Estimation device, estimation system, estimation method, and program therefor

Abstract

To more efficiently estimate characteristics of a power storage device while increasing accuracy.SOLUTION: An estimation device for estimating characteristics of a power storage device and including a control part which includes the steps of: using an estimation model constructed on the basis of a continuation factor of a plurality of points at the time of discharging and / or charging of the power storage device whose characteristics are known and a factor after pause of the plurality of points after the stop of discharging and / or charging; acquiring the continuation factor of the plurality of points of the power storage device to be estimated and the factor after pause of the plurality of points; and estimating the characteristics of the power storage device from an estimation model using the acquired continuation factor and the factor after pause as explanatory variables.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] This specification discloses an estimation device, an estimation system, an estimation method, and a program therefor. [Background technology]

[0002] Conventionally, as a method for estimating deterioration of an electric storage device, for example, a method has been proposed in which a lithium-ion secondary battery is discharged, and the time, current, and power required for the voltage to reach a lower limit voltage are measured, and a degraded battery is identified using these measurements and a specified threshold (see, for example, Patent Document 1). Another method for estimating deterioration of an electric storage device has been proposed in which the charging resistance and discharging resistance of the lithium-ion secondary battery are calculated using a current sensor and a voltage sensor, and a degraded battery is identified based on the relationship between the charging resistance and the discharging resistance (see, for example, Patent Document 2). Another method for estimating deterioration of an electric storage device has been proposed in which the internal resistance is calculated from the voltage and current of the lithium-ion secondary battery after charging and discharging, and the degraded battery is identified based on the internal resistance (see, for example, Patent Document 3). Another method for estimating deterioration of an electric storage device has been proposed in which the charging voltage and discharging voltage of the lithium-ion secondary battery are measured using a pulse current, and a degraded battery is identified based on the relationship between the charging resistance and the discharging resistance (see, for example, Patent Document 4). Furthermore, a method has been proposed for estimating the deterioration of an electricity storage device, in which an external resistor is connected in series and a predetermined discharge current is passed through the external resistor for a certain period of time to determine whether the battery is deteriorated from the voltage value (see, for example, Patent Document 5). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2020-102343 [Patent Document 2] Japanese Patent Application Publication No. 2020-101470 [Patent Document 3] Japanese Patent Application Publication No. 2020-85599 [Patent Document 4] Japanese Patent Publication No. 2020-20715 [Patent Document 5] Japanese Patent Application Laid-Open No. 2012-132758 Summary of the Invention [Problem to be solved by the invention]

[0004] However, in the above-mentioned Patent Document 1, only the voltage when the power storage device is discharged is used for estimation, and the characteristics are estimated using only one data point when the threshold value is reached after discharge, which leaves problems with the estimation accuracy. Also, in Patent Documents 2 to 4, each resistance value is calculated from the voltage and current when the power storage device is charged and discharged, but this is still insufficient, leaving problems with the estimation accuracy. Also, in Patent Document 5, a deteriorated battery is determined from the voltage value when a predetermined discharge current is passed through an external resistance for a certain period of time, but this is still not sufficient, leaving problems with the estimation accuracy.

[0005] The present disclosure has been made in consideration of such problems, and its main purpose is to provide a novel estimation device, estimation system, estimation method, and program therefor that can estimate the characteristics of an energy storage device more efficiently while improving accuracy. [Means for solving the problem]

[0006] As a result of intensive research to achieve the above-mentioned object, the inventors have discovered that by constructing an estimation model for the characteristics of an energy storage device through machine learning using multiple continuation factors during charging and discharging, including discharging and charging, and multiple post-rest factors after charging and discharging are stopped, and estimating the characteristics of an energy storage device with unknown characteristics from this estimation model, it is possible to estimate the characteristics of the energy storage device more efficiently and with greater accuracy, and have completed the invention disclosed in this specification.

[0007] That is, the estimation device disclosed in this specification is An estimation device for estimating a characteristic of an electricity storage device, a control unit that uses an estimation model constructed based on a plurality of continuous time factors during discharging and / or charging of an electricity storage device whose characteristics are known and a plurality of post-pause factors after the discharging and / or charging is stopped, acquires the continuous time factors and the post-pause factors of a plurality of points of an electricity storage device that is an estimation target, and estimates characteristics of the electricity storage device from the estimation model using the acquired continuous time factors and the post-pause factors as explanatory variables; It is equipped with the following.

[0008] The estimation system disclosed in the present specification comprises: a measuring device that discharges and / or charges an electricity storage device and measures the duration factor and the post-pause factor; The above-mentioned estimation device, The control unit acquires the continuation factor and the post-pause factor of the power storage device from the measurement device.

[0009] The estimation method disclosed herein comprises: An estimation method for estimating a characteristic of an electricity storage device, comprising: a step of using an estimation model constructed based on a plurality of continuous time factors during discharging and / or charging of an electric storage device whose characteristics are known and a plurality of post-rest factors after the discharging and / or charging is stopped, acquiring the continuous time factors and the post-rest factors at a plurality of points of an electric storage device to be estimated, and estimating characteristics of the electric storage device from the estimation model using the acquired continuous time factors and the post-rest factors as explanatory variables; It includes:

[0010] The program disclosed in this specification causes one or more computers to implement the steps of the estimation method described above. This program may be recorded on a computer-readable recording medium (e.g., a hard disk, a ROM, a FD, a CD, a DVD, etc.), or may be distributed from one computer to another via a transmission medium (a communication network such as the Internet or a LAN), or may be transmitted in any other form. [Effects of the Invention]

