Battery management system and operation method thereof

Abstract

To provide a battery management system (BMS) that does not require addition of hardware and does not drop the reliability of contactor state determination results due to sensor errors or failures.SOLUTION: A battery management system (BMS) is provided that includes a contactor 11 coupled to a battery, a battery management unit 100 that outputs a control signal for controlling the state of the contactor 11 to the contactor 11, a signal processing unit 200 that receives an RF (Radio Frequency) signal generated at the time of state transition of the contactor 11 and generates a plurality of signal strength values respectively corresponding to a plurality of frequency bands based on the RF signal, and a contactor state estimation unit 300 that receives the plurality of signal strength values, inputs the plurality of signal strength values to a pre-trained neural network model, and generates an estimate estimating the state of the contactor 11.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to a battery management system and a method of operation thereof. [Background technology]

[0002] A battery management system (BMS) for vehicles is an important component that manages the performance and safety of a battery pack in an electric vehicle (xEV: x electrical vehicle). The vehicle battery management system (BMS) monitors the charge state, voltage, temperature, etc. of the battery and controls the charging and discharging of the battery based on the monitor results. The vehicle battery management system (BMS) also detects abnormal conditions of the battery and takes protective measures if necessary.

[0003] A contactor is one of the components that plays an important role in a vehicle battery management system (BMS) and controls the electrical connection between the battery and the vehicle's electrical system. The contactor switches between an open state and a closed state to control the charging and discharging of the battery. A noise signal is generated during the contactor's state switching process, and the noise signal can be used to detect the normal and abnormal states of the contactor.

[0004] In the conventional method, the state of the contactor is sensed using a physical sensor that measures the position, current, voltage, etc. of the contactor to determine the state of the contactor. However, the conventional method requires additional hardware, and there are problems in that the reliability of the contactor state determination result is reduced due to sensor errors or failures.

[0005] The above information disclosed in such Background of the Invention is intended to enhance the understanding of the background of the present invention and, therefore, may include information that does not constitute prior art. Summary of the Invention [Problem to be solved by the invention]

[0006] The present invention provides a battery management system (BMS) and a method thereof, which detects an RF (radio frequency) signal generated when a contactor changes state, and determines the state of the contactor based on a signal strength value generated from the RF signal and a pre-trained neural network model.

[0007] However, the technical problems that the present invention aims to solve are not limited to the problems described above, and other problems not mentioned will be clearly understood by those skilled in the art from the description of the invention described below. [Means for solving the problem]

[0008] A battery management system (BMS) according to an embodiment of the present invention for solving the above technical problems includes a contactor connected to a battery, a battery management unit that outputs a control signal for controlling a state of the contactor to the contactor, a signal processing unit that receives a radio frequency (RF) signal generated when a state of the contactor is changed and generates a plurality of signal strength values ​​corresponding to a plurality of frequency bands based on the RF signal, and a contactor state estimation unit that receives the plurality of signal strength values, inputs the plurality of signal strength values ​​to a pre-trained neural network model, and generates an estimation value for estimating a state of the contactor.

[0009] According to one example, the battery management unit outputs a trigger signal to the signal processing unit in response to the output of the control signal, and the signal processing unit responds to the trigger signal and captures the RF signal.

[0010] According to another example, the signal processing unit includes an antenna that receives the RF signal, an RF tuner that outputs a band signal of a preset frequency band from the RF signal, an analog-to-digital converter (ADC) that converts the band signal into a digital signal, a digital signal processor (DSP) that performs a Fourier transform on the digital signal into a frequency domain signal and generates the plurality of signal strength values ​​from the frequency domain signal, and a signal processing unit controller that controls the antenna, the RF tuner, the analog-to-digital converter (ADC), and the digital signal processor (DSP) in response to the trigger signal and transmits the plurality of signal strength values ​​to the contactor state estimation unit.

[0011] According to yet another example, the pre-set frequency band is characterized by being equal to or greater than 100 kHz and equal to or less than 5 MHz.

[0012] According to yet another embodiment, the widths of the frequency bands are identical in logarithmic scale.

[0013] According to yet another embodiment, the widths of the plurality of frequency bands are the same.

[0014] According to yet another example, the neural network model is a convolutional neural network (CNN) model, and is characterized in that it is pre-trained using learning data including a plurality of signal strength values ​​generated by preprocessing the state of the contactor that is switched by a control signal of the battery management unit and an RF signal generated when the contactor switches states.

[0015] According to yet another embodiment, the control signals include a turn-on signal for switching the contactor to an open state and a turn-off signal for switching the contactor to a closed state.

[0016] According to yet another example, the pre-trained neural network model is characterized in that it outputs the estimated value based on the input signal strength values ​​and a reliability score for the estimated value.

[0017] According to yet another example, the battery management unit detects whether or not the state of the contactor is abnormal based on the estimated value and the reliability of the estimated value.

[0018] According to yet another example, the battery management unit is characterized in that it judges the state of the contactor to be an abnormal state for at least one of the closed state based on the state estimation value of the contactor in response to the turn-on signal, the open state based on the state estimation value of the contactor in response to the turn-off signal, and the weld state based on the state estimation value of the contactor.

[0019] According to one embodiment of the present invention for solving the technical problem, a method for operating a battery management system (BMS) including a contactor connected to a battery includes the steps of: outputting a control signal for controlling a state of the contactor to the contactor; receiving an RF signal generated when the contactor changes state; generating a plurality of signal strength values ​​corresponding to a plurality of frequency bands based on the RF signal; and inputting the plurality of signal strength values ​​into a pre-trained neural network model to generate an estimate for estimating a state of the contactor.

