JP6342612B2 - Secondary battery diagnostic device and secondary battery diagnostic method - Google Patents

Description

  The present invention relates to a secondary battery diagnostic device and a secondary battery diagnostic method, and more specifically, for example, deterioration factors and capacities of secondary batteries such as lithium ion secondary batteries (hereinafter referred to as “secondary batteries”). The present invention relates to a secondary battery diagnostic device and a secondary battery diagnostic method for determining a battery state.

Currently, high-capacity secondary batteries such as lithium-ion secondary batteries are already widely used as power sources for electronic devices such as notebook personal computers, mobile phones, and video cameras. Furthermore, in the future, secondary batteries having a higher capacity and higher capacity density will be more widely used in applications such as hybrid vehicles, electric vehicles, and home-installed power storage devices as compared to conventional electronic devices.
These high-capacity, high-capacity secondary batteries not only have a larger capacity than conventional secondary batteries, but also have a predetermined capacity over a long period of time under conditions where the temperature environment is extremely severe, such as in a car or outdoors. It is required to maintain a state that satisfies the specifications.

However, when a secondary battery is left in a high charge state (fully charged state or near this state) in a high temperature environment for a long time or is used in a state in which excessive charge / discharge cycles are repeated, the battery characteristics deteriorate. May occur.
When such a battery has deteriorated, it may not be possible to recover the battery capacity as in the specification even if it is charged by a regular method, or the battery may not be able to extract power as in the specification. There is.

When the secondary battery deteriorates in this way, it is necessary to take measures such as using it in accordance with the usage method according to the degree of deterioration or replacing it with a new one. Therefore, in order to use a secondary battery over a long period of time in a severe temperature environment, establishment of a method for estimating the degree of deterioration of the secondary battery with practically sufficient accuracy is required.
However, at present, a method for estimating the degree of deterioration with sufficient accuracy in the actual use of a secondary battery has not yet been established. In general, the degree of deterioration of battery characteristics is roughly estimated. I'm just doing it.

In addition, a method has been proposed that can calculate the degree of deterioration with a certain degree of accuracy, but the device is too complex to be applied in the actual use of a secondary battery, and it requires a long time measurement. It is difficult to use due to problems such as lack of accuracy.
The main methods for estimating the degree of deterioration of the secondary battery proposed so far include the following.

(1) A method of measuring the internal resistance of the secondary battery. For example, the internal resistance related value related to the internal resistance of the secondary battery by a predetermined method is compared with the correspondence relation between the internal resistance related value and the battery state that has been grasped in advance. judge.
(2) A method of measuring the voltage at the time of discharging stop and at the time of stopping charging by giving a charge / discharge cycle of discharging, discharging stop, charging, and charging stop. For example, a unit charge / discharge cycle in which the secondary battery is discharged for a predetermined time, paused after stopping the discharge, charged with current, paused after stopping the charging for a predetermined time is repeated once or continuously several times, and the last unit charge is repeated. A voltage difference between the voltage at the time of stopping the discharge and the voltage at the time of stopping the charging in the discharge cycle is obtained, and a predetermined set value set in advance is compared with this voltage difference to determine the deterioration state of the secondary battery.

  (3) An alternating current impedance method for measuring alternating current impedance by sequentially applying a plurality of alternating currents having different frequencies. For example, the deterioration state of the battery is determined based on the internal impedance of the battery. This AC impedance method is one of the methods for measuring the internal resistance of a battery, and generates an equivalent circuit based on a complex impedance obtained by applying AC to the battery to calculate the internal resistance.

  (4) A method of measuring current behavior after switching to constant voltage charging in the constant current constant voltage charging method. For example, the behavior of the current flowing in the lithium ion secondary battery after the closed circuit voltage of the lithium ion secondary battery reaches the second specified voltage value by constant current charging and the charging method is switched to constant voltage charging is measured. Estimate the degree of deterioration of the secondary battery.

  An energy storage device including a battery pack in which a plurality of secondary batteries are connected in series is used, for example, as a power supply device for HEV (Hybrid Electric Vehicle) or UPS (Uninterruptable Power Systems). The HEV power supply device is required to have a high voltage of 100 V or higher. As a secondary battery used for this power supply device, a battery using a nickel metal hydride (Ni-MH) battery or a lithium ion (Li-ion) battery has been put into practical use. However, the average voltage per cell of these electricity storage elements is only about 1.2 V for nickel metal hydride batteries and about 3.6 V for lithium ion batteries. Therefore, in order to configure a 100 V power supply device, it is necessary to connect 83 cells in a nickel metal hydride battery and 28 cells in a lithium ion battery in series.

When charging / discharging is repeated in such an energy storage device in which a large number of cells are connected in series, the voltage of each cell varies. Causes of variations include variations in inter-cell capacitance and internal resistance, differences in operating temperature due to cell placement, and the like. Generally, the greater the number of cells connected in series, the greater the voltage difference between the cells.
When the cell is a lithium ion battery, the overcharge and / or overdischarge protection circuit limits the maximum voltage during charging and the minimum voltage during discharging. For this reason, when the voltage between cells connected in series is varied, the following problems occur.

  That is, when charging, when one cell reaches the maximum voltage, the charging is stopped, so the charging is terminated while the voltage of the other cell is low, and the maximum capacity of the entire device is reached. Charging stops before reaching. On the other hand, the reverse phenomenon occurs during discharge. At the time of discharge, the discharge is stopped when one cell reaches the minimum voltage. Therefore, the discharge ends with the voltage of the other cell being high, and the discharge is performed while leaving energy in the cell. It will stop.

  Even if the cell is not a lithium ion secondary battery, the protection circuit may stop charging / discharging if necessary, but even if this is not the case, the voltage between cells connected in series varies. Since this is the same, when the cell voltage rises above the withstand voltage during repeated use after charging to near the withstand voltage, a cell whose life is deteriorated occurs. Since the life of an energy storage device in which a large number of cells are connected in series is limited by the cell having the shortest life, the generation of a cell whose voltage is higher than other cells due to charge / discharge is not preferable.

For example, in Patent Document 1, the battery capacity is estimated from the value of the imaginary part of the AC impedance in a nickel-based secondary battery.
In Patent Document 2, the battery capacity is estimated from the absolute value of the AC impedance or the value of the real part.
Moreover, in patent document 3, the battery capacity is estimated from the absolute value or the real part value at a frequency at which the value of the imaginary part of the AC impedance is close to zero.

Moreover, in patent document 4, regarding NiH batteries, the batteries are classified into four areas of a normal area and three deterioration areas based on an internal resistance-related value or an AC impedance-related electric quantity. In addition, a method for obtaining the maximum output density from the value of AC impedance is disclosed.
Moreover, in patent document 5, the present deterioration state is estimated only in the secondary battery degraded by cycle degradation based on the impedance measurement data by the alternating current impedance method, and the discharge capacity is estimated.

JP-A-8-250159 JP-A-8-43507 JP-A-8-43506 JP 2007-87772 A JP 2011-133443 A

However, in Patent Documents 1, 2, and 3, the deterioration of the secondary battery exhibits different deterioration states depending on storage deterioration, cycle deterioration, and temperature environment of each deterioration. That is, in order to accurately estimate the current deterioration state, it is necessary to correctly estimate the deterioration factors.
Patent Document 4 does not disclose how to estimate the discharge capacity. Moreover, it does not disclose the relationship between the degradation area, the AC impedance value for each area, and the maximum output density.

  In Patent Document 5, the deterioration state of the secondary battery varies depending on the temperature environment where the deterioration proceeds even if it is cycle deterioration. Furthermore, the deterioration caused by the storage deterioration exhibits a deterioration state different from the cycle deterioration. That is, in order to accurately estimate the current deterioration state, it is necessary to correctly estimate the deterioration factors. The above-described patent document does not disclose a method for estimating the current degradation state and estimating the discharge capacity after estimating the degradation factor.

  The present invention has been made in view of such a problem, and an object of the present invention is to provide a secondary battery diagnostic apparatus that determines a battery state such as a deterioration factor and a capacity based on complex impedance measurement data by an AC impedance method, and It is to provide a secondary battery diagnostic method.

The present invention has been made to achieve such an object, and the invention according to claim 1 is directed to a secondary battery diagnostic apparatus for determining a battery state of a secondary battery, in a data collection measurement secondary battery. A reference data holding unit (2) for storing reference data in which information on deterioration factors and information on complex impedance values of the data collection measurement secondary battery are associated with each other, and a target for determining a battery state An impedance calculation unit (3) that calculates complex impedance of a measured secondary battery (1), the reference data stored in the reference data holding unit (2), and the impedance calculation unit (3) the information of the value itself of the calculated complex impedance, the deterioration factor estimating unit that estimates a deterioration factor of the measured secondary battery (1) and (4), the deterioration factor estimating unit (4 To diagnose determined based on the estimation result, characterized in that it comprises a diagnosis determination unit (5) for outputting a result of the diagnosis.

The invention according to claim 2 is the invention according to claim 1, wherein the complex impedance is at least two types of complex impedances.
The invention according to claim 3 is the invention according to claim 2, wherein the complex impedance is a complex impedance calculated using different frequencies.
The invention according to claim 4 is the invention according to claim 3, wherein the complex impedance is different from a real part or an imaginary part of the complex impedance when measured at a specific frequency and the specific frequency. It is a real part or an imaginary part of a complex impedance when measured at a frequency.

According to a fifth aspect of the present invention, there is provided a secondary battery diagnostic method for determining a battery state of a secondary battery, wherein information on deterioration factors of the data collection measurement secondary battery, and complex impedance of the data collection measurement secondary battery. Storing the reference data associated with the information of the value itself in the reference data holding unit (2), and calculating the impedance of the complex impedance of the secondary battery (1) to be measured, which is a target for determining the battery state The step of calculating by the unit (3), the reference data stored in the reference data holding unit (2), and the information of the complex impedance value itself calculated by the impedance calculation unit (3), Based on the step of estimating the deterioration factor of the secondary battery (1) to be measured by the deterioration factor estimating unit (4) and the estimation result of the deterioration factor estimating unit (4) To diagnose determination by the diagnostic determination unit (5), characterized in that it has a step of outputting the result of the diagnosis.

The invention according to claim 6 is the invention according to claim 5 , wherein the complex impedance is at least two types of complex impedances.
The invention according to claim 7 is the invention according to claim 6 , wherein the complex impedance is a complex impedance calculated using different frequencies.
The invention according to claim 8 is the invention according to claim 5, 6 or 7 , wherein the complex impedance is a real part or an imaginary part of a complex impedance when measured at a specific frequency, and the specific impedance. It is a real part or an imaginary part of complex impedance when measured at a frequency different from the frequency.

  ADVANTAGE OF THE INVENTION According to this invention, the secondary battery diagnostic apparatus and secondary battery diagnostic method which determine battery states, such as a deterioration factor and a capacity | capacitance, based on the complex impedance measurement data by an alternating current impedance method are realizable.