[0011] The estimation device, estimation system, estimation method, and program thereof disclosed herein can evaluate the degree of degradation of an energy storage device more efficiently and with improved accuracy. The reason why the present disclosure achieves such an effect is believed to be as follows. For example, three main degradation modes of energy storage devices such as lithium-ion secondary batteries are known: "positive electrode degradation," "negative electrode degradation," and "capacity mismatch between positive and negative electrodes." Therefore, at least three explanatory variables are required to accurately estimate the degree of degradation. Meanwhile, since the characteristics of an energy storage device, such as the degree of degradation, are reflected in changes in internal resistance associated with degradation, they can be estimated more efficiently and accurately from the time changes in factors reflecting the internal resistance during charging and discharging, and the time changes in factors after charging and discharging are stopped. In this way, the present disclosure can evaluate the characteristics of an energy storage device more accurately in a short time, for example, without performing capacity measurement, which takes a relatively long time to evaluate, or without adjusting the remaining capacity (SOC) of the energy storage device. Therefore, the present disclosure can estimate the characteristics of an energy storage device more efficiently and with improved accuracy. [Brief explanation of the drawings]

[0012] [Figure 1] FIG. 1 is an explanatory diagram showing an example of an estimation system 10. [Figure 2] 10 is a flowchart showing an example of an estimation model construction processing routine. [Figure 3] 4 is a flowchart showing an example of a deterioration determination processing routine. [Figure 4] 4 is an explanatory diagram of a display screen 40 showing the diagnosis result of the object 14 to be judged. FIG. [Figure 5] FIG. 10 is a graph showing an example of voltage changes over time for a new battery and a deteriorated battery. [Figure 6] Voltage change during discharge and after stopping discharge of a new battery at SOC 100%. [Figure 7] FIG. 10 is a graph showing an example of the discharge time tcF and voltage VcF after the discharge of a new battery is stopped. [Figure 8]Relationship diagram of prediction error for SOC according to explanatory variables. [Figure 9] Relationship between the number of explanatory variables and RMSECV and RCV 2. [Figure 10] A graph showing the relationship between the measured and predicted discharge capacity values using 10 explanatory variables. [Figure 11] Cole-Cole plot of AC impedance. [Figure 12] A graph showing the relationship between measured and predicted resistance values using 10 explanatory variables. [Figure 13] Voltage change during charging and after stopping charging of a new battery at SOC 10%. DETAILED DESCRIPTION OF THE INVENTION

[0013] (estimation device) An embodiment of an estimation device disclosed in this specification will be described below with reference to the drawings. FIG. 1 is a schematic explanatory diagram showing an example of an estimation system 10. The estimation system 10 is provided, for example, in a collection facility that collects used electricity storage devices 13, and is a system that estimates characteristics of the electricity storage devices 13. Examples of the characteristics of the electricity storage devices 13 include one or more of the capacity and resistance of the electricity storage devices 13. The characteristics of the electricity storage devices 13 may also include the degree of deterioration. The estimation system 10 estimates the characteristics of the collected electricity storage devices 13 and determines whether they can be reused or recycled. Reusable electricity storage devices 13 are reconditioned and shipped, while recyclable electricity storage devices 13 are disassembled and recycled. The estimation system 10 includes a measuring device 15 and an estimation device 20. The estimation system 10 exchanges information between the measuring device 15 and the estimation device 20 via a network 12, such as a LAN or the Internet.

[0014] The power storage device 13 includes, for example, a hybrid capacitor, a pseudo electric double layer capacitor, an alkali metal secondary battery such as lithium or sodium, an alkali metal ion battery, an air battery, and the like. Among these, as the power storage device 13, a lithium secondary battery, particularly a lithium ion secondary battery is preferable. Here, the power storage device 13 is mainly described as being a lithium ion secondary battery. The power storage device 13 may include, for example, a positive electrode, a negative electrode, and an ion conductive medium interposed between the positive electrode and the negative electrode and conducting carrier ions. The positive electrode may include, as a positive electrode active material, a sulfide containing a transition metal element, an oxide containing lithium and a transition metal element, and the like. The positive electrode active material is, for example, a lithium manganese composite oxide having a basic composition formula of Li (1-x) MnO2 (0 < x < 1, etc., the same below) or Li (1-x) Mn2O4, etc., a lithium cobalt composite oxide having a basic composition formula of Li (1-x) CoO2, etc., a lithium nickel composite oxide having a basic composition formula of Li (1-x) NiO2, etc., a lithium nickel cobalt manganese composite oxide having a basic composition formula of Li (1-x) Ni a Co b Mn cLithium nickel cobalt manganese composite oxides with a formula such as O2 (a+b+c=1) can be used. The term "basic composition formula" means that other elements may be included. The negative electrode may contain a carbon material or a lithium-containing composite oxide as the negative electrode active material. Examples of the negative electrode active material include inorganic compounds such as lithium, lithium alloys, and tin compounds, carbon materials capable of absorbing and releasing lithium ions, composite oxides containing multiple elements, and conductive polymers. Examples of carbon materials include cokes, glassy carbons, graphites, non-graphitizable carbons, pyrolytic carbons, and carbon fibers. Among these, graphites such as artificial graphite and natural graphite are preferred. Examples of composite oxides include lithium titanium composite oxide and lithium vanadium composite oxide. The ion-conducting medium can be, for example, an electrolyte solution containing a supporting salt. Examples of the supporting salt include lithium salts such as LiPF6 and LiBF4. Examples of solvents for the electrolyte solution include carbonates, esters, ethers, nitriles, furans, sulfolanes, and dioxolanes, which can be used alone or in combination. Specific examples of carbonates include cyclic carbonates such as ethylene carbonate, propylene carbonate, vinylene carbonate, butylene carbonate, and chloroethylene carbonate, and chain carbonates such as dimethyl carbonate, ethyl methyl carbonate, diethyl carbonate, ethyl-n-butyl carbonate, methyl-t-butyl carbonate, di-i-propyl carbonate, and t-butyl-i-propyl carbonate. The ion-conducting medium can be a solid ion-conducting polymer, an inorganic solid electrolyte, a mixture of an organic polymer electrolyte and an inorganic solid electrolyte, or an inorganic solid powder bound by an organic binder. The solid electrolyte and the power storage device 13 may also include a separator between the positive electrode and the negative electrode.

