Battery management device, battery management method, and battery pack

Abstract

Provided are a battery management device, a battery management method, and a battery pack. A battery management device uses a plurality of observation data sets associated with external variables observable from outside a battery cell and a plurality of desired data sets associated with internal variables unobservable from outside the battery cell to set at least one of the plurality of external variables as a valid external variable for each internal variable. The observation data sets associated with the respective valid external variables are used for machine training of a sub-multilayer perceptron required to estimate the respective internal variables.

Description

Technical Field

[0001] The present disclosure relates to a technique for analyzing a correlation between an external variable and an internal variable depending on an internal chemical state of a battery cell.

[0002] This application claims the benefit of Korean Patent Application No. 10-2019-0147025, filed with the Korean Intellectual Property Office on Nov. 15, 2019, the disclosure of which is incorporated herein by reference in its entirety. Background Art

[0003] Recently, the demand for portable electronic products such as laptop computers, cameras, and mobile phones has increased rapidly, and with the widespread development of electric vehicles, storage batteries for energy storage, robots, and satellites, much research is being conducted on high-performance batteries that can be repeatedly recharged.

[0004] Currently, commercially available batteries include nickel-cadmium batteries, nickel-metal hydride batteries, nickel-zinc batteries, lithium batteries, etc. Among them, lithium batteries have little or no memory effect, and thus, due to their advantages of being rechargeable whenever convenient, having a very low self-discharge rate, and high energy density, they have received more attention compared to nickel-based batteries.

[0005] Due to repeated charging and discharging, a battery cell gradually deteriorates over time. As the battery cell deteriorates, the internal chemical state of the battery cell also changes. However, an internal variable (e.g., conductivity of a positive electrode active material) indicating the internal chemical state of the battery cell cannot be observed outside the battery cell.

[0006] Meanwhile, the internal chemical state of a battery cell causes a change in an external variable (e.g., voltage and temperature of the battery cell) that can be observed outside the battery cell. Thus, there have been attempts to estimate an internal variable based on an external variable.

[0007] However, some internal variables and some external variables may have a very low correlation. Thus, without considering the correlation between an internal variable and an external variable, an internal variable estimated using an external variable may be greatly different from the actual internal chemical state of the battery cell. Summary of the Invention

[0008] Technical problem

[0009] The present disclosure is designed to solve the above problems, and thus, the present disclosure relates to providing a battery management device, a battery management method, and a battery pack for analyzing a correlation between an internal variable and an external variable of a battery cell.

[0010] The present disclosure also relates to providing a battery management device, a battery management method, and a battery pack, in which external variables having at least a certain level of correlation with each internal variable are used in modeling a sub-multi-layer perceptron necessary for estimating each internal variable.

[0011] These and other objects and advantages of the present disclosure will be understood from the following description and will be apparent from the embodiments of the present disclosure. Additionally, it will be readily understood that the objects and advantages of the present disclosure can be achieved by the means set forth in the appended claims and their combinations.

[0012] Technical solution

[0013] A battery management device according to an aspect of the present disclosure includes: a memory unit configured to store a first observation data set to a p-th observation data set and a first expected data set to an n-th expected data set, where each of p and n is an integer of 2 or greater; and a control unit operably coupled to the memory unit. The first observation data set to the p-th observation data set are respectively associated with p external variables observable outside the battery cell. Each of the first observation data set to the p-th observation data set includes a predetermined number of input values. The first expected data set to the n-th expected data set are respectively associated with a first internal variable to an n-th internal variable, where the first internal variable to the n-th internal variable depends on the internal chemical state of the battery cell and is not observable outside the battery cell. Each of the first expected data set to the n-th expected data set includes the same number of target values as the predetermined number. The control unit is configured to set m observation data sets extracted from the first observation data set to the p-th observation data set through data filtering as a first input data set to an m-th input data set. m is an integer of 1 or greater and p or less. The control unit is configured to set at least one of the first external variable to the m-th external variable as an effective external variable for each of the first internal variable to the n-th internal variable based on the first input data set to the m-th input data set and the first expected data set to the n-th expected data set. The first external variable to the m-th external variable are m external variables among the p external variables that are associated with the m observation data sets.

[0014] The control unit may be configured to determine a multiple correlation coefficient between the q-th observation data set and the (q + 1)-th observation data set when q is an integer from 1 to p - 1. The control unit may be configured to set the q-th observation data set as one of the first input data set to the m-th input data set when the absolute value of the multiple correlation coefficient is less than a predetermined filtering value.

[0015] The control unit may be further configured to store a main multi-layer perceptron that defines the correspondence between the first to the m-th external variables and the first to the n-th internal variables. The control unit may be configured to obtain the first to the n-th output data sets from the first to the n-th output nodes included in the output layer of the main multi-layer perceptron by supplying the first to the m-th input data sets to the first to the m-th input nodes included in the input layer of the main multi-layer perceptron. Each of the first to the n-th output data sets may include the same number of result values as a predetermined number. The control unit may be configured to determine the first to the n-th error factors based on the first to the n-th output data sets and the first to the n-th expected data sets. The control unit may be configured to determine the first to the n-th reference values by comparing each of the first to the n-th error factors with a threshold error factor.

[0016] The control unit may be configured to: when j is an integer from 1 to n, determine the j-th error factor to be equal to the error ratio between the j-th output data set and the j-th expected data set.

