Parameter estimation system, parameter estimation device, and parameter estimation method

Abstract

In the present invention, a controller executes: an acquisition process for acquiring measurement data obtained by measuring the battery of a vehicle, and usage history data including usage history of the vehicle; a first state amount estimation process for estimating, on the basis of the measurement data and the usage history data, the current state amount of the battery as a first state amount; a second state amount estimation process for estimating, by using a battery model, the current state amount of the battery as a second state amount; and a parameter estimation process for estimating an internal parameter through a data assimilation process for subjecting the first state amount and the second state amount to data assimilation.

Description

Parameter Estimation System, Parameter Estimation Device, and Parameter Estimation Method

[0001] The present invention relates to a parameter estimation system, a parameter estimation device, and a parameter estimation method for estimating internal parameters of a battery model.

[0002] There is known a technique of calculating an actually measured value of the degradation degree of a secondary battery module by comparing the current capacity and internal resistance of the secondary battery module with an initial value, and correcting a prediction formula for calculating a predicted value of the degradation degree using the output data of the secondary battery module based on the actually measured value, and calculating the predicted value by the corrected prediction formula (Patent Document 1).

[0003] Japanese Patent Application Laid-Open No. 2020-119658

[0004] The technique described in Patent Document 1 is a technique of correcting a prediction model so that the predicted value of the degradation degree matches the actually measured value when the difference between the predicted value of the degradation degree and the actually measured value of the degradation degree widens beyond a predetermined value. However, since the parameters of the prediction model are in a deviated state without being updated to match the current state of the battery to be estimated, the technique described in Patent Document 1 has a problem that it cannot prevent the error of the predicted value of the degradation degree from becoming large itself.

[0005] The problem to be solved by the present invention is to provide a parameter estimation system, a parameter estimation device, and a parameter estimation method capable of estimating the internal parameters of a battery model so that the error of the predicted value becomes smaller.

[0006] The present invention solves the above problems by acquiring measurement data obtained by measuring a vehicle battery and usage history data including the usage history of the vehicle, estimating the current state quantity of the battery as a first state quantity based on the measurement data and the usage history data, estimating the current state quantity of the battery as a second state quantity using a battery model, and estimating internal parameters by a data assimilation process of data assimilation of the first state quantity and the second state quantity.

[0007] According to the present invention, the internal parameters of the battery model can be estimated so that the error of the predicted value becomes smaller.

[0008] Figure 1 is a block diagram of a parameter estimation system according to the first embodiment of the present invention. Figure 2 is a diagram showing an example of data assimilation processing in this embodiment. Figure 3 is a diagram showing an example of internal parameter estimation in this embodiment. Figure 4 is a flowchart showing the control procedure of the parameter estimation method executed by the controller according to this embodiment. Figure 5 is a block diagram of a parameter estimation system according to the second embodiment of the present invention. Figure 6 is a block diagram of a parameter estimation system according to the third embodiment of the present invention. Figure 7 is a block diagram of a parameter estimation system according to the fourth embodiment of the present invention.

[0009] Embodiments of the parameter estimation system, parameter estimation apparatus, and parameter estimation method according to the present invention will be described with reference to the drawings.

[0010] <<First Embodiment>> Figure 1 is a block diagram showing a parameter estimation system according to the first embodiment of the present invention. As shown in Figure 1, the parameter estimation system 1000 includes a vehicle 1. The parameter estimation system 1000 is a system that estimates the internal parameters of a battery model 108, which will be described later, on the vehicle 1. The battery model 108 is a model for estimating the state quantities of a battery mounted on the vehicle 1. Details will be described later. Note that the parameter estimation system according to this embodiment may be a system that estimates the internal parameters of the battery model 108 on a server (cloud), as will be described later, or it may be a system that estimates the internal parameters of the battery model 108 by distributing the functions between the server and the vehicle.

[0011] Vehicle 1 is an electric vehicle or hybrid vehicle, or any other electric vehicle equipped with a battery. Vehicle 1 is equipped with an ECU 10 and a secondary battery 11. The secondary battery 11 supplies power as the battery of Vehicle 1. In addition to the ECU 10 and the secondary battery 11, Vehicle 1 is equipped with sensors for detecting the state of the secondary battery 11, sensors for detecting the external conditions of the secondary battery 11, a communication device for communicating with a server, a motor that serves as the drive source for the vehicle, auxiliary equipment, etc. Sensors for detecting the state of the secondary battery 11 include, for example, an EIS (Electrochemical Impedance Spectroscopy) sensor 12, a current sensor 13, a voltage sensor 14, a temperature sensor 15, etc. Sensors for detecting the external conditions of Vehicle 1 include, for example, an ambient temperature sensor 16, a position sensor 17, a charging sensor 18, etc. Vehicle 1 is the vehicle to be estimated. That is, the state quantities of the secondary battery 11 installed in Vehicle 1 are estimated. ECU 10 is an example of the "parameter estimation device" described in the claims. Secondary battery 11 is an example of the "battery" described in the claims.

[0012] The ECU 10 is, for example, a battery ECU that controls the secondary battery 11 of the vehicle 1. The ECU 10 includes an on-board controller 100. The on-board controller 100 is connected to on-board sensors and the like via a CAN communication network. The on-board controller 100 is an example of the "controller" described in the claims. The on-board controller 100 includes a processor (computer) having hardware and software, and this computer includes a ROM that stores a control program, a CPU that executes the control program stored in the ROM, and RAM that functions as an accessible storage device. The storage location of the control program is not limited to ROM, but may be any storage medium that the computer can read, and the location of the storage medium may be inside or outside the on-board controller 100. Note that the ECU 10 is not limited to one ECU, but may be composed of multiple ECUs.