[0015] Measuring device 15 is a device that discharges and / or charges electricity storage device 13 and measures the voltage at that time. This measuring device 15 includes a thermostatic chamber 16 that accommodates electricity storage device 13 as evaluation target 14 at a measurement temperature T, and a charger / discharger 17 that is electrically connected to evaluation target 14 and applies a constant current or discharges at a constant current. Measuring device 15 may also be a device that calculates resistance from the current and voltage. Measuring device 15 outputs the measurement results to estimation device 20 via network 12.

[0016] The estimation device 20 is a device that performs processing to estimate the characteristics of the object 14 to be determined, whose characteristics are unknown, from the results of measuring factors of the object 14 to be determined, such as the degree of degradation, capacity, and internal resistance, using an estimation model obtained from the results of measuring an electricity storage device whose characteristics are known. The estimation device 20 includes a control unit 21, a storage unit 22, an input device 28, and a display device 29. The input device 28 includes a mouse, a keyboard, and the like for performing various inputs. The display device 29 displays a screen, and is, for example, a liquid crystal display.

[0017] The control unit 21 is configured as a microprocessor centered on a CPU, and controls the entire device. The control unit 21 also has the following functional blocks: an acquisition unit that acquires measurement results from the measurement device 15, an estimation unit that estimates the characteristics of the power storage device 13 from the acquired measurement results and an estimation model 35, a deterioration derivation unit that derives a deterioration level from the estimated results of the capacity and / or resistance, and a determination unit that determines whether the power storage device 13, which is the processing target 14, can be reused from the obtained deterioration level. These functional blocks are realized, for example, by the control unit 21 executing a deterioration determination program 36 or the like.

[0018] The storage unit 22 is configured as a large-capacity storage device such as an HDD, and stores known test results 23, a machine learning program 30, a tolerance range 31, an estimation model 35, a degradation determination program 36, and the like. The known test results 23 are measurement results of the power storage device 13, whose characteristics are known. The known test results 23 are used to construct the estimation model 35, and the measurement results include a duration factor 25, a post-pause factor 26, and a degradation level 27. The duration factor 25 includes multiple factors in a range from the start of discharging and / or charging of the power storage device until a predetermined time has elapsed. This factor may be, for example, the voltage or resistance during discharging and / or charging of the power storage device. Here, the case where this factor is voltage will be mainly described. The post-pause factor 26 includes multiple factors in a range from the stop of discharging and / or charging of the power storage device until a predetermined time has elapsed after discharging and / or charging. The predetermined times for the duration factor 25 and the post-pause factor 26 may be empirically determined within a range in which a good estimation result is obtained. The accuracy of the characteristics of the power storage device 13 can be improved by estimating them using one or more of the following: a plurality of factors included in the duration of discharging and a plurality of factors included in the rest period after discharging is stopped; or a plurality of factors included in the duration of charging and a plurality of factors included in the rest period after charging is stopped. For ease of explanation, hereinafter, this "discharging and / or charging" will also be simply referred to as "charging and discharging." The degradation degree 27 can be defined, for example, as the percentage of the maximum discharge capacity that has decreased due to degradation after continuous or intermittent use, assuming that the initial maximum discharge capacity is 100%. Alternatively, the degradation degree 27 can be defined as the percentage of the resistance that has increased due to degradation after continuous or intermittent use, assuming that the initial resistance is 100%.

[0019] The machine learning program 30 is executed by the control unit 21 and is a program that uses machine learning to construct an estimation model 35 from known test results 23. For example, statistical analysis software such as R, Python (registered trademark), SQL, and Excel (registered trademark) can be used for the machine learning, and one or more of linear regression, kernel ridge, support vector, XGBoost, neural network (nnet), and random forest (RF) can be used as a method. As a machine learning method, random forest is preferable because it has higher accuracy in characteristic estimation. The estimation model 35 is constructed using, as explanatory variables, a continuation factor 25 at multiple points and a post-rest factor 26 at multiple points contained in the known test results 23. Furthermore, the estimation model 35 is preferably constructed based on known test results 23 obtained by measuring an electricity storage device 13 that has been degraded through multiple degradation processes. Since the electricity storage device 13 has three degradation modes, for example, "positive electrode degradation," "negative electrode degradation," and "capacity mismatch between positive and negative electrodes," it is preferable to adopt degradation processes that represent each of these modes. Examples of the deterioration process of the electricity storage device 13 include high-rate charge / discharge cycle treatment, high-temperature charge / discharge cycle treatment, and long-term storage treatment at a high SOC and high temperature. Here, a "high rate" is preferably 1 C or higher, and more preferably 2 C or higher. A "high rate" may be 10 C or lower. A "high temperature" is preferably 40°C or higher, more preferably 60°C or higher, and may be 80°C or lower. A "high SOC" is preferably 50% or higher, more preferably 80% or higher, and may be 100% or lower. A "long-term storage" is preferably storage for 2 days or longer, more preferably storage for 20 days or longer, and may be storage for 300 days or shorter. Preferably, the estimation model 35 is constructed based on known test results 23 obtained by measuring the electricity storage device 13 after deterioration and adjusting the SOC to multiple points.

