KR102472040B1 - The Battery Characteristic Measuring System and Method Based on Learned Inference Model - Google Patents

Abstract

The present invention relates to a battery characteristic derivation system based on a learned inference model. More specifically, the battery characteristic derivation system operates for a battery provided in a vehicle equipped with a powertrain that uses electricity as power, measures voltage, current, and temperature of each cell comprised in the battery, and measures impedance information according to the frequency of the cell using an inference model learned based on the voltage, current, and temperature. Based on the impedance information measured in the plurality of cells comprised in the battery, the temperature of each of the plurality of cells and defective cell can be diagnosed, and the measurement can be performed without a separate measuring device and input signal source. Accordingly, the impedance information of a corresponding cell can be measured even while driving.

Description

Battery characteristic measuring system and method based on learned inference model {The Battery Characteristic Measuring System and Method Based on Learned Inference Model}

The present invention is a battery characteristic derivation system based on a learned inference model, and more specifically, operates for a battery provided in a vehicle equipped with a powertrain using electricity as power, and in each cell included in the battery The voltage, current, and temperature of the battery can be measured, and impedance information according to the frequency of the cell can be measured using the inference model learned based on this, based on the impedance information measured from a plurality of cells included in the battery. Based on the learned inference model, which can diagnose the temperature and defective cells of each of the plurality of cells, and can be measured without a separate measuring device and input signal source, the impedance information of the corresponding cell can be measured even while driving It relates to a battery characteristic derivation system.

Recently, the transition from an internal combustion engine vehicle to an electric vehicle is gradually accelerating, and along with this, research on motors and batteries used in electric vehicles is being actively conducted. In particular, the battery in an electric vehicle is an important part that greatly affects the mileage and output of the electric vehicle, and many efforts are being made to install a battery with a large capacity in a limited space called a vehicle.

On the other hand, as described above, along with research on installing a battery with the largest capacity in a limited space, many studies on diagnosing and managing batteries in order to increase the lifespan and output efficiency of batteries disposed in vehicles have been conducted. Considering that the life expectancy of existing internal combustion engine vehicles is more than 10 years, the demand for batteries for electric vehicles that can be used for a long time is reasonable. skill is essential.

Most of the electric vehicle batteries currently being developed and marketed are lithium-ion batteries, which are secondary batteries. In the case of lithium-ion batteries, they use liquid electrolytes and have the advantage of good energy efficiency. . In order to compensate for these disadvantages, it is important to accurately diagnose the battery and quickly diagnose the defective cell, and as a conventional technique for diagnosing the battery, as in Korean Patent Publication No. 10-2017-0035229, it is included in electric vehicles. There is a technology related to a battery performance management system and method using an electric vehicle charging station, such as a battery diagnosis technology for diagnosing the state of a battery pack and Korean Patent Publication No. 10-2021-0119329.

On the other hand, in order to accurately diagnose the individual state of each cell included in the battery, it is preferable to measure impedance information according to the frequency of the corresponding cell, but the conventional battery impedance information measurement technology connects an input signal source to the battery, Impedance information according to frequency had to be measured while modulating the frequency of the input signal, and the vehicle could not be operated with a measuring device capable of measuring the impedance. In addition, there was difficulty in measuring accurate impedance information due to noise generated from power trains such as motors and generators or other vehicles themselves during driving. Therefore, by measuring the voltage, current, and temperature of the current electric vehicle battery even while driving, impedance information according to frequency can be measured for each cell of the battery, and the state of the battery can be diagnosed in real time even when the electric vehicle is running. technology is required.

Republic of Korea Patent Publication No. 10-2017-0035229 (2017.03.30.) Republic of Korea Patent Publication No. 10-2021-0119329 (2021.10.05.)

The present invention is a battery characteristic derivation system based on a learned inference model, and more specifically, operates for a battery provided in a vehicle equipped with a powertrain using electricity as power, and in each cell included in the battery The voltage, current, and temperature of the battery can be measured, and impedance information according to the frequency of the cell can be measured using the inference model learned based on this, based on the impedance information measured from a plurality of cells included in the battery. Based on the learned inference model, which can diagnose the temperature and defective cells of each of the plurality of cells, and can be measured without a separate measuring device and input signal source, the impedance information of the corresponding cell can be measured even while driving It is an object of the present invention to provide a battery characteristic derivation system.

In order to solve the above problems, an embodiment of the present invention is a battery characteristic derivation system based on a learned inference model provided in a battery installed in a vehicle equipped with a powertrain using electricity as power and operating, the battery An output signal measurement unit for measuring an output signal including voltage information and current information output from a cell included in the; a temperature measuring unit for measuring the temperature of the cell; and an impedance information calculation unit that inputs input information including the output signal and temperature of the cell to the learned artificial neural network-based inference model and calculates impedance information, wherein the input information is obtained during actual driving of the vehicle. It includes time-series data of voltage information and current information according to time in a certain sliding window section, wherein the impedance information is impedance information for each frequency.

In one embodiment of the present invention, the input information may further include time-series data of temperature according to time in a certain sliding window section obtained during actual driving of the vehicle.