[0013] The in-vehicle controller 100 includes, as functional blocks, a measurement data acquisition unit 101, a usage history data acquisition unit 102, a first state variable estimation unit 103, a parameter storage unit 104, a model data generation unit 105, a second state variable estimation unit 106, and an internal parameter estimation unit 107. The in-vehicle controller 100 executes each of the above functions through the cooperation of software for realizing each of the above functions or executing each of the above processes with the hardware described above. More specifically, these functions are realized by the CPU executing a control program recorded in ROM. In this embodiment, the functions of the in-vehicle controller 100 are divided into seven blocks and the functions of each functional block are described, but the functions of the in-vehicle controller 100 do not necessarily have to be divided into eight blocks, and may be divided into six or fewer functional blocks, or eight or more functional blocks. In addition, the in-vehicle controller 100 may have functions other than those described above.

[0014] The measurement data acquisition unit 101 performs an acquisition process to acquire measurement data measured by the secondary battery 11. In the acquisition process, the measurement data acquisition unit 101 acquires measurement data from each sensor mounted on the vehicle 1 at regular intervals. The measurement data acquisition unit 101 may process the acquired measurement data by preprocessing or the like. The measurement data includes data related to the state of the secondary battery 11, for example, internal information of the secondary battery 11. The internal information of the secondary battery 11 includes at least the EIS value of the secondary battery 11. The data related to the state of the secondary battery 11 includes data on current, voltage, and temperature in the secondary battery 11. The measurement data acquired by the measurement data acquisition unit 101 is output to the first state quantity estimation unit 103.

[0015] The usage history data acquisition unit 102 performs an acquisition process to acquire usage history data of the vehicle 1. In the acquisition process, the usage history data acquisition unit 102 acquires usage history data from each sensor mounted on the vehicle 1 at regular intervals. The usage history data acquisition unit 102 may process the acquired usage history data by preprocessing or the like. The usage history data includes external information of the vehicle 1. The external information of the vehicle 1 is information relating to the history of external conditions during the use of the vehicle 1, and includes at least one of the following: the outside temperature of the vehicle 1, the position of the vehicle 1, and the type of charge when the vehicle 1 is being charged. The usage history data acquired by the usage history data acquisition unit 102 is output to the first state quantity estimation unit 103.

[0016] The first state quantity estimation unit 103 performs a first state quantity estimation process to estimate the current state quantity of the secondary battery 11 as the first state quantity based on the measurement data and the usage history data. In the first state quantity estimation process, the first state quantity estimation unit 103 generates inference data using the measurement data and the usage history data, and estimates the first state quantity of the secondary battery 11 based on the inference data. The inference data is data that includes the measurement data and the usage history data. The first state quantity estimated by the first state quantity estimation unit 103 is output to the internal parameter estimation unit 107. In this embodiment, the first state quantity is treated as an "observed value" estimated from observation data including the measurement data and the usage history data. As will be described later, the in-vehicle controller 100 corrects the second state quantity (predicted value) of the secondary battery 11 estimated from the battery model 108 based on the observation data obtained from the vehicle 1. The inference data may be generated by the measurement data acquisition unit 101 or the usage history data acquisition unit 102, or by another functional unit such as the inference data generation unit. In such cases, the generated inference data is output to the first state quantity estimation unit 103.

[0017] Here, an example of the first state quantity estimation process will be described. The first state quantity estimation unit 103 takes inference data, which includes measurement data and usage history data, as input data and inputs the input data to a trained model. The first state quantity estimation unit 103 estimates the first state quantity by having the trained model output output data that includes the first state quantity. The trained model is a model that has been trained to output output data that includes the first state quantity based on input data that includes inference data. The trained model is composed of a neural network including an input layer consisting of one or more neurons, an output layer, and at least one hidden layer. Input data including inference data is input to the input layer, and output data that includes the first state quantity is output from the output layer. For training the model, training data is used in which input data including inference data and output data including the first state quantity are associated. The first state quantity estimation unit 103 may further include a storage unit such as a database for storing the trained model. The first state quantity estimation unit 103 estimates the first state quantity using the trained model stored in the storage unit.

[0018] The parameter storage unit 104 performs storage processing to store the internal parameters of the battery model. The parameter storage unit 104 is, for example, a database that stores internal parameters. The parameter storage unit 104 stores the internal parameters of the battery model 108 of the secondary battery 11 estimated by the internal parameter estimation unit 107, which will be described later. The parameter storage unit 104 also stores the internal parameters of the battery models of secondary batteries of other electric vehicles (other vehicles) that are different from vehicle 1. In other words, the parameter storage unit 104 stores information about the battery models of multiple electric vehicles, including vehicle 1 and other vehicles. The information about the battery models includes the battery type of the secondary battery installed in each electric vehicle and the parameter values ​​of each internal parameter that defines the battery model.

[0019] The model data generation unit 105 executes a model generation process to generate model data including a battery model of the secondary battery 11. In the model generation process, the model data generation unit 105 first selects initial values ​​for each internal parameter of the battery model. Next, for each internal parameter, the model data generation unit 105 generates a parameter ensemble (set of parameters) in which multiple parameter values ​​with variation are grouped together based on the initial value. The parameter ensemble for each internal parameter is represented by a probability distribution of the multiple parameter values ​​that the internal parameter can take. For example, the probability distribution of the internal parameter is a normal distribution of parameter values ​​with the initial value as the mean. Then, the model data generation unit 105 generates multiple different parameter sets based on the parameter ensemble for each internal parameter. Each of the multiple parameter sets contains the parameter values ​​of each internal parameter. The model data generation unit 105 generates a model ensemble (set of models) as model data, which includes multiple battery models defined by each of the multiple parameter sets. The generated model data is output to the second state quantity estimation unit 106. In the second state quantity estimation unit 106, the battery model included in the model data is used as the battery model 108 to estimate the second state quantity.

[0020] Here, we will explain an example of a method for selecting initial values ​​for internal parameters. The model data generation unit 105 acquires battery type information for the secondary battery 11 (the secondary battery 11 mounted on the vehicle 1) that is the subject of parameter estimation. The model data generation unit 105 identifies a battery model of the same type as the secondary battery 11 that is the subject of parameter estimation from the information on battery models of other electric vehicles stored in the parameter storage unit 104, and selects the parameter values ​​of each internal parameter of the identified battery model as the initial values ​​for each internal parameter.