[0020] According to one example, the method further includes extracting a trigger signal in response to output of the control signal, and capturing the RF signal in response to the trigger signal.

[0021] According to another example, the method further includes receiving the RF signal, outputting a band signal of a preset frequency band from the RF signal, converting the band signal into a digital signal, Fourier transforming the digital signal into a frequency domain signal and generating the plurality of signal strength values ​​from the frequency domain signal, and extracting the plurality of signal strength values ​​in response to the trigger signal.

[0022] According to yet another example, the control signal may include a turn-on signal for switching the contactor to an open state and a turn-off signal for switching the contactor to a closed state.

[0023] According to yet another example, the pre-trained neural network model may generate the estimate based on the input signal strength values ​​and a confidence score for the estimate.

[0024] According to yet another embodiment, the method further includes a step of detecting whether or not the state of the contactor is abnormal based on the estimated value and the reliability of the estimated value.

[0025] According to yet another example, the method includes a step of determining that the state of the contactor is an abnormal state for at least one of the closed state based on the state estimate of the contactor in response to the turn-on signal, the open state based on the state estimate of the contactor in response to the turn-off signal, and the weld state based on the state estimate of the contactor.

[0026] As a technical means for achieving the above technical object, a computer program according to one aspect of the present invention is stored on a medium to cause a computing device to execute the above-mentioned operation method of a battery management system (BMS). Effect of the Invention

[0027] According to the present invention, compared to a method of detecting the state of a contactor using an existing physical sensor, a large amount of hardware is not required, and an abnormal state of a contactor can be detected more accurately by learning an RF signal pattern in advance. Also, in a vehicle battery management system (BMS), the state of a contactor can be monitored in real time and an abnormal state of the contactor can be detected quickly, thereby improving the safety and reliability of the battery management system (BMS).

[0028] However, the effects that can be obtained through the present invention are not limited to the effects described above, and other technical effects not mentioned will be clearly understood by those skilled in the art from the description of the invention described below. [Brief description of the drawings]

[0029] [Figure 1] FIG. 1 is a diagram illustrating a schematic of a battery management system (BMS) according to an embodiment of the present invention. [Diagram 2] 1 is a diagram illustrating an operation process of a battery management system (BMS) according to an embodiment of the present invention. [Diagram 3] 1 is a schematic block diagram of a computing device for performing a method of operating a battery management system (BMS) according to an embodiment of the present invention. [Figure 4] 1 is a flowchart illustrating an operation method of a battery management system (BMS) according to an embodiment of the present invention. [Figure 5A] FIG. 2 is a diagram illustrating an exemplary structure of a pre-trained artificial neural network model according to an embodiment of the present invention. [Figure 5B] FIG. 2 illustrates a schematic diagram of an example of a pre-trained artificial neural network model, according to an embodiment of the present invention. [Figure 5C] 1 is a schematic illustration of a training process for a pre-trained artificial neural network model according to an embodiment of the present invention; DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0030] The following drawings attached to this specification illustrate preferred embodiments of the present invention and, together with the detailed description of the invention described below, serve to further understand the technical concepts of the present invention. Therefore, the present invention should not be interpreted as being limited only to the matters described in such drawings.

[0031] Hereinafter, a preferred embodiment of the present invention will be described in detail with reference to the accompanying drawings. Prior to this, the terms and words used in the present specification and claims should not be interpreted as being limited to their general or dictionary meanings, but should be interpreted according to the meaning and concept that corresponds to the technical idea of ​​the present invention, based on the principle that the inventor can appropriately define the concept of the term in order to best describe his / her invention. Therefore, it should be understood that the embodiments described in the present specification and the configurations illustrated in the drawings are only some of the most preferred embodiments of the present invention, and do not represent the technical idea of ​​the present invention, and therefore, at the time of filing this application, there may be various equivalents and modifications that can replace them. In addition, when used in the present specification, "comprise, include" and / or "comprising, including" specify the presence of the mentioned shapes, numbers, steps, operations, members, elements, and / or groups thereof, and do not exclude the presence or addition of one or more other shapes, numbers, operations, members, elements, and / or groups. In addition, when describing an embodiment of the present invention, the words "may" and "also" are intended to include "one or more embodiments of the present invention."

[0032] It should be understood that even if the terms "first," "second," etc., are used to describe various elements, the elements are not limited by these terms. These terms are used only to distinguish one element from another element, and it should be understood that a first element can also be a second element, unless specifically stated to the contrary.

[0033] Throughout the specification, unless specifically stated to the contrary, each element may be in the singular or in the plural.

[0034] Throughout the specification, "A and / or B" means A, B, or A and B, unless specifically stated to the contrary. That is, "and / or" includes any and all combinations of the listed items. "C through D" means at least C and at most D, unless specifically stated to the contrary.

[0035] FIG. 1 illustrates a schematic diagram of a battery management system (BMS) according to one embodiment of the present invention.

[0036] Referring to FIG. 1, a battery management system (BMS) 1000 includes a contactor 11 connected to a battery, a battery management unit 100, a signal processing unit 200, and a contactor state estimation unit 300.

[0037] A battery management system (BMS) is an important component that manages the performance and safety of a battery pack in an electric vehicle (xEV: x electrical vehicle). The battery management system (BMS) monitors the charge state, voltage, temperature, etc. of the battery, and can control the charging and discharging of the battery based on the monitoring results.