[0020] The estimation model 35 preferably uses, as explanatory variables, a duration factor of 2 to 6 points and a post-pause factor of 2 to 6 points. Considering estimation accuracy, the explanatory variables must have 3 points or more. Furthermore, if the explanatory variables have a low number of points, they will be unable to fully express multiple deterioration modes, resulting in reduced prediction accuracy. On the other hand, if the explanatory variables have a high number of points, they will cause overlearning, resulting in reduced prediction accuracy. The explanatory variables, including the duration factor 25 and the post-pause factor 26, preferably have a total of 4 points or more, more preferably 6 points or more, and even more preferably 8 points or more. Furthermore, the explanatory variables are preferably 12 points or less, more preferably 11 points or less, and may be 10 points or less. Furthermore, the estimation model 35 preferably uses, as explanatory variables, a duration factor 25 corresponding to the elapsed time including multiple points with different digits from the start of discharging and / or charging of the power storage device 13, and a post-pause factor 26 corresponding to the elapsed time including multiple points with different digits from the stop of discharging and / or charging of the power storage device 13. For the duration factor 25 and the post-pause factor 26, using factors (voltages) corresponding to elapsed time at equal intervals on a log scale can improve estimation accuracy, compared to using factors (voltages) corresponding to elapsed time at equal intervals in absolute value. For example, the duration factor 25 and the post-pause factor 26 preferably include one point at the "0" of the elapsed time, one point at two decimal places, one point at one decimal place, one point at one digit, one point at two digits, one point at three digits, etc. It should be noted that "equally spaced on a log scale" includes not only perfectly equal intervals, but also intervals with a margin and that are approximately equal.

[0021] In addition, the explanatory variables are the voltage Vc at the start of discharge and / or charge. 0 , the time tc from the start of discharge and / or charging used for estimation F Voltage Vc after F When time t is 0 to tc F Divide the interval into n parts (tc 1 , tc 2 , …tc n-1 , tc n ), the voltage Vc at each time 1 ~Vc n and the voltage Vs when discharging and / or charging is stopped 0, the time ts after stopping the discharge and / or charging used for estimation F Voltage Vs after time has elapsed F When time t is 0 to ts F The time is divided into m parts, and the voltage Vs at each part is 1 ~Vs m In this case, with respect to the number of explanatory variables, the above n is preferably 1≦n≦5, more preferably 2≦n≦5, and even more preferably 2≦n≦3. Also, the above m is preferably 1≦m≦5, more preferably 2≦m≦5, and even more preferably 2≦m≦3. When n=3 and m=3, the explanatory variables are Vc 0 , Vc 1 , Vc 2 , Vc 3 , Vc F , Vs 0 , Vs 1 , Vs 2 , Vs 3 , Vs F 10 pieces. Also, time tc F is 60 seconds or less tc F ≦1000 seconds is preferable, and 60 seconds ≦tc F It is more preferable that the time ts is ≦300. F is 40 seconds or less ts F ≦1260 seconds is preferred, and 40 seconds ≦ts F ≦300 is more preferred.

[0022] The deterioration determination program 36 is executed by the control unit 21, and is a program that estimates the characteristics of the determination target 14 based on the measurement results of the determination target 14 and the estimation model 35, and determines whether the determination target 14 can be reused. This deterioration determination program 36 uses, for example, a plurality of continuation factors 25 and a plurality of post-suspension factors 26 of the determination target power storage device 13 acquired from the measuring device 15 as explanatory variables to estimate the characteristics of the power storage device from the estimation model 35, and then determines the degree of deterioration from the estimation result. The deterioration determination program 36 determines the degree of deterioration using an allowable range 31. The allowable range 31 can be, for example, a threshold value of the degree of deterioration that is empirically determined within a range that allows reuse.

[0023] (Estimation method) Next, the operation of the estimation device 20 of this embodiment configured as described above, in particular the estimation method executed by the estimation device 20, will be described. This estimation method may include, for example, a construction step of constructing an estimation model 35 by machine learning, an estimation step of estimating the characteristics of the electricity storage device to be evaluated from the estimation model 35, and a determination step of determining the degree of degradation from the estimation result. Note that in this estimation method, the construction step may be omitted by using an already constructed estimation model 35. Furthermore, the determination step may be omitted for the purpose of acquiring the characteristics of the electricity storage device.

[0024] (Building steps) Here, first, a process for constructing the estimation model 35 will be described. In this construction process, a user prepares a plurality of power storage devices 13 with known characteristics, a measurement device 15 measures factors (voltages) of charging and discharging the power storage devices 13 and after the charging and discharging are stopped, and the estimation device 20 constructs the estimation model 35. FIG. 2 is a flowchart showing an example of an estimation model construction process routine executed by the control unit 21 of the estimation device 20. This routine is stored in the storage unit 22 and executed in response to an instruction from the user. When the control unit 21 executes this routine, first, it acquires settings of the continuation time factor and the post-pause factor (S100). The control unit 21 calculates the time tc of the continuation time factor used as an explanatory variable. F and the time ts of the post-rest factor F For example, the time tc during discharge and / or charging is obtained. F is 0 seconds, 0.01 seconds, 0.2 seconds, 5 seconds, 100 seconds, and the time ts after discharging and / or charging is stopped. F may be set to five points: 0 seconds, 1 second, 4 seconds, 15 seconds, and 60 seconds.

