How to estimate the state of charge of a battery

Abstract

The method for estimating the state of charge of a battery includes the steps of setting an initial state of charge value and an initial Kalman error covariance value, receiving G parameter estimates, a present current value, and a present voltage value of the battery, inputting the G parameter estimates, the present current value, and the present voltage value to an extended Kalman filter, updating the present state of charge value and the present Kalman error covariance value of the battery, and outputting the present state of charge value.

Description

[Technical Field]

[0001] The present invention relates to a method for estimating the charge state of a battery in use. [Background technology]

[0002] Compared to other energy storage devices, batteries are easily applicable to electrical devices and, due to their relatively high energy and power density characteristics, are widely used not only in portable electronic devices but also in electric vehicles (EVs) and hybrid electric vehicles (HEVs) powered by electrical sources. In particular, when high output is required, battery packs consisting of multiple battery cells connected in series and parallel can be used.

[0003] Battery management is crucial for the energy-efficient and safe use of electrical devices powered by batteries or battery packs, and accurate estimation and diagnosis of the battery state are essential. Commonly used estimation values ​​include state of charge (SOC), state of health (SOH), and power limit estimation (PLE).

[0004] Conventional states of charge (SOC) are expressed by factors such as battery charge amount and remaining battery capacity, and are defined as the percentage of the current capacity relative to the battery's full charge capacity. Methods for estimating the SOC include integrating the amount of charge released or added after measuring it with a current sensor, utilizing the relationship between open-circuit voltage (OCV) and the SOC, and estimating the SOC using a battery model.

[0005] A method that uses a current sensor to measure the amount of charge released or added and estimate the state of charge (SOC) estimates the current SOC by adding the value obtained by dividing the integrated current by the full charge capacity to the initial SOC. This method, also known as Ah-counting or coulomb counting, is widely used because of its simplicity, but it is affected by the accuracy of the current sensor.

[0006] The method of measuring the open-circuit voltage (OCV) and estimating the state of charge (SOC) utilizes the OCV-SOC relationship unique to each battery. This method is highly reliable because this relationship is known to remain largely unchanged even as the battery deteriorates. However, measuring the OCV requires leaving the battery in a state of zero current for an extended period. Therefore, it is not possible to measure the OCV while the battery is in use, making it difficult to accurately predict the state of charge (SOC).

[0007] When using a battery model, the impact of errors (noise) in current sensors can be minimized, and the state of charge (SOC) can be estimated in real time, even if the battery does not have long periods of rest. Examples of battery models include equivalent circuit models (ECMs) and physics-based models. Equivalent circuit models can provide insights into what happens inside battery cells, and the parameters used in these models do not actually have physical meaning. Physics-based models are more accurate than equivalent circuit models, but they have problems with complexity and convergence.

[0008] Thus, conventional State of Charge (SOC) estimation methods have problems such as requiring complex battery models or necessitating the discontinuation of battery use. Accurate SOC estimation is essential to prevent battery overcharging and over-discharging and to perform cell balancing. However, with the current integration method, accuracy cannot be guaranteed due to current measurement errors. For example, even assuming that the current sensor has an error of only 0.1A, using an electric vehicle for 8 hours results in a SOC estimation error of 0.8Ah, and an SOC estimation error of more than 5Ah per week. If the battery capacity is 100Ah, the error can reach as high as 5%.

[0009] In State of Charge (SOC) estimation, not only accuracy but also low computational burden and fast calculation speed are important. When the internal state of a battery can be accurately estimated and controlled, the safety and performance of the battery pack improve relative to its price and weight, making it applicable not only to automobiles but also to transportation methods such as aviation and various other fields. [Overview of the Initiative] [Problems that the invention aims to solve]

[0010] The problem that this invention aims to solve is to provide a method for accurately estimating the state of charge (SOC) of a battery in real time using the battery's voltage and current values. This invention provides a method for accurately estimating the state of charge (SOC) of a battery in real time using G-parameters that indicate the internal state of the battery. The state of charge (SOC) estimation method of this invention is incorporated into a battery management system (BMS) and executed by the battery management system, enabling the estimation of the actual state of charge (SOC) of a battery in use. [Means for solving the problem]

[0011] A method for estimating the charge state of a battery according to one aspect of the present invention includes the steps of setting an initial charge state value and an initial Kalman error covariance value, receiving an estimated G-parameter value, current value, and current voltage value of the battery, inputting the estimated G-parameter value, current value, and current voltage value into an extended Kalman filter to update the current charge state value and the current Kalman error covariance value of the battery, and outputting the current charge state value. The method for estimating the charge state of a battery according to one aspect can also be performed by a computer device. [Effects of the Invention]

[0012] The battery charge state estimation method according to various embodiments of the present invention offers significant improvements over conventional methods in terms of cost, scalability, and adaptability. While existing battery model-based charge state estimation methods have been difficult to apply to battery management systems (BMS) due to their complexity, the charge state estimation method of the present invention can be implemented in actual battery management systems (BMS).

[0013] Furthermore, while conventional methods require stopping battery use to improve the accuracy of charge state estimation, the charge state estimation method of the present invention can estimate the charge state of a battery in real time while it is actually in use. Moreover, while the accuracy of charge state estimation in conventional methods is limited by current sensor errors, the charge state estimation method of the present invention can minimize the influence of current sensor errors by using a filter. In addition, the charge state estimation method of the present invention can be used universally not only for battery cells and battery packs but also for battery systems. [Brief explanation of the drawing]

[0014]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

[0015] The advantages, features, and methods for achieving them of the present invention will become clear with reference to the embodiments described in detail, along with the accompanying drawings. However, the present invention should be understood not to be limited to the embodiments presented below, but to be embodied in a variety of different forms, including all transformations, equivalents, or substitutions that fall within the spirit and technical scope of the present invention. The embodiments presented below are provided to fully illustrate the disclosure of the present invention and to fully inform those skilled in the art of the invention of its scope. Where a specific description of the present invention relates to the relevant prior art and would obscure the gist of the invention, such detailed description is omitted.

[0016] The terms used in this application are used solely to describe specific embodiments and are not intended to limit the invention. A singular expression includes plural expressions unless the context clearly indicates otherwise. In this application, terms such as “includes” or “having” should be understood to indicate the existence of features, numbers, stages, operations, components, parts, or combinations thereof described in the specification, and not to preemptively exclude the possibility of the existence or addition of one or more other features, numbers, stages, operations, components, parts, or combinations thereof. Terms such as the first and second are also used to describe various components, but these components are not limited by the aforementioned terms. The aforementioned terms are used solely for the purpose of distinguishing one component from others.

[0017] Embodiments of the present invention will be described in detail below with reference to the attached drawings. In the description with reference to the attached drawings, identical or corresponding components will be given the same drawing number, and redundant descriptions relating to them will be omitted.

[0018] Figure 1 illustrates a schematic configuration diagram of a battery system according to one embodiment.

[0019] Referring to Figure 1, the battery system 100 also includes a battery 110, a voltage measuring unit 120, a current measuring unit 130, a microprocessor 140, and a memory 150.