[0038] The contactor 11 is one of the components that play an important role in the battery management system (BMS) 1000. For example, the contactor 11 may control an electrical connection between the battery and the vehicle's electrical system. The contactor 11 may switch states to control charging and discharging of the battery. For example, the state of the contactor 11 may be switched to an open state or a closed state. In the process of switching the state of the contactor 11, a radio frequency (RF) signal SIG RF The RF signal SIG RF may also contain noise. 、 The RF signal SIG containing the noise RF can be used to sense the normal and abnormal states of the contactor 11.

[0039] The battery management unit 100 is a component that controls the operation of the contactor 11 and outputs a signal for diagnosing the state of the contactor 11 to the signal processing unit 200 and the contactor state estimation unit 300. The battery management unit 100 outputs a control signal SIG CTR to the contactor 11. CTR The control signal SIG also includes a turn-on signal for switching the contactor 11 to an open state and a turn-off signal for switching the contactor 11 to a closed state. CTR In response to the output of TRG For example, the trigger signal SIG TRG is the RF signal SIG generated from contactor 11. RF The signal includes an instruction to sense and sample the battery. The battery management unit 100 also includes a microcontroller unit (MCU).

[0040] The battery management unit 100 may detect whether the state of the contactor 11 is abnormal based on the estimated state value of the contactor 11 and the reliability of the estimated value. For example, the state of the contactor 11 according to the estimated state value of the contactor 11 is at least one of a closed state, an open state, and a welded state. If the state of the contactor 11 is a welded state, the battery management unit 100 may determine that the state of the contactor 11 is an abnormal state.

[0041] The signal processor 200 is also a software defined radio (SDR). The SDR is a wireless communication system implemented through software instead of components implemented by existing analog hardware (e.g., mixers, filters, amplifiers, modulators / demodulators, sensors, etc.). The SDR allows a single piece of hardware to capture, demodulate, and access a wide range of radio frequency RF signals. For example, the SDR may capture a wide frequency spectrum and selectively analyze a specific portion of the RF signal. The signal processor 200 (e.g., SDR) detects the RF signal SIG generated at the state transition of the contactor 11 in the frequency domain. RF The RF signal SIG RF The RF signal SIG may include a noise pattern, and the noise pattern may reflect the operating state of the contactor 11. That is, the noise pattern may have a unique pattern when the contactor 11 is in an open state, a closed state, or a welded state. The signal processor 200 outputs the RF signal SIG RF Based on this, a plurality of signal strength values ​​SS corresponding to a plurality of frequency bands may be generated.

[0042] The signal processing unit 200 outputs a trigger signal SIG TRG In response to the RF signal SIG RF The battery management unit 100 may capture a control signal SIG CTRWhen outputting the trigger signal SIG TRG The components and operation of the signal processor 200 will be described in more detail with reference to FIG.

[0043] The contactor state estimation unit 300 may receive a plurality of signal strength values ​​SS and input the plurality of signal strength values ​​SS to a pre-trained neural network model to generate an estimate for estimating the state of the contactor 11. According to an example, the neural network model may be a convolutional neural network (CNN) model. In the present invention, the neural network model may refer to an artificial neural network that is generated by a processor for a predetermined purpose and trained by a machine learning technique or a deep learning technique. The structure of such a neural network will be described later with reference to FIG. 5 (FIGS. 5A, 5B, and 5C). The contactor state estimation unit 300 may also include a microcontroller unit (MCU). The contactor state estimation unit 300 may further include a Noise Pattern Processing Unit.

[0044] According to one embodiment, the neural network model is configured to control the control signal SIG CTR The state of the contactor 11 that is switched by and the RF signal SIG that is generated when the contactor 11 changes state RF The neural network model may be trained in advance using training data including a plurality of signal strength values ​​SS generated by preprocessing the signal strength values ​​SS and the contactor 11 state estimation unit 300. The neural network model may output an estimate SEV based on the input signal strength values ​​SS and a reliability score of the estimate SEV. The contactor state estimation unit 300 may transmit the contactor 11 state estimate SEV and the reliability score data of the estimate SEV to the battery management unit 100.

[0045] FIG. 2 is a schematic diagram illustrating an operation process of a battery management system (BMS) according to an embodiment of the present invention.

[0046] Referring to Figure 2, the frequency-based RF signal SIG generated when the contactor changes state is RF It recognizes noise patterns in the RF signal and uses a pre-trained artificial neural network to RF By analyzing the noise pattern of the contactor 11, the condition of the contactor 11 can be diagnosed.

[0047] The battery management system (BMS) 1000 also includes a contactor 11 connected to a battery, a battery management unit 100, a signal processing unit 200, and a contactor state estimation unit 300.

[0048] The contactor 11 is one of the components that play an important role in the battery management system (BMS) 1000. For example, the state of the contactor 11 can be an open state or a closed state. In the process of changing the state of the contactor 11, the RF signal SIG RF may occur.

[0049] The battery management unit 100 outputs a control signal SIG CTR The control signal SIG CTR The control signal SIG also includes a turn-on signal for switching the contactor 11 to an open state and a turn-off signal for switching the contactor 11 to a closed state. CTR In response to the output of TRG The battery management unit 100 can output the trigger signal SIG TRG to the contactor state estimation unit 300.