In one embodiment of the present invention, the battery management system is composed of one master BMS and a plurality of slave BMS, the master BMS is connected to a plurality of slave BMS, the slave BMS is a plurality of battery modules constituting the battery an output signal measuring unit connected to each of the output signals output from one or more cells included in the battery module; a temperature measuring unit for measuring the temperature in the corresponding cell; and an impedance information calculation unit that calculates impedance information in the corresponding cell, wherein the master BMS receives a plurality of impedance information from each of the plurality of slave BMSs and a battery analysis unit that analyzes the state of the corresponding battery. can do.

In one embodiment of the present invention, the inference model includes a CNN-based artificial neural network module, and through the artificial neural network module, an input in a certain sliding window section received from the output signal measurement unit and the temperature measurement unit Impedance information according to the frequency of the output signal output to the cell in the sliding window period may be calculated by machine learning the information.

In one embodiment of the present invention, the inference model includes an LSTM-based artificial neural network module, and through the artificial neural network module, an input in a certain sliding window section received from the output signal measurement unit and the temperature measurement unit Impedance information according to the frequency of the output signal output to the cell in the sliding window period may be calculated by machine learning the information.

In order to solve the above problems, an embodiment of the present invention is a battery characteristic derivation method based on a learned inference model operating in a battery installed in a vehicle equipped with a powertrain using electricity as power, which is included in the battery An output signal measurement step of measuring an output signal including voltage information and current information output from a cell to be; a temperature measurement step of measuring the temperature in the cell; and an impedance information calculation step of calculating impedance information by inputting input information including the output signal and temperature of the cell to the learned artificial neural network-based inference model, wherein the input information is obtained during actual driving of the vehicle. It includes time-series data of voltage information and current information according to time in a certain sliding window section, wherein the impedance information is impedance information for each frequency.

According to an embodiment of the present invention, impedance information can be measured for all cells included in a battery of a vehicle, so that defective cells having problems such as poor charging or leakage current can be easily found. Through this, by quickly removing the defective cells, the failure rate of the battery can be reduced and the lifespan of the battery can be extended.

According to an embodiment of the present invention, as a system installed in a BMS system in a vehicle, it is possible to achieve an effect of measuring impedance information of each cell included in a corresponding battery even when charging or driving.

According to an embodiment of the present invention, it is possible to generate a lot of learning data through a battery model that calculates the impedance of a corresponding battery according to a power pattern used in a vehicle and environmental conditions, and an inference model learned from the a lot of learning data By using the battery, it is possible to achieve an effect of measuring impedance information of each cell included in the battery in real time.

According to an embodiment of the present invention, by using the learned inference model, the effect of measuring more accurate voltage and current for a corresponding cell by minimizing the influence of signals generated by complex factors inside and outside the vehicle can be exerted. can

According to an embodiment of the present invention, by measuring the voltage and current of each cell included in the battery, as well as the impedance information according to the temperature of the cell, the effect of performing a more accurate diagnosis on the cell can exert

1 schematically illustrates a configuration of a system for deriving battery characteristics and a method for deriving battery characteristics according to an embodiment of the present invention.
2 schematically illustrates the configuration of a battery installed in an electric vehicle and a battery management system mounted on the battery according to an embodiment of the present invention.
3 schematically illustrates a process of calculating impedance information according to an embodiment of the present invention.
4 schematically illustrates a battery model generating learning data according to an embodiment of the present invention.
5 schematically illustrates a process of calculating impedance information of a corresponding cell using a learned inference model according to an embodiment of the present invention.
6 schematically illustrates a graph of an intercept frequency of a battery cell according to an equivalent circuit and temperature of a battery cell according to an embodiment of the present invention.
7 schematically illustrates a graph showing a relationship between temperature and impedance information according to an embodiment of the present invention.
8 schematically illustrates impedance information of a battery cell according to the frequency and temperature of an input signal according to an embodiment of the present invention.

In the following, various embodiments and/or aspects are disclosed with reference now to the drawings. In the following description, for purposes of explanation, numerous specific details are set forth in order to facilitate a general understanding of one or more aspects. However, it will also be appreciated by those skilled in the art that such aspect(s) may be practiced without these specific details. The following description and accompanying drawings describe in detail certain illustrative aspects of one or more aspects. However, these aspects are exemplary and some of the various methods in principle of the various aspects may be used, and the described descriptions are intended to include all such aspects and their equivalents.

Moreover, various aspects and features will be presented by a system that may include a number of devices, components and/or modules, and the like. It should also be noted that various systems may include additional devices, components and/or modules, and/or may not include all of the devices, components, modules, etc. discussed in connection with the figures. It must be understood and recognized.

"Example", "example", "aspect", "exemplary", etc., used herein should not be construed as preferring or advantageous to any aspect or design being described over other aspects or designs. . The terms '~unit', 'component', 'module', 'system', 'interface', etc. used below generally mean a computer-related entity, and for example, hardware, hardware It may mean a combination of and software, software.

Also, the terms "comprises" and/or "comprising" mean that the feature and/or element is present, but excludes the presence or addition of one or more other features, elements and/or groups thereof. It should be understood that it does not.

