Method for determining the state of charge of a battery used in an electric vehicle

Abstract

A method for determining the state of charge (SOC) of a battery. The method may employ a transformation technique that linearizes a nonlinear model by converging to a near-measured value, thereby accurately estimating the state of charge without affecting computation time or system load. The transformation technique employs an adaptive unscented Kalman filter (UKF).

Description

Background Art

[0001] The present disclosure relates to a method for determining the state of charge of a battery.

[0002] Batteries serve as an energy source for many electrical systems, particularly in hybrid electric vehicles (HEVs) and electric vehicles (EVs). In these vehicles, the battery interacts with other components through a battery management system (BMS) to provide power to the vehicle and meet the vehicle's energy needs while maintaining the safety of the electrical systems. The battery is typically a high-voltage (HV) battery. Plug-in hybrid electric vehicles (PHEVs, HEVs) and pure electric vehicles rely on batteries as auxiliary and primary energy sources, respectively, to propel the vehicle. Therefore, the available energy from the battery must be tracked to prevent the battery from being overcharged or under-discharged. To ensure safety and maximize the use of the available energy from the battery, such vehicles employ a BMS that controls the battery's functions and determines its performance.

[0003] The reliability of these electrical systems is highly dependent on the health and safety of the batteries, and therefore highly dependent on the ability of the BMS to provide operating data that allows peak performance to be achieved without compromising the health and safety of the batteries. Without fast and accurate battery models for use by the BMS, controlling and monitoring the batteries installed in an HEV or EV would be more challenging. Models are used to estimate battery metrics, including state of charge (SOC), state of health (SOH), state of energy (SOE), and state of power (SOP). In addition, these models are used to help the BMS implement battery control, real-time observation, parameter estimation, and battery optimization functions.

[0004] The state of charge (SOC) measures the remaining available capacity of a battery in real time, as it is charged by an external charging or regeneration mechanism and discharged during driving. The SOC determines the battery's ability to deliver the required energy, and accurate SOC estimation prevents battery degradation due to overcharging and discharging. Typically, the SOC is the ratio of the current battery capacity to the maximum battery capacity.

[0005] It is crucial for the BMS to accurately determine the battery state of charge in real time. By doing so, the BMS can ensure maximum performance output from the battery while minimizing the effects of shortening battery life.

[0006] The BMS should be able to estimate the battery's state of charge under any given condition. If the SOC determined by the BMS is inaccurate, for example, if the SOC is underestimated, this can lead to overcharging of the battery. Overcharging the battery can potentially lead to dangerous events, such as thermal runaway. On the other hand, if the SOC is overestimated, the battery's overall capacity will be limited, thereby impacting performance and allowing less energy to be drawn from the battery.

[0007] A typical method for determining SOC is based on the integral of the current over a period of time, multiplied by the inverse of the battery's remaining capacity. Other methods for determining SOC using estimation techniques, such as Kalman filters and extended Kalman filters (EKF), are limited due to inaccuracies caused by the nonlinear behavior of the estimated battery (KF) and / or due to the linearization method used (EKF). Furthermore, advanced methods such as learning algorithms (e.g., machine learning, neural networks) have high computational complexity that cannot be applied to real-time SOC estimation.

[0008] It would be desirable to find a method for determining SOC that includes the use of nonlinear data modeling and provides updates in real time. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] The features, objects, and advantages of the present disclosure will become more apparent from the detailed description set forth below when considered in conjunction with the accompanying drawings, in which like reference numerals designate correspondingly throughout, and in which:

[0010] Figure 1 An electric vehicle having an exemplary powertrain including a battery.

[0011] Figure 2 is a simplified circuit diagram of an exemplary embodiment of a two-branch RC model equivalent circuit for a battery.

[0012] Figure 3 is a block diagram of a battery management system including a controller and a plurality of modules for executing a method for estimating a state of charge of a battery.