[0050] The battery management unit 100 may detect whether the state of the contactor 11 is abnormal based on the state estimate value SEV of the contactor 11 and the reliability of the estimate value SEV. The battery management unit 100 may determine that the state of the contactor 11 is abnormal when the state of the contactor 11 according to the state estimate value SEV of the contactor 11 in response to a turn-on signal is a closed state. The battery management unit 100 may determine that the state of the contactor 11 is abnormal when the state of the contactor 11 according to the state estimate value SEV of the contactor 11 in response to a turn-off signal is an open state. The battery management unit 100 may determine that the state of the contactor 11 is abnormal when the state of the contactor 11 according to the state estimate value SEV of the contactor 11 is a welded state.

[0051] The battery management unit 100 also includes a microcontroller unit (MCU). The microcontroller unit (MCU) of the battery management unit 100 can send an operating command to the contactor 11. For example, the microcontroller unit (MCU) can transmit an operating command to the power system to activate the contactor 11. For example, the operating command can be "contactor ON" or a command indicating a specific voltage / current level.

[0052] The microcontroller unit (MCU) of the battery management unit 100 transmits the RF signal SIG RF Trigger signal SIG to start sensing TRG For example, the signal processing unit 200 may also include a Noise Pattern Processing Unit. For example, the trigger signal SIG TRG is also a byte code such as "START_NOISE_DETECTION". The microcontroller unit (MCU) of the battery management unit 100 transmits an operation command to the contactor 11 and also transmits a trigger signal SIG TRG For example, a trigger signal SIG TRG contains a time stamp and the RF signal SIGRF The trigger signal SIG TRG For example, the trigger signal SIG TRG The data may be transmitted via a serial peripheral interface (SPI) protocol, but is not limited thereto, and any communication protocol that transmits data or commands from a battery management system (BMS) may be used.

[0053] A microcontroller unit (MCU) of the battery management system (BMS) may adjust system parameters based on analysis data according to the estimated state value of the contactor 11. For example, the microcontroller unit (MCU) may diagnose the state of the contactor 11 based on the analysis data, and if necessary, determine a protective measure for the battery management system (BMS). For example, the protective measure may include setting a maximum noise amplitude threshold value, adjusting the operating voltage of the contactor 11, etc. However, the above content is merely an example, and the present invention is not limited by the above content.

[0054] The signal processing unit 200 also includes an antenna 210, an RF tuner 220, an analog-to-digital converter (ADC) 230, a digital signal processor (DSP) 240, and a signal processing unit controller 250. The signal processing unit 200 is also a software defined radio (SDR). In the following, in the embodiment of FIG. 2, the signal processing unit 200 will be described assuming that it is an SDR. The signal processing unit 200 receives a trigger signal SIG from the battery management unit 100. TRG In response to the change in state of the contactor 11, an RF signal SIG RF The signal processor 200 may capture a trigger signal SIG TRG At the same time, the RF signal SIG received from contactor 11 RFFor example, the signal processor 200 may start sampling the RF signal SIG RF The signal processing unit 200 may sample the RF signal SIG RF , extracting a plurality of signal strength values ​​SS, and generating a state estimate value SEV of the contactor 11 based on the plurality of signal strength values ​​SS. RF The signal processor 200 may generate noise data such as an analog voltage value by sampling the signal. The components included in the signal processor 200 will now be described in detail.

[0055] The antenna 210 receives an RF signal SIG RF The antenna 210 can receive the RF signal SIG generated when the contactor 11 changes state. RF The antenna 210 may receive the RF signal SIG RF into an electrical signal and transmit it to the RF tuner 220. For example, the antenna 210 may receive a wideband frequency signal having a frequency between 1 kHz and 1 GHz.

[0056] The RF tuner 220 outputs an RF signal SIG RF In the predefined frequency band, the band signal SIG BD For example, the RF tuner 220 may output an RF signal SIG RF Receives the RF signal SIG RF The RF tuner 220 can downconvert the frequency of the band signal SIG having a frequency of 100 kHz to 5 MHz inclusive among the RF signals having a frequency of 1 kHz to 1 GHz inclusive. BD According to one example, the widths of the frequency bands may be the same in log scale. According to another example, the widths of the frequency bands may be the same. The RF tuner 220 may output the downconverted band signal SIGBD may be transmitted to the analog-to-digital converter 230.

[0057] The analog-to-digital converter (ADC) 230 converts the band signal SIG BD The digital signal SIG DIG The analog-to-digital converter 230 may convert the analog form of the band signal SIG BD is sampled and the digital signal SIG DIG For example, the analog-to-digital converter 230 may have a 30 MSPS (mega samples per second) spec. The analog-to-digital converter 230 with a sampling rate of 30 MSPS may extract 30 million signal samples per second. According to the Nyquist sampling theory, the sampling process performed by the sampling rate of 30 MSPS is also sufficient to capture signals having frequencies greater than 100 kHz and less than 5 MHz. The analog-to-digital converter 230 may convert the band signal SIG BD The analog-to-digital converter 230 may sample the digital signal SIG DIG may be communicated to the digital signal processor 240.

[0058] 2, the signal processing unit 200 may further include a digital oscillator, a digital mixer, and a band-pass filter. For example, the digital oscillator, the digital mixer, and the band-pass filter may be located between the analog-to-digital converter (ADC) 230 and the digital signal processor (DSP) 240.

[0059] The digital oscillator and digital mixer generate the digital signal SIG DIGThe bandpass filter is a filter that passes only signals between specific frequencies. The bandpass filter converts the frequency of the digital signal SIG received from the analog-to-digital converter 230. DIG For example, the band-pass filter may be a low-pass filter, but this is merely an example and the above content is not intended to limit the present invention.