[0020] The battery 110 is the part that stores power and includes a plurality of battery cells 111 that are electrically connected to each other between the first terminal 101 and the second terminal 102. The battery cells 111 can be connected in series, in parallel, or in a combination of series and parallel. Each battery cell 111 has the same capacity and can discharge and charge the same amount of current. However, in practice, the internal states of each battery cell 111 can be different. For example, each battery cell 111 may have different internal resistances and electromotive forces. Each battery cell 111 may have different G-parameter values ​​and H-parameter values. The battery 110, which includes the battery cells 111, may also have its own G-parameter values ​​and H-parameter values.

[0021] In this invention, the method for estimating the charge state of the battery 110 will be described as a basis, but the charge state estimation method of this invention can also be applied when estimating the charge state of each battery cell 111.

[0022] The battery cell 111 may also include rechargeable secondary batteries. For example, the battery cell 111 may include nickel-cadmium batteries, lead-acid batteries, nickel-metal hydride (NiMH) batteries, lithium-ion batteries, lithium polymer batteries, etc. The number of battery cells 111 constituting the battery 110 is also determined by the required capacity, output voltage, and output current of the battery 110.

[0023] Figure 1 shows a single battery 110, but multiple batteries 110 can be connected in parallel and / or series, and are also connected to a load and / or charging device via the first terminal 101 and the second terminal 102. Although not shown in Figure 1, the battery 110 is also in use when connected to a load and / or charging device. Furthermore, the charge state estimation method of the present invention can also be applied when estimating the charge state of the entire group of multiple batteries 110 connected in parallel and / or series.

[0024] The battery system 100 is also a battery pack or battery module containing at least one battery cell 111. The battery system 100 is also a system containing at least one battery pack or battery module.

[0025] The voltage measurement unit 120 can periodically generate the voltage value of the battery 110 by measuring the voltage of both electrodes at each preset sampling period Ts. In another example, the voltage measurement unit 120 can periodically generate the voltage value of each battery cell 111 by measuring the voltage of each battery cell 111 at each preset sampling period Ts. If the battery cells 111 are connected in parallel, the voltage measurement unit 120 can measure the voltage of only one battery cell 111 and determine that all battery cells 111 connected in parallel have the same voltage value.

[0026] The sampling period Ts is simply an example of 1 second. However, the sampling period Ts can also be set to other times, such as 0.1 seconds, 0.5 seconds, 2 seconds, 5 seconds, or 10 seconds. The sampling period Ts can be appropriately set by the electrical system connected to the battery system 100. The voltage value currently measured is referred to as the current voltage and is denoted as V(t). The voltage value measured before the sampling period Ts is referred to as the voltage immediately preceding and is denoted as V(t-1).

[0027] The current measuring unit 130 can measure the current flowing through the battery 110 at each sampling period Ts and periodically generate the current value of the battery 110. In another example, the current measuring unit 130 can measure the current of each battery cell 111 at each sampling period Ts and periodically generate the current value of each battery cell 111. When the battery cells 111 are connected in series, the current measuring unit 130 can measure the current of only one battery cell 111 and determine that all battery cells 111 connected in series have the same value.

[0028] The current value measured by the current measuring unit 130 is displayed as positive (+) when it is the charging current and negative (-) when it is the discharge current. The currently measured current value is called the current value and is displayed as I(t), and the current value measured before the sampling period Ts is called the current value immediately before and is displayed as I(t-1). The voltage measuring unit 120 and the current measuring unit 130 are synchronized with each other and can measure the voltage and current of the battery 110, respectively, at the same timing.

[0029] The microprocessor 140 can estimate the charge state of the battery 110. The microprocessor 140 estimates the initial charge state (SOC). est (0) and the initial value of the Kalman error covariance P k Set (0) and estimate the G parameter of battery 110 G est (t), the current value I(t) and the current value V(t) are received, and the G parameter estimate G is applied to the extended Kalman filter. est Input (t), current value I(t) and current value V(t), and current charge state (SOC) of battery 110. est (t) and the current value of the Kalman error covariance P k (t) and update the current charge status SOC est It can also be configured to output (t).

[0030] The microprocessor 140 numerically calculates the estimated value G est (t) of the G parameter and the estimated value H est (t) of the H parameter that indicate the current state of the battery 110 in real time from the voltage value of the battery 110 provided by the voltage measurement unit 120 and the current value of the battery 110 provided by the current measurement unit 130. The G parameter is a parameter indicating the sensitivity of the voltage with respect to the current change of the battery 110, and the H parameter is a parameter indicating the effective potential determined by the local equilibrium potential dispersion and the resistance distribution in the battery 110.

[0031] The microprocessor 140 uses an adaptive filter to generate the estimated value G est (t) of the G parameter and the estimated value H est (t) of the H parameter of the battery 110 in real time from the voltage value and the current value of the battery 110. The adaptive filter is also a filter using the recursive least squares method (RLS) or a filter using the weighted least squares method (WLS). In this specification, an embodiment in which the microprocessor 140 uses a recursive least squares (RLS) filter to generate the estimated value G est (t) of the G parameter and the estimated value H est (t) of the H parameter of the battery 110 in real time will be described in detail, but the present invention is not limited thereto.

[0032] The microprocessor 140 inputs the estimated value G est (t) of the G parameter, the current current value I(t), and the current voltage value V(t) of the battery 110 generated using an adaptive filter into an extended Kalman filter, and the current state of charge SOC est(t) can be generated in real time. The microprocessor 140 can utilize coefficient data stored in memory 150. This coefficient data is also generated from the open circuit voltage (OCV)-state of charge (SOC) relationship predetermined for the battery 110.

[0033] Microprocessor 140, Battery 110, Current Charge State SOC est In generating (t) in real time, only simple operations such as arithmetic are used, and this is also included in the battery system 100 or a battery management system (BMS) installed in the battery pack. In another example, the charge state estimation method according to this embodiment is also performed by a microcontroller or ECU (electronic control unit) in the battery management system (BMS) of an electric vehicle. In yet another example, the charge state estimation method according to this embodiment is also performed by an integrated controller of an energy storage system. In yet another example, the charge state estimation method according to this embodiment is also performed by a processor of a server connected by communication to the battery system or energy storage system.

[0034] The memory 150 can store the instruction words and data necessary for the microprocessor 140 to perform the charge state estimation method according to this embodiment. According to this embodiment, based on the voltage value and current value of the battery 110 generated at each sampling period Ts, the estimated G parameter value of the battery 110 G est (t) is generated, and the G parameter estimate G est Based on (t), the current value I(t) and the current value V(t), the current charge state (SOC) is calculated. estSince (t) is generated, the memory 150 stores the current voltage, current, and current of the battery 110, but no other voltage data or current data is stored in the memory 150. Because the memory 150 does not need to store a large amount of instruction words and data, it can also be implemented by a small-sized memory. For example, the memory 150 can also be implemented by the memory within the microprocessor 140.