[0025] Next, the control unit 21 starts discharging and / or charging the power storage device 13, whose characteristics are known, and then measures the duration tc F The control unit 21 acquires the measurement result of the factor (voltage) according to the above from the measurement device 15 (S110). In addition, the control unit 21 acquires the measurement result of the factor (voltage) according to the above from the measurement device 15 (S110).F The control unit 21 acquires measurement results of factors (voltages) corresponding to the factors from the measuring device 15 (S120). Next, it determines whether measurement results of the continuation factors and post-pause factors of all the power storage devices 13 have been acquired (S130). If not all measurement results have been acquired, the process from S110 onward is repeated. That is, the next power storage device 13 is set, and the measuring device 15 repeatedly measures the voltage during charging and discharging and the rest voltage after the charging and discharging are stopped. On the other hand, when all measurement results have been acquired, the control unit 21 constructs an estimation model 35 by machine learning, stores it in the storage unit 22 (S140), and terminates this routine. The voltage change of a deteriorated power storage device 13 does not have a specific relationship depending on the state of deterioration, and even if the capacity deterioration is the same, the time change of the voltage differs depending on the deterioration process (see FIG. 5 described later). Here, machine learning is used to construct an estimation model 35 that can more accurately estimate the characteristics of the power storage device 13 from the continuation factors 25 and post-pause factors 26. In addition, the control unit 21 can use, for example, statistical analysis software R, Python (registered trademark), SQL, Excel (registered trademark), etc. to construct the estimation model 35, and can use one or more of linear regression, kernel ridge, support vector, XGBoost, neural network (nnet), and random forest (RF), preferably random forest, as a method.

[0026] (Estimation step) In this estimation step, a plurality of continuation factors during discharging and / or charging of the power storage device to be determined and a plurality of post-rest factors after the discharge and / or charging are stopped are acquired, and the acquired factors (voltage) are used as explanatory variables to estimate the characteristics (capacity and resistance) of the determination target 14 from the estimation model 35. As described above, the estimation model 35 is constructed based on a plurality of continuation factors during discharging and / or charging of the power storage device 13 whose characteristics are known, and a plurality of post-rest factors after the discharge and / or charging are stopped. The control unit 21 uses the above-mentioned time tc as the continuation factors 25 and post-rest factors 26 of the determination target 14. F , and time ts FObtain a factor (voltage) according to the above.

[0027] (Decision step) In this determination step, the deterioration level of the power storage device to be determined is acquired based on the estimation result of the characteristic of the power storage device 13 estimated in the estimation step, and a process of determining this deterioration level is performed. For example, if the characteristic is battery capacity, the deterioration level according to the result of estimating the battery capacity of the determination target 14 is acquired, and based on whether it is within the allowable range 31, it is determined whether the determination target 14 should be adjusted and reused, or disassembled and recycled.

[0028] 3 is a flowchart showing an example of a deterioration determination processing routine executed by the control unit 21. This routine is stored in the storage unit 22 and executed in response to a user's instruction. After constructing the estimation model 35, the user connects the object 14 to be determined to the charger / discharger 17, and then executes this routine. When this routine starts, the control unit 21 of the estimation device 20 first acquires the settings of the duration factor and the post-pause factor (S200). As in S100, the control unit 21 determines the time tc of the duration factor used as an explanatory variable. F and the time ts of the post-rest factor F Next, the control unit 21 acquires the set value of tc F The control unit 21 acquires the measurement result of the factor (voltage) corresponding to the characteristic unknown from the measurement device 15 (S210). F The control unit 21 obtains from the measurement device 15 the measurement results of the factors (voltage) corresponding to the charging / discharging, and the continuation factors and post-pause factors after the charging / discharging is stopped (S220). After obtaining the factors during charging / discharging and the continuation factors and post-pause factors after the charging / discharging is stopped, the control unit 21 uses the factors as explanatory variables to estimate the characteristics of the object 14 to be determined from the estimation model 35 (S230), and obtains the degree of deterioration from the estimation result (S240). The control unit 21 derives the characteristics of the object 14 to be determined that best fit the explanatory variables of all the measurement points using the estimation model 35, and derives the degree of deterioration corresponding to the characteristics. Note that the explanatory variables can use the values described above for the estimation device.

[0029] Next, the control unit 21 determines whether the acquired deterioration level is within a predetermined allowable range (S250). This allowable range is set empirically based on, for example, a lower limit value at which the power storage device 13 of the evaluation target 14 can be reused through readjustment. When the deterioration level of the evaluation target 14 is within the predetermined allowable range, the control unit 21 outputs a message indicating that the device can be reused (S260). On the other hand, when the deterioration level of the evaluation target 14 is outside the predetermined allowable range, the control unit 21 outputs a message indicating that the device should be recycled (S270). FIG. 4 is an explanatory diagram of a display screen 40 that displays the diagnosis result of the evaluation target 14. As shown in FIG. 4, in addition to recycle and reuse, a rebuild decision, which involves adjustment and reuse, may be made, and information such as the remaining capacity and the degree of resistance increase may be added to the diagnosis result.