It is a block block diagram for demonstrating Example 1 of the secondary battery diagnostic apparatus which concerns on this invention. It is a block block diagram for demonstrating Example 2 of the secondary battery diagnostic apparatus which concerns on this invention. It is a figure which shows the flowchart for demonstrating the secondary battery diagnostic method of this invention. It is a figure which shows the flowchart for demonstrating the data collection of a data collection measurement secondary battery. It is a figure which shows the relationship between 100 Hz complex impedance imaginary part ((omega | ohm)) and 0.1Hz complex impedance real part ((omega | ohm)) of all the degradation factors. It is a figure which shows the relationship of 0.1 Hz complex impedance imaginary part ((omega | ohm)) of 45 degreeC cycle and 60 degreeC cycle, and 4 Hz complex impedance imaginary part ((omega | ohm)). Relation between 100Hz complex impedance imaginary part (Ω) and 1,000Hz complex impedance real part (Ω) of 25 ° C cycle excluding 45 ° C cycle and 60 ° C cycle, 25 ° C storage, 45 ° C storage and 60 ° C storage FIG. Furthermore, it is a figure which shows the relationship of the complex impedance imaginary part ((omega | ohm)) of 4Hz of a 25Hz cycle except 45 degreeC preservation | save, 45 degreeC preservation | save, and 60 degreeC preservation | save, and the complex impedance real part ((ohm)) of 1,000 Hz. Furthermore, it is a figure which shows the relationship between the 25-degree C cycle except the 60 degreeC preservation | save, the complex impedance imaginary part ((omega | ohm)) of 0.1 Hz of 45 degreeC preservation | save, and the complex impedance real part ((ohm)) of 0.1 Hz. It is a figure which shows the flowchart for demonstrating the content of FIG. 5 thru | or FIG. It is a figure which shows the flowchart which investigates the degradation factor estimation precision of 8-dimensional feature-value data. It is a figure which shows the relationship between the complex impedance imaginary part ((omega | ohm)) of 4Hz and the complex impedance real part ((omega | ohm)) of 100 Hz of all the degradation factors. It is a figure which shows the relationship between the complex impedance imaginary part ((ohm)) of 100 Hz of 60 degreeC cycle and 60 degreeC preservation | save, and the complex impedance real part ((ohm)) of 0.1 Hz. Relationship between the complex impedance imaginary part (Ω) of 4 Hz and the real part of complex impedance (Ω) of 1,000 Hz for the 25 ° C cycle, 45 ° C cycle, 25 ° C storage and 45 ° C storage excluding the 60 ° C cycle and 60 ° C storage FIG. Furthermore, it is a figure which shows the relationship between the complex impedance imaginary part ((omega | ohm)) of 4Hz complex impedance of 100 Hz, and a 25 Hz cycle except a 25 degreeC cycle, a 45 degreeC cycle, and 45 degreeC preservation | save. Furthermore, it is a figure which shows the relationship between the 25-degree cycle except a 45 degreeC cycle, the complex impedance imaginary part ((ohm)) of 0.1 Hz of 45 degreeC preservation | save, and the complex impedance imaginary part ((ohm)) of 4 Hz. It is a figure which shows the flowchart which investigates deterioration factor estimation precision. It is a figure which shows the relationship between the complex impedance imaginary part (Ω) of 4 Hz and the complex impedance imaginary part (Ω) of 10 Hz of all deterioration factors. It is a figure which shows the relationship between the complex impedance imaginary part ((omega | ohm)) of 4Hz of complex impedance of 4 Hz of a 60 degreeC cycle and 60 degreeC preservation | save. The relationship between the complex impedance imaginary part (Ω) of 4 Hz and the complex impedance imaginary part (Ω) of 10 Hz for 25 ° C cycle, 45 ° C cycle, 25 ° C storage and 45 ° C storage excluding 60 ° C storage and 60 ° C storage is shown. FIG. Furthermore, it is a figure which shows the relationship between the 45-degree C cycle except a 25 degree-C cycle, 25 degreeC preservation | save, and the 4Hz complex-impedance imaginary part ((omega | ohm)) and the complex impedance real part ((ohm)) of 100 Hz. Furthermore, it is a figure which shows the relationship between the 45-degree C cycle except 25 degreeC preservation | save, the 10-Hz complex impedance imaginary part ((omega | ohm)) of 45 degreeC preservation | save, and the 100-Hz complex impedance imaginary part ((ohm)). It is a figure which shows the flowchart which investigates deterioration factor estimation precision. It is a figure which shows the relationship between the complex-impedance imaginary part (Ω) of 0.1 Hz and the complex-impedance real part (Ω) of 0.1 Hz of all deterioration factors. 45 ° C cycle excluding 25 ° C cycle, 60 ° C cycle, 25 ° C storage, 45 ° C storage and 60 ° C storage 100Hz complex impedance imaginary part (Ω) and 100Hz complex impedance real part (Ω) FIG. It is a figure which shows the relationship between the complex impedance imaginary part ((ohm)) of 4 Hz of 45 degreeC preservation | save and 60 degreeC preservation | save, and the complex impedance real part ((ohm)) of 4 Hz. Furthermore, it is a figure which shows the relationship between the complex impedance imaginary part (Ω) of 4 Hz and the complex impedance real part (Ω) of 4 Hz of 45 ° C cycle excluding 45 ° C storage and 60 ° C storage, 60 ° C cycle and 25 ° C storage. . Furthermore, it is a figure which shows the relationship between the complex impedance imaginary part (Ω) of 0.1 Hz and the complex impedance real part (Ω) of 0.1 Hz of 45 ° C. cycle and 60 ° C. cycle excluding storage at 25 ° C. It is FIG. (1) which shows the relationship between 4Hz complex impedance real part ((omega | ohm)) and a capacity | capacitance maintenance factor (%). It is a figure (the 2) which shows the relationship between 4Hz complex impedance real part ((omega | ohm)) and a capacity | capacitance maintenance factor (%). FIG. 6 is a diagram (part 3) illustrating a relationship between a real part (Ω) of a complex impedance of 4 Hz and a capacity retention rate (%). It is FIG. (The 4) which shows the relationship between 4Hz complex-impedance real part ((omega | ohm)) and a capacity | capacitance maintenance factor (%). It is FIG. (5) which shows the relationship between 4Hz complex-impedance real part ((omega | ohm)) and a capacity | capacitance maintenance factor (%). It is FIG. (6) which shows the relationship between 4Hz complex-impedance real part ((omega | ohm)) and a capacity | capacitance maintenance factor (%). It is FIG. (7) which shows the relationship between 4Hz complex-impedance real part ((omega | ohm)) and a capacity | capacitance maintenance factor (%). It is FIG. (The 8) which shows the relationship between 4Hz complex impedance real part ((omega | ohm)) and a capacity | capacitance maintenance factor (%). It is a figure (the 1) which shows the relationship between 4Hz complex impedance imaginary part ((omega | ohm)) and a capacity | capacitance maintenance factor (%). It is a figure (the 2) which shows the relationship between 4Hz complex impedance imaginary part ((omega | ohm)) and a capacity | capacitance maintenance factor (%). It is a figure which shows the relationship between the complex impedance imaginary part ((ohm)) of 0.1Hz and the complex impedance imaginary part ((ohm)) of 100Hz after 4.10V rest of all the deterioration factors. It is a figure which shows the relationship between the complex impedance imaginary part ((omega | ohm)) of 4 Hz after 3.20V rest of all the degradation factors, and the complex impedance imaginary part ((omega | ohm)) of 50.2 Hz. It is FIG. (1) which shows the flowchart for demonstrating isolation | separation of a deterioration factor. FIG. 6 is a (second) diagram illustrating a flowchart for explaining separation of deterioration factors. It is a figure which shows the relationship between the complex impedance imaginary part ((omega | ohm)) of 1 Hz after the rest of 4.10V of all degradation factors, and the complex impedance real part ((omega | ohm)) of 1,000 Hz. It is a figure which shows the relationship between the complex impedance imaginary part (Ω) of 0.1 Hz and the complex impedance imaginary part (Ω) of 100 Hz after 4.10V rest of 50 ° C cycle excluding the 25 ° C cycle and 50 ° C storage. It is a figure which shows the relationship between the complex impedance imaginary part ((omega | ohm)) of 1Hz and the complex impedance real part ((omega | ohm)) of 1000 Hz after 3.20V rest of all the degradation factors. It is a figure which shows the relationship between the complex impedance imaginary part ((ohm)) of 4Hz complex impedance of 4Hz after 3.20V rest of 50 degreeC cycle and 50 degreeC preservation | save except 25 degreeC cycle, and 50.2Hz. FIG. 11 is a third diagram illustrating a flowchart for explaining separation of deterioration factors; FIG. 6 is a (fourth) diagram illustrating a flowchart for explaining separation of deterioration factors; FIG. 6 is a diagram (part 1) illustrating a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a capacity retention rate (%). FIG. 5 is a diagram (part 2) illustrating a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a capacity maintenance ratio (%). FIG. 6 is a diagram (part 3) illustrating a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a capacity retention rate (%). It is FIG. (4) which shows the relationship between the real part ((omega | ohm)) of 1000 Hz complex impedance, and a capacity | capacitance maintenance factor (%). It is FIG. (5) which shows the relationship between the real part ((omega | ohm)) of 1000 Hz complex impedance, and a capacity | capacitance maintenance factor (%). It is FIG. (1) which shows the relationship between the real part ((omega | ohm)) of 1000 Hz complex impedance, and a power capacity maintenance factor (%). It is the figure (the 2) which shows the relationship between the real part ((ohm)) of 1000 Hz complex impedance, and a power capacity maintenance factor (%). FIG. 6 is a diagram (No. 3) showing a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a power capacity retention rate (%). FIG. 6 is a diagram (part 4) illustrating a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a power capacity retention rate (%). It is FIG. (5) which shows the relationship between the real part ((omega | ohm)) of 1000 Hz complex impedance, and a power capacity maintenance factor (%). FIG. 6 is a diagram (part 1) illustrating a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a residual current capacity (Ah) at 4.0V. FIG. 9 is a diagram (part 2) illustrating a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a residual current capacity (Ah) at 4.0 V; FIG. 6 is a diagram (part 3) illustrating a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a residual current capacity (Ah) at 4.0 V; FIG. 6 is a diagram (part 4) illustrating a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a residual current capacity (Ah) at 4.0 V; FIG. 10 is a diagram (No. 5) showing a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a remaining current capacity (Ah) at 4.0 V;

Embodiments of the present invention will be described below with reference to the drawings.
In the present invention, reference data is created by degrading a plurality of secondary batteries under various conditions, collecting data such as complex impedance and capacity for each degradation factor, and associating information related to the data. A secondary battery for performing such data collection is referred to as a data collection measurement secondary battery. In addition, a secondary battery that is used for actual use and diagnoses information on deterioration factors and capacity is referred to as a measured secondary battery. As a matter of course, the secondary battery for data collection and measurement and the secondary battery to be measured are secondary batteries of the same product made by the same material, the same design and the same manufacturing process.

The information on the capacity in the present invention is a general term for an index for grasping the capacity of the secondary battery, such as a current capacity, a power capacity, or a capacity maintenance rate during charging or discharging.
The “current capacity” is a general term for a charging current capacity and a discharging current capacity. The “charging current capacity” indicates the amount of electricity stored while the battery is changed from a fully discharged state to a fully charged state. In addition, the “fully charged state” means a state where a specified charging voltage is reached. The “completely discharged state” means a state where a specified discharge voltage is reached.

  “Discharge current capacity” refers to the amount of electricity released while the battery is in a fully discharged state. The “power capacity” is a general term for a charge power capacity and a discharge power capacity. “Charge power capacity” refers to the amount of power stored while the battery is in a fully charged state to a fully charged state. “Discharge power capacity” refers to the amount of power released while the battery is in a fully discharged state.

FIG. 1 is a block diagram for explaining a first embodiment of a secondary battery diagnostic apparatus according to the present invention. In the figure, reference numeral 1 denotes a secondary battery to be measured, 2 denotes a reference data holding unit, 3 denotes an impedance calculation unit, 4 denotes a deterioration factor estimation unit, and 5 denotes a diagnosis determination unit.
The secondary battery diagnostic device according to the first embodiment is a secondary battery diagnostic device that determines a battery state of a secondary battery during charging or discharging.

The reference data holding unit 2 stores reference data in which information on deterioration factors of the data collection measurement secondary battery and information on the complex impedance of the data collection measurement secondary battery are associated with each other. The impedance calculation unit 3 calculates the complex impedance of the secondary battery 1 to be measured.
Further, the deterioration factor estimation unit 4 determines the deterioration factor of the secondary battery 1 to be measured based on the reference data stored in the reference data holding unit 2 and the output information of the complex impedance calculated by the impedance calculation unit 3. To be estimated. The diagnosis determination unit 5 performs diagnosis determination based on the estimation result of the deterioration factor estimation unit 4 and outputs the diagnosis result.

The complex impedance in the present invention is preferably at least two types of complex impedance. Here, the at least two types of complex impedances include, for example, complex impedances calculated using different frequencies, complex impedances calculated using different voltages, and combinations of real or imaginary parts of complex impedances. . There is no particular limitation as long as it matches the definition of at least two types of complex impedances, but complex impedances calculated using different frequencies are preferable.
The complex impedance is more preferably a real part or an imaginary part of the complex impedance when measured at a specific frequency and a real part or an imaginary part of the complex impedance when measured at a frequency different from the specific frequency. .

FIG. 2 is a block diagram for explaining a second embodiment of the secondary battery diagnostic apparatus according to the present invention. Reference numeral 6 in the figure denotes a capacity estimation unit. In addition, the same code | symbol is attached | subjected to the component which has the same function as FIG.
The reference data holding unit 2 in the secondary battery diagnostic device of the second embodiment includes information on the deterioration factor of the data collection measurement secondary battery, information on the complex impedance of the data collection measurement secondary battery, and further data collection measurement. Reference data associated with information on the capacity of the secondary battery is stored.
Further, the capacity estimation unit 6 uses the reference data, the output information from the deterioration factor estimation unit 4, and the output information of the complex impedance calculated by the impedance calculation unit 3 to provide information on the capacity of the secondary battery 1 to be measured. Is estimated.

FIG. 3 is a flowchart for explaining the secondary battery diagnosis method of the present invention.
First, charge / discharge data of a secondary battery for data collection and measurement deteriorated under a predetermined condition (deterioration factor) and complex impedance data by EIS (Electrochemical Impedance Spectrocopy) measurement are obtained, and these measurement data are obtained. Data collection by saving is performed (step S1). Next, a feature amount is determined from the complex impedance collected by data collection, and feature amount data is created (step S2). Next, learning data and evaluation data are created from the feature data, and an evaluation deterioration factor model is created from the learning data (step S3). Next, the estimation accuracy of the evaluation deterioration factor model is evaluated using the evaluation data (step S4). Next, it is determined whether the estimated accuracy is equal to or higher than a predetermined accuracy (step S5). Next, if it is below accuracy, it returns to step S2, and if it is above accuracy, each evaluation degradation factor model is adopted as a degradation factor model (discriminant function) (step S6). Next, the characteristic amount is measured in the secondary battery to be measured whose battery state is to be determined (step S7). Next, the deterioration factor model and the feature quantity of the secondary battery to be measured are compared (step S8). Next, the deterioration factor of the secondary battery to be measured is estimated from the comparison result (step S9). After the feature amount data is created in step S2, the relationship data between the feature amount data and the information related to the capacity of the data collection and measurement secondary battery is created for each deterioration factor (step S10). Next, the capacity of the secondary battery to be measured is estimated from the relationship data between the estimated deterioration factor of the secondary battery to be measured and the information on the capacity of the data collection and measurement secondary battery created for each deterioration factor (step S11). .

  Hereinafter, specific functions of the secondary battery diagnostic device of the present invention will be described sequentially with reference to the drawings.

<Data collection of the secondary battery to be measured>
Data collection with a high current capacity (2.5 Ah) 18650 cylindrical battery (hereinafter referred to as “A battery”) for a commercially available notebook PC battery using lithium cobalt oxide (LiCoO ₂ ) as a positive electrode material and a carbon-based material as a negative electrode material To implement. The standard environment was room temperature (25 ° C.), and the conditions (deterioration factors) for promoting deterioration were the following six types.
1) Cycle degradation under 25 ° C environment (25 ° C cycle degradation)
2) Cycle degradation in a 45 ° C environment (45 ° C cycle degradation)
3) Cycle degradation under 60 ℃ environment (60 ℃ cycle degradation)
4) Fully charged storage deterioration at 25 ° C (25 ° C storage deterioration)
5) Fully charged storage deterioration at 45 ° C (45 ° C storage deterioration)
6) Full charge storage deterioration under 60 ° C environment (60 ° C storage deterioration)

  Cycle deterioration is a deterioration condition that repeats charging and discharging under a predetermined temperature environment. Charging is CC (constant current) charging at 1C, and if the battery voltage reaches 4.2V, switch to CV (constant voltage charging) to control the current so that the battery voltage is maintained at 4.2V. CC-CV charging was terminated when the current reached 50 mA (hereinafter, CC-CV charging is referred to as “1 CCC-4.2 V 50 mA aborted CV charging”).

The discharge was performed at 0.5 C, and the discharge was terminated when the battery voltage reached 3.0 V (hereinafter, CC discharge was referred to as “0.5 C 3.0 V aborted CC discharge”).
The charge / discharge profile for cycle deterioration repeats “1CCC-4.2V50mA censored CV charge” and “0.5C3.0V censored CC discharge” with a predetermined period of rest (a state in which neither charging nor discharging is performed). It was.

  Specifically, first, a pause of 3 hours (waiting for the test battery to reach the same temperature as the predetermined temperature environment) is performed, and then “1CCC-4.2V50 mA censored CV charge” is performed for 10 minutes. This is a charge / discharge profile that repeats “charge-pause-discharge-pause” 25 times of “0.5 C 3.0 V censored CC discharge” after 10 pauses. Here, the test battery in the present invention is a general term for a data collection measurement secondary battery and a measurement secondary battery.

The cycle-degraded test battery is transferred to the standard environment every time the cycle deterioration charge / discharge profile under a predetermined temperature environment is performed once, and the EIS based on the predetermined capacity measurement charge / discharge profile and the predetermined EIS profile. The complex impedance at each frequency was measured.
The charge / discharge profile for the capacity measurement of cycle deterioration is a pause of 3 hours (waiting for the test battery to reach the same temperature as the standard environment), and then “1CCC-4.2V50mA censored CV charge”. After 10 minutes of rest, “0.5C3.0V censored CC discharge” is performed, after 10 minutes of rest, “1CCC-4.2V50mA censored CV charge” is performed, and 10 minutes of rest, “3 hours rest− It is a charge / discharge profile which performs "charge-pause-discharge-pause-charge-pause" once.

The fully charged storage deterioration is deterioration in which a test battery that is fully charged under a standard environment is moved to a predetermined temperature environment and left for a certain number of days.
The fully charged state of the A battery is a state immediately after charging by “1CCC-4.2V50 mA censored CV charging” under the standard environment. Similarly, the complete discharge state is a state immediately after the discharge by the “0.5 C3.0 V censored CC discharge” in the standard environment.