[0035] This invention presents a method for estimating the charge state of a battery using G-parameters and H-parameters, which are parameters indicating the current state of the battery. The charge state estimation method according to this invention is relatively easy to implement, even to the extent that it can be performed in a battery management system (BMS), and can have high accuracy without additional operating conditions.

[0036] The G-parameter is a state variable that indicates the sensitivity of a battery cell's terminal voltage to changes in current applied to the battery cell during use, and it has units of resistance. The H-parameter is the effective potential determined by the local equilibrium potential distribution and resistance distribution within the battery cell during use. The G-parameter and H-parameter of a battery cell can be quantified using theoretical models and explicit correlation equations between battery material properties and design variables. The G-parameter and H-parameter of a battery cell will be explained below.

[0037] In a battery cell, we can assume that the voltage V and current I have a relationship such that V = f(I; x, p). Here, x is a physical quantity that describes the internal state of the battery cell, and p is a parameter.

[0038] The function f is a nonlinear implicit function, and if the function f can be separated into a rapidly changing quantity g and a gradually changing quantity h, then the voltage V and current I can be expressed as V = g(I;x,p) + h(I;x,p).

[0039] If we assume that there exists a function G(I;x,p) = ∂g / ∂I that changes gradually with respect to the current I, then the voltage V and current I can also be expressed as V = G(I;x,p) × I + H(I;x,p). Here, ∂G / ∂I and ∂H / ∂I have very small values. In other words, if the above assumption is satisfied, then since G and H are functions that change slowly with respect to the current I, the function f that shows the nonlinear relationship between voltage V and current I can also be expressed as a quasi-linear relationship, as described above.

[0040] Here, G is called the G parameter, and H is called the H parameter. Current I is the charge / discharge current, and U eq If this is the equilibrium potential of the battery cell, then the discharge overvoltage can be calculated using the G parameter G and the H parameter H, and U eq -V = -G × I + (U eq It can also be expressed as -H).

[0041] Here, -G × I is the overvoltage generated when the battery conducts current through its terminals, and includes the reaction kinetic polarization and the resistive polarization of electrons and ions. (U eq -H) is an overvoltage that arises when the local thermodynamic equilibrium state within the battery deviates from the equilibrium state of the overall system. That is, (U eq -H) indicates inefficiencies caused by thermodynamic non-uniformity within the battery, and if the internal system of the battery reaches a thermodynamic equilibrium state, the H parameter H is equal to the equilibrium potential U eq It will be the same as this.

[0042] The charging state estimation method according to an embodiment of the present invention directly extracts the battery's G parameter G from the battery's voltage and current values, and uses the G parameter G to estimate the battery's charging state.

[0043] Figure 2 illustrates an internal configuration diagram for performing a charging state estimation method according to one embodiment.

[0044] Referring to Figure 2 in conjunction with Figure 1, the microprocessor 140 also includes a GH estimator 142 and an extended Kalman filter 144. The memory 150 may store a lookup table 152 generated from a predetermined open-circuit voltage (OCV)-state of charge (SOC) relationship for the battery 110. The lookup table 152 can store coefficient data corresponding to the state of charge value. Since the open-circuit voltage (OCV)-state of charge (SOC) relationship is nonlinear, coefficient data corresponding to the open-circuit voltage (OCV)-state of charge (SOC) relationship is stored in the lookup table 152, and the extended Kalman filter 144 can use the coefficient data stored in the lookup table 152 to estimate the state of charge of the battery 110. The coefficient data may include first coefficient data, or it may include second and third coefficient data.

[0045] The voltage measurement unit 120 in Figure 1 generates a voltage value V of the battery 110 and provides it to the GH estimator 142 and the extended Kalman filter 144. The voltage value V includes the current voltage value V(t) and the voltage immediately preceding V(t-1). As the sampling period Ts elapses, the current voltage value V(t) becomes the voltage immediately preceding V(t-1), and the new voltage value becomes the current voltage value V(t).

[0046] The current measurement unit 130 in Figure 1 generates a current value I of the battery 110 and provides it to the GH estimator 142 and the extended Kalman filter 144. The current value I includes the current value I(t) and the current value immediately preceding I(t-1). As the sampling period Ts elapses, the current value I(t) becomes the current value immediately preceding I(t-1), and the new current value becomes the current value I(t).

[0047] The GH estimator 142 uses an adaptive filter to estimate the G parameter of the battery 110 based on the voltage value V and the current value I. est This generates the adaptive filter, for example, a filter that utilizes recursive least squares (RLS). The estimated G-parameter of battery 110 is G. estThis is a numerical value representing the G parameter, which indicates the voltage sensitivity of the battery 110 to current changes. For example, the GH estimator 142 uses the adaptive filter to estimate the H parameter of the battery 110 based on the voltage value V and the current value I. est Generates the estimated H parameter of battery 110. est This is a numerical value representing the H parameter, which indicates the effective potential determined by the local equilibrium potential distribution and resistance distribution within the battery 110. The operation of the GH estimator 142 will be explained in more detail below, with reference to Figure 4.

[0048] The extended Kalman filter 144 uses the G-parameter estimate G generated by the GH estimator 142. est , receive voltage value V and current value I, estimate G parameter value G est Based on the voltage value V and current value I, the state of charge value SOC is calculated. est The extended Kalman filter 144 can generate the coefficient data stored in the lookup table 152. The extended Kalman filter 144 generates the G-parameter estimate G est Each time a new voltage value V and current value I are input, the charge state value SOC is updated. est To output the charge status value (SOC) in real time, est The Kalman error covariance value can be updated.

[0049] According to one embodiment, a first-order estimate of the charge state and a first-order estimate of the Kalman error covariance can be calculated based on the current value I(t). Coefficient data, current value I(t), and G parameter estimate G are stored in the lookup table 152. est Based on this, the voltage estimate and the current Kalman gain can be calculated. Based on the primary charge state estimate, the current Kalman gain, the current voltage V(t), and the voltage estimate, the current charge state can be updated. Also, the primary Kalman error covariance estimate, the current Kalman gain, the coefficient data, and the G parameter estimate G est Based on this, the current value of the Kalman error covariance can be updated.

[0050] The operation of the extended Kalman filter 144 will be explained in more detail below, with reference to Figure 3.

[0051] Figure 3 illustrates a flowchart illustrating the operation of an extended Kalman filter in estimating the charge state according to one embodiment.

[0052] Referring to Figures 1 to 3, the operation of the extended Kalman filter 144 is carried out by the microprocessor 140.

[0053] Microprocessor 140 is the initial state of charge SOC est (0) and the initial value of the Kalman error covariance P k (0) is received, initial charge state SOC est (0) and the initial value of the Kalman error covariance P k (0) can be set (S10). Initial charge state SOC est (0) and the initial value of the Kalman error covariance P k (0) represents the state of charge (SOC) immediately before charging. est (t-1) and the value P immediately before the Kalman error covariance. k Used as (t-1). The user estimates the charge state of battery 110 and the initial charge state value (SOC). est Enter (0) or any value to set the initial state of charge (SOC). est (0) can be entered. The user can input the initial value P of the Kalman error covariance. k You can enter any value by setting it to (0).

