Economic efficiency estimation device and economic efficiency estimation method using battery charge / discharge curve

Abstract

To estimate a charge-discharge curve and estimate an efficiency degradation state for various secondary batteries to perform charge prioritization for optimizing economical efficiency for an aggregate of the plurality of secondary batteries.SOLUTION: An economical efficiency estimation device derives an approximate curve of an initial open end voltage function and an initial impedance function to set a prior estimation function to measure a charge-discharge voltage and a charge-discharge current. A state estimation unit uses a Kalman filter to estimate state parameters including an SOC value, a polarization voltage and internal impedance. The device uses the state parameters to determine an estimation error between a prior estimation voltage and an actual measurement voltage, determines a post estimation value with a correction formula according to a Gaussian function using the calculated estimation error, a learning ratio set for the current and a correction width as a term, adjusts the prior estimation function based on the post estimation value, and sets a new prior estimation function to estimate a charge-discharge curve. The device estimates an economical index on the basis of a charge-discharge power amount of a secondary battery estimated from the charge-discharge curve.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] An embodiment of the present invention relates to an economic efficiency estimation device and an economic efficiency estimation method that estimate economic efficiency and efficiency degradation using an estimated charge / discharge curve of a storage battery. [Background technology]

[0002] In recent years, variable renewable energy sources such as solar power and wind energy have been increasing. Due to this increase in energy, various proposals have been made for utilizing surplus energy generated by fluctuations in power demand. Storage batteries that can be repeatedly charged and discharged, such as lithium-ion storage batteries, are becoming candidates for utilizing this surplus energy.

[0003] By charging surplus energy, these lithium-ion batteries can be used to cut peak demand for electricity, store nighttime electricity for use during the day, and as an emergency power source for disaster preparedness. Utilizing these lithium-ion batteries is being considered to optimize power usage when time-of-use tariffs are applied, thereby achieving economic benefits.

[0004] In addition, the lithium-ion batteries to be used include not only unused ones but also ones that have been used as a power source for various devices and then reused. Therefore, it is important to understand the deterioration state of lithium-ion batteries. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Laid-Open No. 2015-224927 Summary of the Invention [Problem to be solved by the invention]

[0006] When multiple lithium-ion batteries are used as an assembly, it is necessary to consider not only the degradation state of each individual lithium-ion battery but also its charge / discharge efficiency characteristics. The various lithium-ion batteries used in the assembly have different types, manufacturers, and reuse histories, and their charge / discharge efficiency characteristics will also vary. Naturally, operating highly efficient lithium-ion batteries reduces energy loss and increases economic benefits. In other words, when multiple lithium-ion batteries are used as an assembly, it is necessary to perform not only capacity degradation diagnosis but also efficiency degradation diagnosis.

[0007] This efficiency degradation diagnosis requires estimating the charge-discharge cycle energy efficiency, which is defined as the ratio of discharge energy to charge energy. This charge-discharge energy can be calculated using the full charge capacity (FCC), open circuit voltage (OCV), internal impedance (Z), and state of charge (SOC). Therefore, to calculate the charge-discharge energy, functions of the OCV value and Z value against the SOC value are required.

[0008] OCV estimation methods can be broadly divided into two groups. The first group uses pseudo OCV (pOCV). In the pOCV measurement process, a lithium-ion battery is generally charged and discharged once at a constant current of less than 10-hour rate (c / 10) from full discharge to full charge, and the OCV is calculated as the average charge / discharge voltage. The second group is the constant current intermittent titration (GITT) method. GITT is a square-wave current charge / discharge technique with a long relaxation time between each wave. The Z value can be estimated by dividing the difference between the charge / discharge voltage and the OCV value by the current. In other words, the Z value can be estimated using either the pOCV or GITT process.

[0009] However, the conditions for applying the pOCV value and GITT include a very long constant current period, which is quite different from the configuration of an assembly of multiple lithium-ion batteries, and therefore it is not suitable as a method for estimating the OCV value during operation.

[0010] Therefore, an object of the present invention is to provide an economic efficiency estimation device and an economic efficiency estimation method that use the charge / discharge curve of a storage battery to estimate the economic efficiency and efficiency degradation of various storage batteries, including reused storage batteries, and assemblies of multiple storage batteries. [Means for solving the problem]

[0011] In order to achieve the above object, an economic efficiency estimation device using charge / discharge curves according to an embodiment of the present invention includes a function derivation unit that derives approximate curves of an open circuit voltage function and an impedance function that are the subject of an initial correction and sets them as pre-estimated functions; a charging unit that detects the state of charge of a secondary battery and charges the secondary battery within a range up to a predetermined upper charge voltage limit; a discharging unit that electrically connects a load to the secondary battery and discharges power from the secondary battery; a measurement unit that measures charge / discharge voltages and charge / discharge currents at regular intervals after the charging or discharging unit starts charging or discharging; and a measurement unit that measures state parameters including a value of the charging rate, and a polarization voltage and internal impedance of the secondary battery. and an estimation calculation processing unit that calculates an ex-post estimation value by correcting the open circuit voltage value and the impedance value using a Gaussian function according to a correction formula using a Gaussian function with a predetermined learning rate L and correction width σ as terms, corrects the ex-post estimation value to set a new ex-post estimation function based on the ex-post estimation value, and estimates a charge and discharge curve, and an economic efficiency index is estimated based on the charge and discharge power amount of the secondary battery estimated from the charge and discharge curve.