A fully charged battery that has been fully charged is transferred to the standard environment every time it is left in a predetermined temperature environment for 5 days, and each capacity is measured by a predetermined capacity measurement charge / discharge profile and each EIS measurement by a predetermined EIS profile. The complex impedance of frequency was measured.
The charge / discharge profile for measuring the capacity of the fully charged battery is initially suspended for 3 hours (waiting for the test battery to reach the same temperature as the standard environment), and then "0.5C3.0V censored CC discharge" After 10 minutes of rest, “1CCC-4.2V50mA censored CV charge” is performed. After 10 minutes of rest, “0.5C3.0V censored CC discharge” is performed. After 10 minutes of rest, “1CCC- This is a charge / discharge profile in which “4.2 V 50 mA censored CV charge” is performed, and “3 hour pause-discharge-pause-charge-pause-discharge-pause-charge-pause”, which is a 10-minute pause, is performed once.

During charging and discharging in the capacity measurement profile, charging current capacity and voltage and discharging current capacity and voltage are measured, respectively. Accordingly, the charge current capacity, discharge current capacity, charge power capacity, and discharge power capacity due to deterioration can be measured. These capacities decrease as the deterioration progresses.
In addition, in order to eliminate the influence of the deterioration factor, the measured value of the capacity used was the discharge and charging in the latter half of each capacity measurement profile.

  The predetermined EIS profile means that the fully charged test battery is discharged by 0.5C, and when the voltage of the test battery reaches the predetermined voltage, the EIS measurement is performed after a pause of 15 minutes. When 0.5C is discharged again and the EIS measurement is repeated after a pause of 15 minutes when the voltage of the test battery reaches the next predetermined voltage, and the test battery voltage reaches 3.0V For example, the EIS profile is such that the discharge is terminated and the operation is terminated after a pause of 15 minutes.

  The voltage of the test battery that stopped discharging for performing EIS was set to five voltages of 4.0V, 3.8V, 3.6V, 3.4V, and 3.2V. In addition, in order to specify the EIS measurement conditions, EIS measurement after a 15-minute pause after reaching the voltage of each test battery is referred to as “4.0 V EIS after pause”, “3.8 V EIS after pause”, “3 . EIS after 6V pause, "3.4V EIS after pause," "3.2V EIS after pause." Note that when the discharge is stopped after reaching a specific voltage, the voltage of the test battery increases rapidly due to the internal resistance of the test battery and its change, and then gradually increases, and is generally stabilized in about 15 minutes. . Therefore, the voltage of the test battery when measuring the EIS after resting at each voltage is different from the resting voltage.

  The discharge rate needs to be fixed in data collection, but is not limited to 0.5C discharge. The voltage of the test battery to be stopped needs to be fixed in data collection, but may be a value other than 4.0V, 3.8V, 3.6V, 3.4V, and 3.2V. There is no need to limit it to one. Also, the pause from the time when the voltage of the test battery is reached until the EIS measurement is sufficient as long as the voltage of the test battery is generally stable, and is not limited to 15 minutes. Furthermore, it may be a profile in which measurement is performed after the test battery reaches a certain voltage by charging instead of discharging.

The measurement frequency range of the EIS is 0.1 Hz to 1,000 Hz, and the measurement is performed at a frequency that is 10 equal parts in a logarithmic value with 10 as the base between each digit. Specifically, 1.00 × 10 ⁿ Hz, 1.26 × 10 ⁿ Hz, 1.59 × 10 ⁿ Hz, 2.00 × 10 ⁿ Hz, 2.52 × 10 ⁿ Hz, 3.17 × 10 ⁿ Hz, 3.99 × 10 ⁿ Hz, 5.02 × 10 ⁿ Hz, 6.32 × 10 ⁿ Hz, 7.96 × 10 ⁿ Hz (n = −1, 0, 1, 2, 3). .

After the EIS measurement, the cycle-deteriorated test batteries resume their respective cycle degradations, and the fully-charged degraded test batteries are fully charged and then resume their respective full-charge degradations. The above-described deterioration and measurement were repeated, and data collection was completed when the current capacity reached approximately 70% of the initial value before the deterioration (current capacity maintenance ratio 70%).
However, the test battery with 60 ° C. cycle deterioration failed before it reached 70%, so data collection was terminated when the failure occurred. Further, since the deterioration at 25 ° C. is slow, the data collection is until the point when the rate reaches approximately 80%. Furthermore, the measurement interval of storage deterioration at 25 ° C. is set to an interval of about one month to two months from the middle.

  In addition, C, such as 0.5C discharge of 1C charge, is a relative current value with respect to the current capacity of the battery. 1C is defined as a current value at which the battery can be discharged for one hour from a fully charged state to a fully discharged state by constant current discharge. Therefore, 0.5C is half of 1C in the case of the same battery, and is a current value that takes 2 hours to discharge. Since the rated current capacity of the A battery is 2.5 Ah, 1C is 2.5 A and 0.5 C is 1.25 A.

As the deterioration progresses, the current capacity decreases, but the value of 1C is determined with respect to the rated current capacity (initial state before deterioration) and is not changed.
1C was determined based on the rated current capacity, but for each test battery, the charge current capacity, discharge current capacity, charge power capacity, and discharge power capacity were fully charged under standard conditions in the initial state before deterioration. It is measured using the charge / discharge profile of capacity measurement. The current capacity maintenance rate, which is the rate of current capacity reduction, is calculated based on this initial value. The complex impedance at each frequency is also measured for each test battery using an EIS profile in the initial state before deterioration. In the embodiments of the present invention, unless otherwise specified, the discharge capacity is used as the capacity, the discharge current capacity is used as the current capacity, and the discharge current capacity maintenance ratio is used as the capacity maintenance ratio.

FIG. 4 is a flowchart for explaining data collection of the secondary battery for data collection and measurement, and is a detailed explanation of S1 of FIG.
First, the initial capacity of a new data collection / measurement secondary battery before deterioration is measured in a standard environment (step S21). Next, the data collection and measurement secondary battery is deteriorated under a plurality of predetermined conditions (deterioration factors) (step S22). Next, the secondary battery for data collection and measurement is periodically transferred to a standard environment, and the capacity and the complex impedance of each frequency are measured by EIS measurement using a predetermined EIS profile (step S23). Next, the degree of deterioration (capacity maintenance rate = (capacity / initial capacity) × 100) is calculated from the measured capacity and initial capacity (step S24). Next, the capacity, the degree of deterioration, and the measured complex impedance of each frequency are stored (step S25). Next, it is determined whether or not the degree of deterioration is equal to or greater than a predetermined threshold (step S26). Next, if it is equal to or greater than the threshold value, the process returns to step S22, and if it is equal to or less than the threshold value, the data collection is completed (step S27).

<Create two-dimensional feature data>
The data of the following frequency was extracted after 4.0V rest with respect to the data of the capacity maintenance ratio of 93% to 70% among the values of the complex impedance collected by the A battery.
0.1 Hz, 4 Hz (3.99 Hz), 10 Hz, 50.2 Hz, 100 Hz, 1,000 Hz
Regarding the above data, "the real part value and the real part value of different frequencies", "the imaginary part value and the imaginary part value of different frequencies" of the complex impedance, which is a characteristic amount for separating the degradation factors to be separated , In order to select a set of “value of real part and imaginary part of different frequencies” or “value of real part and imaginary part of same frequency”, X axis is imaginary part of impedance and Y axis is impedance 15 types of graphs (all different combinations of the imaginary part and the real part of each frequency), 15 types of graphs with the X axis as the real part of impedance and the Y axis as the real part of impedance (with the real part of each frequency) 15 types of graphs (all combinations of imaginary part and imaginary part of each frequency) with X axis as imaginary part of impedance and Y axis as imaginary part of impedance To determine the feature quantity.

However, since the imaginary part is treated with a value obtained by multiplying the value of the imaginary part of the measured complex impedance by minus 1, conventionally, the value of the imaginary part shown in a graph or the like is opposite to the actual value.
Here, the range of the number of actual measurement data samples and the capacity maintenance rate for each deterioration factor, which is data of the capacity maintenance rate of 93% to 70%, is as follows, which is a total of 133 samples.
25 ° C. cycle: 19 samples (capacity maintenance ratio: 91.9% to 71.1%)
45 ° C. cycle: 25 samples (capacity retention: 90.4% to 71.0%)
60 ° C. cycle: 18 samples (capacity maintenance rate: 92.9% to 73.2%)
Storage at 25 ° C .: 32 samples (capacity maintenance rate: 93.0% to 81.1%)
Storage at 45 ° C .: 28 samples (capacity maintenance rate: 92.1% to 69.1%)
Storage at 60 ° C .: 11 samples (capacity maintenance rate: 92.8% to 70.2%)

  FIG. 5 is a diagram showing the relationship between the 100 Hz complex impedance imaginary part (Ω) and the 0.1 Hz complex impedance real part (Ω) of all degradation factors, and FIG. 6 shows the 45 ° C. cycle and the 60 ° C. cycle. It is a figure which shows the relationship between the complex impedance imaginary part ((ohm)) of 0.1 Hz, and the complex impedance imaginary part ((omega | ohm)) of 4 Hz.

  As a result of the observation, from the imaginary part of complex impedance of 100 Hz and the real part of complex impedance of 0.1 Hz, “a set of 45 ° C. cycle and 60 ° C. cycle” and other “25 ° C. cycle, 25 ° C. storage, 45 ° C. storage and 60 ° C. storage” It is suggested that the “set” can be distinguished (separated) (broken line in FIG. 5). In other words, the imaginary part of complex impedance of 100 Hz and the real part of complex impedance of 0.1 Hz are “groups of 45 ° C. cycle and 60 ° C. cycle” and other “groups of 25 ° C. cycle, 25 ° C. storage, 45 ° C. storage and 60 ° C. storage”. It can be seen that this is a feature amount for separating "."

  Taking out only the “45 ° C. cycle and 60 ° C. cycle set” after separation and seeing the relationship between the complex impedance imaginary part of 0.1 Hz and the complex impedance imaginary part of 4 Hz, “45 ° C. cycle” and “60 ° C. cycle” Can be separated (broken line in FIG. 6). That is, it can be seen that the complex impedance imaginary part of 0.1 Hz and the complex impedance imaginary part of 4 Hz are feature quantities for separating the “45 ° C. cycle” from the “60 ° C. cycle”.

FIG. 7 shows the complex impedance imaginary part (Ω) of 100 Hz and the complex impedance real part of 1,000 Hz (Ω) of 25 ° C. cycle except 45 ° C. cycle and 60 ° C. cycle, 25 ° C. storage, 45 ° C. storage and 60 ° C. storage. FIG.
Only the "25 ° C cycle, 25 ° C storage, 45 ° C storage and 60 ° C storage set" after the separation of "45 ° C cycle and 60 ° C cycle" is taken out, and 100Hz complex impedance imaginary part and 1,000Hz complex impedance Looking at the relationship of the real part, it is suggested that “25 ° C. storage” can be separated from other “sets of 25 ° C. cycle, 45 ° C. storage and 60 ° C. storage” (broken line in FIG. 7). In other words, the complex impedance imaginary part of 100 Hz and the complex impedance real part of 1,000 Hz are characteristic quantities for separating “25 ° C. storage” and other “sets of 25 ° C. cycle, 45 ° C. storage and 60 ° C. storage”. I understand that.

FIG. 8 is a diagram showing the relationship between a complex impedance imaginary part (Ω) of 4 Hz and a complex impedance real part (Ω) of 1,000 Hz of 25 ° C. cycle excluding storage at 25 ° C., 45 ° C. storage and 60 ° C. storage. is there.
Taking out only the “25 ° C. cycle, 45 ° C. storage and 60 ° C. storage pair” after separation at “25 ° C.” separation, and looking at the relationship between the complex impedance imaginary part of 4 Hz and the complex impedance real part of 1,000 Hz, “60 It is suggested that “° C. storage” can be separated from other “sets of 25 ° C. cycle and 45 ° C. storage” (dashed line in FIG. 8). That is, it can be seen that the complex impedance imaginary part of 4 Hz and the complex impedance real part of 1,000 Hz are characteristic quantities for separating “60 ° C. storage” from other “groups of 25 ° C. cycle and 45 ° C. storage”.

FIG. 9 is a diagram showing the relationship between the 25-degree C cycle excluding storage at 60 ° C., the complex impedance imaginary part (Ω) of 0.1 Hz and the real impedance part (Ω) of 0.1 Hz stored at 45 ° C. .
Taking out only the “set of 25 ° C. cycle and 45 ° C. storage” after separation at “60 ° C. storage” and looking at the relationship between the imaginary part of 0.1 Hz complex impedance and the real part of 0.1 Hz complex impedance, “25 ° C. cycle” ”And“ 45 ° C. storage ”are suggested (broken line in FIG. 9). That is, it can be seen that the complex impedance imaginary part of 0.1 Hz and the complex impedance real part of 0.1 Hz are feature quantities for separating “25 ° C. cycle” and “45 ° C. storage”.

  FIG. 10 is a flowchart for explaining the contents of FIGS. 5 to 9. As shown in FIG. 10 with the two-dimensional feature amount, it can be seen that the six types of deterioration factors of the A battery can be separated in five steps from step S31 to step S35. That is, in step S31, “100 Hz complex impedance imaginary part and 0.1 Hz complex impedance real number” are used as feature quantities, “45 ° C. cycle and 60 ° C. cycle pair” and other “25 ° C. cycle, 25 ° C. storage, 45 ° C. Separate the storage and 60 ° C storage set.

In step S32, “45 ° C. cycle” and “60 ° C. cycle” are separated using “0.1 Hz complex impedance imaginary part and 4 Hz complex impedance imaginary part” as a feature quantity.
In step S33, “100 ° complex impedance imaginary part and 1,000 Hz complex impedance real part” are separated into “25 ° C. storage” and other “25 ° C. cycle, 45 ° C. storage and 60 ° C. storage” as feature quantities. .

  In step S 34, “60 ° C. storage” and other “sets of 25 ° C. cycle and 45 ° C. storage” are separated using “4 Hz complex impedance imaginary part and 1,000 Hz complex impedance real part” as feature quantities. In step S35, “25 ° C. cycle” and “45 ° C. storage” are separated using “0.1 Hz complex impedance imaginary part and 0.1 Hz complex impedance real part” as the feature quantity.

  As described above, “100 Hz complex impedance imaginary part and 0.1 Hz complex impedance real part”, “0.1 Hz complex impedance imaginary part and 4 Hz complex impedance imaginary part”, “100 Hz complex impedance imaginary part and 1” , 000 Hz complex impedance real part "," 4 Hz complex impedance imaginary part and 1,000 Hz complex impedance real part "," 0.1 Hz complex impedance imaginary part and 0.1 Hz complex impedance real part " It was created as two-dimensional feature data after 4.0V pause.