[0017] The control unit may be configured to: when the j-th error factor is less than the threshold error factor, set the j-th reference value to be equal to a first predetermined value.

[0018] The control unit may be configured to: when the j-th error factor is equal to or greater than the threshold error factor, set the j-th reference value to be equal to a second predetermined value. The second predetermined value is less than the first predetermined value.

[0019] The control unit may be configured to: when i is an integer from 1 to m and j is an integer from 1 to n, determine whether to set the i-th external variable among the first to the m-th external variables as the valid external variable for the j-th internal variable based on the i-th input data set, the j-th expected data set, and the j-th reference value. The control unit may be configured to, when the i-th external variable is set as the valid external variable for the j-th internal variable, use the i-th input data set as training data to learn the sub multi-layer perceptron associated with the j-th internal variable.

[0020] The control unit may be configured to: determine the multiple correlation coefficient between the i-th input data set and the j-th expected data set. The control unit may be configured to, when the absolute value of the multiple correlation coefficient is greater than the j-th reference value, set the i-th external variable as the valid external variable for the j-th internal variable.

[0021] The control unit may be configured to: when the absolute value of the multiple correlation coefficient is equal to or less than the j-th reference value, set the i-th external variable as the invalid external variable for the j-th internal variable.

[0022] A battery pack according to another aspect of the present disclosure includes a battery management device.

[0023] A battery management method according to still another aspect of the present disclosure includes: storing a first observed data set to a p-th observed data set and a first desired data set to an n-th desired data set, where each of p and n is an integer of 2 or greater, the first observed data set to the p-th observed data set are respectively associated with p external variables observable outside the battery cell, each of the first observed data set to the p-th observed data set includes a predetermined number of input values, the first desired data set to the n-th desired data set are respectively associated with a first internal variable to an n-th internal variable unobservable outside the battery cell, the first internal variable to the n-th internal variable depend on the internal chemical state of the battery cell, and each of the first desired data set to the n-th desired data set includes the same number of target values as the predetermined number; setting m observed data sets extracted from the first observed data set to the p-th observed data set through data filtering as a first input data set to an m-th input data set, where m is an integer of 1 or greater and p or less; and setting at least one of the first external variable to the m-th external variable as an effective external variable for each of the first internal variable to the n-th internal variable based on the first input data set to the m-th input data set and the first desired data set to the n-th desired data set. The first external variable to the m-th external variable are m external variables among the p external variables associated with the m observed data sets.

[0024] Setting the m observed data sets as the first input data set to the m-th input data set includes: when q is an integer from 1 to p - 1, determining a multiple correlation coefficient between the q-th observed data set and the (q + 1)-th observed data set; and when the absolute value of the multiple correlation coefficient is less than a predetermined filtering value, setting the q-th observed data set as one of the input data sets in the first input data set to the m-th input data set.

[0025] Beneficial effect

[0026] According to at least one of the embodiments of the present disclosure, the correlation between the internal variables and the external variables of the battery cell can be analyzed.

[0027] According to at least one of the embodiments of the present disclosure, when the correlation between two external variables is equal to or greater than a predetermined level, an observed data set of only one of the two external variables is extracted, and the extracted observed data set can be used in the modeling of the sub-multilayer perceptron.

[0028] According to at least one of the embodiments of the present disclosure, only external variables having at least a certain level of correlation with each internal variable can be used in the modeling of the sub-multilayer perceptron necessary for estimating each internal variable.

[0029] The effects of the present disclosure are not limited to the effects mentioned above, and those skilled in the art will clearly understand these and other effects from the appended claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The drawings illustrate preferred embodiments of the present disclosure and, together with the following detailed description of the present disclosure, are used to provide a further understanding of the technical aspects of the present disclosure. Therefore, the present disclosure should not be construed as being limited to the drawings.

[0031] Figure 1 is a diagram exemplarily showing the configuration of a battery pack including a battery management device according to an embodiment of the present disclosure.

[0032] Figure 2 is exemplarily showing by Figure 1 the main multi-layer perceptron used by the battery management device.

[0033] Figure 3 is a diagram exemplarily showing a sub multi-layer perceptron.

[0034] Figure 4 is a flowchart exemplarily showing a first method that can be executed by the Figure 1 battery management device.

[0035] Figure 5 is a flowchart exemplarily showing a second method that can be executed by the Figure 1 battery management device.

[0036] Figure 6 is a flowchart exemplarily showing a third method that can be executed by the Figure 1 battery management device. DETAILED DESCRIPTION

[0037] Hereinafter, preferred embodiments of the present disclosure will be described in detail with reference to the drawings. Before the description, it should be understood that the terms or words used in the specification and the appended claims should not be construed as being limited to the general meaning and the dictionary meaning. Instead, on the basis of the principle that allows the inventor to appropriately define the terms to obtain the best description, they should be interpreted based on the meanings and concepts corresponding to the technical aspects of the present disclosure.

[0038] Therefore, the embodiments described herein and the illustrations shown in the drawings are only the most preferred embodiments of the present disclosure, but are not intended to fully describe the technical aspects of the present disclosure. Therefore, it should be understood that various other equivalents and modifications may have been made thereto at the time of filing the application.

[0039] Terms including ordinal numbers such as "first", "second", etc. are used to distinguish one element from another among various elements, but are not intended to limit the elements by the terms.

[0040] Unless the context clearly indicates otherwise, it should be understood that the term "comprising", as used in this specification, specifies the presence of the stated elements, but does not preclude the presence or addition of one or more other elements. Additionally, as used herein, the term "control unit" refers to a processing unit for at least one function or operation, and this can be implemented by hardware and software alone or in combination.