[0021] The second state quantity estimation unit 106 performs a second state quantity estimation process using the battery model 108 to estimate the current state quantity of the secondary battery 11 as the second state quantity. The battery model 108 is a battery model that represents the internal state of the secondary battery 11 of the vehicle 1. The second state quantity estimation unit 106 includes the battery model 108. The battery model 108 is a physical model that models the changes in the internal state of the secondary battery 11 based on the current state of the secondary battery 11. The internal parameters of the battery model 108 are parameters that indicate the internal state of the secondary battery 11. The internal parameters include, for example, parameters that determine the SEI (Solid Electrolyte Interphase) formation rate inside the secondary battery 11, and parameters that determine the crack generation rate of the active material inside the secondary battery 11 with respect to the cycle. Parameters that determine the SEI formation rate include, for example, the reaction rate constant and the activation energy. Parameters that determine the crack generation rate of the active material include, for example, the Li diffusion coefficient in the active material, the activation energy, the fracture strength of the active material, and the crack initiation density. When the internal parameters are estimated by the internal parameter estimation unit 107 (described later), the battery model 108 is updated with the estimated internal parameters.

[0022] The battery model 108 calculates the state variables (second state variables) of the secondary battery 11 using internal parameters. In this embodiment, the second state variables are treated as "predicted values" estimated based on the battery model 108. As described later, the in-vehicle controller 100 corrects the second state variables (predicted values) estimated from the battery model 108 based on observed data. The second state variable estimation unit 106 estimates the second state variables using the battery model included in the model data when model data including the battery model is input from the model data generation unit 105, for example. The second state variables estimated by the second state variable estimation unit 106 are output to the internal parameter estimation unit 107.

[0023] In this embodiment, the state variables (first state variable and second state variable) of the secondary battery 11 are, for example, the State of Health (SOH) of the secondary battery 11. Alternatively, the state variables of the secondary battery 11 may be the EIS value of the secondary battery 11. Alternatively, the state variables of the secondary battery 11 may be the DCR (Direct Current Resistance) value of the secondary battery 11. The data assimilation process described later is performed using the SOH, EIS value, or DCR value of the secondary battery 11 as the assimilation target. When the EIS value or DCR value of the secondary battery 11 is used as the state variable, the battery model 108 may include an equivalent circuit model for calculating the EIS value or DCR value.

[0024] In this embodiment, the second state quantity estimation unit 106 updates the internal parameters of the battery model 108 using the internal parameters estimated by the internal parameter estimation unit 107. Then, the second state quantity estimation unit 106 estimates the second state quantity using the battery model 108 after the internal parameters have been updated. The estimated second state quantity is stored in a secondary battery database (not shown). The in-vehicle controller 100 may further include an information presentation function based on the stored second state quantity. For example, the in-vehicle controller 100 may present the user with the status of the secondary battery 11 and the results of the determination of whether reuse, repurposing, or recycling is necessary based on the status of the secondary battery 11. The in-vehicle controller 100 may also predict the degradation of the secondary battery 11 using a battery degradation prediction model based on the stored second state quantity and present the degradation prediction results to the user.

[0025] The internal parameter estimation unit 107 performs a parameter estimation process to estimate internal parameters by performing a data assimilation process that assimilates the first state variable and the second state variable. This eliminates the difference between the first state variable and the second state variable. The data assimilation process updates the predicted values ​​(probability distribution of the second state variable) of the battery model 108 based on the observed values ​​(probability distribution of the first state variable). The internal parameters estimated by the internal parameter estimation unit 107 are output to the parameter storage unit 104.

[0026] The internal parameter estimation unit 107 performs data assimilation processing using Bayesian sequential computation. The internal parameter estimation unit 107 predicts the probability distribution of the second state variable of the battery model 108 at a certain point in time (for example, time t) one step ahead, based on the probability distribution of the second state variable of the battery model 108 at a certain point in time (for example, the initial time t-1). In this embodiment, time t is, for example, the current time. The internal parameter estimation unit 107 obtains the probability distribution of the first state variable at time t from the first state variable input from the first state variable estimation unit 103. Then, the internal parameter estimation unit 107 calculates the probability distribution of the second state variable in the state where the first state variable at time t has been obtained, using the probability distribution of the first state variable at time t and the probability distribution of the second state variable at time t. That is, the probability distribution of the predicted value in the state where the observed value has been obtained is calculated. In this embodiment, the predicted value is not a prediction of the second state variable at a future time, but a prediction of the second state variable at the current time. The internal parameter estimation unit 107 then estimates the internal parameters based on the probability distribution of the second state variable after data assimilation processing. For example, the internal parameter estimation unit 107 estimates the second state variable with the highest likelihood and estimates the internal parameters of the battery model 108 that calculates this second state variable as the internal parameters of the battery model 108 that represent the current state of the secondary battery 11. This makes it possible to estimate the internal parameters that represent the current internal state of the battery in each individual vehicle.

[0027] Here, an example of data assimilation processing using Bayesian sequential computation will be explained using Figure 2. Figure 2 is a diagram showing an example of data assimilation processing in this embodiment. In the graph of Figure 2, the horizontal axis represents time, and the vertical axis represents state variables. In Figure 2, the probability distributions of the state variables are shown in chronological order. In Bayesian sequential computation, each time the probability distribution of the first state variable, which is an observed value, is obtained, the internal parameter estimation unit 107 predicts the probability distribution of the second state variable for the current period from the probability distribution of the second state variable from the previous period, and calculates the probability distribution of the second state variable for the current period in the state in which the first state variable (observed value) was obtained. In the example of Figure 2, first, the internal parameter estimation unit 107 obtains the probability distribution P1 of the second state variable at time t-1. The probability distribution P1 is p(x ｔ－１ |１：ｔ－１ ) is expressed as follows. If time t-1 is the initial time, the probability distribution of the second state variable estimated by the initial battery model 108 is used as the probability distribution of the second state variable at time t-1. The initial battery model 108 is a battery model defined by the initial values ​​of the internal parameters. Next, the internal parameter estimation unit 107 calculates the probability distribution P2 of the second state variable at time t (the current period). The probability distribution P2 is expressed as p(x ｔ | １：ｔ－１ It is expressed as p(x). The internal parameter estimation unit 107 also obtains the probability distribution P3 of the first state variable at time t. The internal parameter estimation unit 107 then calculates the probability distribution P4 of the second state variable at time t in the state in which the first state variable has been obtained. The probability distribution P4 is p(x ｔ | １：ｔ It is represented as follows:

[0028] Furthermore, the internal parameter estimation unit 107 may perform Bayesian sequential calculations using one of the following for filtering: a particle filter, a Kalman filter, or an ensemble Kalman filter.