In addition, terms including ordinal numbers, such as first and second, may be used to describe various components, but the components are not limited by the terms. These terms are only used for the purpose of distinguishing one component from another. For example, a first element may be termed a second element, and similarly, a second element may be termed a first element, without departing from the scope of the present invention. The terms and/or include any combination of a plurality of related recited items or any of a plurality of related recited items.

In addition, in the embodiments of the present invention, unless otherwise defined, all terms used herein, including technical or scientific terms, are generally understood by those of ordinary skill in the art to which the present invention belongs. has the same meaning as Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning in the context of the related art, and unless explicitly defined in the embodiments of the present invention, an ideal or excessively formal meaning not be interpreted as

1 schematically illustrates a configuration of a battery characteristic derivation system 1000 and a battery characteristic derivation method according to an embodiment of the present invention.

As shown in FIG. 1, a battery characteristic derivation system 1000 based on a learned inference model 1400 that is provided in a battery installed in a vehicle equipped with a powertrain using electricity as power and operates, which is included in the battery an output signal measurement unit 1100 for measuring an output signal including voltage information and current information output from the cell; a temperature measurement unit 1200 for measuring the temperature in the cell; and an impedance information calculation unit 1300 that inputs input information including an output signal and temperature of the cell to the learned artificial neural network-based inference model 1400 and calculates impedance information, wherein the input information includes Includes time-series data of voltage information and current information according to time in a certain sliding window section obtained during actual driving of the vehicle, the impedance information is impedance information for each frequency,

The input information further includes time-series data of temperature according to time in a predetermined sliding window section obtained during actual driving of the vehicle.

Schematically, FIG. 1 (a) shows the configuration of a battery characteristic derivation system 1000, and FIG. 1 (b) shows a battery characteristic derivation method.

Specifically, the battery characteristic derivation system 1000 of the present invention is provided and operated in a battery installed in a vehicle equipped with a power train using electricity as power. As a vehicle equipped with a powertrain that uses electricity as power, an electric vehicle (EV) that uses only 100% of the power of the powertrain mounted on the vehicle (Electric Vehicle, EV); Mild Hybrid Electric Vehicles (MHEVs) that use electricity for only a portion of the power of a powertrain mounted on a vehicle; and Plug-in Hybrid Electric Vehicles (PHEVs); A fuel cell electric vehicle (FCEV), which drives a motor using electricity obtained by reacting hydrogen with oxygen in the air; Include etc. In other words, the battery to be described later corresponds to a battery installed in a vehicle including the above-described EV, MHEV, PHEV, and FCEV, and driven by electricity.

The battery characteristic derivation system 1000 of the present invention is connected to each cell included in the battery installed in the vehicle, measures voltage information, current information, and temperature in the corresponding cell, and the voltage information and current information Impedance information of the corresponding cell is calculated through , and temperature.

In more detail, the output signal measuring unit 1100 of the derivation system 1000 is connected to the positive and negative electrodes of the battery cell to measure an output signal including voltage information and current information of the corresponding battery cell. S10) is performed. The voltage information includes time-series data corresponding to the output voltage of the battery cell over time, and the current information includes time-series data corresponding to the output current of the battery cell over time.

In addition, the temperature measurement unit 1200 of the derivation system 1000 is directly or indirectly connected to the battery cell to perform a temperature measurement step (S20) of measuring the temperature of the battery cell. The temperature corresponds to the temperature of the battery cell over time, and according to an embodiment of the present invention, the temperature of the battery cell is directly measured by a temperature sensor directly attached to the battery cell, or According to another embodiment, a metal material attached to the battery cell; Alternatively, the temperature of the battery cell may be indirectly estimated by measuring the temperature of the temperature transmitting material.

In the prior art, there were many cases in which the impedance of the battery cell was calculated using only the voltage and current information of the battery cell, but the battery currently used in vehicles is a device that generates electricity through a chemical reaction, and in this process, heat Because it accompanies entry and exit, battery characteristics and temperature are closely related. Therefore, in the present invention, the temperature of the corresponding cell is measured together to calculate the impedance information of the battery cell, and the temperature of the battery cell will be described in more detail later in the description of FIGS. 5 to 8 .

The impedance information calculation unit 1300 of the derivation system 1000 includes voltage information, current information, and temperature of the corresponding battery cell measured by the output signal measurement unit 1100 and the temperature measurement unit 1200. Based on the input information, an impedance information calculation step (S20) of calculating impedance information according to the frequency of the corresponding battery cell is performed. In the process of calculating the impedance information, the impedance information calculation unit 1300 calculates the impedance information using the inference model 1400. On the other hand, as an embodiment of the present invention, the inference model 1400, as shown in FIG. 1, the output signal measurement unit 1100, the temperature measurement unit 1200, and the impedance information calculation unit ( 1300), and in another embodiment of the present invention, the inference model 1400 includes the output signal measurement unit 1100, the temperature measurement unit 1200, and the impedance information calculation unit. It may be located in a separate space separated from the space where the 1300 is located. The space means a hardware configuration that means one device or one module, and may mean a software configuration that performs separate software functions. A more detailed description of the impedance information calculation step (S20) will be described later in the description of FIGS. 5 to 8. The above-described battery characteristic derivation system 1000 preferably functions together with a battery management system described later, but as another embodiment of the present invention, a system on chip (SoC) that performs the functions of the present invention or is programmed to perform It is implemented as a chip device or module and can be installed and operated in the BMS system of the battery.