[0013] Figure 4 is Figure 3 A flowchart of the method performed by the controller and modules. DETAILED DESCRIPTION

[0014] As used herein, the term "a" shall mean one or more than one. The term "plurality" shall mean two or more than two. The term "another" is defined as a second or more. The terms "include" and / or "have" are open ended (e.g., comprising). As used herein, the term "or" shall be interpreted as inclusive or to mean any one or any combination. Thus, "A, B, or C" means "any of the following: A; B; C; A and B; A and C; B and C; A, B, and C." Exceptions to this definition occur only when a combination of elements, functions, steps, or acts are inherently mutually exclusive in some way.

[0015] As used herein, vector notation is defined as follows:

[0016]

[0017]

[0018]

[0019] Reference throughout this document to "one embodiment," "some embodiments," "an embodiment," or similar terms indicates that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. Thus, the appearances of these phrases in various places throughout this specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments without limitation.

[0020] As required, detailed embodiments of the present invention are disclosed herein; however, it should be understood that the disclosed embodiments are merely exemplary of the present invention, which may be embodied in various and alternative forms. The drawings are not necessarily drawn to scale; certain features may be exaggerated or minimized to show details of particular components. Therefore, the specific structural and functional details disclosed herein should not be interpreted as limiting, but merely as a representative basis for teaching one skilled in the art to variously employ the present invention.

[0021] Figure 1 An exemplary electric vehicle 1 having a powertrain 100 is shown. The exemplary vehicle includes an internal combustion engine 10, a generator 11, and one or more electric motors for rotating the wheels of the vehicle. The internal combustion engine 10 drives the generator 11 to generate electric power for a battery 12 and an electric motor 14 / 14'. A generator inverter 16 for the generator 11 may also be provided. A gearbox 13 is provided to provide the gear ratio required by the vehicle. Power to the electric motor is communicated via an inverter 15 / 15', which converts the provided DC power to the AC power required by the electric motor 14 / 14'. The inverter 15 / 15' may include multiple phases corresponding to each phase of the electric motor 14 / 14'. The systems and methods described below may be used with devices such as Figure 1 The battery 12 shown in FIG.

[0022] In one embodiment, the estimation method for determining the SOC uses a second-order RC equivalent circuit model based on the battery (e.g., Figure 2 The battery parameters are calculated and the voltage is estimated based on the circuit model. Figure 2 is a circuit diagram depicting an exemplary embodiment of the RC model 200 used in the SOC estimation method 300. In one embodiment, all RC parameters (including but not limited to R0, C1, R1, C2, R2, and open circuit voltage (OCV)) can be generated by fitting the parameters to data obtained from tests performed on the battery cell 215. In one embodiment, R0 is determined as the state of charge (SOC) of the battery cell 215, the temperature T of the battery cell 215, and the voltage of the battery cell 215.cell and the state of health (SOH) of the battery cell 215. In one embodiment, OCV is determined as the state of charge (SOC) of the battery cell 215, the temperature T cell and the state of health (SOH) of the battery cell 200. In one embodiment, C1, R1, C2, and R2 are determined as the state of charge (SOC) of the battery cell 215, the equivalent circuit current I, the temperature T of the battery cell 215, and the state of health (SOH) of the battery cell 200. cell and the state of health (SOH) of the battery cell 200.

[0023] Figure 3 is a block diagram of multiple modules that can be used in a BMS 300 to perform a method for determining or estimating the SOC of a vehicle battery. The system includes a controller that includes an SOC estimation module 330. The system also includes lookup tables for various battery parameters, which can be created and stored in a lookup table module 310. The system can also include a module 320 for estimating or determining the remaining capacity of the battery based on the battery's SOH. Furthermore, the system includes a module 340 for performing noise covariance updates for the SOC estimation. The term covariance refers to the joint variability of two or more random variables.

[0024] The above modules can be integrated into a single controller or microprocessor. Alternatively, one or more modules can be individually packaged in one or more microprocessors or controllers.