[0041] Further, throughout the specification, it should be further understood that when an element is referred to as being "connected to" another element, it can be directly connected to the other element or there can be intervening elements.

[0042] Figure 1 FIG. is an exemplary diagram showing the configuration of the battery pack 1 including the battery management device 100 according to the present disclosure. Figure 2 is an exemplary diagram showing Figure 1 the main multi-layer perceptron used by the battery management device 100, and Figure 3 is an exemplary diagram showing the sub multi-layer perceptron.

[0043] Referring to Figure 1 , the battery pack 1 includes battery cells 10, switches 20, and a battery management device 100.

[0044] The battery pack 1 is installed on an electrically powered device such as an electric vehicle to supply the electrical energy required to drive the electrically powered device. The battery management device 100 is arranged to be electrically connected to the positive and negative terminals of the battery cells 10.

[0045] The battery cells 10 can be lithium-ion cells. The battery cells 10 can include any type capable of being repeatedly charged and discharged, and are not limited to lithium-ion cells.

[0046] The battery cells 10 can be electrically coupled to an external device through the power terminals (+, -) of the battery pack 1. The external device can be, for example, an electrical load (e.g., a motor), a direct current (DC)-alternating current (AC) inverter, and a charger for an electric vehicle.

[0047] The switch 20 is installed on the current path connecting the positive terminal of the battery cell 10 to the power terminal (+) or the current path connecting the negative terminal of the battery cell 10 to the power terminal (-). When the switch 20 is in the open operating state, the battery cell 10 stops charging and discharging. When the switch 20 is in the closed operating state, the battery cell 10 is allowed to charge and discharge.

[0048] The battery management device 100 includes an interface unit 110, a memory unit 120, and a control unit 130. The battery management device 100 may further include a sensing unit 140.

[0049] The sensing unit 140 includes a voltage sensor 141, a current sensor 142, and a temperature sensor 143. The voltage sensor 141 is configured to measure the voltage across the battery cell 10. The current sensor 142 is configured to measure the current flowing through the battery cell 10. The temperature sensor 143 is configured to measure the temperature of the battery cell 10. The sensing unit 140 can send sensing data indicating the measured voltage, the measured current, and the measured temperature to the control unit 130.

[0050] The interface unit 110 can be coupled to an external device to enable communication therebetween. The external device can be, for example, a charging system, an electrical load, and a mobile device. The interface unit 110 is configured to support wired communication or wireless communication between the control unit 130 and the external device. The wired communication can be, for example, controller area network (CAN) communication, and the wireless communication can be, for example, ZigBee or Bluetooth communication. The interface unit 110 can include output devices such as a display or a speaker to provide the result of each operation performed by the control unit 130 in a form recognizable by the user. The interface unit 110 can include input devices such as a mouse and a keyboard to receive data from the user.

[0051] The memory unit 120 is operably coupled to at least one of the interface unit 110, the control unit 130, or the sensing unit 140. The memory unit 120 can store the result of each operation performed by the control unit 130. The memory unit 120 can include at least one type of storage medium such as, for example, a flash memory type, a hard disk type, a solid state disk (SSD) type, a silicon disk drive (SDD) type, a multimedia card micro type, a random access memory (RAM), a static random access memory (SRAM), a read only memory (ROM), an electrically erasable programmable read only memory (EEPROM), or a programmable read only memory (PROM).

[0052] The memory unit 120 is configured to store various data required to estimate the internal chemical state of the battery cell 10. Specifically, the memory unit 120 stores a first number of observation data sets and a second number of desired data sets. For example, information including the first observation data set to the p-th observation data set and the first desired data set to the n-th desired data set is received by the interface unit 110 from an external device. The memory unit 120 can also store a main multi-layer perceptron. Hereinafter, it is assumed that p indicates the first number and is an integer of 2 or greater, and n indicates the second number and is an integer of 2 or greater.

[0053] The first observation dataset to the p-th observation dataset are each associated with p external variables observable outside the battery cell 10. Each observation dataset includes a predetermined third number (e.g., 10,000) of input values. For example, the third number of input values for one of the p external variables is included in the q-th observation dataset.

[0054] The control unit 130 is operably coupled to at least one of the interface unit 110, the memory unit 120, or the sensing unit 140. The control unit 130 can be implemented in hardware using at least one of the following: application specific integrated circuit (ASIC), digital signal processor (DSP), digital signal processing device (DSPD), programmable logic device (PLD), field programmable gate array (FPGA), microprocessor, or an electrical unit for performing other functions.

[0055] The control unit 130 extracts m observation datasets from the first observation dataset to the p-th observation dataset through data filtering, and sets the m observation datasets as the first input dataset to the m-th input dataset according to the extraction order. m is an integer greater than or equal to 1 and less than or equal to p. The process of extracting the m observation datasets will be described below with reference to Figure 4 Describe the process of extracting the m observation datasets.

[0056] In the specification, X i represents the i-th input dataset associated with the i-th external variable, and X i (k) represents the k-th input value of the i-th input dataset.

[0057] For example, when m = 16, the first external variable to the sixteenth external variable can be defined as follows.