<Creation of degradation factor model for evaluation of 2D feature data and evaluation of estimation accuracy>
In order to evaluate the degradation factor estimation accuracy in the two-dimensional feature amount after the 4.0 A pause of the A battery, the learning data for creating the degradation factor model for evaluation and the degradation model for evaluation created by the learning data Evaluation data for evaluating the degradation factor estimation accuracy was created. Then, a deterioration factor model (discriminant function) for evaluation by SVM (Support Vector Machine; support vector machine) was created from the learning data. SVM (support vector machine) is one of identification methods using supervised learning, and can be applied to pattern recognition and regression analysis. This support vector machine is one of the learning models with the best recognition performance among many currently known methods. Degradation factors were estimated (separated) from the evaluation data using the created degradation factor model, and the degradation factor estimation accuracy was evaluated based on the degree of accuracy. Details will be described below for each step. The learning data is selected in consideration of the following three points.

1) The number of feature values of learning data is made substantially the same for the set of two degradation factors to be separated.
2) In the case of a set of deterioration factors including a plurality of deterioration factors, the number of feature amounts of each deterioration factor is set to be approximately proportional to the number of data of each deterioration factor.
3) Select so that the degree of deterioration (capacity maintenance rate) corresponding to the feature amount included in the learning data for each deterioration factor is not biased.

All data not selected as learning data were used as evaluation data. In addition, because we looked at the degradation factor separation accuracy for each step of each step, not the cumulative degradation factor separation accuracy, samples that were erroneously separated in the previous step are not considered in the next step, and are independent in each step. Evaluated. All subsequent degradation factor separation accuracy is evaluated by this policy.
Step S31: “A set of 45 ° C. cycle and 60 ° C. cycle” and “A set of 25 ° C. cycle, 25 ° C. storage, 45 ° C. storage and 60 ° C. storage” are separated, and the characteristic amount is “100 Hz complex impedance imaginary part and 0. The learning data, evaluation data, and evaluation deterioration factor model at “1 Hz complex impedance real number” were created as follows.
The characteristic amount data of “45 ° C. cycle” was 25 samples, and the characteristic amount data of “60 ° C. cycle” was 18 samples, for a total of 43 samples. On the other hand, the feature value data for “25 ° C. cycle” is 19 samples, the feature value for “25 ° C. storage” is 32 samples, the feature value data for “45 ° C. storage” is 28 samples, and the feature value data for “60 ° C. storage” is The data was 11 samples with a total of 90 samples.

  The data of the characteristic amount of the “45 ° C. cycle” arranged by the capacity retention rate values were selected in total from 13 samples having the larger capacity retention rate, and used as learning data. Similarly, the data of “60 ° C. cycle” feature values arranged in accordance with the capacity retention rate values were selected in total for 9 samples from every other capacity retention rate, and used as learning data. A total of 22 samples were used as learning data for “a set of 45 ° C. cycle and 60 ° C. cycle”, and the remaining 21 samples were used as evaluation data.

  The feature value data of the “25 ° C. cycle” arranged by the capacity retention rate values were extracted at equal intervals from the larger capacity retention rate, and a total of 5 samples were selected and used as learning data. Similarly, the data of “25 ° C. storage” feature values arranged by the capacity maintenance ratio values are extracted at equal intervals as much as possible from the larger capacity maintenance ratio, and a total of nine samples are selected. did. Similarly, a total of 8 samples of “45 ° C. storage” feature values were selected and used as learning data. Similarly, a total of three samples of “60 ° C. storage” feature values were selected and used as learning data. A total of 25 samples were used as learning data for a “25 ° C. cycle, 25 ° C. storage, 45 ° C. storage and 60 ° C. storage set”, and the remaining 65 samples were used as evaluation data.

  Deterioration factors for evaluation by SVM with 22 samples of learning data for "45 ° C cycle and 60 ° C cycle set" and 25 samples of learning data for "25 ° C cycle, 25 ° C storage, 45 ° C storage and 60 ° C storage set" A model (discriminant function) was created. Next, 21 samples for evaluation of “a set of 45 ° C. cycle and 60 ° C. cycle” and evaluation data of “a set of 25 ° C. cycle, 25 ° C. storage, 45 ° C. storage and 60 ° C. storage” 65 samples, a total of 86 samples Degradation factors were estimated using the degradation factor model created for them. As a result, one sample of the evaluation data was erroneously estimated, and the degradation factor estimation accuracy was 98.8%.

  Step S32: The “45 ° C. cycle” and the “60 ° C. cycle” are separated, and the feature data is learning data, evaluation data, and evaluation deterioration factor model for “0.1 Hz complex impedance imaginary part and 4 Hz complex impedance imaginary part”. Was made as follows. 13 samples of learning data of “45 ° C. cycle” selected in step S31 were used as learning data, and the remaining 12 samples were used as evaluation data. Similarly, 9 learning data samples of “60 ° C. cycle” selected in step S51 were used as learning data, and the remaining 9 samples were used as evaluation data.

  A deterioration factor model (discriminant function) for evaluation by SVM was created with 13 samples of learning data for “45 ° C. cycle” and 9 samples of learning data for “60 ° C. cycle”. Next, degradation factors were estimated using degradation factor models created for a total of 21 samples of 12 samples for evaluation of “45 ° C. cycle” and 9 samples of evaluation data for “60 ° C. cycle”. As a result, two samples of the evaluation data were erroneously estimated, and the deterioration factor estimation accuracy was 90.5%.

Step S33: “25 ° C. storage” and other “25 ° C. cycle, 45 ° C. storage and 60 ° C. storage set” are separated, and the feature quantity is “100 Hz complex impedance imaginary part and 1,000 Hz complex impedance real part” The learning data, evaluation data, and deterioration factor model for evaluation were prepared as follows.
“25 ° C. storage” features 32 samples, while “25 ° C. cycle” feature data is 19 samples, “45 ° C. storage” feature values data is 28 samples, “60 ° C. storage” features The quantity data is 11 samples with a total of 58 samples.

Select the data of the feature value of “Storage at 25 ° C.” arranged by the capacity maintenance rate value, and select 16 samples in total from the one with the larger capacity maintenance rate, and use this as learning data, and the remaining 16 samples for evaluation Data.
The feature value data of the “25 ° C. cycle” arranged by the capacity retention rate values were extracted at equal intervals from the larger capacity retention rate, and a total of 5 samples were selected and used as learning data. Similarly, the feature value data of “stored at 45 ° C.” arranged by the capacity maintenance rate values are extracted at equal intervals from the larger capacity maintenance rate, and a total of 8 samples are selected. did. Similarly, a total of three samples of “60 ° C. storage” feature values were selected and used as learning data. A total of 16 samples were used as learning data for “a set of 25 ° C. cycle, 45 ° C. storage and 60 ° C. storage”, and the remaining 42 samples were used as evaluation data.

  A degradation factor model (discriminant function) for evaluation by SVM was created with 16 learning data samples of “25 ° C. storage” and 16 learning data samples of “25 ° C. cycle 45 ° C. storage and 60 ° C. storage set”. Next, degradation was performed using degradation factor models created for a total of 58 samples of 16 samples for evaluation of “25 ° C. storage” and 42 samples of evaluation data for “25 ° C. cycle, 45 ° C. storage and 60 ° C. storage” set. The factors were estimated. As a result, 4 samples of the evaluation data were erroneously estimated, and the deterioration factor estimation accuracy was 93.1%.

Step S34: The “store at 60 ° C.” and the other “set of 25 ° C. cycle and 45 ° C. storage” are separated, and the feature value is the learning data in “the complex impedance imaginary part of 4 Hz and the complex impedance real part of 1,000 Hz” The evaluation data and the deterioration factor model for evaluation were created as follows.
The feature quantity of “storage at 60 ° C.” is 11 samples, while the data of feature quantity of “25 ° C. cycle” is 19 samples, and the data of feature quantity of “storage at 45 ° C.” is 28 samples, for a total of 47 samples.

Select all 6 samples of “60 ° C storage” feature values arranged by capacity retention rate, from the one with the largest capacity retention rate, and use this as learning data, and the remaining 5 samples for evaluation Data.
The feature value data of the “25 ° C. cycle” arranged by the capacity retention rate values were extracted at equal intervals from the larger capacity retention rate, and a total of 5 samples were selected and used as learning data. Similarly, the feature value data of “stored at 45 ° C.” arranged by the capacity maintenance rate values are extracted at equal intervals from the larger capacity maintenance rate, and a total of 8 samples are selected. did. A total of 13 samples were used as learning data for the “25 ° C. cycle and 45 ° C. storage set”, and the remaining 34 samples were used as evaluation data.

  A degradation factor model (discriminant function) for evaluation by SVM was created with 6 samples of learning data of “60 ° C. storage” and 16 samples of learning data of “a set of 25 ° C. cycle and 45 ° C. storage”. Next, degradation factors were estimated using degradation factor models created for a total of 40 samples of 6 samples for evaluation of “storage at 60 ° C.” and 34 samples of evaluation data for “25 ° C. cycle and 45 ° C. storage”. . As a result, no sample was erroneously estimated in the evaluation data, and the degradation factor estimation accuracy was 100%.

Step S35: “25 ° C. cycle” and “45 ° C. storage” are separated, and the feature value is learning data, evaluation data, and evaluation deterioration in “0.1 Hz complex impedance imaginary part and 0.1 Hz complex impedance real part”. The factor model was created as follows.
The learning data 5 samples of the “25 ° C. cycle” selected in step S34 were used as learning data, and the remaining 14 samples were used as evaluation data. Similarly, 8 samples of learning data “stored at 45 ° C.” selected in step S34 were used as learning data, and the remaining 20 samples were used as evaluation data.

  A deterioration factor model (discriminant function) for evaluation by SVM was created with 5 samples of learning data for “25 ° C. cycle” and 8 samples of learning data for “45 ° C. storage”. Next, deterioration factors were estimated using deterioration factor models created for a total of 34 samples of 14 samples for evaluation of “25 ° C. cycle” and 20 samples of evaluation data for “45 ° C. storage”. As a result, no sample was erroneously estimated in the evaluation data, and the degradation factor estimation accuracy was 100%.

<Creation of 8-dimensional feature data, creation of deterioration factor model for evaluation of 8-dimensional feature data, and evaluation of estimation accuracy>
The two-dimensional feature amount determined by the A battery is “100 Hz complex impedance imaginary part and 0.1 Hz complex impedance real part” “0.1 Hz complex impedance imaginary part and 4 Hz complex impedance imaginary part” “100 Hz complex impedance imaginary part” Part and 1,000 Hz complex impedance real part "" 4 Hz complex impedance imaginary part and 1,000 Hz complex impedance real part "" 0.1 Hz complex impedance imaginary part and 0.1 Hz complex impedance real part " .

  That is, 0.1 Hz complex impedance, 4 Hz complex impedance, 100 Hz complex impedance, 1,000 Hz complex impedance, ie, 0.1 Hz, 4 Hz, 100 Hz, and 1,000 Hz complex impedance, and EIS of each deterioration factor. It shows that the characteristics can be separated. Therefore, it can be seen that 8-dimensional data including the imaginary part and the real part of the complex impedance of these four frequencies can be used as a feature quantity for the degradation factor separation (estimation).

FIG. 11 is a diagram illustrating a flowchart for examining the degradation factor estimation accuracy of the 8-dimensional feature data. A degradation model for evaluation by SVM is created, and the degradation factor estimation accuracy is examined in the steps shown in FIG.
Step S71: “A set of 45 ° C. cycle and 60 ° C. cycle” and “A set of 25 ° C. cycle, 25 ° C. storage, 45 ° C. storage and 60 ° C. storage” are separated, and feature amounts are “0.1 Hz, 4 Hz, 100 Hz, 1 ”8-dimensional data consisting of complex impedance of 4 frequencies of 1,000 Hz”.

The creation method (selection method) of the learning data and the evaluation data was performed in the same manner as in step 31 for the two-dimensional feature amount.
Deterioration factors for evaluation by SVM with 22 samples of learning data for "45 ° C cycle and 60 ° C cycle set" and 25 samples of learning data for "25 ° C cycle, 25 ° C storage, 45 ° C storage and 60 ° C storage set" A model (discriminant function) was created. Next, 21 samples for evaluation of “a set of 45 ° C. cycle and 60 ° C. cycle” and evaluation data of “a set of 25 ° C. cycle, 25 ° C. storage, 45 ° C. storage and 60 ° C. storage” 65 samples, a total of 86 samples Degradation factors were estimated using the degradation factor model created for them. As a result, one sample of the evaluation data was erroneously estimated, and the degradation factor estimation accuracy was 98.8%.

Step S72: “45 ° C. cycle” and “60 ° C. cycle” are separated, and the feature quantity is “8-dimensional data consisting of complex impedances of four frequencies of 0.1 Hz, 4 Hz, 100 Hz, and 1,000 Hz”.
The creation method (selection method) of the learning data and the evaluation data was performed in the same manner as in step 32 for the two-dimensional feature amount.

  A deterioration factor model (discriminant function) for evaluation by SVM was created with 13 samples of learning data for “45 ° C. cycle” and 9 samples of learning data for “60 ° C. cycle”. Next, degradation factors were estimated using degradation factor models created for a total of 21 samples of 12 samples for evaluation of “45 ° C. cycle” and 9 samples of evaluation data for “60 ° C. cycle”. As a result, no sample was erroneously estimated in the evaluation data, and the degradation factor estimation accuracy was 100%.

Step S73: “25 ° C. storage” and other “25 ° C. cycle, 45 ° C. storage and 60 ° C. storage set” are separated, and feature amounts are “4 Hz complex impedance of 0.1 Hz, 4 Hz, 100 Hz, and 1,000 Hz” It is “8-dimensional data consisting of”.
The creation method (selection method) of the learning data and the evaluation data was performed in the same manner as in step 33 for the two-dimensional feature amount.

  A degradation factor model (discriminant function) for evaluation by SVM was created with 16 learning data samples of “25 ° C. storage” and 16 learning data samples of “25 ° C. cycle 45 ° C. storage and 60 ° C. storage set”. Next, degradation was performed using degradation factor models created for a total of 58 samples of 16 samples for evaluation of “25 ° C. storage” and 42 samples of evaluation data for “25 ° C. cycle, 45 ° C. storage and 60 ° C. storage” set. The factors were estimated. As a result, 5 samples of the evaluation data were erroneously estimated, and the deterioration factor estimation accuracy was 91.8%.

Step S74: “60 ° C. storage” and other “25 ° C. cycle and 45 ° C. storage set” are separated, and the feature quantity is “8 dimensions consisting of complex impedances of four frequencies of 0.1 Hz, 4 Hz, 100 Hz, and 1,000 Hz”. Data ".
The creation method (selection method) of the learning data and the evaluation data was performed in the same manner as in step S34 for the two-dimensional feature amount.

