Estimation device, estimation method, and computer program

Abstract

To provide an estimation device, an estimation method, and a computer program which can accurately estimate deterioration in a power storage element.SOLUTION: An estimation device 101 includes: an acquisition section 201 for acquiring time-series data of an SOC in a power storage element; a decomposition section 202 for decomposing a waveform of variation in the SOC in the time-series data into waveform components in a plurality of frequency regions; and an estimation section 204 for estimating deterioration in the power storage element on the basis of the intensity and a deterioration factor of each waveform component.SELECTED DRAWING: Figure 8

Description

[Technical field]

[0001] The present invention relates to an estimation device, an estimation method, and a computer program for estimating deterioration of a storage element. [Background technology]

[0002] Energy storage elements that can store electrical energy and supply it as a power source when needed are used. Energy storage elements are applied to portable devices, power supply devices, transportation equipment including automobiles and railways, and industrial equipment including aviation, space, and construction. It is important to constantly grasp the storage capacity of the energy storage element so that the stored energy can be used when needed and in the amount needed. It is known that energy storage elements deteriorate mainly chemically depending on time and frequency of use. Therefore, the usable energy decreases depending on time and frequency of use. It is important to grasp the deterioration state of the energy storage element in order to use the energy when needed and in the amount needed. So far, technologies have been developed to estimate the deterioration of the energy storage element. There is a demand for a technique that can accurately estimate the state of health (SOH) of an energy storage element even if the state of charge (SOC) of the energy storage element varies in various ways.

[0003] As shown in Patent Document 1, the inventors have found that the degradation value is large when the SOC fluctuation range around a certain SOC is large, and that even if the SOC fluctuation range is the same, the degradation value changes depending on the central SOC. They have developed an estimation device that estimates the degradation of a storage element based on the magnitude of this SOC fluctuation and the central SOC. In the estimation device of Patent Document 1, the degradation value is estimated to be large according to the magnitude of the SOC fluctuation. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent No. 6428957 Summary of the Invention [Problem to be solved by the invention]

[0005] Conventional methods may not be able to estimate the deterioration of an energy storage element with sufficient accuracy. Even if the SOC fluctuation range is small, degradation may occur when charging at a high rate. There is a demand for a method to better estimate degradation even for complex SOC fluctuation patterns and fluctuation patterns with a small SOC fluctuation range.

[0006] An object of the present invention is to provide an estimation device, an estimation method, and a computer program capable of accurately estimating the deterioration of a storage element. [Means for solving the problem]

[0007] An estimation device according to one embodiment of the present invention includes an acquisition unit that acquires time series data of an SOC of a storage element, a decomposition unit that decomposes a waveform of the SOC fluctuations in the time series data into frequency components, and an estimation unit that estimates deterioration of the storage element based on the frequency components.

[0008] An estimation method according to one aspect of the present invention includes acquiring time-series data of an SOC of an energy storage element, The waveform of the fluctuation in the SOC in the time-series data is decomposed into frequency components, and the deterioration of the power storage element is estimated based on the frequency components.

[0009] A computer program according to one embodiment of the present invention causes a computer to execute a process of acquiring time series data of SOC of an energy storage element, decomposing a waveform of the SOC fluctuation in the time series data into frequency components, and estimating deterioration of the energy storage element based on the frequency components. Effect of the Invention

[0010] In the present invention, the deterioration of the energy storage element can be estimated with high accuracy. [Brief description of the drawings]

[0011] [Figure 1] FIG. 13 is a diagram showing an example of change in degradation value of a power storage element due to energization with respect to the fluctuation range of SOC. [Diagram 2] FIG. 13 is a diagram showing an example of a change in degradation value of a power storage element due to current conduction with respect to a central SOC. [Diagram 3] 1 is a graph showing the relationship between time and SOC when the SOC is varied by Δ0.5%, Δ1.5%, Δ5%, Δ20%, and Δ30%. [Figure 4A] This is a graph showing the waveform in FIG. 3 decomposed into waveform components in a plurality of frequency domains through Fourier transformation. [Figure 4B] This is a graph showing the waveform in FIG. 3 decomposed into waveform components in a plurality of frequency domains through Fourier transformation. [Figure 4C] This is a graph showing the waveform in FIG. 3 decomposed into waveform components in a plurality of frequency domains through Fourier transformation. [Figure 4D] This is a graph showing the waveform in FIG. 3 decomposed into waveform components in a plurality of frequency domains through Fourier transformation. [Figure 4E] This is a graph showing the waveform in FIG. 3 decomposed into waveform components in a plurality of frequency domains through Fourier transformation. [Diagram 5] 1 is a graph showing the relationship between time and SOC when the SOC is varied by Δ1.5% and Δ20%. [Figure 6] This is a graph showing the waveform in FIG. 5 decomposed into waveform components in a plurality of frequency domains through Fourier transformation. [Figure 7] FIG. 2 is a diagram illustrating a configuration of a monitoring device. [Figure 8] FIG. 1 is a diagram illustrating a configuration of an estimation device. [Figure 9] 10 is a flowchart showing a procedure of a process performed by the estimation device to calculate a degradation value based on time-series data. [Figure 10] 1 is a graph showing the relationship between time and SOC when the SOC is varied by Δ0.5%, Δ1.5%, Δ5%, Δ20%, and Δ30%. [Figure 11] This is a graph showing the waveform in FIG. 10 decomposed into waveform components in a plurality of frequency domains through Fourier transformation. [Figure 12]1 is a graph showing the relationship between the number of days and cycle deterioration when the SOC is varied by Δ0.5%, Δ1.5%, Δ5%, Δ20%, Δ30%, and Δ100%. [Figure 13] 1 is a graph showing the relationship between the amplitude spectrum, the frequency, and the degradation coefficient. [Figure 14] 13 is a graph showing the relationship between the amplitude spectrum, the frequency, and the degradation coefficient when interpolated by an interpolation calculation based on k in each graph of FIG. 12. [Figure 15] 13 is a graph showing the relationship between estimated values ​​and actual measured values. [Figure 16] 1 is a graph showing time series data of SOC. [Figure 17] FIG. 11 is a block diagram showing an example of a configuration of an estimation device according to a second embodiment. [Figure 18] FIG. 11 is an explanatory diagram showing an example of a record layout of a teacher data DB. [Figure 19] FIG. 2 is a schematic diagram showing a configuration of a degradation coefficient output model. [Figure 20] 10 is a flowchart illustrating an example of a processing procedure for generating a degradation coefficient output model by a control unit. [Figure 21] 4 is a flowchart defining an operation procedure when the estimation device estimates deterioration of a storage element. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0012] (Outline of the embodiment) The estimation device of the embodiment includes an acquisition unit that acquires time series data of SOC in a storage element, a decomposition unit that decomposes a waveform of the SOC fluctuation in the time series data into frequency components, and an estimation unit that estimates deterioration of the storage element based on the frequency components.