[0054] The extended Kalman filter 144 of the microprocessor 140 calculates the G parameter estimate G for each sampling period Ts. est (t), the current value I(t) and the current value V(t) can be received (S20). The microprocessor 140 receives the G parameter estimate G generated by the GH estimator 142. est(t) can be input to the extended Kalman filter 144. The microprocessor 140 can receive the current voltage V(t) from the voltage measurement unit 120 and the current current I(t) from the current measurement unit 130.

[0055] Microprocessor 140 is a state-of-the-art (SOC) just before charging. est (t-1), current value I(t), sampling period Ts, and maximum capacity Q of battery 110. max Based on this, the primary estimate of the charging state SOC est - (t) The microprocessor 140 can calculate the information related to the sampling period Ts and the maximum capacity Q of the battery 110. max Information related to this can be saved. (State of Charge (SOC) value immediately before charging) est (t-1) is the current state of charge (SOC) before the sampling period Ts. est Corresponds to (t). Primary estimate of charge state. SOC est - (t) teeth, SOC est - (t) =SOC est (t-1)+I(t)×Ts / Q max It can also be calculated by [method / method].

[0056] The microprocessor 140 calculates the Kalman error covariance value P. k (t-1), sampling period Ts, maximum capacitance Q max and processor noise σ w Based on this, the linear estimate of the Kalman error covariance P k - (t) The following can be calculated (S40): Processor noise σ w This is set by the user according to the system specifications, for example, 10 -6 and 10 -1 It is also set to a value between σ. Processor noise σ wFor example, it can also be set to 0.0002. Kalman error covariance value P k (t-1) is the current value P of the Kalman error covariance prior to the sampling period Ts. k Corresponds to (t). Kalman error covariance linear estimate. P k - (t) teeth, P k - (t) =P k (t-1)+ (Ts / Q max ) 2 ×σ w It can also be calculated by [method / method].

[0057] The microprocessor 140 receives coefficient data, current value I(t), and G parameter estimate G. est Based on (t), the estimated voltage V est (t) can be calculated (S50).

[0058] According to one embodiment, the microprocessor 140 calculates the primary charge state value in step (S30) from the coefficient data stored in the lookup table 152. SOC est - (t) The first coefficient value C(t) corresponding to this can be extracted.

[0059] The coefficient data is also generated from the open circuit voltage (OCV) - state of charge (SOC) relationship determined in advance for the battery 110. For example, in FIG. 5, a curve showing the open circuit voltage (OCV) related to the state of charge (SOC) is illustrated. As shown in FIG. 5, the open circuit voltage (OCV) related to the state of charge (SOC) is non-linear. The look-up table 152 corresponds to the curve of the open circuit voltage (OCV) related to the state of charge (SOC) illustrated in FIG. 5, and the data of the open circuit voltage (OCV) related to the state of charge (SOC) is stored. Since the look-up table 152 must be stored in the memory 150, the open circuit voltage data values corresponding to some state of charge data values are stored. For example, some state of charge data values are also, for example, 0.01, 0.05, 0.1, 0.2, 0.5, 0.8, 0.9, 0.95, 0.99, and the open circuit voltage data values corresponding to the state of charge data values of 0.01, 0.05, 0.1, 0.2, 0.5, 0.8, 0.9, 0.95, and 0.99 are also stored as the coefficient data. According to another example, some state of charge data values are also multiples of, for example, 0.05 or 0.1.

[0060] The first coefficient value C(t) is the estimated state of charge at the first order SOC est - (t) to the adjacent state of charge data value N ε SOC est - (t) and the state of charge data value N ε SOC est - (t) corresponding open circuit voltage data value OCV(N ε SOC est - (t) is also determined based on. For example, the first coefficient value C(t) is C(t) = OCV(N ε SOC est - (t) / N ε ​​​​​SOC est - (t) It is also determined by ]. For example, the primary estimate of the charge state SOC est - (t) If it is 0.11, then the charge state data value N adjacent to 0.11 ε [ SOC est - (t) ] is also 0.1. If the open-circuit voltage data value corresponding to the charge state data value of 0.1 is 3, then the first coefficient value C(t) is also 3 / 0.1, which is 30, and such a value is also stored in memory 150 as coefficient data. Charge state data value N ε [ SOC est - (t) ] is the primary estimate of the charge state in the coefficient data. SOC est - (t) It is also the charge status data value closest to it.

[0061] Microprocessor 140 has a first coefficient value C(t) and a primary estimate of the charge state. SOC est - (t) , G parameter estimate G est Based on (t) and the current value I(t), the estimated voltage V est (t) can be calculated. Voltage estimate V est (t) is V est (t) = C(t) × SOC est - (t) +G est It can also be calculated by (t) × I(t). C(t) × SOC est - (t) This is a primary estimate of the charge state. SOC est - (t) This corresponds to the open-circuit voltage value, G est(t) × I(t) corresponds to the voltage drop due to the resistance component of battery 110.

[0062] According to another embodiment, the microprocessor 140 calculates a primary charge state from coefficient data stored in the lookup table 152 in step (S30). SOC est - (t) The corresponding second coefficient value C1(t) and third coefficient value E(t) can be extracted. The lookup table 152 contains the primary estimated value of the charge state. SOC est - (t) Coefficient data including the second coefficient value C1(t) and the third coefficient value E(t) can be stored.

[0063] The coefficient data is also determined from the open-circuit voltage (OCV)-state of charge (SOC) relationship predetermined for battery 110. Referring to the SOC-OCV curve in Figure 5, which shows the open-circuit voltage (OCV) related to the state of charge (SOC), the second coefficient value C1(t) and the third coefficient value E(t) are the first-order estimate of the state of charge in the SOC-OCV curve in Figure 5. SOC est - (t) Adjacent charge status data value N ε [ SOC est - (t) The gradient and OCV intercept of the linear function tangent to the point corresponding to ] can be determined, respectively. For example, the charge state data value N ε [ SOC est - (t) ] However, the SOC in Figure 5 a In this case, the second coefficient value C1(t) and the third coefficient value E(t) are, in the SOC-OCV curve, SOC a The gradient C of the linear function (TLa) tangent to the point corresponding to this. a and OCV intersection E a These can be determined as follows. Another example is the charge status data value N. ε [ SOCest - (t) ] However, the SOC in Figure 5 b In this case, the second coefficient value C1(t) and the third coefficient value E(t) are, in the SOC-OCV curve, SOC b The linear function TL tangent to the point corresponding to b Gradient C b and OCV intersection E b Each of these can be determined accordingly.