[0030] After S260 or S270, the control unit 21 stores the determination result of S250 and determines whether or not there is a next power storage device 13 to be determined (S280). If there is a next power storage device 13 to be determined, the control unit 21 repeatedly executes the processes from S200 onwards, whereas if there is no next power storage device 13 to be determined in S280, the control unit 21 ends this routine. In this way, the estimation device 20 uses the continuation factor 25 during charging and discharging of the power storage device 13 and the post-pause factor 26 after the charging and discharging are stopped to estimate the characteristics of the power storage device 14 from the estimation model 35, thereby deriving the degree of deterioration and determining whether or not the power storage device 13 can be reused.

[0031] The estimation system 10 of the present embodiment described above can evaluate the degree of deterioration of an electric storage device more efficiently and with improved accuracy. The reason why the estimation system 10 achieves such an effect is presumed to be as follows. For example, three main degradation modes of electric storage devices such as lithium-ion secondary batteries are known: “positive electrode degradation,” “negative electrode degradation,” and “capacity discrepancy between positive and negative electrodes.” Therefore, at least three explanatory variables are required to accurately estimate the degree of deterioration. Meanwhile, characteristics of an electric storage device, such as the degree of deterioration, are reflected in changes in internal resistance associated with deterioration. Therefore, the degree of deterioration can be estimated more efficiently and accurately from the time changes in factors such as voltage during charging and discharging, which reflect the internal resistance, and the time changes in factors after charging and discharging are stopped. In this way, the estimation system 10 can evaluate the characteristics of an electric storage device more accurately in a short time, for example, without performing capacity measurement, which takes a relatively long time to evaluate, or without adjusting the remaining capacity (SOC) of the electric storage device. Therefore, the estimation system 10 can estimate the characteristics of an electric storage device more efficiently and with improved accuracy.

[0032] It goes without saying that the present disclosure is not limited to the above-described embodiments, and can be embodied in various forms as long as they fall within the technical scope of the present disclosure.

[0033] For example, in the above-described embodiment, the estimation system 10 includes the measuring device 15 and the estimating device 20, but is not particularly limited to this, and the measuring device 15 may be omitted by acquiring voltage changes during charging and discharging from an external device. This estimating device 20 also uses the continuation time factors 25 including multiple points and the post-pause factors 26 including multiple points to estimate the characteristics of the object 14 to be determined from the estimation model 35, and therefore, it is possible to estimate the characteristics of the electricity storage device more efficiently while improving accuracy.

[0034] In the above-described embodiments, the present disclosure has been described as the estimation system 10, the estimation device 20, and the estimation method. However, for example, the present disclosure may be implemented as a program for executing the estimation method. This program may be recorded on a computer-readable recording medium (e.g., a hard disk, a ROM, a FD, a CD, a DVD, etc.), may be distributed from one computer to another via a transmission medium (a communication network such as the Internet or a LAN), or may be transmitted and received in any other form. By executing this program on one computer or by having multiple computers share and execute each step, each step of the above-described estimation method is executed, thereby obtaining the same effects as the estimation method. [Example]

[0035] Below, examples in which the estimation method and estimation device of the present disclosure are specifically examined will be described as experimental examples. Experimental examples 3 and 6 to 9 correspond to working examples of the present disclosure, and experimental examples 1 and 2 correspond to comparative examples.

[0036] Panasonic NCR18650B lithium-ion batteries were used to study the diagnosis of battery degradation. The degraded batteries were prepared by repeatedly charging at 2C between 2.5V and 4.2V at 20°C and discharging at 0.1C (20°C-2C cycle battery), repeatedly charging at 0.5C and discharging at 0.5C at 60°C (60°C-0.5C cycle battery), and storing the batteries for a long period (60°C-stored battery) after adjusting the remaining capacity (State of Charge) to 100% (4.2V). The long-term storage period was 20 or 215 days. For each battery, the discharge capacity was measured using an Asuka Battery 5V / 10A-80CH at 20°C with a CCCV discharge of 0.1C. This was used as an indicator of the degree of degradation. Additionally, for each degraded battery, CC discharge was performed at 0.1C for 30 minutes in the SOC range of 5 to 100% in 5% increments, after which the discharge was stopped. The voltage obtained during this process was used as an explanatory variable to predict the capacity as a battery characteristic. Figure 5 shows an example of the change in voltage over time for a new battery and a degraded battery. As shown in Figure 5, even with the same degree of capacity degradation, the change in voltage over time was found to differ depending on the degradation process.

[0037] Figure 6 shows the measurement results of the voltage change during discharge of a new battery at SOC = 100% and immediately after the discharge was stopped. The horizontal axis shows the logarithmic scale of time, and it was found that the change in voltage over time can be observed even in a short period of time. Figure 7 shows the change in voltage over time tc after the discharge of a new battery was stopped. F and voltage Vc F As shown in FIG. 7, the discharge time tc F and the voltage Vc at that time F First, define the discharge time from 0 to tc F The time is divided into n parts, and the time from the start of discharge to the beginning of discharge is calculated in order of shortest time. 1 , tc 2 , …, tc n-1 , tc n Similarly, the voltage at each time is Vc 1 , Vc 2 , …, Vc n-1 , Vc nSimilarly, the time ts after the discharge is stopped is used for diagnosis. F and the voltage Vs F and the discharge time is defined as 0 to ts F The time is divided into m parts, and the time from the start of discharge to the beginning of discharge is 1 , ts 2 , …, ts m-1 , ts m and the voltage at each time is Vs 1 , Vs 2 , …, Vs m-1 , Vs m Vc defined in this way 1 ~Vc n , Vc F and the voltage Vc at the start of discharge 0 , Vs 1 ~Vs m , Vc F and the voltage Vs when discharge stops 0 Using these as explanatory variables, machine learning was applied to the resulting database to predict capacity. Data was created for 20 conditions with different SOCs, including 7 new batteries, 21 batteries cycled at 20°C and 2C, 12 batteries cycled at 60°C and 0.5C, and 8 batteries stored at 60°C, and a prediction model was built using 960 data points.