[0013] According to the above configuration, since the waveform of the SOC fluctuation is decomposed into frequency components, it is possible to detect a waveform with a large fluctuation but a long period (fluctuation time) and a waveform with a small fluctuation but a short period. A waveform with a large fluctuation but a long period has a large intensity (spectral intensity, amplitude spectrum in the case of Fourier transform) and a low-frequency waveform component (frequency component). A waveform with a small fluctuation but a short period has a small intensity and a high-frequency waveform component. If the fluctuation is large but the period is very long, the deterioration is small. If the fluctuation is small but the period is very short, the deterioration is large. If the storage element is a lithium-ion secondary battery and rapid charging is performed frequently on the high SOC side, lithium electrodeposition may occur at the negative electrode, resulting in large deterioration. Although the waveform of the SOC fluctuation differs depending on the characteristics of the storage element and the user's usage, according to the above configuration, any waveform can be detected and the deterioration of the storage element can be estimated well.

[0014] In the above-described estimation device, the acquisition unit may acquire time-series data of SOC for a predetermined period, and the decomposition unit may process a function of a time change in SOC fluctuation for the predetermined period.

[0015] According to the above configuration, the time change of SOC fluctuation in a predetermined period is acquired and the time change of the SOC fluctuation is analyzed. Deterioration is not estimated sequentially from the SOC fluctuation data every moment, but the time change of SOC fluctuation in a predetermined period is acquired and deterioration is estimated in a batch processing manner. What SOC change contributes to deterioration in the time change of the SOC fluctuation width in a certain time width is analyzed, and deterioration can be estimated well.

[0016] In the above-described estimation device, the decomposition unit may decompose a waveform of the SOC fluctuation in the time-series data into waveform components in a plurality of frequency domains, and the estimation unit may estimate the deterioration based on an intensity of each waveform component and a deterioration coefficient.

[0017] According to the above configuration, degradation is estimated based on the magnitude of SOC fluctuation and the degradation coefficient. While taking into consideration that the amount of change in SOH increases in accordance with the fluctuation range of SOC, degradation can be accurately estimated based on the actual complex fluctuation pattern, and also considering waveforms with a very small fluctuation range.

[0018] The above-described estimation device may further include a specification unit that specifies a deterioration coefficient based on the waveform components, and the estimation unit may estimate the deterioration based on the deterioration coefficient specified by the specification unit.

[0019] According to the above configuration, the deterioration coefficient corresponding to the SOC fluctuation range and the SOC center and used for calculating the deterioration is specified based on the waveform components. The deterioration coefficient is specified based on the distribution of the waveform components in the frequency domain, the intensity and frequency of the waveform components of the high peaks, etc., and the deterioration can be easily estimated.

[0020] In the above-described estimation device, the specifying unit may specify the deterioration coefficient based on a relationship between an intensity and a frequency of a waveform component and the deterioration coefficient.

[0021] According to the above configuration, the relationship between the intensity and frequency of the waveform component and the degradation coefficient is obtained in advance, and the degradation coefficient is specified based on this relationship, making it possible to easily and satisfactorily estimate degradation.

[0022] In the above-described estimation device, the determination unit may determine the deterioration coefficient by weighting waveform components with greater intensity in comparison with waveform components with smaller intensity.

[0023] Weighting is performed based on the distribution of the waveform components in the frequency domain, the intensity and frequency of the waveform components with high peaks, etc., and the deterioration coefficient is determined based on the amplitude and frequency of the waveform components obtained by weighting, so that deterioration can be accurately estimated.

[0024] In the above-described estimation device, the determination unit may input the waveform components acquired by the decomposition unit to a learning model that outputs a degradation coefficient when a waveform component is input, and determine a degradation coefficient.

[0025] According to the above configuration, the deterioration coefficient is specified taking into consideration waveform components with low spectral intensity and high frequency, and deterioration can be properly estimated.

[0026] The estimation method according to the embodiment includes acquiring time series data of SOC of a storage element, decomposing a waveform of the SOC fluctuation in the time series data into frequency components, and estimating deterioration of the storage element based on the frequency components. The waveform of the SOC fluctuation in the time series data may be decomposed into waveform components in a plurality of frequency domains, and estimating deterioration of the storage element based on the intensity and deterioration coefficient of each waveform component.