[0064] Microprocessor 140 has a second coefficient value C1(t) and a primary estimate of the charge state. SOC est - (t) , G parameter estimate G est Based on (t), the current value I(t) and the third coefficient value E(t), the estimated voltage V est (t) can be calculated. Voltage estimate V est (t) is V est (t) = C1(t) × SOC est - (t) +G est It can also be calculated by (t) × I(t) + E(t). C(t) × SOC est - (t) +E(t) is the primary estimate of the charge state. SOC est - (t) This corresponds to the open-circuit voltage value, G est (t) × I(t) corresponds to the voltage drop due to the resistance component of battery 110.

[0065] Microprocessor 140: Coefficient data, Kalman error covariance linear estimate P k - (t) and G parameter estimate G est Based on (t), the current value of the Kalman gain L k (t) can be calculated (S60).

[0066] The microprocessor 140 obtains a primary estimate of the charge state from the coefficient data stored in the lookup table 152. SOC est - (t) A first coefficient value C(t) corresponding to this can be extracted. In one embodiment, the first coefficient value C(t) is the same as the first coefficient value C(t) extracted in step (S50), or in another embodiment, it is the same as the second coefficient value C1(t) extracted in step (S50).

[0067] Microprocessor 140 is the first coefficient value C(t), the Kalman error covariance first-order estimate. P k - (t) , G parameter estimate G est (t) and measurement noise σ v Based on this, the current value of the Kalman gain L k (t) can be calculated. Measurement noise σ v This is set by the user according to the system specifications (for example, the voltage measurement unit 120 and the current measurement unit 130), for example, 10 -1 and 10 -3 It is also set to a value between σ. v For example, it can also be set to 0.01. Kalman gain current value L k (t) is L k (t) = C(t) × P k - (t) / [ C(t) 2 × P k - (t) + G est (t) 2 ×σ v It can also be calculated by [ ].

[0068] The microprocessor 140 calculates the primary charge state in step (S30). SOC est - (t), the current value of the Kalman gain L calculated in step (S60) k (t), current voltage V(t) received in step (S20), and estimated voltage V calculated in step (S50). est Based on (t), the current state of charge (SOC) est (t) can be updated and output (S70). Current charge status SOC est (t) is SOC est (t) SOC est - (t) +L k (t) × (V(t) - V est It is also calculated by (t). The current charge state (SOC) calculated in this way. est (t) is also output as an estimated value of the charge state of battery 110. After the next sampling period Ts, the current charge state SOC is also output. est Since (t) must be updated and output in real time, in step (S80), the current value of the Kalman error covariance P k (t) is updated.

[0069] Microprocessor 140 calculates the Kalman error covariance first-order estimate. P k - (t) , Kalman payoff current value L k (t), coefficient data and G parameter estimates G est Based on (t), the current value of the Kalman error covariance P k (t) can be updated (S80).

[0070] The microprocessor 140 obtains a primary estimate of the charge state from the coefficient data stored in the lookup table 152. SOC est - (t) A first coefficient value C(t) corresponding to this can be extracted. In one embodiment, the first coefficient value C(t) is the same as the first coefficient value C(t) extracted in step (S50), or in another embodiment, it is the same as the second coefficient value C1(t) extracted in step (S50).

[0071] The microprocessor 140 calculates the first-order estimate of the Kalman error covariance in step (S40). P k - (t) , the current value of the Kalman gain L calculated in step (S60) k (t), first coefficient value C(t), G parameter estimate G received in step (S20). est (t), and measurement noise σ v Based on this, the current value of the Kalman error covariance P k (t) can be calculated. Current value P of Kalman error covariance. k (t) is P k (t) P k - (t) - L k (t) 2 ×[ C(t) 2 × P k - (t) + G est (t) 2 ×σ v It can also be calculated by [ ].

[0072] When the sampling period Ts has passed (S90), the microprocessor 140 proceeds to step (S20) and generates a new G parameter estimate G est (t), a new current value I(t) and a new voltage value V(t) can be received (S20). Previous G parameter estimate G est (t) is the G parameter value immediately before G est At (t-1), the previous current value I(t) becomes the current value immediately before I(t-1), and the previous voltage value V(t) becomes the voltage value immediately before V(t-1). The newly received G parameter estimate G est Based on (t), the current value I(t) and the current value V(t), a new current charge state value SOC is obtained by repeating steps (S20) to (S80). est (t) is calculated and output in real time.

[0073] Figure 4 illustrates a flowchart illustrating the operation of calculating G-parameter values ​​using a GH estimator according to one embodiment.

[0074] The operation of the GH estimator 142 is carried out by the microprocessor 140. The microprocessor 140 can estimate the G-parameter and H-parameter values ​​of the battery 110 using adaptive filters.

[0075] Referring to Figures 1, 2, and 4, when the microprocessor 140 utilizes a recursive least squares (RLS) filter, the microprocessor 140 determines the initial state vector Θ of the battery 110. est (0) = [G est (0);H est (0) and the initial value of the covariance matrix P(0)=[P1(0);P2(0)] can be set (S110). Initial value of the state vector Θ est (0) and the initial value of the covariance matrix P(0) are the same as the value Θ immediately before the state vector. est (t-1) and the covariance matrix value P(t-1) are used. The current state vector value Θ est Since (t) and the current value of the covariance matrix P(t) will converge later, the user can set any value to the initial value of the state vector Θ. est (0) and the initial value P(0) of the covariance matrix can be input. For example, the initial value Θ of the state vector. est (0) is Θ est (0) = [G est (0);H est (0)]=[1;1] is set, and the initial value of the covariance matrix P(0) is also set to P(0)=[P1(0);P2(0)]=[1;1]. In this example, the initial value of the state vector Θ est Although both (0) and the initial value of the covariance matrix P(0) are initialized to 1, this is illustrative, and they can be initialized to other values ​​as well.

[0076] The microprocessor 140 can receive the current voltage V(t) and current I(t) of the battery 110 (S120). The voltage measurement unit 120 and the current measurement unit 130 sense the voltage and current of the battery 110, respectively, at each sampling period Ts, and can provide the voltage value and current value to the microprocessor 140 at each sampling period Ts. The microprocessor 140 can determine the current or most recently received voltage value as the current voltage value V(t), and the current or most recently received current value as the current current value I(t). The current voltage value V(t) and current value I(t) of the battery 110 received before the sampling period Ts become the voltage value immediately before V(t-1) and the current value immediately before I(t-1), respectively.

[0077] The microprocessor 140 calculates the current state vector Θ based on the current voltage V(t) and current I(t) received at each sampling period Ts. est (t)=[G est (t);H est The current value Θ of the state vector of battery 110 can be updated. est (t) is the estimated G parameter of battery 110 G est (t) and H parameter estimate H est (t) consists of Θ est (t)=[G est (t);H est It is defined as (t)). The current value of the covariance matrix P(t) consists of the first value of the covariance matrix P1(t) and the second value of the covariance matrix P2(t), and is defined as P(t)=[P1(t);P2(t)].