[0058] The first external variable can indicate the time period from the time point when the state of charge (SOC) of the battery cell 10 is equal to the first SOC (e.g., 100%) to the time point when the voltage of the battery cell 10 reaches the first voltage (e.g., 3.0V) through a first test of discharging the battery cell 10 with a current at a first current rate (e.g., 1 / 3C) at a first temperature.

[0059] The second external variable can indicate the voltage of the battery cell 10 at the time point when a second test is performed for a first reference time (e.g., 0.1 second), the second test being to discharge the battery cell 10 with a current at a second current rate (e.g., 200A) at a second temperature starting from the time point when the SOC of the battery cell 10 is equal to the second SOC (e.g., 50%).

[0060] The third external variable can indicate the voltage of the battery cell 10 at the time point when the second test is performed for a second reference time longer than the first reference time (e.g., 1.0 second).

[0061] The fourth external variable may indicate the voltage of the battery cell 10 at a time point when the second test is performed for a third reference time (e.g., 10.0 seconds) longer than the second reference time.

[0062] The fifth external variable may indicate the voltage of the battery cell 10 at a time point when the second test is performed for a fourth reference time (e.g., 30.0 seconds) longer than the third reference time.

[0063] The sixth external variable may include a time period from a time point when the SOC of the battery cell 10 is equal to the third SOC (e.g., 100%) to a time point when the voltage of the battery cell 10 reaches a second voltage (e.g., 3.8V) through a third test of discharging the battery cell 10 with a current at a third current rate (e.g., 1.0C) at a third temperature.

[0064] The seventh external variable may indicate a time period until a time point when the voltage of the battery cell 10 reaches a third voltage (e.g., 3.5V) lower than the second voltage through the third test.

[0065] The eighth external variable may represent the time taken for the voltage of the battery cell 10 to decrease from the second voltage to the third voltage through the third test.

[0066] The ninth external variable may indicate the voltage of the battery cell 10 at a time point when a fourth test of discharging the battery cell 10 with a current at a fourth current rate (e.g., 1.0C) at a fourth temperature is performed for a fifth reference time (e.g., 0.1 seconds) starting from a time point when the SOC of the battery cell 10 is equal to the fourth SOC (e.g., 10%).

[0067] The tenth external variable may indicate the voltage of the battery cell 10 at a time point when the fourth test is performed for a sixth reference time (e.g., 1.0 seconds) longer than the fifth reference time.

[0068] The eleventh external variable may indicate the voltage of the battery cell 10 at a time point when the fourth test is performed for a seventh reference time (e.g., 10.0 seconds) longer than the sixth reference time.

[0069] The twelfth external variable may indicate the voltage of the battery cell 10 at a time point when the fourth test is performed for an eighth reference time (e.g., 30.0 seconds) longer than the seventh reference time.

[0070] The thirteenth external variable may indicate the voltage of the battery cell 10 at a time point when the fourth test is performed for a ninth reference time (e.g., 100.0 seconds) longer than the eighth reference time.

[0071] The fourteenth external variable may indicate the ratio between the voltage change of the battery cell 10 during a first period (e.g., 0.0 to 0.1 seconds) and the voltage change of the battery cell 10 during a second period (e.g., 20.0 to 20.1 seconds) in a fifth test in which the battery cell 10 is charged with a fifth current rate (e.g., 1.0C) at a fifth temperature starting from a time point when the SOC of the battery cell 10 is equal to a fifth state (e.g., 10%).

[0072] The fifteenth external variable may indicate the voltage of the first peak on the differential capacity curve of the battery cell 10 obtained from a sixth test in which the battery cell 10 is charged with a sixth current rate (e.g., 0.04C) at a sixth temperature starting from a time point when the SOC of the battery cell 10 is equal to a sixth SOC (e.g., 0%). When V, dV, and dQ are the voltage, the voltage change, and the capacity change of the battery cell 10 respectively, the differential capacity curve shows the correspondence between V and dQ / dV. The differential capacity curve may be referred to as a "V - dQ / dV curve". The first peak may be the peak with the minimum V among multiple peaks on the differential capacity curve.

[0073] The sixteenth external variable may indicate the difference between the voltage of the first peak and a reference voltage. The reference voltage is the voltage of the first peak of the differential capacity curve obtained when the battery cell 10 is at the beginning of life (BOL) and may be preset.

[0074] The first to nth expected data sets are respectively associated with the first to nth internal variables that are not observable outside the battery cell 10. That is, when the second index j is an integer from 1 to n, the jth expected data set is associated with the jth internal variable. Each internal variable indicates the internal chemical state of the battery cell 10. Each expected data set includes the same number of target values as a third number. For example, the third number of values preset for the jth internal variable are included in the jth expected data set.

[0075] In the specification, Y j represents the jth expected data set, and Y j (k) represents the kth target value of the jth expected data set.

[0076] Y1(k) to Y n (k) are the expected values of the first to nth internal variables when the battery cell 10 has a specific degradation state. X1(k) to X m (k) are the expected values of the first to mth external variables depending on Y1(k) to Y n (k) when the battery cell 10 has a specific degradation state.

[0077] The degradation state of the battery cell 10 varies depending on the environment in which the battery cell 10 is used, and each degradation state can be defined by a combination of the first internal variable to the nth internal variable. When a ≠ b, X1(a) to X m (a) and Y1(a) to Y n (a) are associated with the 'a' degradation state, while X1(b) to X m (b) and Y1(b) to Y n (b) are associated with the 'b' degradation state that is different from the 'a' degradation state of the battery cell 10.