[0027] According to the above configuration, since the waveform of the SOC fluctuation is resolved, it is possible to detect waveforms with large fluctuations but long periods, and waveforms with small fluctuations but short periods. Either waveform can be detected according to the characteristics of the storage element and the user's usage, and deterioration of the storage element can be estimated well.

[0028] A computer program according to an embodiment causes a computer to execute a process of acquiring time series data of an SOC of an energy storage element, decomposing a waveform of the SOC fluctuation in the time series data into frequency components, and estimating deterioration of the energy storage element based on the frequency components. The computer may also execute a process of decomposing a waveform of the SOC fluctuation in the time series data into waveform components in a plurality of frequency domains, and estimating deterioration of the energy storage element based on the intensity and deterioration coefficient of each waveform component.

[0029] Fig. 1 is a diagram showing an example of change in deterioration amount due to current flow through a storage element versus SOC fluctuation range. In Fig. 1, the vertical axis shows the difference between the deterioration amount when a predetermined amount of electricity is passed through and the deterioration amount at a 3% SOC fluctuation range, and the horizontal axis shows the SOC fluctuation range. In FIG. 1, the amount of deterioration due to current flow after a predetermined number of charge / discharge cycles are repeated so that the central SOC is 60% is plotted against the range of SOC fluctuation.

[0030] As shown in Figure 1, the inventors discovered that even if the central SOC is the same, the amount of degradation due to current flow changes when the SOC fluctuation range is different. They discovered that the degradation due to current flow increases according to the magnitude of the SOC fluctuation. The mechanism of this phenomenon has not yet been fully elucidated. The inventors speculate that the larger the fluctuation in SOC, the more significant the expansion (during charging) and contraction (during discharging) of the negative electrode becomes, causing partial destruction of the SEI coating formed on the surface of the negative electrode, which results in greater degradation of the energy storage element due to current flow.

[0031] Fig. 2 is a diagram showing an example of change in deterioration amount due to current flow through a storage element relative to a central SOC. In Fig. 2, the vertical axis shows the difference between the deterioration amount when a predetermined amount of electricity is flowing and the deterioration amount at a central SOC of 10%, and the horizontal axis shows the central SOC, which is the center of SOC fluctuation. Here, the central SOC refers to the center of SOC fluctuation in SOC time-series data.

[0032] In FIG. 2, the amount of deterioration due to current flow after a predetermined number of charge / discharge cycles are repeated so that the SOC fluctuation band is 10% is plotted against the central SOC. Here, an example of the charge / discharge operation will be described. When the central SOC is 10% and charging / discharging is repeated so that the SOC fluctuation range is 10%, it means repeating charging / discharging so that the SOC fluctuates between 0% and 20%. It can be seen that even if the fluctuation amount is the same, the amount of deterioration differs greatly depending on the central SOC.

[0033] Based on the above, SOH needs to be estimated taking into account not only the total amount of change in SOC but also the range of SOC fluctuations and the median SOC. As shown in Patent Document 1, the inventor of the present application has developed an estimation device that focuses on the SOC fluctuation range and estimates the degradation of a storage element based on the magnitude of SOC fluctuation and the central SOC. As described above, in Patent Document 1, the degradation value is set to increase according to the magnitude of SOC fluctuation, thereby improving the accuracy of the estimation. It is desired to detect waveforms with a very small SOC fluctuation range and estimate the degradation value more accurately for complex SOC fluctuation patterns.

[0034] The waveform of the SOC fluctuation during a predetermined acquisition period is decomposed into waveform components in a plurality of frequency domains, and the deterioration of the storage element is estimated based on the strength of the waveform components and the deterioration coefficient. The inventors have found that by decomposing into waveform components, it is possible to simulate an actual complex fluctuation pattern while taking into consideration that the amount of change in SOH increases in accordance with the fluctuation width of the SOC, and to detect waveforms with a plurality of fluctuation widths, including waveforms with a very small fluctuation width, and to more satisfactorily estimate the deterioration, thereby completing the present invention. The present invention can be restated as follows. The inventors have found that the accuracy of degradation estimation may be improved by acquiring the time change of SOC fluctuation over a certain time width (fixed period) and analyzing the time change of SOC fluctuation. That is, instead of sequentially estimating degradation from SOC fluctuation data from moment to moment, this method acquires the time change of SOC fluctuation over a certain time width (fixed period) and estimates degradation in a batch processing (collective processing) manner. The point of the present invention is to analyze what SOC changes contribute to degradation in the time change of SOC fluctuation width over a certain time width and estimate degradation. Therefore, in the time change of SOC fluctuation width over a certain period, the time change of SOC fluctuation is expressed by another variable, and the time change of the original SOC fluctuation width is expressed by another function. A common method for this is frequency conversion, and a widely known method is Fourier transformation. Generally, a fast Fourier transform (FFT) is performed. When expressing the time change of SOC fluctuation over a certain time span (certain period) using another function, it is best to express it using multiple (many) functions. However, because this takes time for analysis (control), it is acceptable to express it using only the function that has the greatest contribution to the time change of SOC fluctuation. Alternatively, an averaging process may be performed on the time change of the SOC fluctuation over a certain time width (fixed period).

[0035] Figure 3 is a graph showing the relationship between time and SOC when SOC is changed by Δ0.5%, Δ1.5%, Δ5%, Δ20%, and Δ30%. The horizontal axis is time [seconds], and the vertical axis is SOC [%]. The graphs for Δ0.5%, Δ1.5%, Δ5%, Δ20%, and Δ30% are shown as a, b, c, d, and e.