[0012] Furthermore, a method for estimating economic efficiency using charge and discharge curves according to an embodiment of the present invention derives approximate curves of an open circuit voltage function and an impedance function to be subjected to an initial correction, and sets the approximate curves as a priori estimation function, detects a state of charge of a secondary battery, charges the secondary battery within a range up to a predetermined upper charge voltage limit, discharges power from the secondary battery using a load connected to the secondary battery, measures a charge and discharge voltage and a charge and discharge current at regular time intervals after charge and discharge start, estimates state parameters including a state of charge, a polarization voltage, and an internal impedance of the secondary battery using a Kalman filter algorithm, calculates estimation errors of the open circuit voltage value and the impedance value from the priori estimation function, the state parameters, and a relational equation between the charge and discharge voltage and the charge and discharge current measured by a measurement unit, corrects the open circuit voltage value and the impedance value using a Gaussian function according to a correction equation using a Gaussian function with a predetermined learning rate L and correction width σ as terms, calculates posterior estimates, corrects the prior estimation function based on the posterior estimates, and sets a new priori estimation function, estimates a charge and discharge curve, and uses the open circuit voltage function and the impedance function to

[0013]

number

[0014] According to the present invention, it is possible to provide an economic efficiency estimation device and an economic efficiency estimation method that use the charge / discharge curve of a storage battery to estimate the economic efficiency and efficiency degradation that can be applied to various storage batteries and assemblies of multiple storage batteries. [Brief explanation of the drawings]

[0015] [Figure 1] FIG. 1 is a block diagram showing the configuration of a power storage device equipped with a device for estimating a charge / discharge curve and economic efficiency of a storage battery according to an embodiment. [Figure 2] FIG. 2 is a block diagram showing an example of the configuration of a device for estimating a charge / discharge curve and economic efficiency of a storage battery. [Figure 3] FIG. 3 is a diagram showing an example of the specifications of a lithium ion storage battery. [Figure 4] FIG. 4 is a diagram conceptually showing an approximate curve for correcting the open circuit voltage OCV. [Figure 5] FIG. 5 is a diagram conceptually showing an equivalent circuit of a lithium ion battery. [Figure 6] FIG. 6 is a diagram showing the relationship between the error index and the condition combination of the learning rate L and the correction width σ. [Figure 7] Figure 7 shows three broad groups based on the speed of graph transformation (the magnitude of the learning rate): "fast (high speed)," "intermediate (medium speed)," and "slow (low speed)." [Figure 8A] FIG. 8A is a diagram showing the characteristics of the image of the initial V^(Soc) reference value in a process of increasing or decreasing the reference value in one direction. [Figure 8B] FIG. 8B shows the characteristics of the image of the initial V^(Soc) reference value due to the process of rotating the reference value around one Soc. [Figure 9A] FIG. 9A is a diagram showing the error index Imaer corresponding to the adjustment coefficient of the group with a high deformation speed. [Figure 9B] FIG. 9B is a diagram showing the error index Imaer corresponding to the adjustment coefficients in the medium deformation speed group. [Figure 9C] FIG. 9C is a diagram showing the error index Imaer corresponding to the adjustment coefficients of the group with a slow deformation speed. [Figure 10A] FIG. 10A is a diagram showing estimated charge / discharge energy values ​​when the first adjustment coefficient is set. [Figure 10B] FIG. 10B is a diagram showing the estimated charge / discharge energy value when the second adjustment coefficient is set. [Figure 10C] FIG. 10C is a diagram showing estimated charge / discharge energy values ​​when the third adjustment coefficient is set. [Figure 10D] FIG. 10D is a diagram showing estimated charge / discharge energy values ​​when the fourth adjustment coefficient is set. [Figure 10E] FIG. 10E is a diagram showing the estimated charge / discharge energy value when the fifth adjustment coefficient is set. [Figure 11] FIG. 11 is a flowchart for explaining a method for estimating a charge / discharge curve and an economical efficiency of a storage battery. [Figure 12] FIG. 12 is a diagram illustrating a Gaussian function. [Figure 13] FIG. 13 is a diagram showing the average error of the charge / discharge curve estimation result of this embodiment. [Figure 14] FIG. 14 is a diagram for explaining the economic merit estimated from the differential voltage of each cycle. DETAILED DESCRIPTION OF THE INVENTION

[0016] Hereinafter, an economic efficiency estimation device and an economic efficiency estimation method for a storage battery according to one embodiment of the present invention will be described with reference to the drawings. 1 shows the configuration of a power storage device equipped with a storage battery economic efficiency estimation device according to this embodiment. This power storage device 1 is mainly composed of a power conditioning system 2, a battery module 3, a battery management unit (BMU: Battery Management Unit) 4, an energy management unit (EMU: Energy Management Unit or EMS: Energy Management system) 5, a charge / discharge curve and economic efficiency estimation device 6, and a battery temperature measurement unit 7. Although not shown, components that are included in a normal power storage device are also included in the power storage device of this embodiment, and detailed description thereof will be omitted.

[0017] The power conditioner 2 functions as a converter, converting power supplied from a power grid 9 (such as an external power company), power supplied from a solar power generation system, or power supplied from the battery module 3 so that it can be used by electrically driven devices including the specific load 8. It may also function as a charger for charging a storage battery. For example, if the specific load 8 is an electrical device powered by AC power, the power conditioner 2 converts DC power supplied from the battery module 3 into AC power. Depending on the electrical device of the specific load 8, the power conditioner 2 may also boost the voltage of the power. Furthermore, the power conditioner 2 not only supplies power to the specific load 8, but can also reduce the consumption of power supplied from the power grid 9 by releasing the energy stored in the battery module 3 when consumption of power supplied from the power grid 9 is at its maximum. In this case, the discharged battery module 3 can be fully charged via the power conditioner 2 when power demand decreases, such as late at night.

[0018] The specific loads 8 to which the storage device 1 supplies power are intended to be devices to which power should be supplied when the power supply from the power grid 9 is stopped (for example, during a power outage), and include, for example, electronic devices such as computers and communication devices, to which power is supplied for power backup purposes.

[0019] The battery module 3 includes a secondary battery (storage battery) 11 that outputs DC current and voltage, a cell monitor unit (CMU: Cell Monitor Unit) 12, and a protection unit 13. The number of battery modules 3 is appropriately set according to the design of the power supply amount for a specific load, etc. To form a high-capacity secondary battery, multiple battery modules 3 may be electrically connected to form a single assembly, i.e., a single battery pack. In this embodiment, a lithium-ion battery is used as an example of the secondary battery 11 that is the subject of charge / discharge curve estimation, but this is not limiting. Similar to lithium-ion batteries, the estimation can be easily applied to batteries with different structures as long as they have a small memory effect and good self-discharge characteristics. For example, it can be applied to nanowire batteries, which are improved versions of lithium-ion batteries.

[0020] The secondary battery 11 of this embodiment is not limited to a battery internal material (electrode material, etc.) or cell structure, and the type of the exterior material may be a cylindrical can type, a rectangular can type, a laminate type, etc. The connection type of the secondary batteries 11 that make up the battery module 3 can be a known connection type such as a single cell, a series battery pack, or a parallel battery pack.