[0078] For example, when n = 14, the first internal variable to the fourteenth internal variable can be defined as follows.

[0079] The first internal variable can be the conductivity of the positive electrode of the battery cell 10.

[0080] The second internal variable can be the ion diffusivity of the positive electrode active material of the battery cell 10.

[0081] The third internal variable can be the rate constant of the exchange current density of the positive electrode active material of the battery cell 10.

[0082] The fourth internal variable can be the ion diffusivity of the negative electrode active material of the battery cell 10.

[0083] The fifth internal variable can be the rate constant of the exchange current density of the negative electrode active material of the battery cell 10.

[0084] The sixth internal variable can be the tortuosity of the negative electrode of the battery cell 10. Tortuosity is the ratio of the distance that an ion travels from one point to another to the straight-line distance between the same two points.

[0085] The seventh internal variable can be the porosity of the negative electrode of the battery cell 10.

[0086] The eighth internal variable can be the ion concentration of the electrolyte of the battery cell 10.

[0087] The ninth internal variable can be the scale factor multiplied by the initial ionic conductivity of the electrolyte of the battery cell 10. The initial ionic conductivity can be a preset value indicating the ionic conductivity of the electrolyte of the battery cell 10 when the battery cell 10 is at BOL.

[0088] The tenth internal variable can be the scale factor multiplied by the initial ion diffusivity of the electrolyte of the battery cell 10. The initial ion diffusivity can be a preset value indicating the ion diffusivity of the electrolyte of the battery cell 10 when the battery cell 10 is at BOL.

[0089] The eleventh internal variable may be the cation transference number of the electrolyte of the battery cell 10. The transference number indicates the fractional contribution of cations (e.g., Li+) to the conductivity of the electrolyte.

[0090] The twelfth internal variable may be the lithium inventory loss (LLI) of the battery cell 10. The LLI indicates the lithium loss in the battery cell 10 compared to BOL.

[0091] The thirteenth internal variable may be the loss of active material (LAM) of the positive electrode of the battery cell 10. The LAM of the positive electrode indicates the loss of the positive electrode active material of the battery compared to BOL.

[0092] The fourteenth internal variable may be the LAM of the negative electrode of the battery cell 10. The LAM of the negative electrode indicates the loss of the negative electrode active material of the battery compared to BOL.

[0093] The first input data set to the m-th input data set and the first expected data set to the n-th expected data set may be preset according to the simulation results of multiple battery cells having the same electrochemical specifications as the battery cell 10 but different degradation states.

[0094] The control unit 130 may use the main multi-layer perceptron 200 stored in the memory unit 120 to determine the first reference value to the n-th reference value required for the modeling of the first sub multi-layer perceptron to the n-th sub multi-layer perceptron.

[0095] Reference Figure 2 , the main multi-layer perceptron 200 includes an input layer 201, a predetermined number of intermediate layers 202, and an output layer 203. In the main multi-layer perceptron 200, the number of nodes (also referred to as "neurons") included in each layer, the connections between the nodes, and the functions of each node included in each intermediate layer may be preset. Predefined training data may be used to set the weights of each connection.

[0096] The input layer 201 includes the first input node I1 to the m-th input node I m . The first input node I1 to the m-th input node I m are respectively associated with the first external variable to the m-th external variable. Each input value included in the i-th input data set X i is provided to the i-th input node I i .

[0097] The output layer 203 includes the first output node O1 to the n-th output node O n . The first output node O1 to the n-th output node O n are respectively associated with the first internal variable to the n-th internal variable.

[0098] When the first input data set X1 to the m-th input data set Xm The k-th input values of each of them are respectively input to the first input node I1 to the m-th input node I m When this happens, the first output data set Z1 to the n-th output data set Z n The k-th result values of each of them are respectively output from the first output node O1 to the n-th output node O n are output.

[0099] The control unit 130 can generate the first output data set Z1 to the n-th output data set Z by repeatedly inputting the m input values in the same order in the first input data set X1 to the m-th input data set X m (e.g., X1(k) to X m (k)) to the first input node I1 to the m-th input node I m Each output data set has the same number of result values as the third number. In the specification, Z n represents the j-th output data set, and Z j (k) represents the k-th result value of the j-th output data set. Among the first output data set to the n-th output data set, the n result values Z1(k) to Z j (k) can be arranged in the same order. n (k).

[0100] The control unit 130 can determine the first error factor to the n-th error factor based on the first output data set to the n-th output data set and the first expected data set to the n-th expected data set. That is, the control unit 130 can determine the j-th error factor by comparing the j-th output data set with the j-th expected data set. Specifically, the control unit 130 can use the following Equation 1 to determine the j-th error factor.

[0101] <Equation 1>

[0102]

[0103] In Equation 1, u is the third number, and F error_j is the j-th error factor. That is, the j-th error factor can be determined to be equal to the error ratio of the j-th output data set to the j-th expected data set.

[0104] The control unit 130 may determine the first reference value to the nth reference value by comparing each of the first error factor to the nth error factor with a threshold error factor. The threshold error factor may be a preset value, such as 3%. Specifically, when the jth error factor is less than the threshold error factor, the control unit 130 may determine the jth reference value to be equal to a first predetermined value. Conversely, when the jth error factor is equal to or greater than the threshold error factor, the control unit 130 may determine the jth reference value to be equal to a second predetermined value. The second predetermined value (e.g., 0.25) may be less than the first predetermined value (e.g., 0.50). After determining the first reference value to the nth reference value, the process using the main multi-layer perceptron 200 may be completed. Alternatively, each of the first reference value to the nth reference value may be preset to the first predetermined value or the second predetermined value. In this case, the process using the main multi-layer perceptron 200 may be omitted.