[0078] The current voltage V(t) and current I(t) of battery 110 are received at each sampling period Ts, and the current state vector Θ is also received. est(t) and the current value of the covariance matrix P(t) are also updated recursively at each sampling period Ts. The microprocessor 140 uses recursive least squares (RLS) to update the current value of the state vector Θ based on the current value of the voltage V(t) and current value I(t) received at each sampling period Ts. est (t) can be updated at each sampling period Ts. Current value Θ of the state vector. est If (t) is updated, the estimated G parameter G est (t) and H parameter estimate H est (t) is also determined.

[0079] The microprocessor 140 calculates the current value of the gain matrix L(t) based on the first forgetting factor λ1 and the second forgetting factor λ2, the current value I(t), and the covariance matrix value P(t-1) immediately prior to the calculation (S130).

[0080] The current value of the gain matrix L(t) is equal to the current value of the state vector Θ. est It is used when calculating (t) and the current value of the covariance matrix P(t). The gain matrix L(t) consists of the first value of the gain matrix L1(t) and the second value of the gain matrix L2(t), and can also be calculated as follows.

[0081]

number

[0082] Here, λ1 is the first forgetting factor and is related to the G parameter. λ2 is the second forgetting factor and is related to the H parameter. The first forgetting factor λ1 and the second forgetting factor λ2 are, respectively, the G parameter estimates G. est (t) and H parameter estimate H est In calculating (t), past voltage and current values ​​are used to estimate the current G parameter G. est (t) and the current H parameter estimate H estThis value shows the effect on (t). The closer the first forgetting factor λ1 and the second forgetting factor λ2 are to 1, the longer the G parameter estimate G will be. est (t) and H parameter estimate H est (t) has an effect on (t), and the closer it is to 0, the shorter the duration of the effect.

[0083] In one example, the first forgetting factor λ1 and the second forgetting factor λ2 are both between 0.9 and 1. In another example, the first forgetting factor λ1 is set to be greater than or equal to the second forgetting factor λ2. For example, the first forgetting factor λ1 may be set to 0.99999 and the second forgetting factor λ2 to 0.95. Such settings also depend on the characteristics of battery 100.

[0084] The inventors of the present invention found that in experiments conducted on a specific battery cell, high reliability results were obtained when the first forgetting factor λ1 and the second forgetting factor λ2 were 0.99999 and 0.95, respectively. However, the above values ​​are illustrative and can be set to different values ​​depending on the characteristics of the battery cell 111. For example, the first forgetting factor λ1 may be set to 0.9999 and the second forgetting factor λ2 may be set to 0.98.

[0085] In other examples, both the first forgetting factor λ1 and the second forgetting factor λ2 are set to 1. In that case, it can be seen that neither the first forgetting factor λ1 nor the second forgetting factor λ2 are applied.

[0086] The microprocessor 140 calculates the current value of the covariance matrix P(t) based on the first forgetting factor λ1 and the second forgetting factor λ2, the current value I(t), the covariance matrix value immediately prior to P(t-1), and the gain matrix value L(t) (S140).

[0087] The current value of the covariance matrix P(t) is used as the value of the covariance matrix immediately prior to calculation P(t-1) when calculating the current value of the gain matrix L(t) after a sampling period Ts. The current value of the covariance matrix P(t) can also be calculated as follows.

[0088]

number

[0089] The microprocessor 140 has a state vector value Θ est (t-1), based on the current value of the gain matrix L(t), the current value of the voltage V(t), and the current value of the current I(t), the current value of the state vector Θ est Calculate (t) (S150).

[0090] The microprocessor 140 has a current value I(t) and a state vector value Θ. est (t-1)=[G est (t-1);H est Based on (t-1), the estimated voltage of battery 110 is V est (t) can be calculated. For example, the estimated voltage V est (t) represents the current value I(t) and the previous value G of the G parameter. est (t-1) and the H parameter value immediately before H est Based on (t-1), V est (t)=G est (t-1)×I(t)+H est It can also be calculated as shown in (t-1).

[0091] The microprocessor 140 calculates the current voltage V(t) and the estimated voltage V est Based on (t), e(t) = V(t) - V est As shown in (t), the voltage error e(t) can be calculated.

[0092] The microprocessor 140 has a state vector value Θ est (t-1), based on the current value of the gain matrix L(t) and the voltage error e(t), the current value of the state vector Θ est(t) can be calculated. For example, the current value Θ of the state vector. est (t) is Θ est (t)=Θ est It can also be calculated as (t-1) + L(t) × e(t). The current value of the state vector Θ est (t) is calculated, and the estimated G parameter G est (t) and H parameter estimate H est (t) is determined.

[0093] The microprocessor 140 determines the G-parameter estimate G in step (S150). est (t) and the current value I(t) can be provided to the extended Kalman filter 144 (S160).

[0094] The microprocessor 140 can repeatedly perform step (S120) or step (S150) for each sampling period Ts (S170).

[0095] State vector current value Θ est The mathematical formula Θ that recursively expresses (t) est (t)=Θ est (t-1)+L(t)×e(t) can also be derived as follows.

[0096] First, the loss function ε to which the first forgetting factor λ1 and the second forgetting factor λ2 are applied is defined as follows:

[0097]

number

[0098] Here, V(i) is the i-th voltage value and I(i) is the i-th current value. V(t) and I(t) are the current voltage and current values, respectively, and V(t-1) and I(t-1) are the voltage and current values ​​immediately before the event, respectively.

[0099] G(i) and H(i) are the actual values ​​of the i-th G parameter and the i-th H parameter, respectively. est (t) and H est (t) represents the current G-parameter estimate and the current H-parameter estimate, respectively.

[0100] The loss function ε is G est (t) and H est When the derivatives of (t) and (t) are both zero, G est (t) and H est The loss function ε is minimized for (t).

[0101] The loss function ε is G est G for which the derivative with respect to (t) is 0. est The answer to (t) is as follows:

[0102]

number

[0103] If we rearrange the aforementioned formula, G est (t) is as follows:

[0104]

number

[0105] The loss function ε is H est H whose derivative with respect to (t) is 0 est The answer to (t) is as follows:

[0106]

number

[0107] If we rearrange the aforementioned formula, H est (t) is as follows:

[0108]

number

[0109] For real-time estimation, the current value Θ of the state vector est Using (t), G obtained above est (t) and H est If we organize (t) in a recursive form, we get the following:

[0110] Θ est (t)=[G est (t);H est (t)]=Θ est (t-1)+L(t)×[V(t)-G est (t-1)×I(t)-H est (t-1)] Estimated voltage V est (t) is V est (t)=G est (t-1)×I(t)+H est The voltage error e(t) is calculated as shown in (t-1), and the voltage error e(t) is e(t) = V(t) - V est Since it is defined as (t), the current value of the state vector Θ est (t) can also be expressed as follows, as mentioned above.

[0111] Θ est (t)=[G est (t);H est (t)]=Θ est (t-1)+L(t)×e(t) Here, the current value of the gain matrix L(t) and the current value of the covariance matrix P(t) are calculated as follows, as described above.

