

[0083] If γ=0, that is, if the voltage information is not considered and only the resistance information is used, only φ1 and φ2 are calculated, and in step 94, φ(k)=(φ1(k) φ2(k)) T In step 96, ε is set to two elements, 1 and 2. If there is no resistance information and only voltage information, φ(k) = (φ3(k) φ4(k)) T and calculate with γ=1.

[0084] We will now discuss the process when φ is only φ2, that is, the resistance and the final value. In this case, γ=1 is set, and P can be calculated using a recursive formula even with a general inverse matrix. The recursive formula in this case is the formula shown in Figures 13 and 14. The same applies if φ is only φ1.

[0085] The maximum eigenvalue λM and its corresponding eigenvector q in Figure 14 may be found by the power method (|q| = 1). When λM > threshold, P(k) = P(k) - λMqTq. The power method finds the maximum eigenvalue of matrix A by setting x(k+1) = Ax(k) / |Ax(k)|, λ(k+1) = |Ax(k)| / |x(k)| (k = 0, 1, ...), giving an initial value x(0) ≠ 0, and successively finding x(k) and λ(k). λ(∞) becomes λM, and x(∞) becomes q. Although it is an iterative calculation, it can be done quickly.

[0086] The OCV of a battery is calculated by subtracting the negative electrode potential from the positive electrode potential of the battery. Therefore, if an, bn, ap, and bp are calculated in the OCV identification unit 5, the OCV function can be expressed by equation (12).

[0087]

number

[0088] Next, the theoretical derivation of the theoretical resistance R will be described. The battery current I is expressed by Butler-Volmer equation (13). Q is the discharge amount, and I0(Q) is a function that changes depending on the discharge amount Q. T is the absolute temperature.

[0089]

number

[0090] As a result, equation (14) is established, and the resistance R is given by equation (15).

number

number

[0091] Since equation (15) differs for each positive and negative electrode, hereafter the positive electrode resistance will be represented as Rp and the negative electrode resistance as Rn. Even if the positive electrode resistance Rp and negative electrode resistance Rn have the same Q, their values will change depending on the current I and temperature T. For this reason, some kind of correction for current and temperature is required. For example, the resistance value can be converted to when the current I is nearly 0 and the temperature is 25°C. Alternatively, it is also possible to only consider data where the current I is around 90A. Note that when the current I is nearly 0, the voltage V barely rises, so the measurement error is large and the data cannot be used as measurement data.

[0092] Next, the positive electrode resistance Rp and the negative electrode resistance Rn are converted into resistance values at 25°C. In the limit of I=0, R in equation (15) becomes equation (16).

number

[0093] Here, let T25 = 273.15 + 25, and let Ea / Rg of the positive electrode and negative electrode be Bp, Bn, and I0 be Ip and In, respectively. If the 25°C standard resistance function of the positive electrode at the limit of current I = 0 is R25(Q), and the 25°C standard resistance function of the negative electrode is Rn25, then we obtain equation (17).

number

[0094] From equation (17), the positive electrode resistance Rp and the negative electrode resistance Rn are expressed as equation (18) using Rn25 and Rp25.

number

[0095] If the fitting coefficients for the positive electrode resistance Rp and the negative electrode resistance Rn are A, B, and C, respectively, the resistance function R is obtained from equations (15) and (18) as shown in equation (19). Note that when polarization resistance is used for R, C = 0.

number

[0096] Here, I is the battery current at the time of measurement. Rp25 and Rn25 are scaled and translated relative to the Q axis. Since Rp25(Q) and Rn25(Q) are functions of Q, they are made into functions using the aforementioned an, bn, ap, and bp. In other words, Rp25(SOC) = Rp(bp-SOC×ap) and Rn25(SOC) = Rn(bn-SOC×an). When I is small, equation (19) can be approximated to equation (20).

[0097]

number

[0098] Furthermore, when Bp≈Bn, it may be approximated by equation (21).

number

[0099] If C=0 (when using a polarized resistor for the resistor), the element in the C part is deleted to accommodate this.

[0100] Next, a specific method of calculating the numerical values for one charge from an incompletely charged state to a fully charged state of the battery in this embodiment will be described.

[0101] The resistance curve shown as the sum of the positive electrode resistance function and the negative electrode resistance function is fitted to the polarization resistance Rpr(k) to find the fitting coefficient. Since the last data is fully charged, SOCi(k) + 100Q(k) / Qmax(k) = 100. Therefore, SOCi(k) = 100 - 100Q(k) / Qmax(k), and there is no need to calculate SOCi(k) numerically from the first and last data. Then, use only φ1, or φ1 and φ2 for φ.