[0021] The battery temperature measurement unit 7 measures the temperature using a temperature sensor (not shown) that is arranged in contact with each secondary battery 11. The usable ambient temperature range for lithium-ion batteries within the device is approximately 5 to 40°C, but if necessary, depending on the installation environment (cold or tropical climate), a battery temperature control mechanism can be installed within the device. This battery temperature control mechanism is composed of a fan and heater that adjusts the temperature within the usable range of the secondary batteries 11 (approximately 5 to 40°C) based on the temperature measured by the battery temperature measurement unit 7, so as not to degrade battery performance when the upper or lower limit of a preset temperature range is exceeded. Of course, if the usable temperature range of the secondary batteries 11 is expanded due to future battery improvements, the device can accommodate all of that temperature range.

[0022] The cell monitoring unit 12 continuously measures the output voltage, current, and temperature of each secondary battery 11, which is a single battery (or single cell), and transmits the measurement results to the battery management unit 4. In particular, under the control of the calculation control unit 14, which will be described later, the charge / discharge voltage is measured at regular intervals during the charge / discharge process for estimating the charge / discharge curve.

[0023] Furthermore, the cell monitoring unit 12 transmits the output voltage, current, and temperature acquired from the secondary battery 11 as monitor information to the battery management unit 4. Based on the received monitor information, the battery management unit 4 determines whether an abnormality such as overcharging, over-discharging, or temperature rise has occurred, and controls the protection unit 13 to stop charging or output (discharging) to the secondary battery 11 to prevent overcharging and over-discharging. If an emergency abnormality occurs due to a failure of the secondary battery 11, the protection unit 13 stops charging or output (discharging) to the secondary battery 11 by electrically cutting off the power.

[0024] Furthermore, the protection unit 13 may have a function to avoid danger by notifying the battery management unit 4 of an abnormality. Note that it is essential to determine the occurrence of an abnormality. This determination function may be installed in either the cell monitoring unit 12 on the battery module 3 side or the battery management unit 4 on the energy storage device 1 side, but it may also be installed in each to enhance safety through dual determination. In dual determination, the order of determination is determined in advance, for example, the cell monitoring unit 12 first determines that an abnormality has occurred, and then the battery management unit 4 determines that an abnormality has occurred a second time. In this case, the determination procedure is usually such that if either of the two determination units determines that an abnormality has occurred, the protection unit 13 performs protective action. Note that, depending on the design concept, it is also possible to configure the protection unit 13 to perform protective action only when both determine that an abnormality has occurred, and to issue a warning if only one of them determines that an abnormality has occurred.

[0025] Furthermore, the battery management unit 4 centrally collects the monitoring information transmitted from the cell monitoring units 12 of each battery module 3 and transmits it to the higher-level energy management unit 5. Based on this monitoring information, the energy management unit 5 instructs the power conditioner 2 to charge and discharge the battery modules 3. The power conditioner 2 controls the charging and discharging of the battery modules 3 according to the instructions.

[0026] The energy management unit 5 is composed of an arithmetic and control unit 14 , a display unit 15 , a server 16 , and an interface unit 17 . The arithmetic control unit 14 has functions equivalent to those of a computer arithmetic processing unit, etc., and instructs the battery management unit 4 or the power conditioner 2 to charge and discharge the battery module 3. In addition, an upper limit charging voltage value and a lower limit discharging voltage value are set in advance for each battery module 3, and based on the monitor information transmitted from the battery management unit 4, the arithmetic control unit 14 instructs the battery management unit 4 or the power conditioner 2 to stop charging or discharging.

[0027] The display unit 15 is configured by, for example, a liquid crystal display unit, and displays the operating status of the power storage device 1, the remaining capacity of the battery module 3 (secondary battery 11), and warnings under the control of the calculation control unit 14. The display unit 15 may also employ a touch panel or the like and be used as an input device.

[0028] The server 16 stores the latest information, such as the operating status of the power storage device 1, monitor information on the battery module 3, etc., and information on charge / discharge curves, transmitted to the energy management unit 5. The interface unit 17 communicates with a centralized management system composed of an externally installed server, etc., via a network 18 such as the Internet.

[0029] Next, a description will be given of the storage battery charge / discharge curve and economic efficiency estimation device 6. Fig. 2 shows an example of the configuration of the storage battery charge / discharge curve and economic efficiency estimation device 6. This charge / discharge curve and economic efficiency estimation device 6 is composed of a charging power supply unit 22, a discharging unit 23, a discharging load unit 24, a measurement unit 25, a time measurement unit 26, an estimation calculation processing unit 27, an SOC calculation unit 28, a function derivation unit 29, and a state estimation unit 30. The estimated charge / discharge curve and economic efficiency index SOEc of each secondary battery 11 are stored in the server 16.

[0030] The charging power supply unit 22 detects the charging state of the secondary battery 11, and outputs a DC current voltage that matches the rating to the secondary battery 11 within the range up to the upper limit charging voltage of the secondary battery 11 to charge the secondary battery 11. This charging power supply unit 22 is provided as a power supply dedicated to estimating the charge / discharge curve and the economic efficiency index SOEc of the secondary battery 11, but a battery charging power supply unit that is normally provided in the power storage device or the power conditioner 2 may also be used. In this embodiment, the pressure measurement unit 25 and the charging power supply unit 22 form a charging unit. In the following description, estimating the economic efficiency index SOEc and economic efficiency estimation have the same meaning.

[0031] The discharge unit 23 includes a discharge load unit 24, and is electrically connected to the secondary battery 11 by operating a switch (not shown), causing a predetermined amount of power (assumed to be a constant current or a constant voltage) to be discharged from the secondary battery 11. This discharge load unit 24 may be a resistor or an electronic load, but it may also be possible to simulate a load and regenerate power to the power grid without providing such a dedicated load.

[0032] The measurement unit 25 measures the DC voltage and DC current output from the battery module 3 (secondary battery 11). The measurement is performed periodically at predetermined time intervals, measuring the DC voltage and DC current output from the battery module 3. Note that the voltage and current measurements can be performed without actually measuring them, by using the voltage and current values ​​contained in the latest monitor information sent from the battery management unit 4 and stored in the server 16 of the energy management unit 5.