  A degradation factor model (discriminant function) for evaluation by SVM was created with 6 samples of learning data of “60 ° C. storage” and 16 samples of learning data of “a set of 25 ° C. cycle and 45 ° C. storage”. Next, degradation factors were estimated using degradation factor models created for a total of 40 samples of 6 samples for evaluation of “storage at 60 ° C.” and 34 samples of evaluation data for “25 ° C. cycle and 45 ° C. storage”. . As a result, no sample was erroneously estimated in the evaluation data, and the degradation factor estimation accuracy was 100%.

Step S75: “25 ° C. cycle” and “45 ° C. storage” are separated, and the feature quantity is “8-dimensional data consisting of complex impedances of four frequencies of 0.1 Hz, 4 Hz, 100 Hz, and 1,000 Hz”.
The creation method (how to select) of the learning data and the evaluation data was performed in the same manner as in step S35 with the two-dimensional feature amount.

  A deterioration factor model (discriminant function) for evaluation by SVM was created with 5 samples of learning data for “25 ° C. cycle” and 8 samples of learning data for “45 ° C. storage”. Next, deterioration factors were estimated using deterioration factor models created for a total of 34 samples of 14 samples for evaluation of “25 ° C. cycle” and 20 samples of evaluation data for “45 ° C. storage”. As a result, no sample was erroneously estimated in the evaluation data, and the degradation factor estimation accuracy was 100%.

From the above results, it was confirmed that (n × 2) -dimensional data composed of complex impedances of n types of frequencies including all the feature quantities selected from the two-dimensional feature quantities is a feature quantity for estimating the deterioration factor. .
Since SVM can only create a model (evaluation function) that divides a sample into two, when estimating (separating) which of m types of deterioration factors, it is necessary to prepare m-1 steps. Appropriate selection of the estimation (separation) order is necessary to improve the estimation accuracy. It is appropriate to determine the order from the two-dimensional feature value studied when selecting the complex impedance of n types of frequencies.

  In addition, since SVM can only create a model (evaluation function) that divides a sample into two, when estimating (separating) which of m types of deterioration factors, it is necessary to prepare m-1 steps. At the same time, the deterioration factor can be estimated in one step as long as the method realizes a model that estimates which of m types of deterioration factors.

<8-dimensional feature value after A battery 3.2V rest>
Among the values of the complex impedance collected by the A battery, data having the same frequency as that after 4.0 V pause as shown below was extracted from data having a capacity maintenance ratio of 93% to 70% after 3.2 V pause.
0.1 Hz, 4 Hz (3.99 Hz), 10 Hz, 50.2 Hz, 100 Hz, 1,000 Hz
Even after 3.2 V pause, as with 4.0 V pause, regarding the above data, 36 types of graphs with the X axis as the imaginary part of impedance and the Y axis as the real part of impedance (of the imaginary part and real part of each frequency) 15 different graphs (all different combinations of real part and real part of each frequency), X axis is imaginary part of impedance, Y axis 15 types (all combinations of different imaginary part and imaginary part of each frequency) were observed.

First, it is a two-dimensional feature whether the EIS characteristic of each deterioration factor can be separated by the complex impedance of four frequencies of 0.1 Hz, 4 Hz, 100 Hz, and 1,000 Hz of the eight-dimensional feature amount that is the feature amount after 4.0V pause. When the amount was examined, it was found that separation was possible as follows.
First, it was found that six types of deterioration factors of the A battery can be separated by two-dimensional feature amounts in five steps from step S91 to step S95 shown below.

FIG. 12 is a diagram showing the relationship between the complex impedance imaginary part (Ω) of 4 Hz and the complex impedance real part (Ω) of 100 Hz, which are all deterioration factors.
In step S91, as shown in FIG. 12, “a complex impedance imaginary part of 4 Hz and a complex impedance real number of 100 Hz” are used as feature quantities, “a pair of 60 ° C. cycle and 60 ° C. storage” and other “25 ° C. cycle, 45 ° C. cycle, It is suggested that separation of “25 ° C. storage and 45 ° C. storage” is possible.

FIG. 13 is a diagram showing the relationship between a 100 Hz complex impedance imaginary part (Ω) and a 0.1 Hz complex impedance real part (Ω) stored at 60 ° C. and 60 ° C. FIG.
In step S92, as shown in FIG. 13, “60 ° C. cycle” and “60 ° C. storage” can be separated using “100 Hz complex impedance imaginary part and 0.1 Hz complex impedance real part” as feature quantities. Is done.

FIG. 14 shows a complex impedance imaginary part (Ω) of 4 Hz and a complex impedance real part of 1,000 Hz (Ω) of 25 ° C. cycle, 45 ° C. cycle, 25 ° C. storage and 45 ° C. storage excluding 60 ° C. and 60 ° C. storage. FIG.
In step S93, as shown in FIG. 14, “4 Hz complex impedance imaginary part and 1,000 Hz complex impedance real part” are stored as “25 ° C.” and other “25 ° C. cycle, 45 ° C. cycle and 45 ° C. storage”. It is suggested that separation of the “set” is possible.

FIG. 15 is a diagram showing a relationship between a complex impedance imaginary part (Ω) of 4 Hz and a complex impedance imaginary part (Ω) of 100 Hz of 25 ° C. cycle, 45 ° C. cycle, and 45 ° C. storage excluding 25 ° C. storage. .
In step S94, as shown in FIG. 15, “45 ° C. cycle” and other “25 ° C. cycle and 45 ° C. storage group” are separated using “4 Hz complex impedance imaginary part and 100 Hz complex impedance imaginary part” as feature quantities. It is suggested that it is possible.

FIG. 16 is a diagram showing the relationship between the 25 ° C. cycle excluding the 45 ° C. cycle, the complex impedance imaginary part (Ω) of 0.1 Hz stored at 45 ° C., and the complex impedance imaginary part (Ω) of 4 Hz.
In step S95, as shown in FIG. 16, it is suggested that “25 ° C. cycle” and “45 ° C. storage” can be separated using “0.1 Hz complex impedance imaginary part and 4 Hz complex impedance imaginary part” as a feature quantity. Is done.

  The two-dimensional feature quantities confirmed above are “4 Hz complex impedance imaginary part and 100 Hz complex impedance real part”, “100 Hz complex impedance imaginary part and 0.1 Hz complex impedance real part”, “4 Hz complex impedance imaginary part and “1,000 Hz complex impedance real part” “4 Hz complex impedance imaginary part and 100 Hz complex impedance imaginary part” “0.1 Hz complex impedance imaginary part and 4 Hz complex impedance imaginary part”, after 4.0 V pause It can be seen that the same eight-dimensional data consisting of the imaginary part and the real part of the complex impedance of four frequencies of 0.1 Hz, 4 Hz, 100 Hz, and 1,000 Hz can be used as the characteristic amount for the degradation factor separation (estimation).

FIG. 17 is a diagram illustrating a flowchart for checking the degradation factor estimation accuracy. The result of examining the degradation factor estimation accuracy of the feature quantity by creating a degradation model for evaluation by SVM and examining it in the steps shown in FIG. 17 is shown below.
Step S91: The characteristic amount data of “60 ° C. cycle” was 18 samples, and the characteristic amount of “60 ° C. storage” was 11 samples, for a total of 29 samples. On the other hand, the feature value data of “25 ° C. cycle” is 19 samples, the feature value data of “45 ° C. cycle” is 25 samples, the feature value of “25 ° C. storage” is 32 samples, and the feature value of “45 ° C. storage” is The data was 28 samples, for a total of 104 samples.

  The data of the feature amount of the “60 ° C. cycle” arranged by the capacity retention rate value was selected in total from 9 samples every other one from the larger capacity retention rate, and used as learning data. Similarly, the data of “60 ° C. storage” feature values arranged in accordance with the capacity retention rate values were selected in total for every other sample from the larger capacity retention rates, and used as learning data. A total of 14 samples were used as learning data for a “set of 60 ° C. cycle and 60 ° C. storage”, and the remaining 15 samples were used as evaluation data.

  The feature value data of the “25 ° C. cycle” arranged by the capacity retention rate values were extracted at equal intervals from the larger capacity retention rate, and a total of 5 samples were selected and used as learning data. Similarly, the feature value data of the “45 ° C. cycle” arranged with the capacity maintenance rate values are extracted at equal intervals from the one with the larger capacity maintenance rate, and a total of 5 samples are selected. did. Similarly, a total of 5 samples of “25 ° C. storage” feature values were selected and used as learning data. Similarly, a total of 6 samples of “45 ° C. storage” feature values were selected and used as learning data. A total of 21 samples were used as learning data for “a set of 25 ° C. cycle, 25 ° C. storage, 45 ° C. storage and 60 ° C. storage”, and the remaining 83 samples were used as evaluation data.

  Deterioration factors for evaluation by SVM with 14 samples of learning data for “60 ° C cycle and 60 ° C storage” and 21 samples of learning data for “25 ° C cycle, 45 ° C cycle, 25 ° C storage and 45 ° C storage” A model (discriminant function) was created. Next, 15 samples for evaluation of “a set of 45 ° C. cycle and 60 ° C. cycle” and 83 samples for evaluation of “a set of 25 ° C. cycle, 25 ° C. storage, 45 ° C. storage and 60 ° C. storage” total 98 samples. Degradation factors were estimated using the degradation factor model created for them. As a result, no sample was erroneously estimated in the evaluation data, and the degradation factor estimation accuracy was 100%.

Step S92: Nine samples of learning data for the “60 ° C. cycle” selected in step S91 were used as learning data, and the remaining nine samples were used as evaluation data. Similarly, 5 samples of learning data “stored at 60 ° C.” selected in step S91 were used as learning data, and the remaining 6 samples were used as evaluation data.
A degradation factor model (discriminant function) for evaluation by SVM was created with 9 samples for learning of “60 ° C. cycle” and 6 samples of learning data for “60 ° C. storage”. Next, degradation factors were estimated using degradation factor models created for a total of 15 samples of 9 samples for evaluation of “60 ° C. cycle” and 6 samples of evaluation data for “60 ° C. storage”. As a result, no sample was erroneously estimated in the evaluation data, and the degradation factor estimation accuracy was 100%.

Step S93: The feature quantity of “25 ° C. storage” is 32 samples, while the feature quantity data of “25 ° C. cycle” is 19 samples, and the feature quantity data of “45 ° C. cycle” is 25 samples, “45 ° C. storage” The feature amount data is 28 samples.
Select the data of the feature value of “Storage at 25 ° C.” arranged by the capacity maintenance rate value, and select 16 samples in total from the one with the larger capacity maintenance rate, and use this as learning data, and the remaining 16 samples for evaluation Data.

  The same five samples as in Step 1 were selected from the feature value data of the “25 ° C. cycle” and used as learning data. Similarly, the same 5 samples as in Step 1 were selected from the feature value data of “45 ° C. cycle” and used as learning data. Similarly, the same 6 samples as in Step 1 were selected from the feature value data of “45 ° C. storage” and used as learning data. A total of 16 samples were used as learning data for “a set of 25 ° C. cycle, 45 ° C. cycle and 45 ° C. storage”, and the remaining 56 samples were used as evaluation data.

  A degradation factor model (discriminant function) for evaluation by SVM was created with 16 learning data samples of “25 ° C. storage” and 16 learning data samples of “25 ° C. cycle 45 ° C. storage and 60 ° C. storage set”. Next, degradation was performed using the degradation factor model created for a total of 72 samples of 16 samples for evaluation of “25 ° C. storage” and 56 samples of evaluation data for “25 ° C. cycle, 45 ° C. storage and 60 ° C. storage” set. The factors were estimated. As a result, no sample was erroneously estimated in the evaluation data, and the degradation factor estimation accuracy was 100%.

Step S94: “45 ° C. cycle” feature value is 25 samples, while “25 ° C. cycle” feature value data is 19 samples, “45 ° C. storage” feature value data is 28 samples, 47 samples in total. is there.
Select all 13 samples of “45 ° C cycle” feature values arranged by capacity retention rate, from the one with the largest capacity retention rate, and use it as learning data. The remaining 12 samples are for evaluation. Data.

  The feature value data of the “25 ° C. cycle” arranged by the capacity retention rate values were extracted at equal intervals from the larger capacity retention rate, and a total of 5 samples were selected and used as learning data. Similarly, the feature value data of “stored at 45 ° C.” arranged by the capacity maintenance rate values are extracted at equal intervals from the larger capacity maintenance rate, and a total of 8 samples are selected. did. A total of 13 samples were used as learning data for the “25 ° C. cycle and 45 ° C. storage set”, and the remaining 34 samples were used as evaluation data.

  A degradation factor model (discriminant function) for evaluation by SVM was created using 13 samples for learning of “45 ° C. cycle” and 13 samples of learning for “set of 25 ° C. cycle and 45 ° C. storage”. Next, degradation factors were estimated using degradation factor models created for a total of 46 samples of 12 samples for evaluation of “45 ° C. cycle” and 34 samples of evaluation data for “25 ° C. cycle and 45 ° C. storage set”. . As a result, no sample was erroneously estimated in the evaluation data, and the degradation factor estimation accuracy was 100%.

Step S95: 5 samples of learning data of “25 ° C. cycle” selected in step S94 were used as learning data, and the remaining 14 samples were used as evaluation data. Similarly, 8 samples of learning data “stored at 45 ° C.” selected in step S94 were used as learning data, and the remaining 20 samples were used as evaluation data.
A deterioration factor model (discriminant function) for evaluation by SVM was created with 5 samples of learning data for “25 ° C. cycle” and 8 samples of learning data for “45 ° C. storage”. Next, deterioration factors were estimated using deterioration factor models created for a total of 34 samples of 14 samples for evaluation of “25 ° C. cycle” and 20 samples of evaluation data for “45 ° C. storage”. As a result, no sample was erroneously estimated in the evaluation data, and the degradation factor estimation accuracy was 100%.

<A battery 3.2V after 6V rest>
The extracted frequencies of 0.1 Hz, 4 Hz (3.99 Hz), 10 Hz, and 50.2 Hz with respect to the value of the complex impedance collected by the A battery after 3.2 V pause with the capacity maintenance rate of 93% to 70%. When observing 100 Hz and 1,000 Hz data, not only four types of 0.1 Hz, 4 Hz, 100 Hz, and 1,000 Hz, which are the same as after 4.0 V pause, but also three types of data of 4 Hz, 10 Hz, and 100 Hz. Even so, it was found that the six types of deterioration factors of the A battery can be separated.

First, by observing a two-dimensional feature amount, A battery is used by using three types of 4 Hz (3.99 Hz), 10 Hz, and 100 Hz, that is, a six-dimensional feature amount in five steps from step S101 to step S105 shown below. It was found that 6 types of deterioration factors can be separated.
FIG. 18 is a diagram illustrating the relationship between the complex impedance imaginary part (Ω) of 4 Hz and the complex impedance imaginary part (Ω) of 10 Hz, which are all deterioration factors.

That is, in step S101, as shown in FIG. 18, “4 Hz complex impedance imaginary part and 10 Hz complex impedance imaginary part” is used as a feature amount, “60 ° C. cycle and 60 ° C. storage pair” and other “25 ° C. cycle, 45 ° Separate the "Circulation cycle, 25 ° C storage and 45 ° C storage set".
FIG. 19 is a diagram illustrating a relationship between a complex impedance imaginary part (Ω) of 4 Hz and a complex impedance imaginary part (Ω) of 100 Hz stored at 60 ° C. and stored at 60 ° C. FIG.