[0038] To estimate the battery capacity, we used the statistical analysis software R and random forest as a machine learning method to build a prediction model with hyperparameters ntree and mtry set to ntree = 10,000 and mtry = 8, respectively. When only the discharge voltage was used as an explanatory variable to predict the capacity (tc F = 100 seconds, n = 28) is used as Experimental Example 1, and when only the voltage after the pause is used (ts F = 60 seconds, m = 59) in Example 2, when both voltages are used (tc F = 100 seconds, n = 28, ts F= 60 seconds, m = 59) as Experimental Example 3, and the results of comparing the capacity prediction results are shown in Table 1. Here, RMSE means the root mean square error between the experimental value and the predicted value, and the lower the RMSE, the higher the prediction accuracy. In this case, 10-fold cross-validation was performed, so 10 prediction models were constructed using training data, and the average RMSE of the 10 test data, i.e., RMSE CV The results are summarized in Table 1. The coefficient of determination R between the experimental values and the predicted values after 10-fold cross-validation was also calculated. CV 2 The R CV 2 The closer to 1, the higher the prediction accuracy. From the results in Table 1, it is found that using both the discharge voltage and the voltage after quiescence results in a lower RMSE CV becomes lower and R CV 2 is closer to 1, so it was found that the prediction accuracy is higher. Figure 8 is a diagram showing the relationship between prediction error and SOC according to explanatory variables. As shown in Figure 8, it is presumed that there are SOC ranges in the discharge voltage and the post-stop voltage that are difficult to predict. On the other hand, if both the discharge voltage and the post-stop voltage are used, as shown in Figure 8, the prediction error decreases across the entire SOC range, and it is presumed that the prediction accuracy will improve because the weak areas of each can be covered. The prediction results that follow use both the discharge voltage and the post-stop voltage as explanatory variables.

[0039] [Table 1]

[0040] Next, we examined the number of explanatory variables. Figure 9 shows the RMSE versus the number of explanatory variables. CV and R CV 2 In Figure 9, the explanatory variables are tc F = 100 seconds, ts F = 60 seconds, and the data was divided evenly on a log scale. As a result, the number of explanatory variables was 10 (n = 3, m = 3), and the RMSE CV shows a minimum, and RCV 2 The model was found to be optimal because it showed a local maximum. It was inferred that a small number of explanatory variables would reduce prediction accuracy due to an inability to fully represent multiple degradation modes, while a large number of explanatory variables would result in overfitting and thus reduced prediction accuracy. Here, capacity was predicted using 10 voltages as explanatory variables: voltages at 0, 0.01, 0.2, 5, and 100 seconds after discharge began, and voltages at 0, 1, 4, 15, and 60 seconds after discharge stopped. Figure 10 shows the relationship between the measured and predicted discharge capacity values using the 10 explanatory variables. Figure 10 also includes a boundary line that includes 90% of the measurement results. As shown in Figure 10, the predicted discharge capacity values were within approximately 90% of the measured values, demonstrating high prediction accuracy.

[0041] Figure 11 is a Cole-Cole plot of the AC impedance of a Panasonic lithium-ion secondary battery (LIB) at 20°C and SOC 50%. The impedance was measured using a Solartron CELLTEST-8T, with an AC voltage amplitude of 5mV. In the Cole-Cole plot of Figure 11, the values are larger at the bottom of the vertical axis, and after the charge transfer resistance of the positive and negative electrodes appears as two arcs, the real part Z' when the imaginary part Z" of the AC impedance reaches its maximum value is taken as the characteristic resistance R of this battery. E It is defined as 20℃ and 50% and is used as an index of deterioration along with the capacity. E The resistance at 20°C and 50% SOC is a resistance component with a relatively fast response, and is an index of resistance that is the sum of the DC resistance and the charge transfer resistance of the positive and negative electrodes. Figure 12 shows the resistance R at 20°C and SOC 50% using the same explanatory variables as in Figure 10. E This shows the relationship between the actual measured value and the predicted value. CV is 0.00751Ω, R CV 2 was 0.816, and it was found that these could also be predicted with high accuracy, demonstrating that predictions can be made from the perspective of two axes: battery capacity and resistance.

[0042] Next, we investigated the range of discharge time and rest time as explanatory variables for diagnosing deteriorated batteries. The results are shown in Table 2. The time tc used for diagnosis F When comparing between 60 seconds, 100 seconds, and 1000 seconds, the RMSE at 100 seconds is CV Since tc is low, 1000 seconds > tc F It was estimated that >60 seconds is desirable. The shorter the diagnostic time, the better. Therefore, 300 seconds >tc is desirable. F It was estimated that the time ts used for diagnosis was >60 seconds. F When comparing 40 seconds, 60 seconds, and 1260 seconds, the RMSE at 60 seconds is CV Because it is low, 1260 seconds > ts F It was estimated that a time of >40 seconds would be desirable. The shorter the diagnostic time, the better. Therefore, 300 seconds is desirable. F It was inferred that a time of >40 seconds would be desirable. As shown in Table 2, it was found that the prediction accuracy could be improved by dividing the data evenly on a logarithmic scale rather than on a real scale. Furthermore, although the test was conducted using a 0.1C discharge this time, it was inferred that a similar prediction could be made in the low current range, as no nonlinear response was observed. Furthermore, the statistical analysis software R was used for machine learning, but environments such as PHP, Matlab, and Excel could also be used, and although random forest (RF) was used as the method, linear regression, neural networks, support vector machines, etc. could also be used.