[0033] The time measurement unit 26 is a timer for measuring the time during which power is being discharged from the battery module 3, and measures the measurement timing. The SOC calculation unit 28 calculates the state of charge (SOC) at the timing when the measurement unit 25 measures the voltage. The SOC value is the value obtained by dividing the charging current Qc [Ah] by the full charge capacity FCC [Ah] at that time. The SOC value of the state of charge in this embodiment is calculated using a Kalman filter, which will be described later. The full charge capacity FCC value can be calculated using an appropriate method, such as the charge / discharge voltage difference method, the AC impedance method, the discharge curve differentiation method, or an optimization filter. For example, it may be calculated by dividing the difference between the accumulated charge amounts at two points in time by the difference between the SOC estimated by the Kalman filter.

[0034] The function derivation unit 29 derives approximate curves of the open circuit voltage function and impedance function to be corrected and sets them as pre-estimated functions. Various curve fitting techniques can be applied to derive the approximate curves, and pre-determined approximate curves can be stored and set as pre-estimated functions. In addition, an initial open circuit voltage function and an initial pre-estimated function can also be derived.

[0035] The estimation calculation processing unit 27 is a calculation processing unit (CPU or the like) that stores a calculation algorithm using a relational expression described later and estimates a charge / discharge curve and economy (economic efficiency index SOEc) based on machine learning and estimated state parameters of the secondary battery 11. This estimation calculation processing unit 27 does not have to be provided exclusively in the device 6 for estimating charge / discharge curve and economy of the secondary battery, and the calculation control unit 14 of the energy management unit 5 can substitute for the processing function.

[0036] The state estimation unit 30 estimates state parameters including the state of charge SOC, open circuit voltage OCV, and internal impedance Z of the secondary battery 11 using a Kalman filter based on an electrochemical model, which in this embodiment is an extended Kalman filter that supports nonlinearity.

[0037] In addition, in this embodiment, an online estimation method using network communication between an external device 19 and a system server 20 that manages multiple other power storage devices 1 can be applied via a network communication network 18 such as the Internet. Note that, in Fig. 1, one power storage device 1 is shown as a representative for ease of explanation, and the number of devices is not limited to one.

[0038] [Kalman filter] The following describes the estimation of a charge / discharge curve used for diagnosing deterioration in the charge / discharge cycle of a lithium-ion battery and estimating economic efficiency using a Kalman filter and a Gaussian function in the state estimation unit 30. Fig. 3 is a diagram showing the specifications of an example of a lithium-ion battery. Fig. 4 is a diagram conceptually showing an approximate curve for correcting the open circuit voltage OCV. Fig. 5 is a diagram conceptually showing an equivalent circuit of a lithium-ion battery.

[0039] In this embodiment, as an example, eight lithium-ion storage battery modules connected in series are charged and discharged, and time-series data is acquired. While cycle tests using small lithium-ion storage battery cells are generally used to perform degradation diagnosis and state estimation, this embodiment also allows for testing by treating a lithium-ion storage battery module made up of large 50 Ah cells as a single battery pack (assembly). During the charge / discharge cycle test, a constant power mode is used to simulate the charging and discharging conditions of the power conditioner 2. Room temperature is desirable for the ambient temperature during the cycle test, but due to the inevitable resistance heat that occurs during charging and discharging, a module temperature range of approximately 20°C to 30°C is set.

[0040] The graph of open circuit voltage OCV and internal impedance Z for a lithium-ion battery shown in Figure 4 was transformed by adding a height-adjusted Gaussian function. That is, the error (estimation error) between the prior estimate (scalar) at the current point of the state of charge SOC and the posterior estimate (scalar) obtained using the Kalman filter was multiplied by the learning rate L to obtain the height of the Gaussian function, which was then added to the prior estimate (vector) from graph learning. The function of the graph transformed in this way is the posterior estimate (vector) from graph learning.

[0041] Next, the error index I maer Define as the mean absolute error rate.

[0042]

number

[0043] From the results shown in Figure 6, the estimation process is -2 In this case, divergence occurs in most cases, and when the correction width σ is reduced, convergence is approached. This is because a large learning rate L and a correction width σ are the E ~ In addition, the tendency of the optimal correction width σ is different when the learning rate L is L=10 -3.8 and L=10 -3.2Here, the combinations of the learning rate L and the correction range σ are roughly classified into three groups, "fast (high speed)," "medium (medium speed)," and "slow (low speed)," which differ in the transformation speed (the magnitude of the learning rate L or the learning speed) of the graph shown in Figure 7. The relationship between the magnitude of the learning rate L and the transformation speed is such that the larger the learning rate L, the faster the transformation speed, and the smaller the learning rate L, the slower the transformation speed.

[0044] Furthermore, when graph transformation by Gaussian function addition and a Kalman filter are used together, if a simple integration is attempted, the learning rate L is set to a value close to 1, but this setting causes the learning to diverge, making it impossible to estimate the charge / discharge energy appropriately. Therefore, in this embodiment, the learning rate L is adjusted to "0.1 or less" and the correction width σ is also adjusted to prevent divergence. Furthermore, it is more preferable that the relationship between the learning rate L and the correction width σ is expressed by the following formula (3), which is a correction formula using a Gaussian function:

[0045]

number

[0046] Next, we artificially adopt the error of the OCV profile data and calculate the C k This assumption means that the initial open-circuit voltage (OCV) is unclear, which is a concern in the case of reused lithium-ion batteries. When starting to use a reused battery, the current OCV value or perhaps the electrode material used in the cell may be known, but the overall open-circuit voltage (V^oc(Soc)) profile is unknown. Therefore, the open-circuit voltage (V^oc(Soc)) must be estimated from limited information about the lithium-ion battery.

[0047] In order to numerically separate the reference value V^oc(Soc) of the initial open circuit voltage OCV from the actual initial Voc(Soc), the following two equations (4) and (5) are used.

[0048]

number

[0049] Figure 8A shows the K sep,1 and K. sep,2 In the image of the initial V^oc(Soc) reference value generated by sep,1 and K. sep,2 In the image of the reference value of the initial V^oc(Soc) generated by

[0050] In these Figures 8A and 8B, the central line A in the figure is the actual V^oc(Soc), and the other lines B, C, D, and E around it are K sep,1 and K. sep,2 Furthermore, by combining these two equations, we can express it as the following equation (6).