[0036] 4A-E are graphs in which the waveform in FIG. 3 is Fourier transformed and decomposed into waveform components in multiple frequency domains. The horizontal axis is frequency [Hz], and the vertical axis is amplitude (amplitude spectrum) [%]. By using Fourier transform, a complex waveform can be decomposed into multiple sine waveforms, and the obtained f(ω) can be shown in a graph with the horizontal axis being angular frequency and the vertical axis being the magnitude (amplitude) of the sine wave at each angular frequency ω. In FIG. 4, the angular frequency is replaced with frequency. The fluctuation width of the waveform in FIG. 3 is related to the amplitude of the waveform components in FIG. 4, and the period of the waveform in FIG. 3 corresponds to the center of gravity of the frequency of some waveform components in FIG. 4. By using Fourier transform, the complex waveform in FIG. 3 is frequency converted to f(ω), and the contribution of f(ω) is expressed as spectrum intensity (peak height or peak area). By acquiring time series data of SOC for a short acquisition period, long-term time series data can be estimated. The usage pattern of the storage element of a user is often almost fixed.

[0037] The graphs in FIGS. 4A to 4E correspond to the waveforms a, b, c, d, and e in FIG. 3, respectively. As shown in Figures 4A to 4E, the waveforms a, b, c, d, and e in Figure 3 are each decomposed into waveform components in a plurality of frequency domains. The peak component of the waveform component corresponding to the waveform a has a frequency of approximately 0.013 and an amplitude of approximately 0.133, as shown in Figure 4A. The peak component of the waveform component in Figure 4E, which corresponds to the waveform e of Δ30% in Figure 3, has a large amplitude because the fluctuation of the waveform e in Figure 3 is large, and a long period, so the frequency is small. The peak component of the waveform component in Figure 4B, which corresponds to the waveform b of Δ1.5%, has a small amplitude because the fluctuation of the waveform b in Figure 3 is small, and a short period, so the frequency is large.

[0038] Figure 5 is a graph showing the relationship between time and SOC when SOC is changed by Δ1.5% and Δ20%. The horizontal axis is time [seconds], and the vertical axis is SOC [%]. Graphs a, b, and c show the graphs for Δ1.5%, Δ20%, and Δ1.5% combined with the graph for Δ20%. Figure 6 is a graph in which the waveform in Figure 5 has been Fourier transformed and broken down into waveform components in multiple frequency domains. The horizontal axis is frequency (Hz) and the vertical axis is amplitude. Figure 6 shows that most of the waveform components correspond to waveform b with a Δ20%, and this waveform component contains a waveform component corresponding to waveform a with a Δ1.5%. Degradation can be estimated based on waveform components with large amplitudes. In other words, degradation can be estimated by weighting based on waveform components with large amplitudes.

[0039] The estimation method according to this embodiment acquires time-series data of the SOC of an energy storage element, and decomposes the waveform of the SOC fluctuation in the time-series data into waveform components in multiple frequency domains. The waveform components are weighted toward waveform components with large amplitudes compared to waveform components with small strengths (amplitudes). The relationship between the amplitude, frequency, and degradation coefficient is obtained in advance. The degradation coefficient k is specified based on the amplitude and frequency of the waveform components of importance obtained by weighting. The degradation value (cycle degradation) is calculated using the following formula (1) based on the specified degradation coefficient k. Degradation value = k × √t (1) Here, t is the elapsed time. The degradation value increases, for example, according to a root law. Increasing according to a root law means that the increment of the degradation value per unit time gradually decreases as time passes (see FIG. 15).

[0040] (Embodiment 1) Hereinafter, the present invention will be specifically described with reference to the drawings showing the embodiment. 7 is a diagram showing the configuration of the monitoring device 151. The monitoring device 151 includes an A / D conversion unit 53, a history creation unit 54, a counter 55, a storage unit 56, a communication unit 57, and an estimation device 101. A current sensor 41, a voltage sensor 42, and a temperature sensor 43 are connected to the monitoring device 151.

[0041] Some of the components included in the monitoring device 151 may be located away from the other components. For example, the estimation device 101 may be located in a remote location and communicate with the communication unit 57. Also, a server located in a remote location and connected to a network may function as the estimation device 101.

[0042] The monitoring device 151 monitors the deterioration of the power storage element to be monitored (a lithium ion secondary battery in this embodiment). The monitoring device 151 may monitor one battery cell, or may monitor a plurality of battery cells (battery assembly) connected in series or in parallel. The monitoring device 151 may constitute a power storage device (battery pack) together with the battery assembly.

[0043] The monitored power storage element is not limited to a non-aqueous electrolyte secondary battery such as a lithium ion secondary battery, and may be another electrochemical cell to which the hypothesis, algorithm, and mathematical model described below are applicable. Hereinafter, the monitored power storage element is also simply referred to as a battery.

[0044] The counter 55 in the monitoring device 151 counts clock pulses generated by an oscillator circuit using a quartz crystal oscillator or the like, and holds the counted value. This count value may indicate the current time.

[0045] The current sensor 41 measures the current charged to the storage element and the current discharged from the storage element, and outputs to the A / D conversion unit 53 an analog signal Ai indicating the measurement result.

[0046] The voltage sensor 42 measures the voltage between the positive and negative electrodes of the power storage element, and outputs to the A / D conversion unit 53 an analog signal Av indicating the measurement result.

[0047] The temperature sensor 43 measures the temperature T at a predetermined location of the power storage element, and outputs to the A / D conversion unit 53 an analog signal At indicating the measurement result.

[0048] The A / D converter 53 converts, for example, analog signals Ai, Av, and At received from the current sensor 41, voltage sensor 42, and temperature sensor 43, respectively, into digital signals Di, Dv, and Dt at each predetermined sampling time.