[0105] The control unit 130 performs a process for determining the first sub multi-layer perceptron to the nth sub multi-layer perceptron.

[0106] The control unit 130 determines the multiple correlation coefficients between each of the first expected data sets to the nth expected data sets and each of the first input data sets to the mth input data sets. The multiple correlation coefficient between the ith input data set and the jth expected data set may be determined according to Equation 2 below.

[0107] <Equation 2>

[0108]

[0109] In Equation 2, N is a third number, and r i,j is the multiple correlation coefficient between the ith input data set and the jth expected data set. When m multiple correlation coefficients are determined for the corresponding expected data set, a total of m×n multiple correlation coefficients may be determined.

[0110] A multiple correlation coefficient r i,j greater than the jth reference value indicates a high correlation between the ith external variable and the jth internal variable. When the absolute value of the multiple correlation coefficient r i,j is greater than the jth reference value, the control unit 130 may determine to use the ith input data set for modeling the jth sub multi-layer perceptron. When the absolute value of the multiple correlation coefficient r i,j is greater than the jth reference value, the control unit 130 may set the ith external variable as an effective external variable for the jth internal variable.

[0111] Conversely, a multiple correlation coefficient r i,j equal to or less than the jth reference value indicates a low correlation between the ith external variable and the jth internal variable. When the multiple correlation coefficient r i,jWhen the absolute value of is equal to or less than the j-th reference value, the control unit 130 may determine not to use the i-th input data set for modeling the j-th sub-multi-layer perceptron. That is, the control unit 130 may exclude the i-th external variable from the valid external variables of the j-th internal variable.

[0112] The control unit 130 may generate a first sub-multi-layer perceptron to an n-th sub-multi-layer perceptron respectively associated with the first internal variable to the n-th internal variable. That is, the j-th internal variable and the j-th sub-multi-layer perceptron are associated with each other. The j-th sub-multi-layer perceptron is used to estimate the value of the j-th internal variable.

[0113] Reference Figure 3 , the j-th sub-multi-layer perceptron 300 j includes an input layer 301 j , a predetermined number of intermediate layers 302 j and an output layer 303 j . The functions of each node included in each intermediate layer 302 j can be preset. The output layer 303 j has a single output node associated with the j-th internal variable.

[0114] The control unit 130 may generate input nodes of the input layer 301 of the j-th sub-multi-layer perceptron in the same number as the number of the set of valid external variables for the j-th internal variable. j Figure 3 Exemplarily, each of the first external variable, the third external variable, and the fifth external variable is set as a valid external variable of the j-th internal variable.

[0115] The input layer 301 of the j-th sub-multi-layer perceptron j has three input nodes I j1 to I j3 . The first input data set, the third input data set, and the fifth input data set respectively associated with the three external variables are provided as training data to each of the three input nodes I j1 to I j3 , and the learning of the j-th sub-multi-layer perceptron 300 j is performed by comparing the result value of the data set W j obtained from the output layer 303 of the j-th sub-multi-layer perceptron with the target value of the j-th expected data set. j

[0116] In the learning of the j-th sub-multi-layer perceptron 300 jAfter the learning is completed, the control unit 130 can use the sensing unit 140 to measure the voltage, current and temperature of the battery cell 10. The control unit 130 can determine the values ​​of the three external variables based on the sensing data from the sensing unit 140. Subsequently, the control unit 130 can input the values ​​of the three external variables into the j-th sub-multilayer perceptron 300 respectively. j The three input nodes I j1 to I j3 To obtain the j-th multilayer perceptron 300 j The output layer 303 j The corresponding result value is the estimated value of the jth internal variable associated with the degradation state corresponding to the values ​​of the three external variables. When the estimated value of the jth internal variable is outside the predetermined jth safety range, the control unit 130 can control the switch 20 to enter the disconnected working state to protect the battery cell 10.

[0117] Figure 4 This is an example of how Figure 1 Flow chart of a first method performed by the battery management device 100. Figure 4 The first method is a data filtering method for extracting at least one of the first observation data set to the pth observation data set as an input data set.

[0118] refer to Figures 1 to 4 In step S410 , the control unit 130 stores the first to p-th observed data sets and the first to n-th expected data sets in the memory unit 120 .

[0119] In step S420 , the control unit 130 sets each of the first index q and the second index m to 1.

[0120] In step S430, the control unit 130 determines the multiple correlation coefficient between the qth observation data set and the q+1th observation data set. For example, when q=1, the multiple correlation coefficient between the first observation data set and the second observation data set is determined. The multiple correlation coefficient between the qth observation data set and the q+1th observation data set can be determined according to the following equation 3. In the specification, P q represents the qth observation data set.

[0121] <Equation 3>

[0122]

[0123] In Equation 3, N is the third number, P q (k) is the kth input value of the qth observation data set, P q+1 (k) is the kth input value of the q+1th observation data set, and R q,q+1is the multiple correlation coefficient between the q-th observed data set and the (q + 1)-th observed data set.