[0051]

number

[0052] Figures 9A, 9B, and 9C show various K sep,1and K. sep,2 The error index I in the charge / discharge energy estimation corresponding to the value of maer Figure 9A shows the K sep,1 and K. sep,2 The error index I corresponding to maer Figure 9B shows the K sep,1 and K. sep,2 The error index I corresponding to maer Figure 9C shows the K sep,1 and K. sep,2 The error index I corresponding to maer This shows:

[0053] 9A, 9B, and 9C and equation (6), K sep,1 is the mean absolute error rate of the charge / discharge energy estimation error, I maer In response to this, K sep,2 had little impact. sep,1 The deviation from 1 in maer This resulted in an increase in maer The slower the deformation speed, i.e., the smaller the learning rate L, the greater the tendency for the error to increase.

[0054] Next, based on these observations, we consider the estimation of the charge and discharge energy for each cycle. sep,1 and K. sep,2 Starting from an arbitrary or uncertain initial OCV value generated from Equation (6), this arbitrary OCV value was generated using Equation (6). In addition, one of the three graph transformation conditions shown in Figure 7 was used. In this study, an estimate starting from the correct open circuit voltage OCV profile was also included for comparison. The results are shown in Figures 10A-10E.

[0055] 10A-10E show charge / discharge energy estimates starting from an arbitrary initial OCV value generated by equation (6). sep,1 =0.8, K sep,210B shows the estimated charge / discharge energy when the second adjustment coefficient K is set to 0.2. sep,1 =0.8 and K sep,2 = 1.8, Fig. 10C shows the third adjustment coefficient K sep,1 =1.2 and K sep,2 = 0.2, Fig. 10D shows the fourth adjustment coefficient K sep,1 =1.2 and K sep,2 = 1.8, Fig. 10E shows the fifth adjustment coefficient K sep,1 =K sep,2 = 1.

[0056] In each of Figures 10A to 10E, line A (charge) and line A' (discharge) show the actual changes in charge and discharge energy for each cycle. The other lines B, C, and D (charge) and dotted lines B', C', and D' (discharge) in the figure represent the change in W^ in the following equation (7). c,d The estimated charge and discharge energy for each cycle is calculated as follows. Specifically, line A represents the actual charge, dotted line A' represents the actual discharge, line B represents the charge with a fast transformation rate (large learning rate L), and dotted line B' represents the discharge with a fast transformation rate. Line C represents the charge with an intermediate transformation rate (medium learning rate L), and dotted line C' represents the medium-rate discharge with an intermediate transformation rate. Line D represents the charge with a slow transformation rate (small learning rate L), and dotted line D' represents the slow discharge.

[0057]

number

[0058] As can be easily seen from Figures 10A to 10E, the transition of the estimated charge / discharge energy value is influenced by Ksep,1 but not by Ksep,2. Therefore, Ksep,1 is a determining factor for the sign of the initial charge / discharge energy estimation error. For example, if Ksep,1 is smaller than 1 (Ksep,1<1), the estimated value of the initial charge / discharge energy starts from a value smaller than the actually measured value. On the other hand, if Ksep,1 is larger than 1, the estimated value starts from a value larger than the actually measured value.

[0059] The number of cycles required for error convergence is affected by the graph deformation speed shown in Figure 7. For the "fast," "medium," and "slow" deformation speed conditions, the number of cycles required for convergence was approximately 20, 40, and 100, respectively. The progression of the estimated values ​​after 100 cycles under all conditions was similar to that when starting from the correct initial open circuit voltage OCV shown in Figure 10E.

[0060] As a result, in this embodiment, the charge / discharge energy per cycle can be estimated and its energy efficiency can be calculated. In this embodiment, even if the estimation is performed for a reused lithium-ion battery and the initial values ​​of the open circuit voltage OCV and internal impedance Z are arbitrary and unknown, the estimation can be performed in the same way as for an unused lithium-ion battery.

[0061] This embodiment has the following two advantageous aspects. First, it offers the potential for Kalman filter-based differential analysis of the open-circuit voltage (OCV). Conventionally, lithium-ion batteries are discharged at low currents and the discharge capacity is differentiated with respect to the discharge voltage, or vice versa. This process aims to reduce polarization during low-current discharges, and the time-series data is approximated by the open-circuit voltage (OCV). This differential analysis is known to be effective in diagnosing the internal degradation state of lithium-ion batteries, for example, the transition of the operating window of the positive and negative electrodes. However, because the required operation differs from actual operation, which requires long discharge times and short-term processing, it is not easy to apply this method to actual reused batteries.

[0062] Secondly, in this embodiment, the open circuit voltage OCV profile can be continuously estimated during operation of the lithium ion battery, which involves degradation, and therefore, real-time differential analysis of the open circuit voltage OCV can be realized.

[0063] [Economic efficiency estimation device] Next, the economic efficiency estimation device using the charge / discharge curve of the storage battery according to this embodiment will be described with reference to the flowchart shown in FIG. First, an initial function derivation process is performed (Step 1). Specifically, an approximation curve is derived from the initial open circuit voltage function Voc(Soc) and the initial impedance function Z(Soc) that are the subject of the first correction. A known approximation method for charge / discharge curves can be used to derive this approximation curve. For example, after a charge / discharge test is performed in advance on the battery module 3, the charge / discharge voltage values ​​and current values ​​in one charge / discharge cycle are measured. An approximation curve is calculated from the charge / discharge voltage values ​​and current values ​​and stored. Then, the charge / discharge curve, the open circuit voltage, and the impedance function are calculated using the following equation (8):

[0064]

number

[0065] Next, the discharge unit 23 electrically connects the secondary battery 11 to the discharge load unit 24 to start discharging, and the time measurement unit 26 starts measuring time (step S2). In this embodiment, the process starts with discharging, but it may also be configured to start with charging. After discharging starts, it is determined whether the voltage is at the lower limit of discharge (step S3). If the determination in step S3 indicates that the lower limit of discharge voltage has been reached (YES), charging is started (step S8). On the other hand, if the lower limit of discharge voltage has not been reached (NO), discharging continues. Either a constant current or a constant power can be selected as the discharging method. After discharging starts, it is determined whether a specified time, which will be described later, has elapsed (step S4). If the determination in step S4 indicates that discharging continues until the specified time has elapsed (NO), and if the specified time has elapsed (YES), the measurement unit 25 measures the discharge voltage V of the secondary battery 11. meas The charge / discharge current I is measured, and the SOC value is calculated by the SOC calculation unit 28 (step S5). The SOC calculation unit 28 calculates the state of charge SOC using the Kalman filter of the state estimation unit 30 described above. At this time, correction is repeatedly performed to improve the accuracy of the ex-post estimation approximate curve G. This is done by correcting the approximate curve using positive and negative voltage values ​​(relative to the pre-estimated value) actually measured at multiple state of charge SOCs.