[0049] The history creation unit 54 stores the count value of the counter 55 at the sampling time and the digital signals Di, Dv, and Dt in the memory unit 56. The memory unit 56 accumulates the sampling time, the current value, the voltage value, and the temperature T for each sampling time. The memory unit 56 stores the number of charge / discharge cycles, and the number of charge / discharge cycles is updated each time charge / discharge is repeated.

[0050] The communication unit 57 may communicate with other devices, such as a main control device (main ECU (Electronic Control Unit)) in the vehicle, a personal computer, a server, a smartphone, and a terminal for servicing the power storage element.

[0051] For example, when the communication unit 57 receives a command to estimate the degradation state of the power storage element from another device, the communication unit 57 outputs the received estimation command to the estimation device 101.

[0052] FIG. 8 is a diagram showing the configuration of the estimation device 101. 8, the estimation device 101 includes a control unit 20, a storage unit 23, and an interface unit 24. The interface unit 24 is configured with, for example, a LAN interface and a USB interface, and communicates with other devices such as a monitoring device 151 by wire or wirelessly. A signal line or a terminal extending from the estimation device 101 to the communication unit 57 may function as an output unit that outputs the estimation result, etc. The communication unit 57 may function as the output unit. When different input data is input to the estimation device 101, a different output is obtained from the output unit. When a different SOC fluctuation range (and / or center SOC) is input to the estimation device 101, the output unit may output a different output (for example, a voltage value, a duty ratio). A display unit (or a notification unit) that displays the output result may be connected to the output unit. The output from the output unit may be displayed on the display unit (or the notification unit) via the communication unit 57.

[0053] A deterioration estimation program 231 for executing a deterioration estimation process described later is stored in the storage unit 23. The deterioration estimation program 231 is provided in a state stored in a computer-readable recording medium 60 such as a CD-ROM, a DVD-ROM, or a USB memory, and is stored in the storage unit 23 by installing it in the estimation device 101. The deterioration estimation program 231 may also be obtained from an external computer (not shown) connected to a communication network and stored in the storage unit 23. The storage unit 23 stores a degradation coefficient DB 232 required for degradation estimation processing. The degradation coefficient DB 232 stores the relationship between the amplitude and frequency of the waveform components and the degradation coefficient for each acquisition period during which time-series data of, for example, the center SOC, temperature, and SOC is acquired. Note that if any of the center SOC, temperature, and acquisition period is constant, for example, the acquisition period is constant, it is not necessary to store the relationship for each acquisition period. For example, for each temperature T and for each SOC center, the time change of the degradation value when the SOC is varied by Δ0.5%, Δ1.5%, Δ5%, Δ20%, and Δ30% is measured by a preliminary test. The degradation coefficient k is calculated based on the measurement results of this test.

[0054] The control unit 20 is configured with, for example, a CPU, a ROM, a RAM, etc., and controls the operation of the estimation device 101 by executing computer programs such as a deterioration estimation program 231 read from the storage unit 23. The control unit 20 functions as a processing unit that executes deterioration estimation processing by reading and executing the deterioration estimation program 231.

[0055] The control unit 20 includes an acquisition unit 201 , a decomposition unit 202 , an identification unit 203 , and an estimation unit 204 . The acquisition unit 201 acquires time-series data of the SOC of the energy storage element. More specifically, when the acquisition unit 21 receives an estimation command from the communication unit 57, it acquires each sampling time and the current value, voltage value, and temperature T at each sampling time from the memory unit 56 in the monitoring device 151 via the interface unit 24 in accordance with the received estimation command. The sampling interval and the acquisition period during which sampling is performed are determined based on the characteristics and usage of the energy storage element. If the sampling interval is short, the waveform may be cut off. If the fluctuation range is large and the cycle is long, it is necessary to extend the acquisition period. In this manner, the acquisition unit 21 acquires from the storage unit 56 the data measured after the start of use of the energy storage element. Alternatively, the acquiring unit 21 may acquire the data from a data file.

[0056] The acquisition unit 21 reserves a memory area for storing data regarding the sampling time, the SOC, and the temperature. The acquisition unit 21, for example, calculates the amount of electricity passed through the power storage element by tallying up the current values ​​at each sampling time, and converts the calculated amount of electricity into a change in SOC. The acquisition unit 21 then calculates the SOC at each sampling time based on the conversion result. The acquisition unit 21 may correct the SOC using, for example, a measured value of the open circuit voltage.

[0057] The decomposing unit 202 performs a Fourier transform on the waveform of the SOC fluctuation in the SOC time-series data acquired by the acquiring unit 21, and decomposes the waveform into waveform components in a plurality of frequency domains.

[0058] The determination unit 203 weights the waveform components having large amplitudes compared to the waveform components having small amplitudes. The weighting is performed based on the peak height of each waveform component, the area of ​​each waveform component, etc. The determination unit 203 determines the deterioration coefficient k based on the amplitude and frequency of the waveform component of importance obtained by performing the weighting.

[0059] The estimation unit 204 calculates a deterioration value, for example, cycle deterioration (amount of decrease in capacity) by the above formula (1) based on the deterioration coefficient k identified by the identification unit 203. The estimation unit 204 may transmit estimation result information indicating the calculated degradation value to another device via the communication unit 57 as a response to the estimation command.

[0060] The estimation unit 204 can estimate the degradation of the energy storage element due to energization based on, for example, a change in the state of the coating in the electrode, which is based on the magnitude of the SOC fluctuation. As the SEI coating is formed, the growth rate of the SEI coating decreases, but if the SOC fluctuation is large and the SEI coating is partially destroyed, the growth rate of the SEI coating increases and the degradation value increases. In this embodiment, the weighting is used to take into account a fluctuation waveform with a large fluctuation range, and also a fluctuation waveform with a small fluctuation range is detected by Fourier transform, so that the degradation estimation is highly accurate.