2 schematically illustrates the configuration of a battery installed in an electric vehicle and a battery management system mounted on the battery according to an embodiment of the present invention.

As shown in Figure 2, the battery management system is composed of one master BMS and a plurality of slave BMS, the master BMS is connected to a plurality of slave BMS, the slave BMS is a plurality of battery modules constituting the battery an output signal measuring unit 1100 connected to each of the battery modules and measuring an output signal output from one or more cells included in the battery module; a temperature measuring unit 1200 that measures the temperature in the corresponding cell; and an impedance information calculation unit 1300 that calculates impedance information in a corresponding cell, wherein the master BMS receives a plurality of impedance information from each of the plurality of slave BMSs and analyzes the state of the corresponding battery. includes;

Specifically, the battery characteristic derivation system 1000 of the present invention is provided and operated in a battery composed of a plurality of battery cells. In general, a battery is composed of cells, modules, and packs. First, a battery cell is a basic unit of a lithium ion battery that can be used while charging and discharging electric energy, and is manufactured by putting a positive electrode, a negative electrode, a separator, and an electrolyte in a rectangular or cylindrical aluminum case. The battery cell must have a high capacity per unit volume or per unit weight to achieve maximum performance in a limited space, require a much longer lifespan than batteries for general mobile devices, withstand impact transmitted during driving, and It is designed to have reliability and stability enough to operate normally at high temperatures below the standard and at low temperatures above a certain standard.

A plurality of cells as described above are bundled in a certain number so as to be protected from external shocks such as heat and vibration, and put into a frame to manufacture a single battery assembly, which is called a battery module. One battery module may be composed of about 6 to about 10 cells, and recently, about 12 to about 48 cells connected in series or parallel may be used. A battery module composed of 12 battery cells usually has a capacity of 2 to 3 kWh.

Several battery modules as described above are gathered and a battery pack is formed by adding a battery management system (BMS) and a cooling device. One battery pack contains 8 to 40 battery modules, and each battery module is connected in series or parallel, or in a mixed form. Usually, one battery pack is installed in one vehicle. As described above, through the modularized battery pack, maintenance and repair of the battery can be facilitated, and a more efficient battery with reduced weight and volume can be used.

On the other hand, FIG. 2 shows a battery in which a total of 30 battery modules are connected in series as an embodiment of the present invention, and each battery module preferably includes a plurality of battery cells as described above. It shows a configuration in which one battery cell is included in one battery module. This is only one embodiment for explanation, but is not limited thereto, and the present invention is also applied to battery modules and battery packs composed of more cells or having a mixture of series and parallel connections.

As shown in FIG. 2 , the battery characteristic derivation system 1000 of the present invention preferably operates together with a BMS included in a battery pack of a vehicle. BMS is a system that controls the state of charge, discharge state, and remaining amount of the battery. It controls the battery to create an optimal operating environment in conjunction with other control systems inside the vehicle, and operates with one master BMS and multiple slave BMS. It consists of The plurality of slave BMSs are individually connected to the master BMS, and each of the plurality of slave BMSs is connected to each battery module of the vehicle. Each slave BMS includes the output signal measurement unit 1100, the temperature measurement unit 1200, the impedance information calculation unit 1300, and the inference model 1400 described in FIG. 1, and the output included in each slave BMS. The signal measurement unit 1100, the temperature measurement unit 1200, and the impedance information calculation unit 1300 measure the voltage information, current information, and temperature of the corresponding battery cell to calculate the impedance information of the corresponding battery cell.

More specifically, according to an embodiment of the present invention, each of the plurality of slave BMSs is connected to each of a plurality of cells in the battery module to measure voltage, current, temperature, and the like of each of the plurality of cells. . For example, when battery module #1 includes battery cell #1 to battery cell #8, slave BMS #1 to slave BMS #8 are connected to the battery module #1, and the slave BMS #1 to slave BMS #8 is connected to battery cell #1 to #8, respectively, and can measure voltage information, current information, and temperature of each corresponding battery cell. In this case, since the impedance calculation unit and the unit performing calculations such as the inference model 1400 perform calculations on only one cell, accurate measurement of individual cells is possible, but the total number of battery cells is greater than a certain level. If it loses, the total system's computational load increases.

Alternatively, according to another embodiment of the present invention, one slave BMS is connected to one battery module, and the slave BMS is connected to each cell included in the corresponding battery module, and the cell included in each corresponding battery module Each voltage, current, and temperature can be individually measured and then aggregated. In other words, for example, when the battery module #1 includes battery cell #1 to battery cell #8, only the slave BMS #1 is connected to the battery module #1, and the slave BMS #1 is the battery. Cells #1 to #8 are connected with wires, respectively, so that voltage information, current information, and temperature of each cell can be measured. The measured voltage information, current information, and temperature of each cell are input to the inference model 1400 of the slave BMS#1 to calculate impedance information of each cell. In this case, since the impedance calculating unit and the unit performing the same calculation as the inference model 1400 perform the calculation for one module, even when the number of battery cells is large, calculation is performed for each module, thereby reducing the calculation load of the system. There are advantages to reducing it.