[0029] Here, an example of internal parameter estimation according to this embodiment will be explained using Figure 3. Figure 3 shows an example of internal parameter estimation in this embodiment. Figures 3(A) to (D) show the procedures for each process, including the generation of model data, estimation of the second state variable using the battery model included in the model data, data assimilation processing based on the second state variable and the first state variable, estimation of the internal parameters of the battery model, and storage of the updated internal parameters. Figure 3(A) shows an example of a parameter ensemble generated by the model data generation unit 105. Figure 3(B) shows an example of a model ensemble generated by the model data generation unit 105. Figure 3(C) schematically shows an example of Bayesian sequential calculation using a particle filter. Figure 3(C) shows the probability distribution of the state variable in time series order. The horizontal axis represents time, and the vertical axis represents the state variable (SOH). Figure 3(D) shows an example of the updated internal parameters of the battery model stored in the database of the parameter storage unit. Note that a fused particle filter may be used for filtering instead of a particle filter.

[0030] First, the model data generation unit 105 generates a parameter ensemble that expresses the variability of each internal parameter as a probability distribution, based on the initial values ​​of each internal parameter. For example, if the battery model 108 is defined by n internal parameters, as shown in Figure 3(A), the parameter ensemble includes probability distributions of N parameter values ​​for each of the n internal parameters, including parameter 1, parameter 2, ..., parameter n. Each internal parameter exhibits variability due to parameters requiring adaptation, unknown parameters (crack density, energy activation, etc.), individual differences, initial conditions, and uncertainty in boundary conditions. The variability of the parameter values ​​of each internal parameter is expressed as a probability distribution.

[0031] Next, the model data generation unit 105 generates a model ensemble as model data, which includes multiple battery models defined by each of the multiple parameter sets. Each internal parameter has N parameter values ​​that vary. Each parameter set contains one of the N parameter values ​​for each internal parameter, so N parameter sets are generated. In the example in Figure 3(B), the model ensemble includes N parameter sets, including parameter set 1, parameter set 2, parameter set 3, ..., parameter set N. Each parameter set contains n internal parameters that define the battery model. The parameter values ​​for each internal parameter included in each parameter set are different for each set. The model data generation unit 105 may generate N battery models defined by each parameter set from the N parameter sets.

[0032] The second state quantity estimation unit 106 estimates N second state quantities (second SOH) using N battery models. In the example in Figure 3, a second SOH ensemble (set of second SOHs) containing N second SOHs is output to the internal parameter estimation unit 107. The internal parameter estimation unit 107 uses the estimated distribution of the N second SOHs as the probability distribution P1 of the second SOH at time t-1 and samples particles according to the probability distribution P1 of the second SOH at time t-1. Time t-1 is one period before the current time (time t). In the example in Figure 3 (C), the distribution of each particle at time t-1 is shown to approximate the probability distribution of the second SOH. Next, the internal parameter estimation unit 107 predicts the distribution of each particle at time t, one period ahead, from the distribution of each particle at time t-1. The predicted particle distribution approximates the probability distribution P2 of the second SOH at time t. Furthermore, the internal parameter estimation unit 107 obtains the first SOH (observed value) and the probability distribution P3 of the first SOH at time t. Based on the probability distribution P3 of the first SOH, the internal parameter estimation unit 107 calculates the likelihood for each particle and calculates the weight of each particle. In the example of Figure 3 (C), the magnitude of the weight is expressed by the size of the particle. Based on the calculated weights, the internal parameter estimation unit 107 updates the particle ensemble (set) by resampling the particles. The internal parameter estimation unit 107 calculates a probability distribution that approximates the updated particle ensemble as the probability distribution P4 of the second state variable at time t in the state where the first state variable has been obtained.

[0033] The internal parameter estimation unit 107 estimates the second state variable with the highest likelihood and estimates the internal parameters of the battery model 108 that calculate the second state variable as the internal parameters of the battery model 108 that represent the current state of the secondary battery 11. The internal parameter estimation unit 107 outputs the estimated internal parameters to the parameter storage unit 104. As shown in Figure 3(D), the parameter storage unit 104 updates the stored internal parameters of the battery model 108 with the estimated internal parameters.

[0034] Furthermore, the internal parameter estimation unit 107 may use a Kalman filter for filtering. For example, after predicting the second state variable (predicted value) at time t (current period), the internal parameter estimation unit 107 calculates a Kalman gain when filtering the second state variable at time t using the first state variable (observed value) at time t, and corrects the second state variable (predicted value) at time t using the Kalman gain. Alternatively, the internal parameter estimation unit 107 may use an ensemble Kalman filter for filtering. For example, the internal parameter estimation unit 107 samples multiple second state variables (predicted values), predicts each second state variable (predicted value) at time t, filters each second state variable at time t using the first state variable (observed value) at time t, and obtains the corrected second state variable (predicted value) at time t.