In step S102, as shown in FIG. 19, “60 ° C. cycle” and “60 ° C. storage” are separated using “4 Hz complex impedance imaginary part and 100 Hz complex impedance imaginary part” as a feature quantity.
FIG. 20 shows a complex impedance imaginary part (Ω) of 4 Hz and a complex impedance imaginary part (Ω) of 10 Hz of 25 ° C. cycle, 45 ° C. cycle, 25 ° C. storage and 45 ° C. storage excluding 60 ° C. and 60 ° C. storage. It is a figure which shows the relationship.

In step S103, as shown in FIG. 20, “4 Hz complex impedance imaginary part and 10 Hz complex impedance imaginary part” are used as feature quantities, and “25 ° C. cycle” and other “45 ° C. cycle, 25 ° C. storage and 45 ° C. storage” set. Is separated.
FIG. 21 is a diagram showing the relationship between the complex impedance imaginary part (Ω) of 4 Hz and the complex impedance real part (Ω) of 100 Hz of 45 ° C. cycle excluding the 25 ° C. cycle, 25 ° C. storage and 45 ° C. storage. .

In step S104, as shown in FIG. 21, “4 Hz complex impedance imaginary part and 100 Hz complex impedance real part” are separated into “25 ° C. storage” and other “45 ° C. cycle and 45 ° C. storage” as feature quantities. .
FIG. 22 is a diagram showing a relationship between a 45 ° C. cycle excluding storage at 25 ° C., a 10 Hz complex impedance imaginary part (Ω) and a 100 Hz complex impedance imaginary part (Ω) after 45 ° C. storage.

In step S105, as shown in FIG. 22, “45 ° C. cycle” and “45 ° C. storage” are separated using “10 Hz complex impedance imaginary part and 100 Hz complex impedance imaginary part” as the feature quantity.
The two-dimensional feature amount confirmed above is “4 Hz complex impedance imaginary part and 10 Hz complex impedance imaginary part” “4 Hz complex impedance imaginary part and 100 Hz complex impedance imaginary part” “4 Hz complex impedance imaginary part and 10 Hz. Complex impedance imaginary part ”,“ 4 Hz complex impedance imaginary part and 100 Hz complex impedance real part ”,“ 10 Hz complex impedance imaginary part and 100 Hz complex impedance imaginary part ”, 3 Hz complex impedance of 4 Hz, 10 Hz, and 100 Hz It can be seen that the 6-dimensional data composed of the imaginary part and the real part of can be used as a feature quantity for separation factor estimation (estimation).

FIG. 23 is a diagram illustrating a flowchart for checking the degradation factor estimation accuracy. The result of examining the degradation factor estimation accuracy of the feature quantity by creating a degradation model for evaluation by SVM and examining it in the steps shown in FIG. 23 is shown below.
Step S101: The learning data and evaluation data creation method (how to select) was performed in the same manner as in Step 1 with 8-dimensional feature values.

  Deterioration factors for evaluation by SVM with 14 samples of learning data for “60 ° C cycle and 60 ° C storage” and 21 samples of learning data for “25 ° C cycle, 45 ° C cycle, 25 ° C storage and 45 ° C storage” A model (discriminant function) was created. Next, 15 samples for evaluation of “a set of 45 ° C. cycle and 60 ° C. cycle” and 83 samples for evaluation of “a set of 25 ° C. cycle, 25 ° C. storage, 45 ° C. storage and 60 ° C. storage” total 98 samples. Degradation factors were estimated using the degradation factor model created for them. As a result, no sample was erroneously estimated in the evaluation data, and the degradation factor estimation accuracy was 100%.

Step S102: The method for creating (selecting) the learning data and the evaluation data was performed in the same manner as in Step 2 with the 8-dimensional feature amount.
A degradation factor model (discriminant function) for evaluation by SVM was created with 9 samples for learning of “60 ° C. cycle” and 6 samples of learning data for “60 ° C. storage”. Next, degradation factors were estimated using degradation factor models created for a total of 15 samples of 9 samples for evaluation of “60 ° C. cycle” and 6 samples of evaluation data for “60 ° C. storage”. As a result, no sample was erroneously estimated in the evaluation data, and the degradation factor estimation accuracy was 100%.

Step S103: The characteristic amount of “25 ° C. cycle” is 19 samples, while the characteristic amount data of “45 ° C. cycle” is 25 samples, the characteristic amount data of “25 ° C. storage” is 32 samples, and “45 ° C. storage” The feature amount data is 28 samples.
Select 9 samples of “25 ° C cycle” feature values arranged by capacity maintenance rate every other one from the largest capacity maintenance rate, and use it as learning data. The remaining 10 samples are used for evaluation. Data.

  The same 5 samples as in Step 1 were selected from the feature value data of “stored at 25 ° C.” and used as learning data. Similarly, the same five samples as those in step S101 were selected from the feature value data of “45 ° C. cycle” and used as learning data. Similarly, the same 6 samples as in step S101 were selected from the feature value data of “stored at 45 ° C.” and used as learning data. A total of 16 samples were used as learning data for “a set of 45 ° C. cycle, 25 ° C. storage and 45 ° C. storage”, and the remaining 69 samples were used as evaluation data.

  A deterioration factor model (discriminant function) for evaluation by SVM was created with 9 samples of learning data for “25 ° C. cycle” and 16 samples of learning data for “25 ° C. cycle 45 ° C. storage and 60 ° C. storage set”. Next, degradation was performed using the degradation factor model created for a total of 79 samples of 10 samples for evaluation of “storage at 25 ° C.” and 69 samples of evaluation data for “25 ° C. cycle, 45 ° C. storage and 60 ° C. storage”. The factors were estimated. As a result, no sample was erroneously estimated in the evaluation data, and the degradation factor estimation accuracy was 100%.

Step S104: “25 ° C. storage” feature values are 32 samples, while “45 ° C. cycle” feature value data is 25 samples, and “45 ° C. storage” feature values data is 28 samples, for a total of 47 samples. is there.
Select the data of the feature value of “Storage at 25 ° C.” arranged by the capacity maintenance rate value, and select 16 samples in total from the one with the larger capacity maintenance rate, and use this as learning data, and the remaining 16 samples for evaluation Data.

  The characteristic amount data of the “45 ° C. cycle” arranged by the capacity retention rate values were extracted at equal intervals as much as possible from the larger capacity retention rate, and a total of 8 samples were selected and used as learning data. Similarly, the feature value data of “stored at 45 ° C.” arranged by the capacity maintenance rate values are extracted at equal intervals from the larger capacity maintenance rate, and a total of 8 samples are selected. did. A total of 16 samples were used as learning data for “45 ° C. cycle and 45 ° C. storage set”, and the remaining 37 samples were used as evaluation data.

  A degradation factor model (discriminant function) for evaluation by SVM was created with 16 learning data samples of “25 ° C. storage” and 16 learning data samples of “45 ° C. cycle and 45 ° C. storage set”. Next, degradation factors were estimated using degradation factor models created for a total of 53 samples of 16 samples for evaluation of “25 ° C. storage” and 37 samples of evaluation data for “45 ° C. cycle and 45 ° C. storage set”. . As a result, no sample was erroneously estimated in the evaluation data, and the degradation factor estimation accuracy was 100%.

Step S105: The data of the feature value of “45 ° C. cycle” is 25 samples. On the other hand, the feature amount data of “45 ° C. storage” is 28 samples.
Select all 12 samples of the feature value of the “45 ° C cycle” arranged by capacity retention rate, from the one with the largest capacity retention rate, and use it as learning data. The remaining 13 samples are for evaluation. Data.

Select 14 samples in total from the one with the larger capacity retention rate, and select 14 samples for the feature value data stored at 45 ° C, arranged by capacity retention rate, and use the remaining 14 samples for evaluation. Data.
A deterioration factor model (discriminant function) for evaluation by SVM was created with 12 samples of learning data for “45 ° C. cycle” and 14 samples of learning data for “45 ° C. storage”. Next, degradation factors were estimated using degradation factor models created for a total of 27 samples of 13 samples for evaluation of “45 ° C. cycle” and 14 samples of evaluation data for “45 ° C. storage”. As a result, no sample was erroneously estimated in the evaluation data, and the degradation factor estimation accuracy was 100%.

<Complex impedance at one frequency>
When creating the two-dimensional feature value data after A battery 4.0V rest, the imaginary part and real part of different frequencies, or the imaginary part and imaginary part of different frequencies were used. However, it is also possible to separate degradation factors by creating two-dimensional feature data using the real part and imaginary part of the same frequency. Specifically, let us look at complex impedance data of 0.1 Hz, 4 Hz, and 100 Hz after A battery 4.0 V pause.

  FIG. 24 is a diagram showing the relationship between the complex impedance imaginary part (Ω) of 0.1 Hz and the real part of complex impedance (Ω) of 0.1 Hz, and FIG. 25 excludes the 25 ° C. cycle. FIG. 26 shows the relationship between the 100 Hz complex impedance imaginary part (Ω) and the 100 Hz complex impedance real part (Ω) of 45 ° C. cycle, 60 ° C. cycle, 25 ° C. storage, 45 ° C. storage and 60 ° C. storage. FIG. 27 is a diagram showing the relationship between the complex impedance imaginary part (Ω) of 4 Hz stored at 45 ° C. and 60 ° C. and the real part (Ω) of 4 Hz complex impedance. FIG. 27 further excludes 45 ° C. storage and 60 ° C. storage. FIG. 28 is a diagram showing the relationship between the 4 Hz complex impedance imaginary part (Ω) and the 4 Hz complex impedance real part (Ω) stored at 45 ° C., 60 ° C. and 25 ° C. Furthermore, it is a figure which shows the relationship between the 45-degree C cycle except a 25 degreeC preservation | save, the complex impedance imaginary part ((omega | ohm)) of 0.1 Hz of a 45-degree cycle, and the complex impedance real part ((ohm)) of 0.1 Hz.

  As shown in FIG. 24, “25 ° C. cycle” and “45 ° C. cycle, 60 ° C. cycle, 25 ° C. storage, 45 ° C. storage and 60 ° C. storage set” can be separated from 1 Hz complex impedance data. After separation of “25 ° C. cycle”, as shown in FIG. 25, “45 ° C. cycle, 60 ° C. cycle, 25 ° C. storage, 45 ° C. storage and 60 ° C. set” is “45 ° C. cycle, 60 ° C. according to 100 Hz complex impedance”. The “cycle and 25 ° C. storage set” and other “45 ° C. storage and 60 ° C. storage set” can be separated. As shown in FIG. 26, “45 ° C. storage” and “60 ° C. storage” can be separated by using the complex impedance of 4 Hz of “a set of 45 ° C. storage and 60 ° C. storage”. Similarly, as shown in FIG. 27, by using the complex impedance of 4 Hz of “45 ° C. cycle, 60 ° C. cycle and 25 ° C. storage”, “25 ° C. storage” and “45 ° C. cycle and 60 ° C. cycle set” can be obtained. Can be separated. Next, as shown in FIG. 28, the “set of 45 ° C. cycle and 60 ° C. cycle” can be separated into “45 ° C. cycle” and “60 ° C. cycle” by using the complex impedance of 0.1 Hz again.

<Capacity maintenance rate estimation by A battery complex impedance measurement>
FIG. 29 is a diagram (part 1) illustrating a relationship between a real part (Ω) of a complex impedance of 4 Hz and a capacity maintenance ratio (%). In the A battery, the real part of the complex impedance of 4 Hz included in the eight-dimensional feature amount after the 4.0 V pause is changed to the data of the real part of the complex impedance of 4 Hz in the data of the capacity maintenance ratio 93% to 70%. FIG. 29 shows a graph in which the corresponding capacity retention rate is taken on the Y axis. The number of samples of actual measurement data and the range of the capacity maintenance ratio for each deterioration factor of the capacity maintenance ratio from 97% to 80% are as follows, and a total of 133 samples.

25 ° C. cycle: 19 samples (capacity maintenance ratio: 91.9% to 71.1%)
45 ° C. cycle: 25 samples (capacity retention: 90.4% to 71.0%)
60 ° C. cycle: 18 samples (capacity maintenance rate: 92.9% to 73.2%)
Storage at 25 ° C .: 32 samples (capacity maintenance rate: 93.0% to 81.1%)
Storage at 45 ° C .: 28 samples (capacity maintenance rate: 92.1% to 69.1%)
Storage at 60 ° C .: 11 samples (capacity maintenance rate: 92.8% to 70.2%)
With these data, the estimation accuracy in the case of estimating the capacity maintenance rate from the real value of 4 Hz complex impedance after 4.0 V discharge suspension is verified.

FIG. 29 shows a graph in which all data are plotted regardless of the deterioration factor and approximated by a cubic curve that minimizes the square error. By using this cubic approximate curve, the capacity maintenance rate of the battery can be estimated with a certain accuracy from the measurement of the 4 Hz impedance after the 4.0 V discharge is stopped.
The capacity retention ratio obtained by substituting the actual capacity value of the sample for which the accuracy (capacity retention ratio estimation accuracy of the approximate curve) was actually measured and the real value of 4 Hz complex impedance after the 4.0 V discharge suspension of the sample into the approximate curve. Is assumed to be twice the standard deviation of ΔC (2σΔC), that is, assuming that ΔC is normally distributed, an estimated value of 95% is obtained by assuming a range of ± percent from the estimated value. Evaluate whether it falls between the errors. The closer 2σΔC is to 0%, the higher the estimation accuracy.

The results of the evaluation are as follows.
All data: 133 samples (2σΔC = 5.83%)
25 ° C. cycle: 19 samples (2σΔC = 7.98%)
45 ° C cycle: 25 samples (2σΔC = 4.02%)
60 ° C. cycle: 18 samples (2σΔC = 3.06%)
Storage at 25 ° C .: 32 samples (2σΔC = 2.93%)
Storage at 45 ° C .: 28 samples (2σΔC = 3.92%)
Storage at 60 ° C .: 11 samples (2σΔC = 4.04%)
The accuracy evaluation value for each degradation factor is the result of calculating 2σΔC for the approximate curve using only the data for each degradation factor.

  FIG. 30 is a diagram (part 2) illustrating a relationship between a real part (Ω) of a complex impedance of 4 Hz and a capacity maintenance ratio (%). FIG. 30 is a graph that is the same graph as FIG. 29, and shows a classification and display of the deterioration factor of each point. As is clear from FIG. 30, when the deterioration factors are different, the relationship between the real part of the 4 Hz complex impedance and the change in the capacity maintenance rate is different. Therefore, the capacity maintenance rate estimation estimated from the measurement of the 4 Hz complex impedance real part is obtained by obtaining a cubic curve that minimizes the square error of the relationship between the real part of the 4 Hz complex impedance and the capacity maintenance rate separately for each deterioration factor. Increases accuracy. The accuracy is specifically shown below.