[0124] In step S440, the control unit 130 determines whether the absolute value of the multiple correlation coefficient R q,q+1 is less than a predetermined filtering value (e.g., 0.75). A value of "yes" for step S440 indicates that the correlation between the q-th observed data set and the (q + 1)-th observed data set is not high enough. Therefore, it is highly necessary to set the q-th observed data set as the input data set. When the value of step S440 is "yes", step S450 is executed. On the contrary, a value of "no" for step S440 indicates that the correlation between the q-th observed data set and the (q + 1)-th observed data set is high enough. Therefore, the need to set the q-th observed data set as the input data set is low. When the value of step S440 is "no", step S470 is executed.

[0125] In step S450, the control unit 130 sets the q-th observed data set as the m-th input data set. In an example where q = 1 and m = 1, the first input data set is equal to the first observed data set. In another example where q = 3 and m = 2, the second input data set is equal to the third observed data set.

[0126] In step S460, the control unit 130 increments the second index m by 1.

[0127] In step S470, the control unit 130 determines whether the first index q is equal to p - 1. p is the number of observed data sets. When the value of step S470 is "no", step S480 is executed. When the value of step S470 is "yes", step S490 is executed.

[0128] In step S490, the control unit 130 sets the (q + 1)-th observed data set as the m-th input data set. Through step S490, among the first observed data set to the p-th observed data set, at least the p-th observed data set is set as the input data set.

[0129] Through Figure 4 steps S430 to S450, when two observed data sets have a strong correlation, only one of the two observed data sets is extracted as the input data set. Compared with the case where all observed data sets are used as input data sets without exception, the amount of computation required for modeling the first sub-multi-layer perceptron to the n-th sub-multi-layer perceptron can be reduced and the computation speed can be increased.

[0130] After determining the first input data set to the m-th input data set by the method of Figure 4 , the control unit 130 may execute the method of Figure 5 . Alternatively, the control unit 130 may not execute according to Figure 4In the case of the method, the first observation dataset to the p-th observation dataset are used as the first input dataset to the m-th input dataset, and in this case, p = m.

[0131] Figure 5 is an exemplary flowchart of a second method that can be executed by Figure 1 the battery management device 100.

[0132] Referring to Figures 1 to 5 , in step S510, the control unit 130 uses the main multi-layer perceptron 200 to obtain the first output dataset to the n-th output dataset from the first input dataset to the m-th input dataset.

[0133] In step S520, the control unit 130 sets the third index j to 1.

[0134] In step S530, the control unit 130 determines the j-th error factor based on the j-th output dataset and the j-th desired dataset.

[0135] In step S540, the control unit 130 determines whether the j-th error factor is less than the threshold error factor. When the value of step S540 is "yes", step S550 is executed. When the value of step S540 is "no", step S560 is executed.

[0136] In step S550, the control unit 130 sets the j-th reference value to be equal to the first predetermined value.

[0137] In step S560, the control unit 130 sets the j-th reference value to be equal to the second predetermined value.

[0138] In step S570, the control unit 130 determines whether the third index j is equal to n. n is the number of desired datasets. When the value of step S570 is "no", step S580 is executed.

[0139] In step S580, the control unit 130 increments the second index j by 1. After step S580, the process returns to step S530.

[0140] After determining the first reference value to the n-th reference value by the method of Figure 5 , the control unit 130 may execute the method of Figure 6 . Alternatively, when the first reference value to the n-th reference value is preset and stored in the memory unit 120 as described above, the control unit 130 may execute the method of Figure 5 without executing the method of Figure 6 .

[0141] Figure 6 is an exemplary illustration of what can be done by Figure 1Flowchart of the third method executed by the battery management device 100. According to Figure 6 The third method according to

[0142] Reference Figures 1 to 6 , in step S610, the control unit 130 sets each of the third index j and the fourth index i to 1.

[0143] In step S620, the control unit 130 determines the multiple correlation coefficient between the i-th input data set and the j-th expected data set.

[0144] In step S630, the control unit 130 determines whether the absolute value of the multiple correlation coefficient is greater than the j-th reference value. When the value of step S630 is "yes", step S640 is executed. When the value of step S630 is "no", step S650 is executed.

[0145] In step S640, the control unit 130 sets the i-th external variable as the valid external variable for the j-th internal variable.

[0146] In step S650, the control unit 130 determines whether the fourth index i is equal to m. m is the number of input data sets. When the value of step S650 is "no", step S660 is executed. When the value of step S650 is "yes", step S670 is executed.

[0147] In step S660, the control unit 130 increments the fourth index i by 1. After step S660, the process returns to step S620.

[0148] In step S670, the control unit 130 determines whether the third index j is equal to n. When the value of step S670 is "no", step S680 is executed.

[0149] In step S680, the control unit 130 increments the third index j by 1 and sets the fourth index i to 1. After step S680, the process returns to step S620.

[0150] The value of step S650 being "yes" indicates that the setting of the valid external variable for the j-th internal variable is completed. Whenever the value of step S650 is "yes", the control unit 130 may generate the j-th sub-multilayer perceptron for estimating the value of the j-th internal variable.

[0151] The embodiments of the present disclosure described above can be implemented not only by devices and methods, but also by a program that executes functions corresponding to the configurations of the embodiments of the present disclosure or a recording medium on which the program is recorded, and such implementations can be easily achieved by those skilled in the art based on the disclosure of the embodiments described above.

[0152] Although the present disclosure has been described above with respect to a limited number of embodiments and drawings, the present disclosure is not limited thereto, and it is obvious to those skilled in the art that various modifications and changes can be made within the technical scope of the present disclosure and the equivalent scope of the appended claims.