[0035] In this embodiment, as described above, the internal parameter estimation unit 107 can update the internal parameters of the battery model 108 based on the current state of the battery of the vehicle 1 (first state). This allows the current internal state of the battery specific to each vehicle to be reflected in the battery model 108 in real time. Furthermore, by updating the internal parameters of the battery model 108 to match the current state of the secondary battery 11, the initial values ​​of the internal parameters of the battery model 108 become more accurate when estimating the future state of the battery. This reduces the error in the predicted value estimated from the battery model 108, thus preventing the error in the future predicted value from increasing over time. As described above, in this embodiment, the future state of the battery can be estimated with high accuracy, tailored to each individual vehicle. The on-board controller 100 may also use the updated battery model 108 to estimate the future second state (predicted value) of the secondary battery 11 and estimate the lifespan of the secondary battery 11. Since the error in the predicted second state is reduced, the lifespan of the secondary battery 11 is estimated with greater accuracy.

[0036] The secondary battery 11 is, for example, a lithium-ion secondary battery and is installed in the vehicle 1. An example of this type of secondary battery 11 is one in which an active material having multiple charge-discharge regions in which the charge-discharge potential changes in steps with the insertion and removal of lithium ions is used as the negative electrode active material. A graphite-based active material containing a graphite structure is preferred as such an active material having multiple charge-discharge regions in which the charge-discharge potential changes in steps with the insertion and removal of lithium ions. The secondary battery 11 is not limited to an electrolyte lithium-ion battery, but may also be an all-solid-state lithium-ion secondary battery. Furthermore, the secondary battery 11 is not limited to a lithium-ion battery, but may also be other types of batteries such as lead-acid batteries.

[0037] The secondary battery 11 is equipped with an EIS sensor 12, a current sensor 13, a voltage sensor 14, and a temperature sensor 15. The EIS sensor 12 measures impedance during discharge and charging by EIS measurement (electrochemical impedance measurement). The measured impedance value (EIS value) is output to the on-board controller 100. The current sensor 13 is a sensor for detecting the input and output current of the secondary battery 11. The current sensor 13 is connected to the wiring connected to the positive or negative terminal of the secondary battery 11. The voltage sensor 14 is a sensor for detecting the voltage between the terminals of the secondary battery 11. The voltage sensor 14 is connected between the wiring connected to the positive and negative terminals of the secondary battery 11. The detected values ​​of the current sensor 13 and the voltage sensor 14 are output to the on-board controller 100. The temperature sensor 15 is a sensor for detecting the temperature of the secondary battery 11. The detected value of this temperature sensor 15 is output to the on-board controller 100. The measured values ​​and detected values ​​from the EIS sensor 12, current sensor 13, voltage sensor 14, and temperature sensor 15 are examples of measurement data.

[0038] Vehicle 1 is equipped with an outside temperature sensor 16, a position sensor 17, and a charging sensor 18. The outside temperature sensor 16 is a sensor that detects the outside temperature of vehicle 1. The position sensor 17 acquires the position of vehicle 1. The position sensor 17 is, for example, a GPS receiver that receives signals from multiple GPS satellites in a GPS system. The charging sensor 18 detects the type of charging at the charging facility that charged the secondary battery 11 of vehicle 1. For example, the charging sensor 18 detects the type of charging at the charging facility by detecting the power output from the charging facility to the secondary battery 11 while the secondary battery 11 is being charged. The charging type identifies whether the charging facility is a normal charging facility, a fast charging facility, or a V2H facility.

[0039] Next, with reference to Figure 4, the processing flow for estimating the internal parameters of the battery model 108 of the secondary battery 11 by the in-vehicle controller 100 will be described. Figure 4 is a flowchart showing the control procedure of the parameter estimation method executed by the controller according to this embodiment. In this embodiment, the in-vehicle controller 100 repeatedly executes the control flow shown in Figure 4 at a constant period.

[0040] In step S1, the in-vehicle controller 100 executes an acquisition process of acquiring measurement data obtained by measuring the secondary battery 11 of the vehicle 1 and usage history data including the usage history of the vehicle 1. In step S2, the in-vehicle controller 100 executes a first state quantity estimation process of estimating the current state quantity of the secondary battery 11 as the first state quantity based on the measurement data and the usage history data. In step S3, the in-vehicle controller 100 executes a second state quantity estimation process of estimating the current state quantity of the secondary battery 11 as the second state quantity using the battery model 108. The in-vehicle controller 100 executes a parameter estimation process of estimating internal parameters by a data assimilation process that assimilates the first state quantity and the second state quantity. In the control flow shown in FIG. 4, after executing the parameter estimation process, the in-vehicle controller 100 may estimate the state quantity of the secondary battery 11 using the battery model 108 updated by the estimated internal parameters and predict the life of the secondary battery 11. The result of the life prediction of the secondary battery 11 is notified to the occupant via the display of the in-vehicle navigation device of the vehicle 1.

[0041] As described above, in the parameter estimation system, parameter estimation device, and parameter estimation method according to the present embodiment, the controller that estimates the internal parameters of the battery model acquires an acquisition process of acquiring measurement data obtained by measuring the vehicle's battery and usage history data including the usage history of the vehicle, a first state quantity estimation process of estimating the current state quantity of the battery as the first state quantity based on the measurement data and the usage history data, a second state quantity estimation process of estimating the current state quantity of the battery as the second state quantity using the battery model, and a parameter estimation process of estimating internal parameters by a data assimilation process that assimilates the first state quantity and the second state quantity. Thereby, the internal parameters of the battery model can be estimated so that the error of the predicted value becomes smaller.

[0042] Further, in the parameter estimation system, parameter estimation device, and parameter estimation method according to the present embodiment, the measurement data includes the internal information of the battery. Thereby, the current state quantity of the battery can be estimated based on the internal information of the battery.

[0043] Also, in the parameter estimation system, parameter estimation device, and parameter estimation method according to the present embodiment, the internal information includes at least the EIS value of the battery. Thereby, the current state quantity of the battery can be estimated based on the EIS value of the battery.

[0044] Also, in the parameter estimation system, parameter estimation device, and parameter estimation method according to the present embodiment, the usage history data includes external information of the vehicle. Thereby, the current state quantity of the battery can be estimated based on the external information of the vehicle in which the battery is mounted.