[0066] Next, after the discharge voltage and discharge current measurements and the SOC value calculation are completed, machine learning processing is performed (step S6). In the machine learning processing, a prior estimation value is calculated using a prior estimation function and actual measurement values ​​using a relational expression described below. Furthermore, the prior estimation value is corrected using the estimation error and a Gaussian function to obtain a posterior estimation value, and a posterior estimation approximation curve is obtained based on the posterior estimation value. In such machine learning processing, the obtained posterior estimation approximation curve is set as a new prior estimation function, and the accuracy of the approximation curve can be improved by repeating the correction.

[0067] Next, it is determined whether the number of repeated corrections has reached a predetermined number of repetitions (step S7). If it is determined that the number of corrections has not reached the number of repetitions (NO), the process returns to step S3 and continues discharging. If discharging is to continue, it is determined again whether the voltage is at the lower limit discharge voltage (step S3), and if it has not reached the lower limit discharge voltage (NO), the processing routine from step S3 to step S7 is repeated. Furthermore, if it is determined in step S7 that the number of corrections has reached the number of repetitions (YES), a charge / discharge curve is estimated based on a relational expression with voltage using approximate curves of the open circuit voltage function and the impedance function (step S14). The estimated charge / discharge curve is temporarily stored in server 16 of energy management unit 5.

[0068] If it is determined in step S3 that the lower limit discharge voltage has been reached (YES), charging is started (step S8). After charging starts, it is determined whether the voltage has reached the upper limit charge voltage (step S9). If it is determined in step S9 that the upper limit charge voltage has not been reached (NO), it is determined whether a predetermined time has elapsed (step S10). On the other hand, if it is determined that the upper limit charge voltage has been reached (YES), the process returns to step S2.

[0069] If the determination in step S10 is YES, after a predetermined time has elapsed, the measurement unit 25 measures the discharge voltage Vc,d and charge / discharge current I of the secondary battery 11, and the SOC calculation unit 28 calculates the state of charge SOC using the Kalman filter of the state estimation unit 30 (step S11). At this time, correction is repeatedly performed to improve the accuracy of the ex-post estimated approximate curve G. This is done by correcting the approximate curve using positive and negative voltage values ​​(relative to the pre-estimated value) actually measured at multiple state of charge SOCs. Furthermore, after the charge voltage and charge current are measured and the SOC value is calculated in step S11, machine learning processing, which will be described later, is performed (step S12).

[0070] Next, it is determined whether the number of repeated corrections has reached a predetermined number of repetitions (step S13). If it is determined in step S13 that the number of repetitions has not been reached (NO), the process returns to step S9 and charging continues. If charging is to be continued, it is determined in step S9 whether the charging voltage is at the upper limit voltage, and if it is not at the upper limit voltage and it is determined in step S13 that the number of repetitions has not been reached, the processing routine from step S9 to step S12 is repeated.

[0071] Here, the specified time for measuring the charge / discharge voltage in step S4 and step S10 will be described. Discharge is initiated by electrically connecting the battery module 3 to the discharge load unit 24 by the discharge unit 23. In this embodiment, the actual measurement value is measured every time a certain time has elapsed since the start of discharge. This certain time must be long enough so that the state of charge (SOC) does not change significantly, and is set to approximately several tens of seconds (e.g., 10 to 80 seconds). Note that this specified time range is merely an example, and may vary depending on the device configuration and measurement characteristics; it is not strictly limited. Furthermore, if the time elapsed since the start of discharge exceeds the specified time range but voltage measurement has not yet begun, discharge is stopped and treated as a measurement error.

[0072] If the determination in step S13 above indicates that the number of repetitions has been reached (YES), the process proceeds to step S14, where a charge / discharge curve is estimated based on a relationship equation with voltage using approximate curves of the open-circuit voltage function and the impedance function. The estimated charge / discharge curve is stored in the server 16 of the energy management unit 5. The stored charge / discharge curve is read from the server 16 upon request and can be used for display on the display unit 15 or for calculating the amount of charge / discharge power. In the charge / discharge curve estimation process, the charge / discharge process, the battery temperature stabilization, or the voltage and current measurements performed during the charge / discharge process may be performed on a single battery. Furthermore, the single battery may be a battery unit in which multiple batteries are connected in parallel or in series. Furthermore, the voltage and current measurements during the charge / discharge process or the battery temperature stabilization may be performed on various storage batteries or on an assembly of multiple storage batteries.

[0073] Next, an economic efficiency estimation process (described later) is performed (step S15), and the full charge capacity FCC, the charge energy Ec, the discharge energy Ed, and the economic merit G for one charge / discharge cycle are estimated in a predetermined charge / discharge cycle, and an economic efficiency index SOEc is estimated. The estimated economic efficiency index SOEc is stored in the server 16 of the energy management unit 5, and then the series of routines is terminated. The stored economic efficiency index SOEc can be read from the server 16 upon request and displayed on the display unit 15.

[0074] [Machine learning processing] Next, the machine learning processing of this embodiment in the above-mentioned steps S6 and S12 will be described. First, the sampled profile data for a plurality of charging rates SOC are expressed as vectors in the following equation (9).

[0075]

number

[0076]

number

[0077]

Number

[0078]

Number

[0079] Next, the algorithm of the Kalman filter will be described. The state estimation unit 6 of the present embodiment uses an adjusted extended Kalman filter for the state estimation of the lithium-ion battery. The state equation and the observation equation of the extended Kalman filter were formulated as a linear discrete-time system.