[0153] Additionally, since those skilled in the art can make many substitutions, modifications, and changes to the present disclosure described above without departing from the technical scope of the present disclosure, the present disclosure is not limited by the above embodiments and drawings, and some or all of the embodiments can be selectively combined to allow various modifications.

Claims

1. A battery management device, comprising: A memory unit configured to store a first to a p-th observed data set and a first to an n-th desired data set, where each of p and n is an integer of 2 or greater; and A control unit operably coupled to the memory unit, wherein the first to the p-th observed data sets are respectively associated with p external variables observable outside the battery cell, each of the first to the p-th observed data sets includes a predetermined number of input values, the first to the n-th desired data sets are respectively associated with first to n-th internal variables, where the first to the n-th internal variables depend on the internal chemical state of the battery cell and are not observable outside the battery cell, each of the first to the n-th desired data sets includes the same number of target values as the predetermined number, and the control unit is configured to: Set m observed data sets extracted from the first to the p-th observed data sets through data filtering as the first to the m-th input data sets, where m is an integer greater than or equal to 1 and less than or equal to p, and Based on the first to the m-th input data sets and the first to the n-th desired data sets, set at least one of the first to the m-th external variables as an effective external variable for each of the first to the n-th internal variables, and where the first to the m-th external variables are m external variables among the p external variables and are associated with the m observed data sets, where the control unit is configured to: When q is an integer from 1 to p - 1, determine the multiple correlation coefficient between the q-th observed data set and the (q + 1)-th observed data set, and When the absolute value of the multiple correlation coefficient is less than a predetermined filtering value, set the q-th observed data set as one of the first to the m-th input data sets.

2. The battery management device according to claim 1, wherein The memory unit is further configured to store a main multi-layer perceptron that defines the correspondence between the first to the m-th external variables and the first to the n-th internal variables, and the control unit is configured to: By providing the first to the m-th input data sets to the first to the m-th input nodes included in the input layer of the main multi-layer perceptron, obtain the first to the n-th output data sets from the first to the n-th output nodes included in the output layer of the main multi-layer perceptron, where each of the first to the n-th output data sets includes the same number of result values as the predetermined number, Determine the first to the n-th error factors based on the first to the n-th output data sets and the first to the n-th desired data sets, and The first reference value to the nth reference value is determined by comparing each of the first error factor to the nth error factor with a threshold error factor.

3. The battery management device according to claim 2, wherein, The control unit is configured to: when j is an integer from 1 to n, determine the jth error factor to be equal to the error ratio between the jth output data set and the jth desired data set.

4. The battery management device according to claim 3, wherein, The control unit is configured to: when the jth error factor is less than the threshold error factor, set the jth reference value to be equal to a first predetermined value.

5. The battery management device according to claim 4, wherein, The control unit is configured to: when the jth error factor is equal to or greater than the threshold error factor, set the jth reference value to be equal to a second predetermined value, and wherein the second predetermined value is less than the first predetermined value.

6. The battery management device according to claim 2, wherein, The control unit is configured to: when i is an integer from 1 to m and j is an integer from 1 to n, determine whether to set the ith external variable among the first external variable to the mth external variable as a valid external variable for the jth internal variable, based on the ith input data set, the jth desired data set, and the jth reference value, and when the ith external variable is set as a valid external variable for the jth internal variable, use the ith input data set as training data to learn a sub-multi-layer perceptron associated with the jth internal variable.

7. The battery management device according to claim 6, wherein, The control unit is configured to: determine a multiple correlation coefficient between the ith input data set and the jth desired data set, and when the absolute value of the multiple correlation coefficient is greater than the jth reference value, set the ith external variable as a valid external variable for the jth internal variable.

8. The battery management device according to claim 7, wherein, The control unit is configured to: when the absolute value of the multiple correlation coefficient is equal to or less than the jth reference value, set the ith external variable as an invalid external variable for the jth internal variable.

9. A battery pack including the battery management device according to any one of claims 1 to 8.

10. A battery management method, comprising: storing a first observed data set to a pth observed data set and a first desired data set to an nth desired data set, wherein each of p and n is an integer of 2 or greater, the first observed data set to the pth observed data set are respectively associated with p external variables observable outside the battery cell, each of the first observed data set to the pth observed data set includes a predetermined number of input values, the first desired data set to the nth desired data set are respectively associated with a first internal variable to an nth internal variable unobservable outside the battery cell, the first internal variable to the nth internal variable depend on the internal chemical state of the battery cell, and each of the first desired data set to the nth desired data set includes the same number of target values as the predetermined number; setting m observed data sets extracted from the first observed data set to the pth observed data set through data filtering as a first input data set to an mth input data set, wherein m is an integer greater than or equal to 1 and less than or equal to p; and Based on the first input data set to the m-th input data set and the first expected data set to the n-th expected data set, at least one of the first external variable to the m-th external variable is set as the effective external variable for each of the first internal variable to the n-th internal variable, wherein, the first external variable to the m-th external variable are m external variables among the p external variables and associated with the m observation data sets, wherein, setting the m observation data sets as the first input data set to the m-th input data set includes: when q is an integer from 1 to p - 1, determining the multiple correlation coefficient between the q-th observation data set and the (q + 1)-th observation data set; and when the absolute value of the multiple correlation coefficient is less than a predetermined filtering value, setting the q-th observation data set as one of the input data sets in the first input data set to the m-th input data set.

Images