[0061] The monitoring device 151 or the estimation device 101 in the monitoring device 151 includes a control unit 20, which reads out from the storage unit 23 and executes a deterioration estimation program 231 including some or all of the steps of the flowcharts shown below.

[0062] FIG. 9 is a flowchart that defines an operation procedure when the estimation device estimates the deterioration of the energy storage element. 9, a situation is assumed in which the control unit 20 of the estimation device 101 receives an estimation command from another device. Hereinafter, a case will be described in which, for example, the central SOC is 60% and charging and discharging are repeated so that the SOC fluctuates between 30% and 90%.

[0063] First, the control unit 20 acquires fluctuation data of the SOC for a predetermined period (S1). The control unit 20 performs a Fourier transform on the waveform of the SOC fluctuation to obtain f(ω), and converts it into a frequency function f(f) (S2). The control unit 20 performs weighting (S3) based on the peak height of each waveform component, the area (integral value) of each waveform component, or the like. The control unit 20 specifies the deterioration coefficient k (S4). The control unit 20 reads out the deterioration coefficient DB 232 and acquires the relationship between the amplitude, frequency, and deterioration coefficient according to the center SOC, temperature, and acquisition period. The control unit 20 specifies the deterioration coefficient kr from the relationship between the amplitude, frequency, and deterioration coefficient based on the amplitude and frequency of the waveform component of importance obtained by performing weighting. The control unit 20 calculates ΔSOH by the above formula (1) based on the identified deterioration coefficient k (S5), and ends the process.

[0064] The specific details will be explained below. Fig. 10 is a graph showing the relationship between time and SOC when the SOC is changed by Δ0.5%, Δ1.5%, Δ5%, Δ20%, and Δ30%. The horizontal axis is time [seconds], and the vertical axis is SOC [%]. In Fig. 10, graphs a, b, c, d, and e show the cases when the SOC is changed by Δ0.5%, Δ1.5%, Δ5%, Δ20%, and Δ30%. Fig. 10 also shows the change in SOC when the storage element is actually used (Example).

[0065] Figure 11 is a graph showing the waveform in Figure 10 decomposed into multiple frequency domain waveform components through Fourier transformation. The horizontal axis is frequency [Hz], and the vertical axis is amplitude spectrum [%]. As in Fig. 6, waveform components corresponding to the waveforms b, c, d, and e in Fig. 10 are shown, and waveform components obtained by decomposing the waveform of the embodiment are also shown. The peak component of the waveform components corresponding to the waveform a in Fig. 10 has a frequency of about 0.013 and an amplitude of about 0.133, and is not shown in Fig. 10.

[0066] The deterioration coefficient k of each ΔSOC is determined by experiment as described above. FIG. 12 is a graph showing the relationship between the number of days and cycle deterioration when the SOC is changed by Δ0.5%, Δ1.5%, Δ5%, Δ20%, Δ30%, and Δ100%. In FIG. 12, the horizontal axis is days [(√days)], and the vertical axis is cycle deterioration [%]. The cycle deterioration is shown as a standard value when the largest deterioration amount in FIG. 12 is set to 100%. The following FIG. 13 to FIG. 15 are also based on this standardization. In FIG. 12, the graphs when the SOC is changed by Δ0.5%, Δ1.5%, Δ5%, Δ20%, Δ30%, and Δ100% are shown as a, b, c, d, e, and f. From formula (1), k = deterioration value / (√t), and k for each fluctuation range can be found by optimization calculations based on the curves a to f.

[0067] Fig. 13 is a graph showing the relationship between the amplitude spectrum, frequency, and deterioration coefficient. In Fig. 13, the x-axis is the amplitude spectrum [%], the y-axis is the frequency [Hz], and the z-axis is the deterioration coefficient [% / (√days)]. The relationship is shown by associating the frequency and intensity of the peak top of each of the waveforms a, b, c, d, and e in Fig. 11 with k of each of the graphs a, b, c, d, and e obtained from Fig. 12. In Fig. 13, the amplitude spectrum is approximately 10.2%, and the amplitude is approximately 2 x 10 -4 If so, the degradation factor k is approximately 3.56.

[0068] Figure 14 shows the relationship between the amplitude spectrum, frequency, and degradation coefficient when interpolated by calculation based on k in each graph in Figure 12. In the case of Figure 14, for waveforms other than waveforms a, b, c, d, and e in Figure 11, k can be obtained by reading the value on the z-axis corresponding to the amplitude spectrum and frequency of the waveform.

[0069] As mentioned above, the control section 20 weights the waveforms in the embodiment of FIG. In Fig. 14, the deterioration coefficient k is calculated by reading the value on the z-axis corresponding to the amplitude spectrum and frequency of the weighted waveform, that is, by reading the z-coordinate of the point ○. Here, k is 0.75.

[0070] Fig. 15 is a graph showing the relationship between the estimated value and the actual measured value. In Fig. 15, the horizontal axis is the total number of days [days], and the vertical axis is the cycle deterioration [%]. The estimated value shows the relationship between the total number of days and cycle deterioration when k = 0.1875 in formula (1). The experimental values ​​are plotted as cycle deterioration at each time point against the total number of days at a plurality of measurement points. From Fig. 15, it was confirmed that the estimation method of this embodiment can perform good estimation.

[0071] According to the estimation method of the embodiment, as shown in FIG. 16 , even when charging and discharging are repeated on the high SOC side and the amplitude of the SOC fluctuation is biased to the positive side, the degradation coefficient k can be obtained by decomposing the waveform into multiple waveform components using Fourier transform, and the degradation value can be accurately estimated.