Hereinafter, for convenience of description, one battery module will be described below on the basis of including only one battery cell, but as described above, the present invention is not limited thereto.

Meanwhile, when impedance information of cells corresponding to each of the plurality of slave BMSs is calculated, the calculated impedance information is transmitted to the master BMS. The master BMS can diagnose a current state of a battery and a defective cell through the impedance information transmitted from a plurality of slave BMSs. The defective cell includes a cell that does not charge properly or is overcharged, does not output normally, or generates leakage current. Measures such as exchanging or repairing the cell by finding such a defective cell , and through this, it is possible to exert an effect of reducing the failure rate of the battery and extending its lifespan. Additionally, the present invention can measure the impedance information of each battery cell in real time even while driving, and synthesize the diagnosis results of each cell included in the battery pack through the above-described BMS system to determine the state of charge of the corresponding battery pack ( State of Charge (SOC), temperature, instantaneous power, lifespan, estimated driving time, driving distance, etc. may be calculated and provided to the user.

3 schematically illustrates a process of calculating impedance information according to an embodiment of the present invention.

Specifically, as shown in FIG. 3 , a composite signal obtained by combining a plurality of signals generated by a plurality of factors is measured inside the vehicle. The composite signal varies depending on various environmental conditions, and the signal generated from the power train occupies the largest portion.

More specifically, when the load of the motor changes while driving, that is, when the vehicle accelerates or decelerates, a signal of 1 mHz to 100 Hz is generated from the motor, and a DC to DC converter of 12 volts (V) or driving the motor or a signal of 2kHz to 300kHz is generated from the inverter that regeneratively brakes. In addition, a signal of 0.5 kHz to 15 kHz is generated in the process of crossing the slot and pole of the motor, that is, in the process of operating the motor. On the other hand, in addition to the frequency generated by the power train, a signal including noise of 1 Hz to 2 kHz generated by the influence of mechanical vibration and wind or noise of 60 Hz to 600 Hz generated during charging is also generated. The plurality of frequency sources of the above-described composite signal correspond to an example derived from experimental data, and a signal in which signals generated from sources other than the above-described sources may be mixed may be further included. Referring to the frequency graph according to the traveling speed shown in FIG. 3 , it can be seen that the frequency of the composite signal varies according to the traveling speed of the vehicle.

When the output signal measurement unit 1100 of the present invention measures the voltage information and current information of the cell to measure the impedance information of the battery cell, the voltage information and current information in which the above-described composite signal is mixed is measured. That is, it becomes difficult to obtain pure voltage information and current information, and for this reason, the conventional technique of measuring impedance information of a battery can be applied only to a battery of a stopped vehicle in order to minimize the above-described composite signal. As a result, the battery state could be provided to the user only through an indirect method, and the battery state according to the ever-changing driving environment could not be accurately diagnosed. For example, even in the same environment where you are running at an average of 80 km/h, driving in a poor road environment; Environments where there is a lot of acceleration and deceleration due to irregular traffic; Since the load applied to the battery is different in an environment where the battery is driven at the same speed and on a highway, measuring the impedance information of the battery while driving can more accurately diagnose the current state of the battery and provide the user with more accurate expected driving. It is possible to exert an effect capable of providing a possible distance and the like.

However, the present invention uses the reasoning model 1400 that has learned a plurality of output data output from the battery model, even if the voltage information and current information measured during driving and the above-described composite signal are mixed, pure voltage information therein. And current information can be obtained, through which impedance information of the corresponding battery cell can be measured, and the state of the corresponding cell can be diagnosed through the measured impedance information. A more detailed description of the reasoning model 1400 will be described later with reference to FIGS. 4 and 5 .

Although FIG. 3 shows a SoC-type device that performs the functions of the present invention, this corresponds to an embodiment of the present invention, and as another embodiment of the present invention, the battery characteristic derivation system 1000 is the BMS system described above. It can be functionally included in and implemented. As shown in FIG. 3, the battery characteristic system, with reference to FIG. Obtained from , and using the inference model 1400 learned from a plurality of learning data, impedance information of the corresponding battery cell is calculated.

The impedance information includes impedance measured by EIS (Electrochemical Impedance Spectroscopy), and when the impedance information is plotted on a complex plane, it is the same as the graph shown in the upper right corner of FIG. 3 . As an embodiment of the present invention, the master BMS preferably analyzes the corresponding battery cell by analyzing the waveform of the impedance information shown on the complex plane.

4 schematically illustrates a battery model generating learning data according to an embodiment of the present invention.