[0043] [Table 2]

[0044] Figure 13 shows the measurement results of voltage changes during charging of a new battery with an SOC of 10% and immediately after charging was stopped. Here, a new battery with an SOC of 10% was used, and CC charging was performed at 0.1 C for 30 minutes, after which the voltage change was measured after charging was stopped. As shown in Figure 13, which is the measurement results of Figure 6 flipped upside down, it was found that it is possible to predict the deterioration of an unknown deteriorated battery during charging, just as it was during discharge as described above.

[0045] It goes without saying that the estimation device, estimation system, estimation method, and program disclosed in this specification are in no way limited to the above-described examples, and can be implemented in various forms as long as they fall within the technical scope of the present disclosure. [Industrial Applicability]

[0046] The estimation device, estimation system, estimation method, and program therefor disclosed in this specification can be used in the technical field of determining deterioration of an electricity storage device. [Explanation of symbols]

[0047] 10 Estimation system, 12 Network, 13 Energy storage device, 14 Judgment target, 15 Measuring device, 16 Thermostatic chamber, 17 Charger / discharger, 20 Estimation device, 21 Control unit, 22 Memory unit, 23 Known deterioration test results, 25 Continuation factor, 26 Post-rest factor, 27 Degradation level, 28 Input device, 29 Display device, 30 Machine learning program, 31 Tolerance range, 35 Estimation model, 36 Degradation judgment program.

Claims

1. An estimation device for estimating a characteristic of an electricity storage device, a control unit that uses an estimation model constructed based on a plurality of continuous time factors during discharging and / or charging of an electric storage device whose characteristics are known and a plurality of post-pause factors after the discharging and / or charging is stopped, acquires the continuous time factors and the post-pause factors of a plurality of points of an electric storage device that is an estimation target, and estimates a capacity as a characteristic of the electric storage device from the estimation model using the acquired continuous time factors and the post-pause factors as explanatory variables, the control unit uses as explanatory variables the duration factor corresponding to the elapsed time including multiple points with different digits from the start of discharging and / or charging of the power storage device, and the post-rest factor corresponding to the elapsed time including multiple points with different digits from the stop of discharging and / or charging of the power storage device, the duration factor having 2 to 6 points and the post-rest factor having 2 to 6 points as explanatory variables, and sets the longest elapsed time tcF used in the estimation from the start of discharging and / or charging to be in the range of 60 seconds≦tcF≦1000 seconds, and the longest elapsed time tsF used in the estimation from the stop of discharging and / or charging to be in the range of 40 seconds≦tsF≦1260 seconds.

2. 2. The estimation device according to claim 1, wherein the characteristic of the power storage device is capacity, and the longest elapsed time tcF used in the estimation from the start of discharging and / or charging is set to a range of 60 seconds≦tcF≦300 seconds, and the longest elapsed time tsF used in the estimation from the stop of discharging and / or charging is set to a range of 40 seconds≦tsF≦300 seconds.

3. The estimation device according to claim 1 or 2, wherein the duration factor and the post-pause factor are a voltage of an electricity storage device.

4. The estimation device according to claim 1 , wherein the control unit acquires the continuation factor and the post-pause factor and constructs the estimation model by machine learning.

5. The estimation device according to claim 4 , wherein the control unit uses a random forest as the machine learning technique.

6. The estimation device according to any one of claims 1 to 5, wherein the control unit estimates characteristics of the power storage device from the estimation model using the continuation factor and the post-pause factor as explanatory variables, and determines a degree of deterioration of the power storage device.

7. a measuring device that discharges and / or charges an electricity storage device and measures the duration factor and the post-pause factor; The estimation device according to any one of claims 1 to 6, The control unit acquires the continuation factor and the post-pause factor of the power storage device from the measurement device.

8. An estimation method for estimating a characteristic of an electricity storage device, comprising: a step of using an estimation model constructed based on a plurality of continuous time factors during discharging and / or charging of an electric storage device whose characteristics are known and a plurality of post-rest factors after the discharging and / or charging is stopped, acquiring the continuous time factors and the post-rest factors at a plurality of points of an electric storage device to be estimated, and estimating a capacity as a characteristic of the electric storage device from the estimation model using the acquired continuous time factors and the post-rest factors as explanatory variables, the step uses, as explanatory variables, the duration factor corresponding to the elapsed time including multiple points with different digits from the start of discharging and / or charging of the electricity storage device, and the post-rest factor corresponding to the elapsed time including multiple points with different digits from the stop of discharging and / or charging of the electricity storage device; the duration factor of 2 to 6 points and the post-rest factor of 2 to 6 points are used as explanatory variables; the longest elapsed time tcF used in the estimation from the start of discharging and / or charging is set to a range of 60 seconds≦tcF≦1000 seconds, and the longest elapsed time tsF used in the estimation from the stop of discharging and / or charging is set to a range of 40 seconds≦tsF≦1260 seconds.

9. A program that causes one or more computers to implement the steps of the estimation method according to claim 8.

Images