[0072] From the above, it has been confirmed that the waveform of the SOC fluctuation can be decomposed into waveform components in a plurality of frequency domains, and the deterioration of the energy storage element can be estimated with high accuracy based on each waveform component and the deterioration coefficient k. By estimating the decrease in the amount of electricity that can be reversibly extracted from the storage element (the capacity degradation value), the internal state of the storage element can be grasped. Since the potential at SOC 100% of the negative electrode can also be known, if the storage element is a lithium-ion secondary battery, the risk of metallic lithium deposition in the negative electrode can also be known. The SOH of the storage element, including this risk, can be monitored. It is also possible to determine how to control the storage element.

[0073] (Embodiment 2) Fig. 17 is a block diagram showing an example of the configuration of an estimation device 101 according to embodiment 2. In the figure, the same components as those in Fig. 8 are given the same reference numerals and detailed description thereof will be omitted. The storage unit 23 of the estimation device 101 according to the second embodiment has the same configuration as the estimation device 101 according to the first embodiment, except that a teacher data DB 233 and a degradation coefficient output model 234 are stored in the storage unit 23.

[0074] 18 is an explanatory diagram showing an example of a record layout of the teacher data DB 233. The teacher data DB 233 stores a large amount of teacher data for generating the degradation coefficient output model 234. The teacher data DB233 stores a No. column, a graph column after Fourier transform, and a degradation coefficient k column. The No. column stores a No. for identifying each teacher data. The graph column after Fourier transform stores a graph after Fourier transform of the time series data of charging and discharging. The graph is a graph of f(f) in which ω of f(ω) is converted to f (frequency (Hz)). The degradation coefficient k column stores the degradation coefficient k corresponding to the graph. For example, if the graph of FIG. 11, which is a Fourier transform of the graph of the embodiment of FIG. 10, is used as the teacher data, the degradation coefficient k calculated from the experimental value data of FIG. 15 is stored in the degradation coefficient k column.

[0075] FIG. 19 is a schematic diagram showing the configuration of the degradation coefficient output model 234. As shown in FIG. The deterioration coefficient output model 234 is a learning model that is expected to be used as a program module that is a part of artificial intelligence software, and can use a multi-layer neural network (deep learning), for example, a convolutional neural network (CNN), but can also use a recurrent neural network (RNN). Other machine learning such as a decision tree, a random forest, and a support vector machine can also be used. The control unit 20 operates to perform an operation on a graph after Fourier transform of the charge / discharge time series data input to the input layer of the deterioration coefficient output model 234 according to an instruction from the deterioration coefficient output model 234, and to output a probability value of the deterioration coefficient k. The deterioration coefficient output model 234 will be described in detail below.

[0076] The control unit 20 constructs a neural network that inputs the acquired graph f(f) and outputs information indicating the degradation coefficient k by learning the feature amount as the degradation coefficient output model 234. The neural network is a CNN, and has an input layer that accepts the input of the graph f(f), an output layer that outputs the probability value of the degradation coefficient k, and an intermediate layer that extracts the feature amount. In FIG. 19, the number of layers of the intermediate layer is three, but is not limited thereto. For example, when the degradation coefficient output model 234 is a CNN, the intermediate layer has a configuration in which a convolution layer and a pooling layer are alternately connected, and finally extracts features while compressing information. In FIG. 19, the convolution layer and the pooling layer are omitted. The output layer has a plurality of neurons that output the identification result of the degradation coefficient k, and outputs a plurality of degradation coefficients k and their probability values ​​based on the features output from the intermediate layer. Specifically, when a softmax function is used, the degradation coefficient k output from the neuron is a plurality of degradation coefficients k and their probability values.

[0077] Although the degradation coefficient output model 234 has been described as being a CNN, an RNN can be used as described above. In an RNN, the intermediate layer of the previous time is used for learning together with the input layer of the next time.

[0078] The control unit 20 performs learning using training data in which a graph after Fourier transform of each number in the training data DB 233 is associated with a degradation coefficient k in each graph.

[0079] The control unit 20 inputs a graph of f(f) which is teaching data to the input layer, and obtains a probability value of the degradation coefficient from the output layer after arithmetic processing in the intermediate layer.

[0080] The control unit 20 compares the identification result output from the output layer with information labeled for the graph of f(f) in the teacher data, i.e., the correct answer value, and optimizes parameters used in the calculation process in the intermediate layer so that the output value from the output layer approaches the correct answer value. The parameters are, for example, weights (coupling coefficients) between neurons, activation function coefficients used in each neuron, etc. The method of optimizing the parameters is not particularly limited, but for example, the control unit 20 optimizes various parameters using an error backpropagation method.

[0081] The control unit 20 performs the above-mentioned process on the graph of f(f) of each piece of teacher data included in the teacher data DB 233, and generates the deterioration coefficient output model 234. When the control unit 20 acquires the graph of f(f) obtained by Fourier transforming the time-series data of charging and discharging, the control unit 20 acquires, using the deterioration coefficient output model 234, the deterioration coefficient k that shows a high probability value based on the probability value of the deterioration coefficient k output by the deterioration coefficient output model 234 and a threshold value, as the deterioration coefficient k to be specified.

[0082] The control unit 20 can re-learn the degradation coefficient output model 234 based on the degradation coefficient k identified by the degradation coefficient output model 234 and the actually measured degradation coefficient k, so as to improve the reliability of the identification result.