Specifically, as shown in FIG. 4, the battery model inputs the environmental condition (Input 1) and the user's power usage pattern (Input 2), and the impedance of the corresponding battery through the output voltage and output current of the corresponding battery. corresponds to the model that produces At this time, the battery model calculates the impedance of the battery in consideration of the standard impedance and the degree of aging of the battery at a specific SOC (state of charge), specific frequency and specific temperature, and input 1 and input corresponding to the number of various cases By inputting 2 to the battery model, a large amount of impedance data set can be obtained. Referring to FIG. 3, the inference model 1400 learns the plurality of impedance data sets obtained as described above, and the waveform of the output voltage, the waveform of the output current, and the temperature of the battery cell are calculated in the pre-learned inference model 1400. If input, it is possible to separate the voltage, current, and temperature information actually output from the battery cell from the composite signal based on the learning data, and the waveform of the output voltage, the waveform of the output current, and the corresponding cell obtained by separating the composite signal. Impedance information of a corresponding battery cell may be measured through temperature.

Additionally, the battery model has a feature of calculating an impedance data set by reflecting the aging state of the battery. In particular, by using both a normal aging model and a specific aging model, various aging states that may occur due to environmental conditions and power usage patterns are derived, and impedances are calculated accordingly to more accurate impedance. dataset can be generated. In addition, through IRET, which tracks entropy in real time, a result value can be derived by considering the change according to temperature.

5 schematically illustrates a process of calculating impedance information of a corresponding cell using a learned inference model 1400 according to an embodiment of the present invention.

As shown in FIG. 5, according to an embodiment of the present invention, the inference model 1400 includes a CNN-based artificial neural network module, and through the artificial neural network module, the output signal measuring unit 1100 And machine learning is performed on the input information received from the temperature measuring unit 1200 in a certain sliding window section, and impedance information according to the frequency of the output signal output to the cell in the sliding window section is calculated.

Meanwhile, according to another embodiment of the present invention, the inference model 1400 includes an LSTM-based artificial neural network module, and the output signal measurement unit 1100 and the temperature measurement unit 1200 are measured through the artificial neural network module. Impedance information according to the frequency of the output signal output to the cell in the sliding window period is calculated by machine learning the input information in a certain sliding window period received from ).

Schematically, the output signal measurement unit 1100 measures voltage information, current information, and temperature according to time in real time while driving from a battery cell, and the impedance information calculation unit 1300 measures the sliding window for a certain time interval T. divide into sections. That is, the impedance information calculating unit 1300 cuts the data from the time t-T to t from the real-time data measured by the output signal measuring unit 1100, and obtains voltage information, current information, and temperature of the corresponding battery cell during the corresponding time. generate information The voltage information, current information, and temperature information include time-series data according to time, and as an embodiment of the present invention, may be in the form of a table having values according to time. As another embodiment of the present invention, FIG. 5 As shown in, it may be in the form of a waveform image itself.

The voltage information, current information, and temperature information of the sliding window section generated by the impedance information calculation unit 1300 are input to the inference model 1400 with reference to FIGS. 1 and 2 . As described above, the reasoning model 1400 corresponds to the reasoning model 1400 pre-learned from a plurality of learning data through the battery model. The voltage information and the current information of the sliding window section are information mixed with composite signals generated by various factors with reference to the description of FIG. 3 . The inference model 1400 may obtain voltage information and current information output from a pure battery by separating the composite signal from voltage information and current information in which composite signals are mixed. Meanwhile, according to an embodiment of the present invention, since temperature information measures the temperature of a corresponding battery cell through a temperature sensor, the data measured by the temperature measurement unit 1200 can be used as it is because the effect of the composite signal is small.

Meanwhile, the reasoning model 1400 learns the learning data derived from the battery model, and then learns voltage information, current information, and temperature generated during driving. In other words, by learning voltage information, current information, and temperature over time measured during driving, the battery state during driving of the vehicle can be more accurately derived.

As an embodiment of the present invention, the inference model 1400 includes a CNN-based artificial neural network module, and through the artificial neural network module, time-series data of voltage information in a sliding window section of a corresponding battery cell and the sliding window section Time-series data of current information in and time-series data of temperature in the sliding window section are machine-learned. It is preferable that the time-series data used at this time corresponds to the form of a waveform image according to time. In addition, the artificial neural network module performs machine learning on a plurality of learning data derived from the battery model.

As another embodiment of the present invention, the inference model 1400 includes an LSTM-based artificial neural network module, and through the artificial neural network module, time-series data of voltage information in the sliding window section of the battery cell, the sliding window section Time-series data of current information in and time-series data of temperature in the sliding window section are machine-learned. It is preferable that the time-series data used at this time corresponds to a table type having the same values according to time. In addition, the artificial neural network module performs machine learning on a plurality of learning data derived from the battery model.

The inference model 1400 calculates impedance information through voltage information and current information from which the composite signal is separated, and temperature information of a corresponding cell. At this time, the impedance information is impedance information according to the frequency of the output signal of the corresponding battery cell, that is, the output signal from which the composite signal is separated. The impedance curve shown in FIG. 5 shows the imaginary part of impedance according to frequency at a specific temperature. The corresponding graph, as an embodiment of the present invention, shows impedance information of a battery cell in a bad state and impedance information of a normal battery cell. Reference impedance information of the corresponding graph corresponds to impedance information of normal battery cells, and measured impedance information of the corresponding graph corresponds to impedance information of defective cells. Referring to this, it can be confirmed that the measured impedance information differs from the reference impedance information in a specific frequency range, in the above embodiment, a range of 10 Hz or less. Meanwhile, as described above, the process of diagnosing the corresponding cell by comparing the reference impedance information with the measured impedance information is preferably performed in the battery analysis unit of the master BMS with reference to FIG. 2 .