[0060] The digital signal processor (DSP) 240 outputs the digital signal SIG DIG , the frequency domain signal SIG FD The digital signal processor 240 may perform a Fourier transform (FT) on the time-related digital signal SIG DIG is decomposed into frequency components, and the frequency domain signal SIG FD The digital signal processor 240 may extract the frequency domain signal SIG FD The digital signal processor 240 may generate a plurality of signal strength values ​​SEV from the plurality of signal strength values ​​SS, which may also include RF signal characteristics (e.g., noise, etc.). The digital signal processor 240 may use various algorithms to generate the digital signal SIG DIG For example, the digital signal processor 240 may process the time domain digital signal SIG DIG is fast Fourier transformed (FFT) to obtain the frequency domain signal SIG FD The digital signal processor 240 may communicate the signal strength values ​​SS to the signal processor controller 250.

[0061] The signal processing unit controller 250 receives the trigger signal SIG TRG 2, in response to which it may control the antenna 210, the RF tuner 220, the analog-to-digital converter (ADC) 230, and the digital signal processor (DSP) 240. For example, the signal processor controller 250 may also be an SDR controller.

[0062] The signal processor controller 250 may transmit a plurality of signal strength values ​​SS to the contactor state estimator 300. The signal processor controller 250 may communicate with a user interface. The signal processor controller 250 may communicate with external networks and hardware. For example, the signal processor controller 250 may transmit a plurality of signal strength values ​​SS to the contactor state estimator 300. RF Based on this, the RF raw signal data including the noise pattern can be generated, and the RF raw signal data can be transmitted to the contactor state estimation unit 300.

[0063] The contactor state estimator 300 may receive a plurality of signal strength values ​​SS and input the plurality of signal strength values ​​SS into a pre-trained neural network model to generate an estimate SEV that estimates the state of the contactor 11. The contactor state estimator 300 may utilize an algorithm to analyze the noise data and estimate the state of the RF signal SIG. RF The contactor state estimation unit 300 may estimate the characteristics, causes, and effects of the noise contained in the RF signal SIG. For example, the noise data may include information such as the average amplitude of the signal and the peak frequency of the noise. The contactor state estimation unit 300 may estimate the characteristics, causes, and effects of the noise contained in the RF signal SIG using a pre-trained artificial neural network. RF The contactor state estimation unit 300 may learn the noise data extracted from the contactor 11 and detect the abnormal state of the contactor 11. The contactor state estimation unit 300 may use an artificial neural network to learn the state data of the contactor 11 and determine whether the state of the contactor 11 is normal or abnormal based on new data (e.g., an RF signal generated from the contactor 11). For example, the artificial neural network may be a convolutional neural network (CNN). The convolutional neural network (CNN) may extract useful features from the frequency-based noise pattern and extract the operating state of the contactor 11 based on the useful features.

[0064] According to an example, the contactor state estimator 300 may transmit the state estimate SEV of the contactor 11 to a battery management system (BMS). For example, the contactor state estimator 300 may transmit the state estimate SEV of the contactor 11 to the battery management system (BMS) via an SPI protocol. For example, the state estimate SEV of the contactor 11 is also an analysis data packet in the format of "Noise Pattern: [data], Frequency: [data], Amplitude: [data]".

[0065] FIG. 3 is a schematic block diagram of a computing device for performing a method of operating a battery management system (BMS) according to an embodiment of the present invention.

[0066] Referring to FIG. 3, a computing device 10 according to one embodiment of the present invention also includes a memory 20 and a processor 30 .

[0067] Memory 20 is a recording medium readable by computing device 10 and may include random access memory (RAM), read-only memory (ROM), and permanent mass storage devices such as disk drives.

[0068] The memory 20 performs a function of temporarily or permanently storing data to be processed by the processor. The memory 20 may include a magnetic storage medium or a flash storage medium, but the scope of the present invention is not limited thereto. For example, the memory 20 may receive data constituting an artificial neural network and temporarily and / or permanently store the data. The memory 20 may store training data for training the artificial neural network, but this is merely an example and the idea of ​​the present invention is not limited thereto.

[0069] The memory 20 may store program code for executing a method of operating a battery management system (BMS) according to an embodiment of the present invention, data required to execute the program code, and data generated in the process of executing the program code, including algorithm codes for Fourier transform (FT) and artificial neural network (ANN) model calculations.

[0070] The memory 20 may store data necessary for performing an operation method of the battery management system (BMS) according to the present invention. For example, the memory 20 may store data for training an artificial neural network. For example, the memory 20 may store, as training data, a plurality of signal strength values ​​generated by preprocessing the state of the contactor switched by the control signal of the battery management unit and the RF signal generated when the state of the contactor is switched.

[0071] The processor 30 generally controls the overall operation of the computing device 10. The processor 30 may be configured to process computer program instructions by performing basic arithmetic, logic, and input / output operations. The processor 30 may receive learning data and generate learning result data from the received learning data through a user-selected pre-processing scheme. For example, the processor 30 may receive contactor RF signal data and generate contactor status data from the contactor RF signal data.

[0072] Such a processor 30 may refer to a data processing device built into hardware having a circuit physically structured to perform a function expressed by a code or instruction included in a program, for example. Examples of such a data processing device built into hardware may include a microprocessor, a central processing unit (CPU), a processor core, a multiprocessor, an application specific integrated circuit (ASIC), and a field programmable gate array (FPGA), but the scope of the present invention is not limited thereto.

[0073] The processor 30 may receive data stored in the memory 20 and transmit the data to the memory 20. Although not shown in the drawings, the computing device 10 may further include a communication module, an input / output device, or a storage device in addition to the memory 20 and the processor 30.