[0102] The parameter calculation unit 3 sets different constraints on the voltage function of the OCV shown in equation (18) depending on whether LFP is used in the positive electrode (constraint 1) or NMC is used in the positive electrode (constraint 2).

[0103] 《Constraint 1》 When graphite is used for the negative electrode and LFP is used for the positive electrode, it is difficult for the parameter calculation unit 3 to identify bp, bn, ap, and an in equation (16) using only the constraints of OCV(0) and OCV(100). For this reason, as shown in FIG. 8A, a certain constraint is added to Vp at OCV(0), for example, 3.4185 V. In other words, a constant positive electrode potential is set as a fitting constraint. Note that although this embodiment has been described using LFP as an example, this constraint can also be applied to a positive electrode having a flat potential curve like LFP.

[0104] This determines bn (bn=Vn -1 (3.4185-OCV(0)) voltage rising part at the end of discharge), bp is determined only by the range (between minimum and maximum) (Vn -1 indicates the inverse function of the function Vn).

[0105] Then, for the OCV (100), a constraint is added as to whether the OCV voltage is due to the initial value of the positive electrode discharge (constraint 1-1) or whether the negative electrode potential is due to the initial value of the discharge (operated as Vp = 3.4185) (constraint 1-2).

[0106] In the case of constraint 1-1, bp-100ap = Vp -1 The constraint is (OCV(100)+0.09V). Here, 0.09V is the voltage where the negative electrode potential of graphite shown in FIG. 6A becomes flat, and it is not a fixed numerical value but indicates a range. As a result, an and ap become unknown variables, bn is fixed, and bp is determined from ap, so bp = 100ap + Vp -1 (OCV(100)+0.09V).

[0107] In the case of constraint 1-2, the constraint is that bn-100an satisfies Vn(bn-100an)=3.4185-OCV(100), where 3.4185>OCV(100)>OCV(0). As a result, the value of bn is determined by the OCV(0) condition, and therefore an is determined. In other words, ap and bp become unknown variables.

[0108] To determine which constraint to adopt, (Constraint 1-1) or (Constraint 1-2), test each and adopt the one with the smaller error. According to constraint 1, fitting can be performed using the electrode resistance even for a positive electrode active material such as LFP, whose positive electrode resistance is almost constant.

[0109] 《Constraint 2》 When graphite is used for the negative electrode and MNC is used for the positive electrode, the parameter calculation unit 3 calculates bp=Vp as described above, since the positive electrode potential Vp decreases monotonically as shown in FIG. 7A. -1 (OCV(0)+Vn(bn)) and bp-100ap=Vp -1 (OCV(100)+Vn(bn-100an)) is added as a constraint. That is, the monotonically decreasing positive electrode potential is set as a constraint for fitting. Note that although NMC is used as an example in this example, this constraint can also be applied to positive electrodes that have a monotonically decreasing potential curve like NMC. For example, LCO, LMO, NCA, etc.

[0110] This allows ap and bp to be removed from the unknowns in equation (16), making it easier to identify the unknowns. Ap and bp can be found from an and bn.

[0111] According to this embodiment, it is possible to estimate the battery state even for a battery using an active material such as an LFP whose OCV is constant relative to the SOC. Furthermore, since it is possible to estimate the state of charge with a small amount of data, it is possible to diagnose the battery even when there is a limit to the communication volume. [Example]

[0112] In the above example, the fitting coefficient and OCV are calculated based on the measured data of voltage, current, and temperature when multiple charging cycles result in full charge. However, in this embodiment, a case where multiple charging cycles do not result in full charge is described. This embodiment can also be implemented when full charge is achieved. In this case, φ is set to φ using φ1 and φ2, or only φ1 or only φ2.

[0113] In this embodiment, the parameter calculation unit 3 defines the SOC of the battery at the start of charging as SOC0(k), the battery resistance (polarization resistance Rpr) at the start of charging as Rpr_b(k), and the battery resistance when charging ends before reaching full charge as Rpr_a(k).The parameter calculation unit 3 then fits a battery resistance curve to the battery resistances Rpr_b(k) and Rpr_a(k) to calculate fitting coefficients.