[0025] The system is configured to probabilistically determine or estimate the state of charge (SOC) based on an innovation sequence given by the voltage estimation error (the difference between the observed value of the variable and the best prediction of that value based on prior information), and to correct the SOC calculated by the model equation. The method for determining the SOC involves two "update" steps of estimating the battery voltage and determining the SOC. The first update step is a measurement update, in which the model predicts the state of charge and battery voltage based on the difference between the estimated value and the true value. The second update step is a time update of the prediction based on the innovation error calculated from the voltage estimation error to correct the predicted state of charge.

[0026] The method for determining SOC can employ a transformation technique that linearizes the nonlinear model used in the method by converging to a close measurement value, thereby accurately estimating the state of charge without affecting calculation time and load on the system. The transformation technique employs an adaptive unscented Kalman filter (UKF) based on a technique known as unscented transformation to linearize the nonlinear model equations, where the initial point determines the probability of convergence of the estimate.

[0027] Additionally, the method for determining SOC can employ a method called Normalized Innovation Sequence (NIS) to update the noise covariance matrix, which measures the estimation bias due to additive noise that may not be Gaussian in nature (whereas the UKF assumes that the noise present in the system is Gaussian). The use of NIS enables the model of battery capacity to adapt to the changing noise covariance and update the state vector.

[0028] The estimation method to determine SOC additionally updates the noise covariance of the estimated state by adapting the estimation error based on a moving average error method (called an "innovation sequence"). This update ensures that the method is robust and accurately estimates the state (ie, SOC).

[0029] Figure 4 Shown in simplified form Figure 3 A flow chart of a method of estimating SOC performed by the BMS 300 system is shown in FIG. As a first step of the method, the system establishes initial values ​​of SOC and covariance error Q in step 410 .

[0030] In step 420, the method generates updated σ points for SOC(x0) and covariance (P0). The updates are generated using the following equations.

[0031]

[0032]

[0033] In step 430, the system performs a time update of the σ point using the following equation.

[0034]

[0035] Where E[x0] is the expected value of x0; λ=α 2 (L+k)-L; α, k are scaling parameters; L=2n+1σ points; and n is the number of state variables (in the main embodiment, n=3).

[0036]

[0037] y k =V ocv,k (SOC K ,T k )+V rc1,k +V rc2,k +I k R 0,k (2)

[0038] where Equation (1) is the state space equation, which consists of equations for each state variable and uses χ k-1As the σ point of the previous time step, y k is the predicted voltage, u k is the control vector representing the data measurement made at time k, V ocv,k Determined by the lookup table, R 0,k is a function of SOC, temperature and current, where f determines The updated value is as follows:

[0039]

[0040]

[0041]

[0042]

[0043]

[0044] Where time constant τ1 = R1*C1, time constant τ2 = R2*C2, and all RC parameters are input from the RC model lookup table, and Ah is the battery cell capacity. Method (1) predicts each σ point χ k The value at the next time step The change of , as a function of the σ points calculated on the state at the current time step, acts on the input at the current time step. Equation (2) is the output equation, which is the equation for the estimated voltage, which is then used to update the estimated state variables.

[0045] It should be noted that, as shown in block 415 , the system is configured to generate σ points based on UKF parameters (eg, α, β, γ, weights (W_c, W_m)).

[0046] In step 432, the system is configured to perform a measurement update and update the state based on the measurement error and the Kalman gain using the following equation:

[0047]

[0048] in is the state of the correction, K is the Kalman gain (a weight function ranging between 0 and 1 that indicates how much weight to assign to the predicted value relative to the measured value), and V measured is the battery voltage measured by the sensor.

[0049] In step 440, the covariance matrix is ​​updated based on the system's innovation sequence, which is also called the error e in the voltage prediction. Figure 3 As shown, module 340 performs a consistency check of the NIS sequence as described below. The consistency check uses the following equation:

[0050] e=V measured -y k

[0051] Q process =KE[ee T ]K T

[0052] where Q process is the covariance of the process noise.