[0080]

Number

[0081]

number

[0082]

number

[0083]

number

[0084]

number

[0085]

number

[0086] The two key steps of the Kalman filter are as follows: First, in the prediction step, the state space vector x^ of the prior prediction is calculated using the following equations (24) and (25). - k+1 and error covariance matrix P k+1 was calculated.

[0087]

number

[0088]

number

[0089] Next, we will explain the extraction of the ex-post estimation values ​​of the open circuit voltage OCV and the internal impedance Z using a Kalman filter, and the ex-ante estimation error for graph transformation using a profile. From the Kalman filter process described above, the a posteriori estimates of the open circuit voltage OCV and the internal impedance Z are calculated using the following equations (29) and (30). The internal impedance Z is assumed to be a DC resistance, and the overall polarization voltage V pol,1(k) +V pol,2(k) +I c,d(k) +V pol,1(k) It is defined as the quotient of Z0(k) and the charge / discharge current.

[0090]

number

[0091]

number

[0092] Prior estimation error V ~ oc(k+1)= V^oc(k+1) - V^oc - (k+1) and Z ~ (k+1) = Z^(k+1) - Z^ - (k+1) is E in Eq. (10) ~ At the same time, S^oc,0 is substituted for S^oc(k+1).

[0093] Next, the estimation of charge and discharge energy will be described. The sampled values ​​of the open circuit voltage OCV and the internal impedance Z corresponding to the vector D were continuously updated by the estimated parameters and the profile graph deformation process using the Kalman filter.

[0094]

number

[0095]

number

[0096]

number

[0097] In addition, V^c(Soc) is the C k Because the initial V^oc(Soc) is used as a reference, the initial V^oc(Soc) must be set artificially. The V^oc(Soc) profiles of commercially available lithium-ion batteries can typically be roughly set for three types of electrodes: Co / Mn / Ni positive electrode type, FePO4 positive electrode type, and Ti negative electrode type. Although the profiles vary slightly depending on the active material configuration chosen by the lithium-ion battery manufacturer, a roughly set initial V^oc(Soc) can be converged using a Kalman filter and continuous learning with profile graph transformation. This suggests that even if a certain degree of uncertainty is initially set, the convergence achieved through continuous learning using a Kalman filter and profile graph transformation is a semi-supervised learning method that leads to the correct solution. Meanwhile, the initial Z^(Soc) is set to 0 by setting x^0 = 0 in the Kalman filter. Furthermore, Z^(Soc) can also be converged in the same way as V^c(Soc).

[0098] Next, an analysis of the estimation error and estimation accuracy of charge / discharge energy will be described. In this embodiment, we mainly discussed the learning rate L, the correction width σ of the graph, and the difference between the initial V^oc(Soc) and the actual open circuit voltage (OCV) profile. To analyze and quantitatively compare, we measured the actual value Wc,d,i of the ith cycle and calculated the corresponding estimation error W ~ c,d,i are calculated from the estimated value W^c,d,i using the following equation (36).

[0099]

number

[0100]

number

[0101]

number

[0102] Generally, a battery's economic merit G (yen / kWh) is highest when it is new and its charge / discharge efficiency (Wh / Wh) is high, and lowest when it is near the end of its life, when degradation has progressed and its full charge capacity FCC (Ah) and charge / discharge efficiency (Wh / Wh) are low. For lithium-ion batteries, the end of their life is typically defined as when their full charge capacity FCC (Ah) has fallen to 60% to 80% of that of a new battery. Here, the state of economy (SOEc), which indicates the current state of a battery's economic merit Gcurr, is defined as 1 when the economic merit Gnew of a new battery is obtained, and 0 when only the economic merit Gend of an end-of-life battery is obtained. This is defined as equation (39) below:

[0103]

number

[0104] The differential voltage Ve can be measured while a stationary lithium-ion battery is in operation. Therefore, if the relationship between the differential voltage Ve, the full charge capacity FCC, and the charge / discharge voltages Vc(SOC) and Vd(SOC) is determined in advance, the economic efficiency index SOEc can be obtained.

[0105] As described above, according to this embodiment, the state estimation of the lithium-ion battery using the Kalman filter can use the estimated state of charge (Soc) to determine (Soc, 0), i.e., the central coordinate of the graph transformation on the SOC axis. Even if an uncertain OCV profile generated by Equation (5) is initially set, accurate convergence of the OCV profile and adaptive estimation of the actual charge / discharge energy of each cycle as shown in Figures 10A-10E can be achieved.

[0106] This embodiment is characterized by its robustness against an uncertain OCV reference function arbitrarily determined by the Kalman filter for estimating the state of a lithium-ion battery. That is, for a reused lithium-ion battery whose origin is unknown, as long as information on one of the following is known: 1) ternary, 2) iron phosphate, or 3) titanate, an initial OCV reference can be provisionally set from publicly known information, even if it is somewhat uncertain. By using the Kalman filter algorithm and a Gaussian function, state parameters including the state of charge (SOC), polarization voltage, and internal impedance can be estimated and converged to appropriate values, and learning for efficiency degradation diagnosis can be converged. In this embodiment, even if the initial values ​​of the open circuit voltage (OCV) and internal impedance (Z) are arbitrary and unknown, estimation can be performed in the same way as for an unused lithium-ion battery. That is, this method can be applied to reused lithium-ion batteries whose origin is unknown.

[0107] Therefore, the power storage device of this embodiment can estimate the SOC value, polarization voltage, and internal impedance of a lithium-ion storage battery, even if it is unused or reused, by using a state estimation unit 30 that employs a Kalman filter in the charge / discharge curve and economic efficiency estimation device 6 and a graph transformation method using Gaussian function addition. Therefore, even if a plurality of reused lithium-ion storage batteries are used as a single assembly, they can be operated with high efficiency, taking into account the characteristics of each individual reused lithium-ion storage battery.

[0108] According to this embodiment, it is possible to estimate a charge / discharge curve with high accuracy, and furthermore, since the charge / discharge curve is estimated from an approximation curve of the open circuit voltage and impedance, it is possible to estimate a charge / discharge curve that can adapt to changes in the current rate. The estimated charge / discharge curve can be used for various purposes, such as calculating the charge / discharge energy E and evaluating the economic efficiency of a storage battery.