[0114] Here, the battery resistance Rpr_b(k) at the start of charging is calculated using equation (1), and the battery resistance Rpr_a(k) at the end of charging is calculated as follows:

[0115] As shown in Figure 4, when the current I falls, the DC voltage component (current I × DC resistance R0) disappears, and the voltage V becomes the sum of the polarization voltage component Rpr_a(k) and the OCV fluctuation ΔOCV. The polarization voltage component Rpr_a(k) gradually decreases after the current I falls, so if the current I, at which the polarization voltage component Rpr_a(k) can be ignored, is set to voltage Vx (equal to OCV) X seconds after the current falls, and the voltage immediately after the current falls to Vs, then Vs = Vpr + Vx holds. Here, if the current immediately before the current falls is set to Is, then Rpr_a(k) can be calculated using equation (22).

[0116]

number

[0117] In the above-mentioned Examples 1 and 2, matching of resistance curves of polarization resistance was described. In this example, the SOC of the battery at the start of charging is set as an unknown quantity, SOC0(k), and an OCV function curve indicating the OCV is fitted to two battery voltages, namely the voltage immediately before charging the battery and the voltage after a sufficient time has passed since charging was completed, to determine the OCV characteristics relative to the SOC.

[0118] That is, the two aforementioned φ are used, φ3 and φ4, and γ is set to 1. In this embodiment, the parameter calculation unit 3 and the OCV identification unit 5 perform integrated processing. [Example]

[0119] In this example, the SOC of the battery at the start of charging is assumed to be an unknown quantity, SOC0(k), and the OCV characteristics for the battery parameters and SOC are calculated based on the battery resistance Rpr calculated from the voltage immediately after the current rises, the voltage and current X seconds after that time, and the voltage and charge amount after a sufficient amount of time has passed since the end of charging. In this case, φ1 and φ4 are used for the aforementioned φ.

[0120] Conversely, when determining the OCV characteristics relative to the battery parameters and SOC based on the voltage immediately before charging the battery and the battery resistance at the end of charging, φ2 and φ3 can be used for the above-mentioned φ. In this embodiment, the parameter calculation unit 3 and the OCV identification unit 5 perform integrated processing. [Example]

[0121] In this embodiment, the parameter calculation unit 3 and OCV identification unit 5 perform integrated processing, and the SOC of the battery at the start of charging is set as an unknown quantity SOC0(k), and the OCV characteristics relative to the battery parameters and SOC are calculated from the battery resistance Rpr_b(k) and voltage OCVb(k) at the start of charging, and the battery resistance Rpr_a(k) and voltage OCVa(k) at the end of charging.

[0122] Specifically, battery resistance Rpr_b(k) is the battery resistance at the start of charging calculated using equation (1), battery resistance Rpr_a(k) is the battery resistance at the end of charging calculated using equation (22), voltage OCVb(k) is the battery voltage immediately before charging, and voltage OCVa(k) is the battery voltage after a sufficient amount of time has passed since charging ended. In this case, all of φ1, φ2, φ3, and φ4 are used for the aforementioned φ.

[0123] In the above, the state detection device 1 of the embodiment receives the voltage, charging current, and temperature of the battery from the BMS 71. However, the BMS 71 may also realize the functions of the parameter calculation unit 3 and the OCV identification unit 5 to determine the battery parameters and OCV of the batteries in the battery pack 7.

[0124] In the above, it has been explained that the state detection device 1 of the embodiment determines the fitting coefficient and OCV from the charging current of the battery, but in a similar manner, the fitting coefficient and OCV can be determined from the voltage and discharging current of the battery. [Explanation of symbols]

[0125] 1. Status detection device 2 Data Acquisition Section 3 Parameter calculation section 4 Battery parameter table (memory section) 5 OCV Identification Unit 6. Deterioration diagnosis and remaining life judgment unit (diagnosis unit) 7 Battery pack 71 BMS 8 Battery parameter BMS transfer section 9 EV data (storage section)

Claims

1. a data acquisition unit that acquires battery information including at least the voltage and current of the secondary battery; a storage unit that stores a resistance function of a positive electrode and a negative electrode that is prepared in advance for each type of secondary battery, and battery parameters including the resistance functions of the positive electrode and the negative electrode; a parameter calculation unit that calculates a fitting coefficient for identifying an OCV based on the battery information acquired by the data acquisition unit from the start of charging to the end of charging or from the start of discharging to the end of discharging in each of a plurality of charge and discharge cycles of the secondary battery; an OCV identification unit that identifies SOC-OCV characteristics based on the fitting coefficients calculated by the parameter calculation unit and potential functions of the positive electrode and the negative electrode stored in the storage unit; Equipped with The parameter calculation unit calculating a battery resistance of a battery that is not fully charged at the start of charging or discharging from the battery information; The charge amount is calculated by integrating the charging current from the start of charging until the battery is fully charged, or the discharging current from the start of discharging until the battery is completely discharged. The fitting coefficient is calculated by fitting a battery resistance function of the battery parameters to data represented by the battery resistance and the charge amount. A battery state detection device characterized by:

2. 2. The battery state detection device according to claim 1, The parameter calculation unit When the battery uses graphite for the negative electrode and a positive electrode active material having a flat potential curve for the positive electrode, the fitting constraint is that the positive electrode potential is constant; When the battery uses graphite for the negative electrode and a positive electrode active material whose potential decreases monotonically for the positive electrode, the monotonically decreasing positive electrode potential is set as a constraint for fitting. A battery state detection device characterized by:

3. a data acquisition unit that acquires battery information including at least the voltage and current of the secondary battery; a storage unit that stores a resistance function of a positive electrode and a negative electrode that is prepared in advance for each type of secondary battery, and battery parameters including the resistance functions of the positive electrode and the negative electrode; a parameter calculation unit that calculates a fitting coefficient for identifying an OCV based on the battery information acquired by the data acquisition unit from the start of charging to the end of charging or from the start of discharging to the end of discharging in each of a plurality of charge and discharge cycles of the secondary battery; an OCV identification unit that identifies SOC-OCV characteristics based on the fitting coefficients calculated by the parameter calculation unit and potential functions of the positive electrode and the negative electrode stored in the storage unit; Equipped with The parameter calculation unit Calculating the battery resistance of the battery at the start of charging or discharging from the battery information; Calculating the battery resistance of the battery at the end of charging or discharging from the battery information; First, with the SOC at the start of charging as an unknown quantity, a battery resistance function of the battery parameters is fitted to data indicated by the battery resistance and the SOC at the start of charging or the start of discharging and data indicated by the battery resistance and the SOC at the end of charging or the end of discharging, to determine the SOC; Next, a battery resistance function of the battery parameters is fitted to data indicated by the battery resistance at the start of charging or discharging and the determined SOC, and data indicated by the resistance at the end of charging or discharging and the determined SOC, to calculate the fitting coefficient. A battery state detection device characterized by:

4. a data acquisition unit that acquires battery information including at least the voltage and current of the secondary battery; a storage unit that stores a resistance function of a positive electrode and a negative electrode that is prepared in advance for each type of secondary battery, and battery parameters including the resistance functions of the positive electrode and the negative electrode; a parameter calculation unit that calculates a fitting coefficient for identifying an OCV based on the battery information acquired by the data acquisition unit from the start of charging to the end of charging or from the start of discharging to the end of discharging in each of a plurality of charge and discharge cycles of the secondary battery; an OCV identification unit that identifies SOC-OCV characteristics based on the fitting coefficients calculated by the parameter calculation unit and potential functions of the positive electrode and the negative electrode stored in the storage unit; Equipped with The parameter calculation unit and the OCV identification unit The SOC at the start of charging or discharging is set as an unknown quantity, and an OCV function curve is fitted based on the voltage of the battery immediately before charging or immediately before discharging and the voltage of the battery after a sufficient time has passed since charging or discharging was completed, to determine the OCV characteristics relative to the SOC. A battery state detection device characterized by:

5. a data acquisition unit that acquires battery information including at least the voltage and current of the secondary battery; a storage unit that stores a resistance function of a positive electrode and a negative electrode that is prepared in advance for each type of secondary battery, and battery parameters including the resistance functions of the positive electrode and the negative electrode; a parameter calculation unit that calculates a fitting coefficient for identifying an OCV based on the battery information acquired by the data acquisition unit from the start of charging to the end of charging or from the start of discharging to the end of discharging in each of a plurality of charge and discharge cycles of the secondary battery; an OCV identification unit that identifies SOC-OCV characteristics based on the fitting coefficients calculated by the parameter calculation unit and potential functions of the positive electrode and the negative electrode stored in the storage unit; Equipped with The parameter calculation unit and the OCV identification unit use an SOC at the start of charging or discharging as an unknown quantity, The battery parameters and the OCV characteristics relative to the SOC are calculated based on the battery resistance calculated from the voltage immediately after the current rises and the voltage and current X seconds after that time, and the voltage and charge amount when a sufficient amount of time has passed since the end of charging or discharging, or Alternatively, fitting is performed based on the voltage immediately before charging or immediately before discharging the battery and the battery resistance at the end of charging or discharging to determine the battery parameters and the OCV characteristics relative to the SOC. A battery state detection device characterized by:

6. a data acquisition unit that acquires battery information including at least the voltage and current of the secondary battery; a storage unit that stores a resistance function of a positive electrode and a negative electrode that is prepared in advance for each type of secondary battery, and battery parameters including the resistance functions of the positive electrode and the negative electrode; a parameter calculation unit that calculates a fitting coefficient for identifying an OCV based on the battery information acquired by the data acquisition unit from the start of charging to the end of charging or from the start of discharging to the end of discharging in each of a plurality of charge and discharge cycles of the secondary battery; an OCV identification unit that identifies SOC-OCV characteristics based on the fitting coefficients calculated by the parameter calculation unit and potential functions of the positive electrode and the negative electrode stored in the storage unit; Equipped with the parameter calculation unit sets an SOC at the start of charging or discharging as an unknown quantity, Calculating the battery resistance of the battery at the start of charging or discharging from the battery information; Calculating the battery resistance of the battery at the end of charging or discharging from the battery information; Fitting is performed based on the voltage immediately before charging or immediately before discharging the battery and the voltage of the battery after a sufficient time has passed since charging was completed, and the battery parameters and the OCV characteristics relative to the SOC are determined. A battery state detection device characterized by:

7. The battery state detection device according to any one of claims 1 to 6, further comprising: determining a degradation state based on the SOC-OCV characteristic identified by the OCV identification unit; The deterioration state of the battery is determined based on the amount of decrease in the Ah capacity of the battery when fully charged from the start of use, or The trend curve of the battery resistance against the operating time of the battery is calculated, and the number of days until the resistance increases to a predetermined value is predicted to determine the remaining life of the battery. Alternatively, a battery state detection device characterized by having a diagnostic unit that performs one of the following: determining a trend curve of the Ah capacity at full charge against the operating time of the battery, predicting the number of days until a predetermined charge amount is reached, and determining the remaining life of the battery.

8. 7. The battery state detection device according to claim 1, The parameter calculation unit performs fitting by a recursive least squares method, a nonlinear Kalman filter, or a quasi-Newton method. A battery state detection device characterized by:

9. acquiring battery information including at least the voltage and current of the secondary battery; a step of fitting a battery resistance function or a voltage function of the battery based on the battery resistance or voltage at the start of charging and at the start of charging, which are obtained from the battery information acquired from the start of charging to the end of charging or from the start of discharging to the end of discharging in each of a plurality of charging or discharging cycles of the secondary battery, to calculate a fitting coefficient for identifying an OCV; Identifying SOC-OCV characteristics based on the fitting coefficients and potential functions of the positive electrode and the negative electrode; Including, calculating a battery resistance of a battery that is not fully charged at the start of charging or discharging from the battery information; The charge amount is calculated by integrating the charging current from the start of charging until the battery is fully charged, or the discharging current from the start of discharging until the battery is completely discharged. The fitting coefficient is calculated by fitting the battery resistance function to the battery resistance and the charge amount. A battery state detection method comprising:

10. The battery state detection method according to claim 9, further comprising: determining a degradation state according to the obtained rate of change of the battery parameter; determining a deterioration state based on a decrease in the Ah capacity of the battery when fully charged from the start of use; A step of calculating a trend curve of the battery resistance of the battery against the operating time of the battery, predicting the number of days until the resistance value increases to a predetermined value, and calculating the remaining life of the battery; Alternatively, a step of determining a trend curve of the Ah capacity at full charge versus the operating time of the battery, predicting the number of days until the predetermined Ah capacity is reached, and determining the remaining life of the battery. A battery state detection method comprising:

11. 10. The battery state detection method according to claim 9, The battery resistance function or voltage function of the battery is fitted by the recursive least squares method using a generalized inverse matrix as the inverse matrix of the variance-covariance matrix, and fitting coefficients for identifying the OCV are calculated. A battery state detection method comprising:

12. The battery state detection method according to claim 11, The battery state detection method is characterized in that the recursive least squares method calculates a generalized inverse matrix using a recursive formula, and further subtracts the maximum eigenvalue and the eigenvector component of the maximum eigenvalue from the inverse matrix of the calculated variance-covariance matrix.

Images