FIG. 31 is a diagram (part 3) illustrating a relationship between a real part (Ω) of a complex impedance of 4 Hz and a capacity retention rate (%).
FIG. 31 shows a graph obtained by plotting only the sample of the 25 ° C. cycle and approximating it with a cubic curve that minimizes the square error. At this time, 2σΔC is 5.95%. This accuracy is a little worse than the estimated accuracy of 5.83% in all data, and the accuracy is higher than the estimated accuracy of 7.98% in the 25 ° C. cycle in the approximate curve using all data.

FIG. 32 is a diagram (No. 4) illustrating a relationship between a real part (Ω) of a complex impedance of 4 Hz and a capacity retention rate (%).
FIG. 32 shows a graph obtained by plotting only a sample at a 45 ° C. cycle and approximating with a cubic curve that minimizes the square error. At this time, 2σΔC is 3.30%. This accuracy is higher than the estimated accuracy of 5.83% for all data, and the approximated curve using all data is higher than the estimated accuracy of 4.02% for 45 ° C. cycle.

FIG. 33 is a diagram (No. 5) showing a relationship between a real part (Ω) of a complex impedance of 4 Hz and a capacity retention rate (%).
FIG. 33 shows a graph obtained by plotting only the sample of the 60 ° C. cycle and approximating it with a cubic curve that minimizes the square error. At this time, 2σΔC is 2.59%. This accuracy is higher than the estimated accuracy of 5.83% for all data, and the estimated accuracy of 3.06% for the 60 ° C. cycle in the approximate curve using all data.

FIG. 34 is a diagram (part 6) illustrating a relationship between a real part (Ω) of a complex impedance of 4 Hz and a capacity retention rate (%).
FIG. 34 shows a graph in which only a sample stored at 25 ° C. is plotted and approximated by a cubic curve that minimizes the square error. At this time, 2σΔC is 1.88%. This accuracy is higher than the estimated accuracy of 5.83% for all data, and the estimated accuracy of 2.93% stored at 25 ° C. for the approximate curve using all data.

FIG. 35 is a diagram (No. 7) illustrating a relationship between a real part (Ω) of a complex impedance of 4 Hz and a capacity retention rate (%).
FIG. 35 shows a graph in which only samples stored at 45 ° C. are plotted and approximated by a cubic curve that minimizes the square error. At this time, 2σΔC is 2.98%. This accuracy is higher than the estimated accuracy of 5.83% for all data, and the estimated accuracy of 3.92% stored at 45 ° C. for the approximate curve using all data.

FIG. 36 is a diagram (No. 8) illustrating the relationship between the real part (Ω) of the complex impedance of 4 Hz and the capacity retention rate (%).
FIG. 36 shows a graph in which only a sample stored at 60 ° C. is plotted and approximated by a cubic curve that minimizes the square error. At this time, 2σΔC is 1.46%. This accuracy is higher than the estimated accuracy of 5.83% for all data, and the estimated curve stored at 60 ° C. is 4.04% for the approximate curve using all data.

From the above results, it can be seen that the estimation accuracy of the capacity maintenance rate is improved by estimating and identifying the deterioration factor.
The complex impedance is not limited to the real part, and the same applies even if an imaginary value is used.
FIG. 37 is a diagram (part 1) illustrating a relationship between an imaginary part (Ω) of a complex impedance of 4 Hz and a capacity retention rate (%). In the A battery, the imaginary part of the complex impedance of 4 Hz included in the 8-dimensional feature amount after the 4.0 V pause is changed to the data of the imaginary part of the complex impedance of 4 Hz, with the capacity maintenance rate of 93% to 70%. FIG. 37 shows a graph in which the corresponding capacity retention rate is taken on the Y axis. 4Hz complex impedance after 4.0V discharge rest in the actual measurement data for each degradation factor of 97% to 80% capacity maintenance rate same as the case of 4Hz complex impedance real value after 4.0V discharge rest, in total 133 samples The estimation accuracy when estimating the capacity maintenance rate from the imaginary value is verified.

FIG. 37 shows a graph obtained by plotting all the data regardless of the deterioration factor and approximating with a cubic curve that minimizes the square error. By using this cubic approximate curve, the capacity maintenance rate of the battery can be estimated with a certain accuracy from the measurement of the 4 Hz impedance after the 4.0 V discharge is stopped.
The capacity maintenance ratio obtained by substituting the capacity maintenance ratio of the sample for which this accuracy (capacity maintenance ratio estimation precision of the approximate curve) was actually measured and the imaginary value of the 4 Hz complex impedance after the 4.0 V discharge suspension of the sample into the approximate curve Is assumed to be twice the standard deviation of ΔC (2σΔC), that is, assuming that ΔC is normally distributed, an estimated value of 95% is obtained by assuming a range of ± percent from the estimated value. Evaluate whether it falls between the errors. The closer 2σΔC is to 0%, the higher the estimation accuracy.

The results of the evaluation are as follows.
All data: 133 samples (2σΔC = 8.42%)
25 ° C. cycle: 19 samples (2σΔC = 8.89%)
45 ° C cycle: 25 samples (2σΔC = 6.05%)
60 ° C. cycle: 18 samples (2σΔC = 4.69%)
Storage at 25 ° C: 32 samples (2σΔC = 3.58%)
Storage at 45 ° C .: 28 samples (2σΔC = 7.40%)
Storage at 60 ° C .: 11 samples (2σΔC = 9.73%)
The accuracy evaluation value for each degradation factor is the result of calculating 2σΔC for the approximate curve using only the data for each degradation factor.

  FIG. 38 is a diagram (part 2) illustrating a relationship between an imaginary part (Ω) of a complex impedance of 4 Hz and a capacity retention rate (%). FIG. 38 is a graph that is the same as FIG. 37 and shows a classification display indicating which deterioration factor causes each point to deteriorate. As is clear from FIG. 38, when the deterioration factors are different, the relationship between the 4 Hz complex impedance imaginary part and the change in the capacity retention rate is different. Therefore, the capacity maintenance rate estimation estimated from the measurement of the 4 Hz complex impedance imaginary part is obtained by obtaining a cubic curve that minimizes the square error of the relationship between the 4 Hz complex impedance imaginary part and the capacity maintenance rate separately for each deterioration factor. Increases accuracy. The accuracy is specifically shown below.

Only a sample at 25 ° C. is plotted, and 2σΔC is 3.09% in an approximate curve with a cubic curve that minimizes the square error. This accuracy is higher than the estimated accuracy of 8.42% in all data, and the estimated accuracy of 8.25% in the 25 ° C cycle in the approximate curve using all data.
In the approximate curve of the cubic curve in which only the sample of the 45 ° C. cycle is plotted and the square error is the minimum, 2σΔC is 3.17%. This accuracy is higher than the estimated accuracy of 8.42% in all data, and the estimated accuracy of 45 ° C cycle in the approximate curve using all data is higher than 6.05%.

In the approximated curve of the cubic curve in which only the sample of the 60 ° C. cycle is plotted and the square error is minimized, 2σΔC is 2.69%. This accuracy is higher than the estimation accuracy of 8.42% in all data, and the estimation accuracy of 4.69% in the 60 ° C. cycle in the approximate curve using all data.
Only a sample stored at 25 ° C. is plotted, and 2σΔC is 2.48% in an approximate curve with a cubic curve that minimizes the square error. This accuracy is higher than the estimated accuracy of 8.42% for all data, and the estimated accuracy of 3.58% stored at 25 ° C. for the approximate curve using all data.

Only a sample stored at 45 ° C. is plotted, and 2σΔC is 5.60% in an approximate curve with a cubic curve that minimizes the square error. This accuracy is higher than the estimated accuracy of 8.42% for all data, and the estimated accuracy of 7.40% stored at 45 ° C. for the approximate curve using all data.
Only a sample stored at 60 ° C. is plotted, and 2σΔC is 2.29% in an approximate curve with a cubic curve that minimizes the square error. This accuracy is higher than the estimated accuracy of 8.42% for all data, and the estimated accuracy of 9.74% stored at 60 ° C. for the approximate curve using all data.
From the above results, it can be seen that the estimation accuracy of the capacity maintenance rate is improved by estimating and identifying the deterioration factor.

<Actual application>
From these results, the capacity of the secondary battery to be measured deteriorated separately from the data collection and measurement secondary battery, using the SVN as the deterioration factor model, using the 8-dimensional feature amount measured after the 4.0 V pause measured with the A battery of this time. The case of estimating the maintenance rate will be described.

  In order to see how accurate the degradation factor estimation is for the feature amount, an evaluation degradation factor model is created from the learning data divided into the learning data and the evaluation data, and the estimation accuracy is checked with the evaluation data. If the estimation accuracy satisfies a predetermined accuracy (for example, 90% or more), this is used as a feature amount for deterioration estimation. If not satisfied, a feature amount that can obtain a predetermined estimation accuracy is searched by a method such as examining combinations of complex impedances of other frequencies or increasing the dimension of the feature amount.

  This time, “8-dimensional data consisting of complex impedances of four frequencies of 0.1 Hz, 4 Hz, 100 Hz, and 1,000 Hz” had an estimation accuracy of over 90%, so this was adopted as a feature value for degradation factor estimation. The evaluation deterioration factor model used sometimes is adopted as a deterioration factor model (discriminant function). Since SVM can only create a model (evaluation function) that divides a sample into two, in order to estimate (separate) which of m types of deterioration factors, it is necessary to prepare m-1 steps. One deterioration factor model needs to be created.

The m−1 deterioration factor models created and adopted in this way are the information about the deterioration factor of the data collection measurement secondary battery of the reference data holding unit 2 in FIGS. 1 and 2, and the data collection measurement secondary battery. This is a specific example of reference data associated with information related to complex impedance.
The characteristic amount (complex impedance of four frequencies of 0.1 Hz, 4 Hz, 100 Hz, and 1,000 Hz) of the secondary battery to be measured is measured. In FIG. 1 and FIG. 2, these measurements are performed by a three impedance calculator.

  First, the measured characteristic amount of the secondary battery to be measured is stored in the degradation factor model in step S111 (not shown), whether the degradation factor is “a set of 45 ° C. cycle and 60 ° C. cycle”, “25 ° C. cycle, 25 ° C. , 45 ° C. storage and 60 ° C. storage ”. Next, when the estimation result is “a set of 45 ° C. cycle and 60 ° C. cycle”, whether the deterioration factor is “45 ° C. cycle” in the deterioration factor estimation model in step S112 is “60 ° C. cycle”. Estimate. Next, the estimation result is estimated to be either “45 ° C. cycle” or “60 ° C. cycle”, and the deterioration factor estimation ends.

  When the deterioration factor is estimated by the deterioration factor model created in step S111, and it is estimated that “a set of 25 ° C. cycle, 25 ° C. storage, 45 ° C. storage and 60 ° C. storage” is used, the feature amount is determined as the deterioration factor model in step S113. Thus, it is estimated whether the deterioration factor is “25 ° C. storage” or “25 ° C. cycle, 45 ° C. storage and 60 ° C. storage”. If the estimation result is “25 ° C. storage”, the deterioration factor estimation ends.

  When the degradation factor is estimated by the degradation factor model created in step S113, and it is estimated that “a set of 25 ° C. cycle, 45 ° C. storage and 60 ° C. storage”, the characteristic amount is represented by the degradation factor model in step S114. It is estimated whether it is “60 ° C. storage” or “25 ° C. cycle and 45 ° C. set”. If the estimation result is “60 ° C. storage”, the degradation factor estimation ends.

If the result of the estimation in step S114 is “a set of 25 ° C. cycle and 45 ° C. storage”, whether the deterioration factor is “25 ° C. cycle” or “45 ° C. storage” in the deterioration factor model of step 5 Is estimated. As a result of the estimation, it is estimated to be either “25 ° C. cycle” or “45 ° C. storage”, and the deterioration factor estimation ends.
In FIG. 1 and FIG. 2, these deterioration factor estimations are performed by the deterioration factor estimation unit 4.

In the first embodiment, diagnosis determination is performed based on the estimation result of the degradation factor estimation unit 4, and the diagnosis result is converted into, for example, a text or the like that can be understood by humans, or converted into an electrical signal to another device, for example. Output from the diagnosis determination unit 5 as an output.
In the second embodiment, the reference data holding unit 2 in FIG. 2 includes information on the deterioration factor of the data collection measurement secondary battery, information on the complex impedance of the data collection measurement secondary battery, and data collection measurement secondary battery. The reference data associated with the information on the capacity of the battery is stored. For example, as the reference data associated with the information on the capacity of the secondary battery for data collection and measurement, the 4 Hz complex impedance for each deterioration factor is stored. It holds an approximate curve of the real part value and capacity retention rate.

  The capacity estimation unit 6 in FIG. 2 calculates the capacity of the secondary battery 1 to be measured based on the reference data, the output information from the deterioration factor estimation unit 4, and the complex impedance output information calculated by the impedance calculation unit 3. presume. Specifically, if the degradation factor of the secondary battery to be measured is estimated, the value of the real part of the 4 Hz complex impedance is substituted into the approximate curve of the capacity maintenance rate approximated from the estimated degradation factor data. Estimate the capacity maintenance rate of the secondary battery to be measured.

For example, if the degradation factor of the secondary battery to be measured is estimated to be “45 ° C. cycle”, the capacity maintenance ratio is calculated by substituting the value of the real part of the 4 Hz complex impedance into the cubic approximation curve shown in FIG. Can be obtained.
In the second embodiment, diagnosis determination is performed based on the estimation result of the capacity estimation unit 6, and the diagnosis result is converted into, for example, a text or the like that can be understood by humans, or converted into an electrical signal to another device, for example. Output from the diagnosis determination unit 5 as an output. The output includes an estimated capacity maintenance rate, a judgment result of whether or not to use the secondary battery to be measured, which is judged based on a combination of the deterioration factor and the capacity maintenance rate, or a replacement timing.

<B battery four-dimensional feature>
Different from the A battery, a ternary positive electrode material (Li (NiCoMn) O ₂ ), a high power low current capacity (1.6 Ah) 18650 cylindrical battery for a commercially available power tool using a carbon-based material for the negative electrode material (hereinafter referred to as “a” battery) , “B battery”). The standard environment was room temperature (25 ° C.), and the conditions (deterioration factors) for promoting the deterioration were the following three types.
1) Cycle degradation under 25 ° C environment (25 ° C cycle degradation)
2) Cycle degradation under 50 ° C environment (50 ° C cycle degradation)
3) Fully charged storage deterioration in a 50 ° C environment (50 ° C storage deterioration)

  Cycle deterioration is a deterioration condition that repeats charging and discharging under a predetermined temperature environment. Charging is CC (constant current) charging at 1C, and if the battery voltage reaches 4.2V, switch to CV (constant voltage charging) to control the current so that the battery voltage is maintained at 4.2V. CC-CV charging was terminated when the current reached 50 mA (hereinafter, CC-CV charging is referred to as “1 CCC-4.2 V 50 mA aborted CV charging”).