6 schematically illustrates a graph of an intercept frequency of a battery cell according to an equivalent circuit and temperature of a battery cell according to an embodiment of the present invention.

Schematically, FIG. 6(a) shows an equivalent circuit of a battery cell, and FIG. 6(b) shows a graph showing a correlation between an intercept frequency calculated from a corresponding equivalent model and temperature.

Specifically, an equivalent circuit refers to a theoretical circuit expressed in a simple form while maintaining all electrical characteristics of a given actual electrical circuit, and is a theoretical circuit configured to facilitate circuit analysis. . The battery cell may be modeled as an equivalent circuit shown in FIG. 6(a). The total impedance Z of the equivalent circuit is as shown in [Equation 1] below.

[Equation 1] is a complex _number that can be separated into a real part and an imaginary part. In other words, the impedance in the intercept frequency means that the conductance component and the inductance component cancel each other in the equivalent circuit of FIG. 6 (a), and is an important index representing the characteristics of the circuit. [Equation 2]

Meanwhile, in [Equation 2], L, R _s , C _kin , and C _d are all values set when designing a battery, that is, constants, and only R _kin operates as the only variable in the above equation. In addition, the corresponding battery cell is a lithium-ion battery and generates power through a chemical reaction. Therefore, the R _kin follows the Arrhenius equation. The Arrhenius equation is an equation representing the relationship between the reaction rate constant and the temperature, and is as shown in [Equation 3] below.

In [Equation 3], A, R, and E _a are all constants, and T corresponds to absolute temperature. That is, k in [Equation 3] is a reaction rate constant, which can be seen as inversely proportional to resistance from an electrochemical point of view. Therefore, R _kin in [Equation 2] has a relationship as shown in [Equation 4] below.

That is, it can be confirmed that the R _kin is a value that changes only with temperature, and since the R _kin changes with temperature, the total impedance Z also varies with temperature. As shown in the graph shown in (b) of FIG. 6, the intercept frequency varies depending on SOC (State Of Charge) and temperature. The intercept frequency when SOC is 100% and the intercept frequency when 20% are large. While there is no difference, it can be seen that there is a large difference from 300 Hz to 1700 Hz depending on the temperature.

Therefore, the equation for calculating the impedance of the battery requires only voltage and current, but it can be confirmed that the temperature has a great effect on the impedance as in the above process. In addition, while researching the present invention, referring to FIG. 5, as a result of inputting temperature-related time-series data together with voltage information and current information into the inference model 1400 to calculate impedance, a more accurate calculated value is derived even while driving. As a result, in the present invention, by inputting the voltage information, current information, and temperature of the battery cell together into the inference model 1400 to derive impedance information, accurate impedance information can be obtained in real time without a separate measuring device while driving. It is possible to exert a calculated effect.

7 schematically illustrates a graph showing a relationship between temperature and impedance information according to an embodiment of the present invention.

Specifically, FIG. 7 is a graph showing that the real component of the impedance of the battery cell satisfies the Arrhenius equation. In the description of FIG. 6, it was confirmed that the R _kin component of the impedance of the battery cell had a temperature dependence, and through the graph shown in FIG. 7, it can be seen that the real component of the impedance Z also has a high temperature dependence. Data in FIG. 7 corresponds to the experimental results, and the solid line corresponds to the Arrhenius fit, which is a set of values satisfying the Arrhenius equation. Referring to FIG. 7 , it can be confirmed that the data, which is the experimental result, generally satisfies the Arrhenius fit. Therefore, if a chemical reaction normally occurs within a battery cell, information on the real component of the corresponding battery cell can be inferred from the temperature, and through this, the inference model 1400 can infer the voltage of the corresponding cell when the temperature information is input. Impedance information can be calculated more accurately than when only information and current information are input. In addition, as an embodiment of the present invention, if the real component of the impedance of the corresponding cell does not follow the Arrhenius fit as a result of measurement, the master BMS may determine that the corresponding battery cell has a problem.

8 schematically illustrates impedance information of a battery cell according to the frequency and temperature of an input signal according to an embodiment of the present invention.

Specifically, FIG. 8 corresponds to data for which the impedance information calculation unit 1300 has calculated final impedance information for a corresponding cell. As shown in FIG. 8, the impedance information of a corresponding cell varies according to the temperature and the frequency of the output signal. When the master BMS receives the impedance information as shown in FIG. 8, it can diagnose the corresponding cell based on this.

More specifically, a round semicircle at the top right of the waveform of the impedance information is a region in which charge movement and/or charge diffusion characteristics appear in the corresponding cell. The master BMS can diagnose the state of the cell by recognizing the size and position of the corresponding region and comparing it with a reference value. As shown in FIG. 8, impedance information at a specific temperature shows a waveform of a certain shape. On the other hand, if impedance information is calculated with only voltage information and current information without real-time temperature data in an environment where temperature changes in real time, such as during driving, since the temperature of the cell can change moment by moment, the measurement is made in an accurate shape as shown in FIG. and the state of the cell cannot be diagnosed through this.