[0112]

number

[0113]

number

[0114] According to this embodiment, a recursive method is used to calculate the G-parameter values, so the memory 150 contains the current voltage V(t), the current current I(t), and the current state vector Θ. est (t), the current value of the covariance matrix P(t), the first forgetting factor λ1, and the second forgetting factor λ2 are stored. Not only is the operation very simple, but it can also be performed on small memory units of a few kilobytes (150). The current value of the state vector Θ est Since (t) and the current value of the covariance matrix P(t) are updated periodically, the voltage fluctuations and current fluctuations of the battery 110 are updated in real time, and the G parameter estimate G est (t) and H parameter estimate H est (t) can be reflected in this.

[0115] Figure 6 illustrates an internal configuration diagram for carrying out a charge state estimation method according to another embodiment.

[0116] Referring to Figure 6, the microprocessor 140 also includes a noise filter 146 in addition to the GH estimator 142 and the extended Kalman filter 144. The GH estimator 142 and the extended Kalman filter 144 have been explained above, so we will avoid repetition.

[0117] According to this embodiment, the sensing voltage value V corresponds to the voltage sensed by the voltage measurement unit 120. sen The signal is input to the noise filter 146, and the noise filter 146 detects the sensing voltage value V sen It can output a voltage value V. The voltage value V includes the voltage immediately before V(t-1) and the voltage at the present time V(t). The sensing current value I corresponds to the current sensed by the current measuring unit 130. sen The signal is input to the noise filter 146, and the noise filter 146 detects the sensing current value I sen It can output a current value I. The current value I includes the current value immediately before I(t-1) and the current value I(t).

[0118] The noise filter 146 senses the voltage value V sen and sensing current value Isen The noise contained in the signal can be removed, and the voltage value V and current value I from which the noise has been removed can be output. The noise filter 146 is also, for example, a low-pass filter. The noise filter 146 is also a moving average filter. The noise filter 146 is also an IIR (infinite impulse response) filter or an FIR (finite impulse response) filter.

[0119] The voltage value V and current value I output from the noise filter 146 are also input to the GH estimator 142. Sensing voltage value V input to noise filter 146 sen The current value I output from the noise filter 146 is also input to the extended Kalman filter 144. The sensing voltage value V is input to the extended Kalman filter 144. sen The voltage and current value I correspond to the current value V(t) and current value I(t), respectively.

[0120] Figure 7 illustrates a flowchart illustrating the operation of the noise filter in another embodiment.

[0121] Referring to Figure 7, the sensing voltage value V corresponds to the voltage of the battery 110 sensed by the voltage measuring unit 120. sen A sensing current value I is generated, corresponding to the current of the battery 110 sensed by the current measuring unit 120. sen This is generated (S210).

[0122] Sensing voltage value V sen and sensing current value I sen These are input to the noise filter 146, respectively, and a noise-filtered voltage value V and a noise-filtered current value I are generated (S220).

[0123] The voltage value V and current value I are transmitted to the GH estimator 142, and the current value I and the sensing voltage value V senIt is transmitted to the extended Kalman filter 144. The sensing voltage value V sen and the current value I input to the extended Kalman filter 144 respectively correspond to the voltage current value V(t) and the current current value I(t).

[0124] FIG. 8 is a graph comparing the state of charge value estimated according to the present invention with the state of charge value of an actual battery, and FIG. 9 is a graph comparing the cell voltage estimated according to the present invention with the cell voltage of an actual battery.

[0125] As shown in FIG. 8 、 The initial state of charge value was set to 0.3, but it can be seen that after approximately 400 seconds, the state of charge value estimated according to the present invention follows the actual state of charge value. Over time, the error from the actual state of charge value is reduced. Since the method according to the present invention uses simple calculations, it can accurately estimate the state of charge of the battery while being driven even in a low-specification processor.

[0126] As shown in FIG. 9, it can be seen that the cell voltage estimated according to the present invention also follows the actual cell voltage. This shows that the estimation algorithm of the present invention is accurate and can be applied to various actual products.

[0127] The idea of the present invention is not limited to the foregoing embodiments. Not only the scope of the claims, but all scopes equivalent to or equivalently changed from the scope of the claims belong to the category of the idea of the present invention.

Explanation of Signs

[0128] 100 Battery system 101 First terminal 102 Second terminal 110 Battery​​

Claims

1. In a method for estimating the battery charge state, The steps include setting the initial charge state value and the initial Kalman error covariance value, The steps include receiving the estimated G-parameter value, current value, and current voltage value of the aforementioned battery, The step involves inputting the estimated G-parameter, current current, and current voltage into the extended Kalman filter and updating the current battery charge state and the current Kalman error covariance, A step of calculating a primary estimate of the charging state based on the current value, The step of calculating the linear estimate of the Kalman error covariance, The steps include receiving coefficient data generated from a predetermined open-circuit voltage (OCV)-state-of-charge (SOC) relationship for the aforementioned battery, A step of calculating a voltage estimate based on the coefficient data, the current current value, and the G parameter estimate, A step of calculating the current Kalman gain value based on the coefficient data, the linear estimate of the Kalman error covariance, and the G-parameter estimate, A step of updating the current charge state value based on the primary charge state estimate, the current Kalman gain value, the current voltage value, and the voltage estimate; A step including updating the current Kalman error covariance value based on the aforementioned linear estimate of the Kalman error covariance, the aforementioned current Kalman gain value, the aforementioned coefficient data, and the aforementioned G-parameter estimate; The step includes outputting the current charge state value, The aforementioned primary estimated charge state SOC est - (t) is equal to the charge state immediately before SOC ｅｓｔ (t-1), current value I(t), sampling period Ts, and maximum capacity Q of the battery. ｍａｘ Using this, SOC est - (t) = SOC ｅｓｔ (t-1)+I(t)×Ts / Q ｍａｘ A method for estimating the battery charge state, characterized by being calculated by [a specific method].

2. The prior Kalman error covariance primary estimated value Pk-(t) is the immediate previous value P of the Kalman error covariance ｋ (t - 1), sampling period Ts, the maximum capacity Q of the battery ｍａｘ and processor noise σ ｗ are used, and Pk-(t) = P ｋ (t - 1) + (Ts / Qmax)2 × σ ｗ The battery state of charge estimation method according to claim 1, characterized in that it is calculated by

3. The step of calculating the aforementioned voltage estimate is: From the coefficient data, a first coefficient value C(t) corresponding to the primary estimated value SOC est - (t) of the charge state is extracted. The first coefficient value C(t), the primary estimated value of the charge state SOC est - (t), and the estimated value of the G parameter G ｅｓｔ Using (t) and the current value I(t), V ｅｓｔ (t)=C(t)×SOC est - (t)+G ｅｓｔ The estimated voltage V is obtained by (t) × I(t). ｅｓｔ The battery charge state estimation method according to claim 1, characterized by including the step of calculating (t).

4. The first coefficient value C(t) is the charge state data value N adjacent to the charge state primary estimate value SOC est - (t). ε [SOC est - (t)], and the charge state data value N ε [SOC est - (t)] corresponds to the open-circuit voltage data value OCV(N ε Using [SOC est - (t)], C(t) = OCV(N ε [SOC est - (t)] / N ε The battery charge state estimation method according to claim 3, characterized by being determined by [SOC est - (t)].