[0083] FIG. 20 is a flowchart showing an example of a processing procedure for generating the degradation coefficient output model 234 by the control unit 20. The control unit 20 reads out the teacher data DB 233, and acquires teacher data in which the graph after the Fourier transform for each No. is associated with the degradation coefficient (step S11).

[0084] The control unit 20 uses the teacher data to generate a degradation coefficient output model 234 that outputs a specified degradation coefficient k when the graph f(f) after the Fourier transform is input (step S12). The control unit 20 stores the generated degradation coefficient output model 234 in the storage unit 23 and ends the series of processes.

[0085] FIG. 21 is a flowchart that defines an operation procedure when the estimation device 101 estimates the deterioration of the energy storage element. First, the control unit 20 acquires fluctuation data of the SOC for a predetermined period (S21). The control unit 20 performs a Fourier transform on the waveform of the SOC fluctuation to obtain f(ω), and converts it into a frequency function f(f) (S22). The control unit 20 inputs the graph of f(f) to the degradation coefficient output model 234 (S23). The control unit 20 specifies the degradation coefficient k (S24). Based on the degradation coefficient k and the probability value output by the degradation coefficient output model 234, the control unit 20 specifies the degradation coefficient k having a probability value equal to or greater than a threshold value as the degradation coefficient k to be specified. The control unit 20 calculates ΔSOH by the above formula (1) based on the identified deterioration coefficient k (S25), and ends the process.

[0086] According to this embodiment, the deterioration coefficient is specified taking into consideration waveform components with low spectral intensity and high frequency, and deterioration can be estimated satisfactorily. Note that a plurality of degradation coefficient output models 234 may be generated according to the median SOC, temperature, or the like. Furthermore, the deterioration coefficient output model may be configured to input time series data of charging and discharging instead of a graph of f(f) and output a deterioration coefficient k.

[0087] In the first and second embodiments, the estimation device 101 is configured to use time series data of SOC, but the time series data of SOC may be ΔSOC calculated by a current integration method or the like, or may be data obtained by adding / subtracting ΔSOC to / from an initial SOC value.

[0088] In the estimation device 101, the estimation unit 204 is configured to calculate a degradation value as an estimation of degradation of the energy storage element, but this is not limited to this. The estimation unit 204 may also calculate a level indicating degradation of the energy storage element, a lifespan of the energy storage element, and a replacement time for the energy storage element.

[0089] In the estimation device 101, the estimation unit 204 may estimate the degradation of the energy storage element based on the sum of the degradation value obtained by this embodiment and the non-energization degradation value Qcnd calculated by the method of Patent Document 1.

[0090] In the above embodiment, the deterioration coefficient k is specified based on the waveforms of the respective fluctuation ranges when the central SOC is 60%, but the present invention is not limited to this.

[0091] The estimation device is not limited to estimating deterioration using a deterioration coefficient output model. When an amplitude spectrum diagram (waveform component), average SOC, average temperature T, SOC change acquisition period, and the previous cycle deterioration amount are input, deterioration may be estimated using a learning model that outputs the cycle deterioration amount during the acquisition period. Furthermore, although the Fourier transform has been used as an example of frequency transform, other transforms (eg, wavelet transform, discrete cosine transform, etc.) may also be used.

[0092] The above-described embodiments are not limiting, and the scope of the present invention is intended to include all modifications within the meaning and scope of the claims. Explanation of symbols

[0093] 20 Control section 201 Acquisition Department 202 Disassembly section 203 Specific section 204 Estimation section 23 Memory section 231 Deterioration Estimation Program 233 Teacher Data DB 234 Degradation coefficient output model 41 Current Sensor 42 Voltage Sensor 43 Temperature Sensor 53 A / D conversion section 54 History Creation Department 55 Counter 56 Memory section 57 Communications Department 60 Recording media 101 Prediction Device 151 Surveillance Device

Claims

1. An acquisition unit that acquires time series data of the SOC of the energy storage element; A decomposition unit that decomposes a waveform of the fluctuation of the SOC in the time series data into frequency components; an estimation unit that estimates deterioration of the power storage element based on the frequency component; An estimation device comprising:

2. The acquisition unit acquires time series data of SOC for a predetermined period, The estimation device according to claim 1 , wherein the decomposition unit processes a function of a time change of the SOC variation for the predetermined period.

3. The decomposition unit decomposes a waveform of the fluctuation in the SOC in the time series data into waveform components in a plurality of frequency domains; The estimation device according to claim 1 , wherein the estimation unit estimates the deterioration based on an intensity of each waveform component and a deterioration coefficient.

4. a determination unit that determines a deterioration coefficient based on the waveform components, The estimation device according to claim 3 , wherein the estimation unit estimates the deterioration based on the deterioration coefficient identified by the identification unit.

5. The estimation device according to claim 4 , wherein the specifying unit specifies the degradation coefficient based on a relationship between an intensity and a frequency of a waveform component and the degradation coefficient.

6. The estimation device according to claim 4 , wherein the determination unit determines the deterioration coefficient by weighting a waveform component having a high intensity in comparison with a waveform component having a low intensity.

7. The estimation device according to claim 4 , wherein the specification unit specifies the degradation coefficient by inputting the waveform component acquired by the decomposition unit into a learning model that outputs a degradation coefficient when a waveform component is input.

8. Acquire time series data of the SOC of the storage element; Decomposing a waveform of the fluctuation of the SOC in the time series data into frequency components; An estimation method for estimating deterioration of the storage element based on the frequency components.

9. Acquire time series data of the SOC of the storage element; Decomposing a waveform of the fluctuation of the SOC in the time series data into frequency components; Deterioration of the power storage element is estimated based on the frequency component. A computer program that causes a computer to carry out processing.

Images