[0045] Also, in the parameter estimation system, parameter estimation device, and parameter estimation method according to the present embodiment, the external information includes at least any one of at least the outside air temperature of the vehicle, the position of the vehicle, and the charging type at the time of charging the vehicle. Thereby, the current state quantity of the battery can be estimated based on the outside air temperature of the vehicle, the position of the vehicle, and the charging type at the time of charging the vehicle.

[0046] Also, in the parameter estimation system, parameter estimation device, and parameter estimation method according to the present embodiment, in the first state quantity estimation process, the controller inputs input data including measurement data and usage history data to a learned model, and outputs output data including the first state quantity from the learned model to estimate the first state quantity. The learned model is a model learned to output output data including the first state quantity based on the input data including the measurement data and the usage history data. Thereby, the first state quantity can be estimated more accurately using the observation information for each vehicle.

[0047] Furthermore, in the parameter estimation system, parameter estimation device, and parameter estimation method of this embodiment, the controller further executes a model generation process to generate model data including a battery model, the second state variable is estimated using the battery model included in the model data in the second state variable estimation process, the initial value of each internal parameter of the battery model is selected, a parameter ensemble is generated for each internal parameter, with a set of multiple parameter values ​​having variation based on the initial value, multiple different parameter sets are generated based on the parameter ensemble for each internal parameter, and a model ensemble including multiple battery models defined by each of the multiple parameter sets is generated as model data, with each of the multiple parameter sets containing the parameter values ​​of each internal parameter. As a result, a model ensemble to be used in data assimilation processing can be generated considering the variation of each internal parameter based on the selected initial parameter.

[0048] Furthermore, in the parameter estimation system, parameter estimation device, and parameter estimation method of this embodiment, the controller identifies a battery model of the same type as the battery of the vehicle whose parameters are to be estimated from among the battery models of other vehicles, and selects the parameter values ​​of each internal parameter of the identified battery model as the initial values ​​of each internal parameter. This reduces the computational load required for initial parameter estimation by using an existing dataset.

[0049] Furthermore, in the parameter estimation system, parameter estimation device, and parameter estimation method of this embodiment, the controller performs data assimilation processing using Bayesian sequential computation during the parameter estimation process. This enables data assimilation between the predicted values ​​of the battery state variables by the model and the observed values ​​of the current state variables of the battery.

[0050] Furthermore, in the parameter estimation system, parameter estimation device, and parameter estimation method of this embodiment, the controller performs Bayesian sequential calculations using one of the particle filter, Kalman filter, or ensemble Kalman filter. This allows data assimilation to be performed using any of the assimilation methods by applying appropriate filtering to the assimilation method.

[0051] Furthermore, in the parameter estimation system, parameter estimation device, and parameter estimation method of this embodiment, the state quantity is the State of Health (SOH) of the battery. This enables data assimilation using the same physical quantity as the observed SOH, and allows for the estimation of the parameters of the physical model.

[0052] Furthermore, in the parameter estimation system, parameter estimation device, and parameter estimation method of this embodiment, the state quantity is the EIS value or DCR value of the battery, and the battery model includes an equivalent circuit model for calculating the EIS value or DCR value. This makes it possible to perform data assimilation using various battery information.

[0053] Furthermore, in the parameter estimation system, parameter estimation device, and parameter estimation method of this embodiment, the controller stores the estimated internal parameters in a database. This allows the model to more accurately estimate the current state quantities of the battery by referring to the internal parameters stored in the database.

[0054] Furthermore, in the parameter estimation system, parameter estimation device, and parameter estimation method of this embodiment, the controller is provided in the vehicle. This allows the internal parameters of the battery model to be estimated in real time by executing each process on the vehicle side.

[0055] The embodiments of the present invention have been described above using an on-board controller mounted in a vehicle as an example of a controller that performs the parameter estimation method, but the present invention is not limited to this embodiment. For example, the controller may be a server controller located on a server outside the vehicle, separate from the on-board controller. The server is a device that can communicate with the vehicle. In the following description, embodiments in which the controller includes an on-board controller and a server controller, and in which each function is realized by the on-board controller and the server controller, and embodiments in which each function is realized by the server controller will be described. A server equipped with a server controller is an example of the "parameter estimation device" described in the claims.

[0056] <<Second Embodiment>> The parameter estimation system according to the second embodiment of the present invention will be described with reference to Figure 5. Figure 5 is a block diagram of the parameter estimation system according to the second embodiment of the present invention. In the second embodiment, the configurations other than those shown below are the same as in the first embodiment, and in the following description, the same configurations and control processes as in the first embodiment will be omitted from the description, but the description of the first embodiment will be appropriately referenced in the omitted descriptions. The difference between the second embodiment and the first embodiment is that the parameter estimation system 1000 includes a server 20, and the server controller 200 provided in the server 20 performs some of the functions of the in-vehicle controller 100 in the first embodiment. Specifically, the functions of the measurement data acquisition unit 101, the usage history data acquisition unit 102, and the first state quantity estimation unit 103 are executed by the in-vehicle controller 100, and the functions of the parameter storage unit 104, the model data generation unit 105, the second state quantity estimation unit 106, and the internal parameter estimation unit 107 are executed by the server controller 200. The server 20 is a server located outside the vehicle 1. The procedure for the parameter estimation method according to the second embodiment is the same as the procedure shown in the flowchart of Figure 4, and is executed by the in-vehicle controller 100 and the server controller 200. In the following description, the same reference numerals as those shown in Figure 1 indicate the same or similar elements, and therefore detailed explanations are omitted.

[0057] In the second embodiment, the vehicle 1 further includes an on-board communication unit 19. The on-board communication unit 19 transmits and receives data with a server communication unit 21 provided in the server 20 via a network constituting a telecommunications network. The on-board controller 100 includes a measurement data acquisition unit 101, a usage history data acquisition unit 102, and a first state quantity estimation unit 103. The on-board controller 100 estimates a first state quantity using the first state quantity estimation unit 103 and transmits the first state quantity to the server communication unit 21 via the on-board communication unit 19. The server 20 includes a server controller 200 and a server communication unit 21. The server controller 200 includes a processor (computer) having hardware and software, and this computer includes a ROM that stores a control program, a CPU that executes the control program stored in the ROM, and RAM that functions as an accessible storage device. The server controller 200 includes a parameter storage unit 104, a model data generation unit 105, a second state variable estimation unit 106, an internal parameter estimation unit 107, and a battery model 108.