5. The step of calculating the aforementioned voltage estimate is: From the coefficient data, the second coefficient value C corresponding to the primary estimated value SOC est - (t) is obtained. １ The steps include extracting (t) and the third coefficient value E(t), The second coefficient value C １ (t), the primary estimated value of the charge state SOC est - (t), the estimated value of the G parameter G ｅｓｔ (t), using the current value I(t) and the third coefficient value E(t), V ｅｓｔ (t) = C １ (t)×SOC est - (t)+G ｅｓｔ The voltage estimate V is calculated as (t) × I(t) + E(t). ｅｓｔ The battery charge state estimation method according to claim 1, characterized by including the step of calculating (t).

6. The second coefficient value C １ (t) and the third coefficient value E(t) are adjacent to the charge state data value N in the curve corresponding to the open circuit voltage (OCV)-state of charge (SOC) relationship, where SOC est - (t) is the primary estimated value of the charge state. ε The battery charge state estimation method according to claim 5, characterized in that the gradient and OCV intercept of a linear function tangent to the point corresponding to [SOC est - (t)] are determined, respectively.

7. The step of calculating the current Kalman gain is as follows: From the coefficient data, a first coefficient value C(t) corresponding to the primary estimated value SOC est - (t) of the charge state is extracted. The first coefficient value C(t), the linear estimate of the Kalman error covariance P k-(t), and the estimated value of the G parameter G ｅｓｔ (t) and measurement noise σ ｖ Using L ｋ (t)=C(t)×P k − (t) / [C(t) 2×P k − (t)+G est (t) 2×σ ｖ ] The current value of the Kalman gain L ｋ The battery charge state estimation method according to claim 1, characterized by including the step of calculating (t).

8. The current charge state SOC ｅｓｔ (t) is the primary estimated value of the charge state SOC est - (t), and the current value of the Kalman gain L ｋ (t), the current voltage V(t), and the estimated voltage V ｅｓｔ Using (t), SOC ｅｓｔ (t)=SOC est - (t)+L ｋ (t)×(V(t)-V ｅｓｔ The battery charge state estimation method according to claim 1, characterized in that it is calculated by (t).

9. The step of calculating the current value of the Kalman error covariance is as follows: From the coefficient data, a first coefficient value C(t) corresponding to the primary estimated value SOC est - (t) of the charge state is extracted. The first-order estimate of the Kalman error covariance P k - (t), and the current value of the Kalman gain L ｋ (t), the first coefficient value C(t), the estimated value of the G parameter G ｅｓｔ (t) and measurement noise σ ｖ Using P ｋ (t) = P k - (t) - L k (t) 2 x [C (t) 2 x P k - (t) + G est (t) 2 x σ ｖ ] The current value of the Kalman error covariance P ｋ The battery charge state estimation method according to claim 1, characterized by including the step of calculating (t).

10. The steps include sensing the voltage and current of the battery at pre-set sampling intervals Ts and periodically generating the voltage and current values ​​of the battery, The battery charge state estimation method according to claim 1, further comprising the step of using an adaptive filter to generate an estimated G-parameter value that quantifies the G-parameter indicating the sensitivity of the battery to changes in current from the voltage value and the current value.

11. The charging state estimation method according to claim 10, characterized in that the adaptive filter is a filter utilizing the recursive least squares method (RLS).

12. The battery charge state estimation method according to claim 10, further comprising the step of using the adaptive filter to generate an estimated H-parameter value, which is a numerical representation of the H-parameter indicating the effective potential determined by the local equilibrium potential scattering and resistance distribution within the battery, from the current voltage value and the current current value.

13. The process further includes setting initial values ​​for the state vector and covariance matrix of the aforementioned battery, The step of periodically generating the voltage and current values ​​of the aforementioned battery is as follows: The steps include generating the voltage and current values ​​of the battery immediately before use, The battery charge state estimation method according to claim 12, characterized by including the step of generating the current voltage value and the current value of the battery after the sampling period Ts.

14. The step of generating the G parameter estimate and the H parameter estimate is: A step of calculating an estimated battery voltage based on the current value and the state vector value immediately preceding the current, The steps include calculating the current value of the gain matrix and the current value of the covariance matrix based on the current value and the covariance matrix immediately preceding, A step of calculating the voltage error based on the current voltage and the estimated voltage, The battery charge state estimation method according to claim 13, comprising the step of generating the G parameter estimate and the H parameter estimate by calculating the current state vector value based on the state vector immediate prior value, the gain matrix current value, and the voltage error.

15. The aforementioned voltage estimate value V ｅｓｔ (t) is the current value I(t) and the G parameter value G immediately before the current change. ｅｓｔ (t-1) and the H parameter value immediately before H ｅｓｔ Using (t-1), V ｅｓｔ (t) = G ｅｓｔ (t-1)×I(t)+H ｅｓｔ The battery charge state estimation method according to claim 14, characterized in that it is calculated by (t-1).

16. The current value Θ of the state vector ｅｓｔ (t) is the value Θ immediately before the state vector. ｅｓｔ (t-1), using the current value L(t) of the gain matrix and the voltage error e(t), Θ ｅｓｔ (t) = Θ ｅｓｔ The battery charge state estimation method according to claim 14, characterized in that it is calculated by (t-1) + L(t) × e(t).

17. The step of generating the G parameter estimate and the H parameter estimate is: The first forgetting factor λ related to the G parameter １ , and the second forgetting factor λ related to the H parameter ２ The battery charge state estimation method according to claim 14, further comprising the step of receiving a signal.

18. The current value of the aforementioned gain matrix is ​​calculated by the following formula: [Math 1] The current value of the covariance matrix is ​​calculated using the following formula: [Math 2] Here, L(t) is the current value of the gain matrix, P(t) is the current value of the covariance matrix, P(t-1) is the previous value of the covariance matrix, I(t) is the current value, and λ １ λ is the first forgetting factor, ２ The battery charge state estimation method according to claim 17, characterized in that is the second forgetting factor.

19. The steps include sensing the voltage and current of the battery at pre-set sampling periods Ts and generating sensing voltage and sensing current values ​​of the battery, The steps include inputting the sensing voltage value and the sensing current value to a noise filter, respectively, to periodically generate the voltage and current values ​​of the battery, The battery charge state estimation method according to claim 1, further comprising the steps of: using an adaptive filter to generate an estimated G-parameter value, which is a numerical representation of the G-parameter indicating the sensitivity of the battery to current changes, from the voltage value and the current value; and an estimated H-parameter value, which is a numerical representation of the H-parameter indicating the effective potential determined by the local equilibrium potential scattering and resistance distribution within the battery.

20. The step of receiving the current current value and the current voltage value is: The current value is the step of receiving the current value, The battery charge state estimation method according to claim 19, characterized in that it includes the step of receiving the sensing voltage value as the current voltage value.

Images