[0074] The operation of processor 30 according to various embodiments is described in further detail below.

[0075] FIG. 4 is a flowchart illustrating an operation method of a battery management system (BMS) according to an embodiment of the present invention.

[0076] 4, a method of operating a battery management system (BMS) may be performed by processor 30 of FIG.

[0077] A control signal for controlling a state of the contactor connected to the battery may be output to the contactor (S10). The control signal may include a turn-on signal for switching the contactor to an open state and a turn-off signal for switching the contactor to a closed state. In step (S10), a trigger signal may be extracted in response to the output of the control signal, and an RF signal may be captured in response to the trigger signal.

[0078] An RF signal generated when a contactor changes state may be received (S20). In step (S20), the RF signal may be received from the contactor, and a band signal in a preset frequency band may be output from the RF signal. Then, the band signal may be converted into a digital signal, and the digital signal may be Fourier transformed into a frequency domain signal, and a plurality of signal strength values ​​may be generated from the frequency domain signal.

[0079] A plurality of signal strength values ​​respectively corresponding to a plurality of frequency bands may be generated based on the RF signal (S30). According to the step (S30), the plurality of signal strength values ​​may be generated in response to a trigger signal.

[0080] The signal strength values ​​may be input to a pre-trained neural network model to generate an estimate of the contactor's state (S40). The pre-trained neural network model may generate a contactor state estimate from the input signal strength values ​​and a confidence score for the estimate. The neural network model is described in further detail in FIG. 5 (FIGS. 5A, 5B, and 5C).

[0081] Step (S40) may further include detecting whether the state of the contactor is abnormal based on the contactor state estimate and the reliability of the estimate. For example, in step (S40), if the contactor state according to the contactor state estimate in response to the turn-on signal is a closed state, the contactor state may be determined to be an abnormal state. In step (S40), if the contactor state according to the contactor state estimate in response to the turn-off signal is an open state, the contactor state may be determined to be an abnormal state. In step (S40), if the contactor state according to the contactor state estimate is a welded state, the contactor state may be determined to be an abnormal state.

[0082] In the following, the structure of an exemplary artificial neural network is described via FIG. 5 (FIGS. 5A, 5B and 5C).

[0083] 5 (FIGS. 5A, 5B, and 5C) are diagrams for explaining an exemplary structure of an artificial neural network (ANN) trained by the processor 30 (FIG. 3) of the present invention. The artificial neural network model illustrated in FIG. 5 is also included in the contactor state estimator 300 of FIG. 1.

[0084] FIG. 5A is a diagram illustrating an exemplary structure of a pre-trained artificial neural network model according to an embodiment of the present invention.

[0085] Referring to FIG. 5A, components of an artificial neural network also include nodes / units, layers, weights, summations, and functions.

[0086] A node (or a unit) is an element that constitutes each layer. A layer is at least one of an input layer, a hidden layer, and an output layer. The input layer is a layer that receives input data. For example, the input layer receives multiple input data (x 1 ,x 2 ,…,x p-1 ,x p ), where p is a natural number equal to or greater than 1. According to one example, the plurality of input data may be a plurality of RF signal data generated from contactors in a battery management system (BMS). The hidden layer may process the input data at least once. The hidden layer may multiply the input data by weights and, for example, generate activation function results for the nodes (h 1 (1) ,h 2 (1) ,…,h m1-1 (1) ,h m1 (1) ), where m is a natural number equal to or greater than 1. The hidden layer may be composed of multiple layers. Although not shown in FIG. 5A, for example, another hidden layer may be composed of multiple nodes (h 1 (2) ,h 2 (2) ,…,h m2-1 (2) ,hm 2 (2) The output layer is also a layer composed of the final hidden layer or input layer, which is multiplied by a weight value (weight) to generate the result value of the output function, for example, 1 ,…,o k ), where k is a natural number equal to or greater than 1. In one example, the result value of the output function is also state data of the contactor generated based on RF signal data generated from the contactor.

[0087] The functions may be at least one of an activation function, an output function, and a loss function. The activation function may be a function that processes the sum of products of a node and a weight, and may calculate the node value of the hidden layer. The weight may indicate the strength of a connection between nodes. The summation may indicate the summation of products of a weight and a node.

[0088] Referring to the example in Figure 5A, the kth hidden node (h k (l) )teeth,

[0089]

number

[0090] where f (l) is the activation function for the l-th hidden layer, and w k (l) is the weight vector, and b k (l) is also the bias for the k-th node in the (l+1)-th layer. The output function is a function that processes the product of the node in the last hidden layer and the weight value. For example, the node value (o 1 ,…,o k ) can be calculated. The loss function is also a function that measures the error between the output function result and the response value for learning the weights.

[0091] According to one embodiment, the artificial neural network model is also a neural network model that is pre-trained with training data for a number of signal strength values ​​generated by pre-processing the contactor states switched by the control signals of the battery management unit and the RF signals generated when the contactor states are switched.

[0092] FIG. 5B illustrates a schematic diagram of an example of a pre-trained artificial neural network model, according to an embodiment of the present invention.

[0093] The artificial neural network according to an embodiment of the present invention may be an artificial neural network based on a convolutional neural network (CNN) model as shown in FIG. 5B. In this case, the CNN model may be composed of a plurality of hidden layers and a classification layer. The CNN model may also be a hierarchical model in which a plurality of calculation layers (convolutional layer, pooling layer) are alternately performed in a plurality of hidden layers, and finally used to extract features of input data (INPUT). In this case, the processor 30 (FIG. 3) according to an embodiment of the present invention may process the learning data by a supervised learning technique to construct or train the artificial neural network model.