[0058] The server communication unit 21 transmits and receives data with the in-vehicle communication unit 19 via a network constituting a telecommunications line network. The server communication unit 21 receives the first state variable from the in-vehicle communication unit 19 and outputs the first state variable to the internal parameter estimation unit 107. The server controller 200 estimates the internal parameters using the internal parameter estimation unit 107. After estimating the internal parameters, the server controller 200 estimates the second state variable using the battery model 108 after the internal parameter update, with the second state variable estimation unit 106. The server controller 200 may also transmit the second state variable and / or information on the lifespan of the secondary battery 11 estimated based on the second state variable to the in-vehicle communication unit 19 via the server communication unit 21. When the in-vehicle controller 100 receives input from the in-vehicle communication unit 19 regarding the second state variable and / or the lifespan of the secondary battery 11, it notifies the occupants of the vehicle 1 of the second state variable and / or the lifespan of the secondary battery 11.

[0059] <<Third Embodiment>> Next, a parameter estimation system according to the third embodiment of the present invention will be described with reference to Figure 6. Figure 6 is a block diagram of the parameter estimation system according to the third embodiment of the present invention. In the third embodiment, the configurations other than those shown below are the same as those in the first and second embodiments. In the following description, the same configurations and control processes as in the first and second embodiments will not be described, but the descriptions of the first and second embodiments will be appropriately referenced in the omitted descriptions. The difference between the third embodiment and the first embodiment is that the parameter estimation system 1000 includes a server 20, and the server controller 200 provided in the server 20 performs some of the functions of the in-vehicle controller 100 in the first embodiment. Specifically, the functions of the measurement data acquisition unit 101 and the usage history data acquisition unit 102 are performed by the in-vehicle controller 100, and the functions of the first state quantity estimation unit 103, parameter storage unit 104, model data generation unit 105, second state quantity estimation unit 106, and internal parameter estimation unit 107 are performed by the server controller 200. The procedure for the parameter estimation method according to the third embodiment is the same as the procedure shown in the flowchart of Figure 4, and is executed by the in-vehicle controller 100 and the server controller 200. In the following description, the same reference numerals as those shown in Figure 1 indicate the same or similar components, and therefore detailed explanations are omitted.

[0060] In the third embodiment, the vehicle 1 further includes an on-board communication unit 19. The on-board communication unit 19 transmits and receives data with the server communication unit 21 via a network constituting a telecommunications network. The on-board controller 100 also includes a measurement data acquisition unit 101 and a usage history data acquisition unit 102. The on-board controller 100 transmits the measurement data and usage history data to the server communication unit 21 via the on-board communication unit 19. If inference data is generated from the measurement data and usage history data in the on-board controller 100, the on-board controller 100 transmits the inference data to the server communication unit 21.

[0061] Server 20 is a server located outside of vehicle 1. Server 20 comprises a server controller 200 and a server communication unit 21. Server controller 200 comprises a first state quantity estimation unit 103, a parameter storage unit 104, a model data generation unit 105, a second state quantity estimation unit 106, and an internal parameter estimation unit 107. Server communication unit 21 transmits and receives data with in-vehicle communication unit 19 via a network constituting a telecommunications network. Server communication unit 21 receives measurement data and usage history data from in-vehicle communication unit 19 and outputs the measurement data and usage history data to the first state quantity estimation unit 103. Based on the measurement data and usage history data, the first state quantity estimation unit 103 estimates the first state quantity and outputs it to the internal parameter estimation unit 107.

[0062] <<Fourth Embodiment>> Next, the parameter estimation system according to the fourth embodiment of the present invention will be described with reference to Figure 7. Figure 7 is a block diagram of the parameter estimation system according to the fourth embodiment of the present invention. In the fourth embodiment, the configurations other than those shown below are the same as those of the first to third embodiments. In the following description, the same configurations and control processes as those of the first to third embodiments will be omitted, but the descriptions of the first to third embodiments will be appropriately referenced in the omitted descriptions. The difference between the fourth embodiment and the first embodiment is that the parameter estimation system 1000 includes a server 20, and the server controller 200 provided in the server 20 executes the functions of the in-vehicle controller 100 in the first embodiment. Specifically, the functions of the measurement data acquisition unit 101, the usage history data acquisition unit 102, the first state quantity estimation unit 103, the parameter storage unit 104, the model data generation unit 105, the second state quantity estimation unit 106, and the internal parameter estimation unit 107 are executed by the server controller 200. Vehicle 1 is equipped with sensors and an on-board communication unit 19 that transmits detection information from each sensor to the server communication unit 21. The procedure for the parameter estimation method according to the fourth embodiment is the same as the procedure shown in the flowchart of Figure 4 and is executed by the server controller 200. In the following description, the same reference numerals as those shown in Figure 1 indicate the same or similar components, and therefore detailed explanations are omitted.

[0063] In the fourth embodiment, the vehicle 1 is equipped with an on-board communication unit 19. The on-board communication unit 19 transmits and receives data with the server communication unit 21 via a network constituting a telecommunications network. For example, the on-board communication unit 19 transmits detection information acquired from the EIS sensor 12, current sensor 13, voltage sensor 14, temperature sensor 15, ambient temperature sensor 16, position sensor 17, and charge sensor 18 to the server communication unit 21. The server 20 is a server located outside the vehicle 1. The server 20 comprises a server controller 200 and a server communication unit 21. The server controller 200 comprises a measurement data acquisition unit 101, a usage history data acquisition unit 102, a first state quantity estimation unit 103, a parameter storage unit 104, a model data generation unit 105, a second state quantity estimation unit 106, and an internal parameter estimation unit 107. The server communication unit 21 transmits and receives data with the on-board communication unit 19 via a network constituting a telecommunications network. The server communication unit 21 receives each detection information acquired by the vehicle 1 from the in-vehicle communication unit 19 and outputs it to the measurement data acquisition unit 101 and the usage history data acquisition unit 102. The measurement data acquisition unit 101 and the usage history data acquisition unit 102 output the measurement data and the usage history data to the first state quantity estimation unit 103, respectively.