On the other hand, in the present invention, by using the inference model 1400 that calculates impedance information according to temperature and learns it, it is possible to accurately calculate the impedance information of a corresponding cell even in a driving environment in which the cell temperature changes in real time. The characteristics and state of the battery can be derived.

According to an embodiment of the present invention, impedance information can be measured for all cells included in a battery of a vehicle, so that defective cells having problems such as poor charging or leakage current can be easily found. Through this, by quickly removing the defective cells, the failure rate of the battery can be reduced and the lifespan of the battery can be extended.

According to an embodiment of the present invention, as a system installed in a BMS system in a vehicle, it is possible to achieve an effect of measuring impedance information of each cell included in a corresponding battery even when charging or driving.

According to an embodiment of the present invention, it is possible to generate a lot of learning data through a battery model that calculates the impedance of a corresponding battery according to a power pattern used in a vehicle and environmental conditions, and an inference model learned from the a lot of learning data By using the battery, it is possible to achieve an effect of measuring impedance information of each cell included in the battery in real time.

According to an embodiment of the present invention, by using the learned inference model, the effect of measuring more accurate voltage and current for the cell by minimizing the influence of signals generated by complex factors inside and outside the vehicle can be exerted. can

According to an embodiment of the present invention, by measuring the voltage and current of each cell included in the battery, as well as the impedance information according to the temperature of the cell, the effect of performing a more accurate diagnosis on the cell can exert

As described above, although the embodiments have been described with limited examples and drawings, those skilled in the art can make various modifications and variations from the above description. For example, the described techniques may be performed in an order different from the method described, and/or components of the described system, structure, device, circuit, etc. may be combined or combined in a different form than the method described, or other components may be used. Or even if it is replaced or substituted by equivalents, appropriate results can be achieved. Therefore, other implementations, other embodiments, and equivalents of the claims are within the scope of the following claims.

Claims (6)

A battery characteristic derivation system based on a learned inference model provided in a battery installed in a vehicle equipped with a powertrain using electricity as power and operating,
an output signal measurement unit for measuring an output signal including voltage information and current information output from a cell included in the battery;
a temperature measuring unit for measuring the temperature of the cell; and
An impedance information calculation unit for calculating impedance information by inputting input information including an output signal and temperature from the cell to the learned artificial neural network-based inference model,
The input information includes time-series data of voltage information, current information, and temperature according to time in a certain sliding window section obtained during actual driving of the vehicle,
The inference model is learned by a plurality of learning data generated by considering a plurality of factors influencing battery characteristics in a virtual battery model, and the learning data is voltage, current, and temperature for a battery having specific impedance information. information, the learned inference model receives the input information and calculates the impedance information,
The impedance information includes an impedance value for each of a plurality of frequencies for a corresponding cell.
delete The method of claim 1,
The battery management system consists of one master BMS and a plurality of slave BMS, the master BMS is connected to a plurality of slave BMS,
The slave BMS includes an output signal measuring unit connected to each of a plurality of battery modules constituting the battery and measuring an output signal output from one or more cells included in the battery module; a temperature measuring unit for measuring the temperature in the corresponding cell; And an impedance information calculation unit for calculating impedance information in the corresponding cell; includes,
The master BMS includes a battery analysis unit that receives a plurality of impedance information from each of the plurality of slave BMS and analyzes a state of a corresponding battery.
The method of claim 1,
The reasoning model includes a CNN-based artificial neural network module, and through the artificial neural network module, machine learning is performed on input information in a certain sliding window section input from the output signal measuring unit and the temperature measuring unit, A battery characteristic derivation system for calculating impedance information according to the frequency of an output signal output to the cell in a window period.
The method of claim 1,
The reasoning model includes an LSTM-based artificial neural network module, and through the artificial neural network module, machine learning is performed on input information in a certain sliding window section input from the output signal measurement unit and the temperature measurement unit, and the sliding A battery characteristic derivation system for calculating impedance information according to the frequency of an output signal output to the cell in a window period.
A method for deriving battery characteristics based on a learned inference model operating in a battery installed in a vehicle equipped with a powertrain using electricity as power,
an output signal measurement step of measuring an output signal including voltage information and current information output from a cell included in the battery;
a temperature measurement step of measuring the temperature in the cell; and
An impedance information calculation step of calculating impedance information by inputting input information including an output signal and temperature from the cell to the learned artificial neural network-based inference model,
The input information includes time-series data of voltage information, current information, and temperature according to time in a certain sliding window section obtained during actual driving of the vehicle,
The inference model is learned by a plurality of learning data generated by considering a plurality of factors influencing battery characteristics in a virtual battery model, and the learning data is voltage, current, and temperature for a battery having specific impedance information. information, the learned inference model receives the input information and calculates the impedance information,
Wherein the impedance information includes an impedance value for each of a plurality of frequencies for a corresponding cell.

Images