[0094] A processor 30 (FIG. 3) according to an embodiment of the present invention may generate a convolution layer for extracting feature values ​​of input data (INPUT) and a pooling layer for combining the extracted feature values ​​to form a feature map. In the convolution layer, feature information required for classification of the input data (INPUT) may be extracted. The processor 30 (FIG. 3) may generate a feature map based on the feature information of the input data (INPUT). A rectified linear unit (ReLu) activation function may be used as the next stage of the convolution layer. The ReLu activation function is a nonlinear function and may increase nonlinearity in a layer that exhibits basic linear characteristics. In the pooling layer, output data of the convolution layer may be input and the size of the output data (activation map) may be reduced or specific data may be emphasized.

[0095] The processor 30 (FIG. 3) according to an embodiment of the present invention may generate a fully connected layer that combines feature maps and prepares to determine the probability that the input data (INPUT) corresponds to each of a plurality of items. The fully connected layer is also a stage that determines classification in the final stage of the CNN model. The fully connected layer may be configured as a fully connected layer that performs flattening to convert each layer into a one-dimensional vector and then connects the layers converted into one-dimensional vectors into one vector. Finally, the class with the highest probability may be classified as an output using a Softmax function. The processor 30 (FIG. 3) according to an embodiment of the present invention may calculate an output layer including an output corresponding to the input data (INPUT).

[0096] FIG. 5C is a schematic illustration of an example of the training process of a pre-trained artificial neural network model, according to an embodiment of the present invention.

[0097] Referring to FIG. 5C, an antenna 210 of a software defined radio (SDR) 200 (FIG. 2) receives an RF signal generated when a contactor changes state, and illustrates a convolutional neural network (CNN) model that uses RF raw signal data as input data. For example, the RF raw signal data may include a noise signal. The RF raw signal data may include a pattern of the noise signal. The convolutional neural network (CNN) model may be trained using a plurality of signal strength values ​​generated by preprocessing the state of the contactor and the RF signal generated when the contactor changes state as training data.

[0098] The Conv1D Layer in FIG. 5C may represent the convolution layer in FIG. 5B. In the Conv1D Layer, a filter may be used to learn the features of the input data (feature learning). For example, in the Conv1D Layer, an RF signal may be received as input data, and a pattern of the RF signal may be learned as a feature of the input data. The ReLU Activation Function in FIG. 5C may represent the ReLU in FIG. 5B. In the ReLU Activation Function, non-linearity may be added to the input data. For example, the CNN model may add non-linearity to the RF signal received as input data, and learn a more complex pattern of the RF signal. The Pooling Layer in FIG. 5C may represent the pooling layer in FIG. 5B. The Pooling Layer may downsample data to prevent overfitting and improve computation efficiency. The above-mentioned Conv1D Layer process, ReLU Activation Function process, and Pooling Layer process may be repeated. The Flatten Layer and the Fully Connected Layer are as described above in FIG. 5B.

[0099] Output Data is output data generated and classified based on input data. In one example, the output data is a state value of a contactor and a confidence score of the state value of the contactor. For example, the state of the contactor state value is any one of open, closed, and weld. The confidence score of the state value of the contactor is a value between 0 and 1.

[0100] The artificial neural network model may receive a plurality of RF signal data and, based on the data learned as described above, may transmit, for example, contactor state estimates (diagnostic data) to a battery management system (BMS).

[0101] The artificial neural network based on the convolutional neural network (CNN) model shown in Figures 5B and 5C is merely exemplary, and the concept of the present invention is not limited thereto. Such an artificial neural network may be stored in the memory as a function coefficient that defines the coefficients of at least one node constituting the artificial neural network, the weights of the nodes, and the relationships with the layers constituting the artificial neural network. It goes without saying that the structure of the artificial neural network may also be stored in the memory as a source code and / or a program.

[0102] The types and / or structures of the artificial neural network described in FIG. 5 (FIG. 5A, FIG. 5B, and FIG. 5C) are exemplary, and the concept of the present invention is not limited thereto. Therefore, various types of models of artificial neural networks may correspond to the "artificial neural network" described throughout the specification. Although not shown in FIG. 5 (FIG. 5A, FIG. 5B, and FIG. 5C), models such as, for example, deep neural network (DNN), convolution neural network (CNN), recurrent neural network (RNN), and bidirectional recurrent deep neural network (BRRDNN) may be used as an artificial neural network according to an embodiment of the present invention, but are not limited thereto.

[0103] The various embodiments of the present invention are not intended to limit the scope of the present invention in any manner. For the sake of brevity, descriptions of conventional electronic configurations, control systems, software, and other functional aspects of the system may be omitted. In addition, the line connections or connecting members between components shown in the drawings are illustrative of functional connections and / or physical or circuit connections, and may be embodied in an actual device by various alternative or additional functional connections, physical connections, or circuit connections. In addition, unless specifically stated as "essential" or "important," a component is not necessarily required for the implementation of the present invention.