[0064] As described above, the parameter estimation system, parameter estimation device, and parameter estimation method in this embodiment include a server located outside the vehicle and capable of communicating with the vehicle. The controller includes an on-board controller installed in the vehicle and a server controller installed in the server. The server controller executes at least one of the following processes: acquisition processing, first state quantity estimation processing, second state quantity estimation processing, and parameter estimation processing. This allows the computational load of each process performed by the controller to be distributed.

[0065] While embodiments of the present invention have been described above, these embodiments are provided to facilitate understanding of the present invention and are not intended to limit it. Therefore, each element disclosed in the above embodiments is intended to include all design modifications and equivalents that fall within the technical scope of the present invention.

[0066] 1000...Parameter estimation system 1...Vehicle 20...Server 100...In-vehicle controller 200...Server controller 101...Measurement data acquisition unit 102...Usage history data acquisition unit 103...First state variable estimation unit 104...Parameter storage unit 105...Model data generation unit 106...Second state variable estimation unit 107...Internal parameter estimation unit 108...Battery model 11...Secondary battery

Claims

1. A parameter estimation system comprising a controller for estimating the internal parameters of a battery model, wherein the controller performs: an acquisition process for acquiring measurement data of a vehicle's battery and usage history data including the vehicle's usage history; a first state quantity estimation process for estimating the current state quantity of the battery as a first state quantity based on the measurement data and the usage history data; a second state quantity estimation process for estimating the current state quantity of the battery as a second state quantity using the battery model; and a parameter estimation process for estimating the internal parameters by a data assimilation process that assimilates the first state quantity and the second state quantity.

2. A parameter estimation system according to claim 1, wherein the measurement data includes internal information of the battery.

3. A parameter estimation system according to claim 2, wherein the internal information includes at least the EIS value of the battery.

4. A parameter estimation system according to any one of claims 1 to 3, wherein the usage history data includes external information of the vehicle.

5. A parameter estimation system according to claim 4, wherein the external information includes at least one of the following: the ambient temperature of the vehicle, the location of the vehicle, and the type of charge used when the vehicle is being charged.

6. A parameter estimation system according to any one of claims 1 to 5, wherein the controller estimates the first state quantity by inputting input data including the measurement data and the usage history data into a trained model in the first state quantity estimation process and causing the trained model to output output data including the first state quantity, and the trained model is a model that has been trained to output output data including the first state quantity based on input data including the measurement data and the usage history data.

7. A parameter estimation system according to any one of claims 1 to 6, wherein the controller further performs a model generation process to generate model data including the battery model, the second state quantity estimation process estimates the second state quantity using the battery model included in the model data, the model generation process selects an initial value for each internal parameter of the battery model, generates a parameter ensemble for each internal parameter, with respect to the initial value, forming a set of multiple parameter values ​​with variation, generates a plurality of different parameter sets based on the parameter ensemble for each internal parameter, generates a plurality of different parameter sets as the model data, and each of the plurality of parameter sets includes the parameter values ​​of each internal parameter.

8. A parameter estimation system according to claim 7, wherein the controller identifies a battery model of the same type as the battery of the vehicle to be parameter estimated from among the battery models of other vehicles, and selects the parameter values ​​of each internal parameter of the identified battery model as the initial values ​​of each internal parameter.

9. A parameter estimation system according to any one of claims 1 to 8, wherein the controller performs the data assimilation process using Bayesian sequential computation in the parameter estimation process.

10. A parameter estimation system according to claim 9, wherein the controller performs the Bayesian successive calculation using one of a particle filter, a Kalman filter, and an ensemble Kalman filter.

11. A parameter estimation system according to any one of claims 1 to 10, wherein the state quantity is the State of Health (SOH) of the battery.

12. A parameter estimation system according to any one of claims 1 to 10, wherein the state quantity is the EIS value or DCR value of the battery, and the battery model comprises an equivalent circuit model for calculating the EIS value or the DCR value.

13. A parameter estimation system according to any one of claims 1 to 12, wherein the controller stores the estimated internal parameters in a database.

14. A parameter estimation system according to any one of claims 1 to 13, wherein the controller is a parameter estimation system provided in the vehicle.

15. A parameter estimation system according to any one of claims 1 to 13, comprising a server located outside the vehicle and capable of communicating with the vehicle, wherein the controller includes an on-board controller provided in the vehicle and a server controller provided in the server, and the server controller performs at least one of the following processes: the acquisition process, the first state quantity estimation process, the second state quantity estimation process, and the parameter estimation process.

16. A parameter estimation device comprising a controller for estimating the internal parameters of a battery model, wherein the controller performs: an acquisition process for acquiring measurement data of a vehicle's battery and usage history data including the vehicle's usage history; a first state quantity estimation process for estimating the current state quantity of the battery as a first state quantity based on the measurement data and the usage history data; a second state quantity estimation process for estimating the current state quantity of the battery as a second state quantity using the battery model; and a parameter estimation process for estimating the internal parameters by a data assimilation process that assimilates the first state quantity and the second state quantity.

17. A parameter estimation method performed by a controller for estimating the internal parameters of a battery model, wherein the controller performs: an acquisition process for acquiring measurement data of a vehicle's battery and usage history data including the vehicle's usage history; a first state quantity estimation process for estimating the current state quantity of the battery as a first state quantity based on the measurement data and the usage history data; a second state quantity estimation process for estimating the current state quantity of the battery as a second state quantity using the battery model; and a parameter estimation process for estimating the internal parameters by a data assimilation process that assimilates the first state quantity and the second state quantity.

Images