[0053] The measurement errors are squared and normalized to give a sequence that is used to check the consistency of the SOC determination made by the controller. In step 440, this sequence is called the Normalized Innovation Square (NIS) and is given by,

[0054]

[0055] Where k is the time step, e is the error, and S is the measurement error variance. An exemplary time step is 1 second. The NIS plot follows a chi-square distribution with mean zero and variance α. In one embodiment, a consistency check is performed by checking whether the NIS sequence lies within a confidence interval, which is calculated as:

[0056]

[0057] where σ 2 is the standard deviation of the NIS sequence, n-1 = dof (degrees of freedom) of the chi-square distribution, and is the chi-square value chosen for the %α confidence level (independent of the χσ point).

[0058] In step 450, the system is configured to check whether NIS is within an acceptable confidence interval (step 450). If NIS is not within the confidence interval, module 340 is configured to update Q for the next time step, such as Figure 4 As shown in step 455.

[0059] If the cycle is complete (see step 460), the system is configured to end the process.

[0060] While the present disclosure has been particularly shown and described with reference to exemplary embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the scope of the claimed embodiments.

[0061] The following list is a non-limiting summary of the above variables, where the status is referred to as SOC:

[0062] x0: initial state

[0063] Initial state (best estimate)

[0064] P0: initial covariance matrix

[0065] k-1: current step size

[0066] k: next time step

[0067] Current status (best estimate)

[0068] P k-1 : current covariance matrix

[0069] χ k-1 : σ point (current step size)

[0070] λ: σ point scaling parameter

[0071] α: σ point scaling parameter, σ point at Distribution around

[0072] k: σ point scaling parameter

[0073] L: the number of σ points

[0074] n: number of states

[0075] Next state (best estimate)

[0076] Average value of status

[0077] χ k : σ point

[0078] y k : Predicted voltage

[0079] V measured : The battery voltage measured by the sensor

[0080] K: Kalman gain

[0081] e: Error in voltage prediction

[0082] Q process : process noise covariance matrix

[0083] S k : Measurement error variance

[0084] NIS k : Normalized Innovation Square

[0085] V ocv,k : Open circuit voltage of RC model

[0086] I k: RC model current

[0087] R 0,k : Resistance of the RC model (components outside the RC parallel circuit)

[0088] R 1,k : Resistance of the RC model (part of the first RC parallel circuit)

[0089] R 2,k : Resistance of the RC model (part of the second RC parallel circuit)

[0090] C 1,k : Capacitor of the RC model (part of the first RC parallel circuit)

[0091] C 2,k : Capacitor of the RC model (part of the second RC parallel circuit)

[0092] V rc1,k : Voltage of the RC model (first RC parallel circuit part)

[0093] V rc2,k : Voltage of the RC model (second RC parallel circuit part)

[0094] τ1: time constant (part of the first RC parallel circuit, τ1 = R1 * C1)

[0095] σ2: variance of the series

[0096] Chi-square value of %α confidence level

Claims

1. A method for determining the state of charge (SOC) of a battery for providing power for propelling an electric vehicle, the method comprising the steps of: establishing an initial value for the SOC and a noise covariance error; generating updated σ points for a best estimate of the SOC and a noise covariance error matrix; determining the SOC based on a nonlinear battery model updated at predetermined time intervals, wherein the method includes linearizing the nonlinear battery model into a linear model using an unscented Kalman filter, and wherein the determination of the SOC is based on a measurement error of a battery voltage and a gain of the unscented Kalman filter; calculating an update to the covariance error matrix based on an innovation sequence related to an error associated with the predicted battery voltage, wherein the innovation sequence is a sequence known as a Normalized Innovation Square (NIS) sequence; performing a consistency check of the NIS sequence by checking whether the NIS sequence is within a predetermined confidence interval; updating the covariance error if the consistency check of the NIS sequence determines that the NIS sequence is outside the confidence interval; and The above steps are repeated after the predetermined time interval has elapsed.

2. The method according to claim 1, wherein The method uses battery parameters based on a two-branch RC model for the performance of the battery.

Images