[0104] In the present specification (particularly, in the claims), the use of the term "said" and similar indicators may be used in both the singular and the plural. In addition, when a range is described in the present invention, it includes the invention to which each individual value within the range is applied (unless otherwise specified), and each individual value constituting the range is described in the detailed description of the invention. Finally, unless the order of steps constituting the method of the present invention is clearly described or there is no description to the contrary, the steps may be performed in any suitable order. The order of the steps described above is not necessarily intended to limit the scope of the present invention. In the present invention, the use of all examples or exemplary terms (e.g., "etc.", etc.) is merely for the purpose of explaining the present invention in detail, and the scope of the present invention is not limited by the above examples or exemplary terms, since it is not limited by the claims. In addition, a person skilled in the art will know that various modifications, combinations and changes can be made within the scope of the claims or their equivalents, depending on design conditions and factors.

[0105] Therefore, the spirit of the present invention is not limited to the above-described embodiments, and not only the scope of the claims, but also all scopes that are equivalent to the scope of the claims or equivalently modified therefrom belong to the scope of the spirit of the present invention. [Explanation of symbols]

[0106] 10 Computing Equipment 11 Antenna 20 Memory 30 processors 100 Battery management unit 200 Signal processing section 210 Antenna 220 RF Tuner 230 Analog-to-Digital Converter 240 Digital Signal Processor 250 Signal Processing Unit Controller 300 Contactor state estimation unit 1000 Battery Management System (BMS)

Claims

1. a contactor coupled to the battery; a battery management unit that outputs a control signal to the contactor for controlling a state of the contactor; a signal processor that receives a radio frequency (RF) signal generated when the contactor changes state, and generates a plurality of signal strength values ​​corresponding to a plurality of frequency bands based on the RF signal; a contactor state estimator that receives the plurality of signal strength values, inputs the plurality of signal strength values ​​into a pre-trained neural network model, and generates an estimate that estimates a state of the contactor.

2. The battery management unit outputs a trigger signal to the signal processing unit in response to the output of the control signal; The battery management system according to claim 1 , wherein the signal processing unit captures the RF signal in response to the trigger signal.

3. The signal processing unit includes: an antenna for receiving the RF signal; an RF tuner that outputs a band signal of a preset frequency band among the RF signals; an analog-to-digital converter (ADC) for converting the band signal into a digital signal; a digital signal processor (DSP) for Fourier transforming the digital signal into a frequency domain signal and generating the plurality of signal strength values ​​from the frequency domain signal; 3. The battery management system of claim 2, further comprising: a signal processor controller responsive to the trigger signal to control the antenna, the RF tuner, the analog-to-digital converter (ADC) and the digital signal processor (DSP) and transmit the plurality of signal strength values ​​to the contactor state estimator.

4. The battery management system according to claim 3 , wherein the preset frequency band is equal to or higher than 100 kHz and equal to or lower than 5 MHz.

5. The battery management system of claim 1 , wherein the widths of the frequency bands are the same in a logarithmic scale.

6. The battery management system according to claim 1 , wherein the widths of the plurality of frequency bands are the same.

7. the neural network model is a convolutional neural network (CNN) model; 2. The battery management system according to claim 1, wherein the contactor state is pre-learned using learning data including a plurality of signal strength values ​​generated by pre-processing the contactor state switched by the control signal of the battery management unit and the RF signal generated when the contactor state is switched.

8. The battery management system of claim 1 , wherein the control signals include a turn-on signal for switching the contactor to an open state and a turn-off signal for switching the contactor to a closed state.

9. The battery management system according to claim 8 , wherein the pre-trained neural network model outputs the estimated value based on the input signal strength values ​​and a reliability score of the estimated value.

10. The battery management system according to claim 9 , wherein the battery management unit detects whether the state of the contactor is abnormal based on the estimated value and a reliability of the estimated value.

11. The battery management unit 11. The battery management system according to claim 10, wherein the state of the contactor is determined to be an abnormal state for at least one of the closed state based on the state estimate value of the contactor in response to the turn-on signal, the open state based on the state estimate value of the contactor in response to the turn-off signal, and a fused state based on the state estimate value of the contactor.

12. 1. A method of operating a battery management system (BMS) including a contactor coupled to a battery, comprising: outputting a control signal to the contactor for controlling a state of the contactor; receiving a radio frequency (RF) signal generated when the contactor changes state; generating a plurality of signal strength values ​​corresponding to a plurality of frequency bands based on the RF signal; and inputting the plurality of signal strength values ​​into a pre-trained neural network model to generate an estimate that estimates a state of the contactor.

13. extracting a trigger signal in response to the output of the control signal; The method of claim 12 , further comprising: capturing the RF signal in response to the trigger signal.

14. receiving the RF signal; outputting a band signal of a preset frequency band among the RF signals; converting the bandpass signal into a digital signal; Fourier transforming the digital signal into a frequency domain signal and generating the plurality of signal strength values ​​from the frequency domain signal; The method of claim 13 , further comprising: extracting the plurality of signal strength values ​​in response to the trigger signal.

15. 13. The method of claim 12, wherein the control signals include a turn-on signal for switching the contactor to an open state and a turn-off signal for switching the contactor to a closed state.

16. 16. The method of claim 15, wherein the pre-trained neural network model generates the estimation value based on the input signal strength values ​​and a reliability score for the estimation value.

17. The method of claim 16, further comprising: detecting whether a state of the contactor is abnormal based on the estimated value and a reliability of the estimated value.

18. 20. The method of claim 17, further comprising determining that a state of the contactor is an abnormal state for at least one of the closed state according to the state estimate value of the contactor in response to the turn-on signal, the open state according to the state estimate value of the contactor in response to the turn-off signal, and a fused state according to the state estimate value of the contactor.

19. A computer program stored on a medium for causing a computing device to carry out a method according to any one of claims 12 to 18.

Images