[0109] Furthermore, if the energy storage device of this embodiment can obtain in real time the economic efficiency index SOEc of a distributed cooperative system, for example, a stationary lithium-ion battery in which a plurality of lithium-ion batteries, including recycled lithium-ion batteries, are installed as a single aggregate, it will be possible to determine charging priorities, in which the more profitable ones will be preferentially selected and operated from among the stationary lithium-ion batteries (distributed batteries) that are being controlled cooperatively, thereby enabling economical operation.

[0110] The present invention is not limited to the above-described embodiments, and various modifications may be made without departing from the spirit of the present invention. Furthermore, various inventions that solve the above problems can be extracted by selecting or combining multiple disclosed constituent elements. [Explanation of symbols]

[0111] 1...energy storage device, 2...power conditioner, 3...battery module, 4...battery management unit, 5...energy management unit, 6...charge / discharge curve and economic efficiency estimation device, 7...battery temperature measurement unit, 8...specific load, 9...power system, 11...secondary battery, 12...cell monitoring unit, 13...protection unit, 14...arithmetic and control unit, 15...display unit, 16...server, 17...interface unit, 18...network communication network, 19...external device, 20...system server, 22...charging power supply unit, 23...discharging unit, 24...discharging load unit, 25...measurement unit, 26...time measurement unit, 27...estimation calculation processing unit, 28...SOC calculation unit, 29...function derivation unit, 30...state estimation unit.

Claims

1. An economic efficiency estimation device using a charge / discharge curve, a function derivation unit that derives approximate curves of the open circuit voltage function and the impedance function that are the subject of the initial correction and sets them as pre-estimated functions; a charging unit that detects the charging state of the secondary battery and charges the secondary battery within a predetermined upper charging voltage limit; a discharge unit that electrically connects a load to the secondary battery and discharges power from the secondary battery; a measuring unit that measures a charge / discharge voltage and a charge / discharge current at regular intervals after the charging unit or the discharging unit starts charging / discharging; a state estimation unit that estimates state parameters including a state of charge value, a polarization voltage, and an internal impedance of the secondary battery using a Kalman filter algorithm; an estimation calculation processing unit that calculates estimation errors of open circuit voltage values ​​and impedance values ​​from the a priori estimation function, the state parameters, and a relational expression between the charge / discharge voltage and the charge / discharge current measured by a measurement unit, corrects the open circuit voltage values ​​and the impedance values ​​using a Gaussian function according to a correction equation using a Gaussian function with a predetermined learning rate L and correction width σ as terms, calculates posterior estimated values, corrects the a priori estimation function based on the posterior estimated values, sets a new a priori estimation function, and estimates a charge / discharge curve; Equipped with An economic efficiency estimation device that estimates an economic efficiency index based on the charge / discharge power amount of the secondary battery estimated from the charge / discharge curve.

2. 2. The economic efficiency estimation device according to claim 1, wherein the open circuit voltage function and the impedance function are differentiable and integrable functions.

3. 2. The economic efficiency estimation device according to claim 1, wherein the Kalman filter includes a differentiation of the open-circuit voltage function in the observation equation, and the open-circuit voltage function before differentiation includes an error.

4. The learning rate L is adjusted to 0.1 or less, and the relationship between the learning rate L and the correction width σ is [Equation 1] 2. The economic efficiency estimation device according to claim 1, wherein a learning rate L and a correction width σ are adopted such that:

5. The correction formula using the Gaussian function is [Equation 2] where M is the deformation vector, l is the order in which the elements of vector M are specified, L is the learning rate, and E ～ 4. The economic efficiency estimation device according to claim 1, wherein: : error of the target physical quantity; Soc: charging rate; (Soc,0): center coordinate on the charging rate axis of the Gaussian function.

6. Using the open circuit voltage function and the impedance function, [Equation 3] However, C fC : full charge capacity value corresponding to each cycle, W^c,d: estimated value of estimated charge / discharge energy, V^oc(Soc): open circuit voltage estimation function, Z^(Soc): impedance estimation function, I^ c,d (Soc): Charge / discharge current, 2. The economic efficiency estimation device according to claim 1, wherein the device estimates charge / discharge energy per cycle.

7. The economic efficiency estimation device according to claim 6, wherein the economic efficiency is estimated using the charge / discharge energy and the electricity rates for each time period.

8. The economic efficiency estimation device according to claim 7 , further comprising: determining charging priorities of distributed storage batteries related to measures to deal with surplus power from variable renewable energy based on the estimated economic efficiency.

9. Approximate curves of the open circuit voltage function and the impedance function to be corrected for the first time are derived and set as pre-estimated functions; Detects the charging state of the secondary battery and charges it within a predetermined upper charging voltage limit. Discharging power from the secondary battery through a load connected to the secondary battery; After charging and discharging starts, the charging and discharging voltage and current are measured at regular intervals. Using a Kalman filter algorithm, state parameters including a state of charge value, a polarization voltage and an internal impedance of the secondary battery are estimated; determining estimation errors of the open circuit voltage value and the impedance value from the pre-estimation function, the state parameters, and a relational expression between the charge / discharge voltage and the charge / discharge current measured by a measurement unit; correcting the open circuit voltage value and the impedance value using a Gaussian function according to a correction expression using a Gaussian function with a predetermined learning rate L and correction width σ as terms to calculate posterior estimates; correcting the pre-estimation function based on the posterior estimates to set a new pre-estimation function, and estimating a charge / discharge curve; Using the open circuit voltage estimation function and the impedance estimation function, [Equation 4] However, C fC : full charge capacity value corresponding to each cycle, W^c,d: estimated value of estimated charge / discharge energy, V^oc(Soc): open circuit voltage estimation function, Z^(Soc): impedance estimation function, I^ c,d (Soc): An economical efficiency estimation method in which the charge / discharge energy per cycle is estimated as the charge / discharge current, and the economical efficiency is estimated using the charge / discharge energy and the electricity rates for each time period.

10. The economic efficiency estimation device of claim 1, wherein the economic efficiency index is calculated by dividing the difference between the current economic merit and the economic merit at the end of the battery's life for one charge / discharge cycle by the difference between the economic merit when the battery is new and the economic merit at the end of the battery's life.

Images