The discharge was performed at 0.5 C, and the discharge was terminated when the battery voltage reached 3.0 V (hereinafter, CC discharge was referred to as “0.5 C 3.0 V aborted CC discharge”).
The charge / discharge profile for cycle deterioration repeats “1CCC-4.2V50mA censored CV charge” and “0.5C3.0V censored CC discharge” with a predetermined period of rest (a state in which neither charging nor discharging is performed). It was.

  Specifically, first, a pause of 3 hours (waiting for the test battery to reach the same temperature as the predetermined temperature environment) is performed, and then “1CCC-4.2V50 mA censored CV charge” is performed for 10 minutes. This is a charge / discharge profile that repeats “charge-pause-discharge-pause” 25 times of “0.5 C 3.0 V censored CC discharge” after 10 pauses. Here, the test battery in the present invention is a general term for a data collection measurement secondary battery and a measurement secondary battery.

The cycle-degraded test battery is transferred to the standard environment every time the cycle deterioration charge / discharge profile under a predetermined temperature environment is performed once, and the EIS based on the predetermined capacity measurement charge / discharge profile and the predetermined EIS profile. The complex impedance at each frequency was measured.
The charge / discharge profile for the capacity measurement of cycle deterioration is a pause of 3 hours (waiting for the test battery to reach the same temperature as the standard environment), and then “1CCC-4.2V50mA censored CV charge”. After 10 minutes of rest, “0.5C3.0V censored CC discharge” is performed, after 10 minutes of rest, “1CCC-4.2V50mA censored CV charge” is performed, and 10 minutes of rest, “3 hours rest− It is a charge / discharge profile which performs "charge-pause-discharge-pause-charge-pause" once.

The fully charged storage deterioration is deterioration in which a test battery that is fully charged under a standard environment is moved to a predetermined temperature environment and left for a certain number of days.
The fully charged state of the B battery is a state immediately after charging by “1CCC-4.2V50 mA censored CV charging” in a standard environment. Similarly, the complete discharge state is a state immediately after the discharge by the “0.5 C3.0 V censored CC discharge” in the standard environment. Since the rated current capacity of the B battery is 1.6 Ah, 1C is 1.6 A, and 0.5 C is 0.8 A.

  The EIS profile is the same as the A battery, in which a fully charged test battery is discharged by 0.5C, and when the voltage of the test battery reaches a predetermined voltage, EIS measurement is performed after a pause of 15 minutes. After that, discharge 0.5C again, and repeat the EIS measurement after a pause of 15 minutes when the voltage of the test battery reaches the next predetermined voltage, and the test battery voltage reaches 3.0V If so, the discharge is stopped, and after a 15-minute pause, the process ends.

  In the B battery, the voltage of the test battery that stopped discharging in order to perform EIS was set to five of 4.10V, 3.80V, 3.60V, 3.45V, and 3.20V. In addition, in order to specify the EIS measurement conditions, the EIS measurement after a 15-minute pause after reaching the voltage of each test battery is referred to as “4.10 V EIS after pause”, “3.80 V EIS after pause”, “3 .EIS after 60 V pause, “3.45 V EIS after pause” and “3.2 V0 EIS after pause”. The data of the following frequencies after 4.10V pause and 3.20V pause were extracted for the data of the capacity maintenance ratio of 93% to 70% among the values of complex impedance collected by B battery.

0.1 Hz, 1 Hz, 4 Hz (3.99 Hz), 10 Hz, 50.2 Hz, 100 Hz, 502 Hz, 1,000 Hz
4. After the 10V and 3.20V pauses, 64 types of graphs with the X axis as the imaginary part of the impedance and the Y axis as the real part of the impedance for the above data (different combinations of the imaginary part and the real part of each frequency) All), 28 types of graphs with X axis as real part of impedance and Y axis as real part of impedance (all different combinations of real part and real part of each frequency), X axis as imaginary part of impedance, Y axis as impedance 28 types of imaginary parts (all combinations of different imaginary parts and imaginary parts of each frequency) were observed.

Here, the range of the number of actual measurement data samples and the capacity maintenance rate for each deterioration factor, which is the data of the capacity maintenance rate of 93% to 70%, is as follows, and is 37 samples in total.
25 ° C. cycle: 8 samples (capacity retention: 93.5% to 84.2%)
50 ° C. cycle: 14 samples (capacity maintenance ratio: 93.8% to 70.5%)
Storage at 50 ° C .: 15 samples (capacity maintenance rate: 93.0% to 81.1%)

  FIG. 39 is a diagram showing the relationship between the complex impedance imaginary part (Ω) of 0.1 Hz and the complex impedance imaginary part (Ω) of 100 Hz after 4.10V rest of all degradation factors, and FIG. It is a figure which shows the relationship between the complex impedance imaginary part ((omega | ohm)) of 4Hz after a 3.20V rest of a degradation factor, and the complex impedance imaginary part ((omega | ohm)) of 50.2Hz.

  As a result of observation, after the 4.10V pause, as shown in FIG. 39, “0.1 Hz complex impedance imaginary part and 100 Hz complex impedance imaginary part”, and after 3.20V pause, as shown in FIG. It can be seen that the degradation factors can be separated by the complex impedance imaginary part and the complex impedance imaginary part of 50.2 Hz, and therefore, after each 4.10V pause, the imaginary part and real number of the complex impedance of 0.1 Hz and 100 Hz, respectively. After the 3.20V pause with 4 dimensions consisting of parts, the data of the imaginary part of the complex impedance of 2 Hz of 4 Hz and 50.2 Hz, and the data of the 4 dimensions consisting of the real part are feature quantities of degradation factor separation (estimation) I found out that The degradation factor estimation accuracy with these feature amounts was confirmed using SVM in the same manner as the A battery.

The learning data for the B battery was selected in consideration of the following as with the A battery.
1) The number of feature values of learning data is made substantially the same for the set of two degradation factors to be separated.
2) In the case of a set of deterioration factors including a plurality of deterioration factors, the number of feature amounts of each deterioration factor is made approximately proportional to the number of data of each deterioration factor.
3) Select so that the degree of deterioration (capacity maintenance rate) corresponding to the feature amount included in the learning data for each deterioration factor is not biased.

It is the same as in the case of the A battery that all data not selected as learning data is used as evaluation data.
FIG. 41 is a diagram (part 1) illustrating a flowchart for explaining separation of deterioration factors.
In the case of 4.10V rest, the degradation factors were separated by the steps as shown in FIG.

Step 121: “25 ° C. cycle” and other “50 ° C. cycle and 50 ° C. storage set” are separated, and the feature quantity is “four-dimensional data consisting of complex impedances of two frequencies of 0.1 Hz and 100 Hz”.
A degradation factor model (discriminant function) for evaluation by SVM is created with 4 samples of learning data of “25 ° C. cycle” and 9 samples of learning data of “set of 50 ° C. cycle and 50 ° C. storage”. Degradation factors were estimated using degradation factor models created for a total of 24 samples of 4 samples of evaluation data and 20 samples of evaluation data of “50 ° C. cycle and 50 ° C. storage set”. As a result, no sample was erroneously estimated in the evaluation data, and the degradation factor estimation accuracy was 100%.

Step 122: “50 ° C. cycle” and “50 ° C. storage” are separated, and the feature quantity is “four-dimensional data consisting of complex impedances of two frequencies of 0.1 Hz and 100 Hz”.
A deterioration factor model (discriminant function) for evaluation by SVM is created with 6 samples of learning data of “50 ° C. cycle” and 7 samples of learning data of “50 ° C. storage”, and 8 samples of evaluation data of “50 ° C. cycle” Degradation factors were estimated using degradation factor models created for a total of 16 samples of 8 samples for evaluation of “50 ° C. storage”. As a result, no sample was erroneously estimated in the evaluation data, and the degradation factor estimation accuracy was 100%.

FIG. 42 is a diagram (part 2) illustrating the flowchart for explaining the separation of the deterioration factors.
3. After 20V rest, degradation factors were separated in steps as shown in FIG.
Step 131: “25 ° C. cycle” and other “sets of 50 ° C. cycle and 50 ° C. storage” are separated, and the feature quantity is “four-dimensional data consisting of complex impedances of two frequencies of 4 Hz and 50.2 Hz”.

  A degradation factor model (discriminant function) for evaluation by SVM is created with 4 samples of learning data of “25 ° C. cycle” and 9 samples of learning data of “set of 50 ° C. cycle and 50 ° C. storage”. Degradation factors were estimated using degradation factor models created for a total of 24 samples of 4 samples of evaluation data and 20 samples of evaluation data of “50 ° C. cycle and 50 ° C. storage set”. As a result, no sample was erroneously estimated in the evaluation data, and the degradation factor estimation accuracy was 100%.

Step 132: “50 ° C. cycle” and “50 ° C. storage” are separated, and the feature quantity is “four-dimensional data consisting of complex impedances of two frequencies of 4 Hz and 50.2 Hz”.
A deterioration factor model (discriminant function) for evaluation by SVM is created with 6 samples of learning data of “50 ° C. cycle” and 7 samples of learning data of “50 ° C. storage”, and 8 samples of evaluation data of “50 ° C. cycle” Degradation factors were estimated using degradation factor models created for a total of 16 samples of 8 samples for evaluation of “50 ° C. storage”. As a result, no sample was erroneously estimated in the evaluation data, and the degradation factor estimation accuracy was 100%.

<8-dimensional feature amount of B battery>
Data of the same frequency as 93% to 70% of data for the capacity maintenance rate of 97% to 70% among the complex impedance values collected after 4.10V and 3.20V of B battery Extracted.
0.1 Hz, 1 Hz, 4 Hz (3.99 Hz), 10 Hz, 50.2 Hz, 100 Hz, 502 Hz, 1,000 Hz
4. After the 10V and 3.20V pauses, 64 types of graphs with the X axis as the imaginary part of the impedance and the Y axis as the real part of the impedance for the above data (different combinations of the imaginary part and the real part of each frequency) All), 28 types of graphs with X axis as real part of impedance and Y axis as real part of impedance (all different combinations of real part and real part of each frequency), X axis as imaginary part of impedance, Y axis as impedance 28 types of imaginary parts (all combinations of different imaginary parts and imaginary parts of each frequency) were observed.

Here, the range of the actual measurement data sample number and the capacity maintenance rate for each deterioration factor, which is the data of the capacity maintenance rate of 97% to 70%, is as follows, which is a total of 46 samples.
25 ° C. cycle: 10 samples (capacity retention: 97.2% to 84.2%)
50 ° C. cycle: 17 samples (capacity maintenance rate: 97.1% to 70.5%)
Storage at 50 ° C .: 19 samples (capacity maintenance rate: 97.2% to 81.1%)

  FIG. 43 is a diagram showing the relationship between the complex impedance imaginary part (Ω) of 1 Hz and the real part of complex impedance (Ω) of 1,000 Hz after 4.10V rest of all degradation factors, and FIG. It is a figure which shows the relationship between the complex impedance imaginary part ((ohm)) of 0.1 Hz complex impedance of 0.1 Hz after a 50 degreeC cycle except a cycle, and 4.10V rest of 50 degreeC preservation | save.

  As a result of the observation, after the rest of 4.10V, from “43 Hz complex impedance imaginary part and 1,000 Hz complex impedance real part”, “25 ° C. cycle” and other “50 ° C. cycle and 50 ° C. storage” 44. From FIG. 44, it was found that “50 ° C. cycle” and “50 ° C. storage” can be separated by “complex impedance imaginary part of 0.1 Hz and complex impedance imaginary part of 100 Hz” from FIG.

  FIG. 45 is a diagram illustrating the relationship between the complex impedance imaginary part (Ω) of 1 Hz and the complex impedance real part (Ω) of 1,000 Hz after 3.20 V rest of all deterioration factors, and FIG. It is a figure which shows the relationship between the complex impedance imaginary part ((ohm)) of 4Hz and the complex impedance imaginary part ((omega | ohm)) of 50.2Hz after 3.20V rest of 50 degreeC cycle except 50 cycles, and 50 degreeC preservation | save.

  3. After the 20V pause, from FIG. 45, “1 Hz complex impedance imaginary part and 1,000 Hz complex impedance real part”, “25 ° C. cycle” and other “50 ° C. cycle and 50 ° C. storage pair” 46 that “50 ° C. cycle” and “50 ° C. storage” can be separated by “4 Hz complex impedance imaginary part and 50.2 Hz complex impedance imaginary part”.

  Therefore, after a pause of 4.10 V, 8-dimensional data consisting of the imaginary part and real part of the complex impedance of four frequencies of 0.1 Hz, 1 Hz, 100 Hz, and 1,000 Hz becomes the characteristic amount for the degradation factor separation (estimation). I understood it. In addition, after the 3.20V pause, 8 dimensional data consisting of the imaginary part and the real part of the complex impedance of 4 frequencies of 1 Hz, 4 Hz, 50.2 Hz, and 1,000 Hz is the characteristic amount of degradation factor separation (estimation). I found out that

FIG. 47 is a diagram (No. 3) illustrating the flowchart for explaining the separation of the deterioration factors.
In the case of after 4.10V rest, the degradation factors were separated by the steps as shown in FIG.
Step 141: “25 ° C. cycle” and other “50 ° C. cycle and 50 ° C. storage set” are separated, and feature quantity is 8 dimensions consisting of complex impedances of four frequencies of 0.1 Hz, 1 Hz, 100 Hz, and 1,000 Hz. Data ".

  A degradation factor model (discriminant function) for evaluation by SVM is created with 5 samples of learning data of “25 ° C. cycle” and 12 samples of learning data of “50 ° C. cycle and 50 ° C. storage pair”. Deterioration factors were estimated using a degradation factor model created for a total of 29 samples of evaluation data 5 samples and a total of 24 samples of evaluation data of “50 ° C. cycle and 50 ° C. storage set”. As a result, one sample of the evaluation data was erroneously estimated, and the degradation factor estimation accuracy was 96.6%.

Step 142: “50 ° C. cycle” and “50 ° C. storage” are separated, and the feature quantity is “8-dimensional data consisting of complex impedances of four frequencies of 0.1 Hz, 1 Hz, 100 Hz, and 1,000 Hz”.
A deterioration factor model (discriminant function) for evaluation by SVM is created with 8 samples of learning data of “50 ° C. cycle” and 9 samples of learning data of “50 ° C. storage”, and 9 samples of evaluation data of “50 ° C. cycle” Deterioration factors were estimated using the degradation factor model created for a total of 19 samples of 10 samples for evaluation of “50 ° C. storage”. As a result, no sample was erroneously estimated in the evaluation data, and the degradation factor estimation accuracy was 100%.

FIG. 48 is a diagram (No. 4) illustrating the flowchart for explaining the separation of the deterioration factors. 3. After 20V rest, degradation factors were separated by steps as shown in FIG.
Step 151: “25 ° C. cycle” and other “50 ° C. cycle and 50 ° C. storage set” are separated, and feature quantity is 8 dimensions consisting of complex impedances of 4 frequencies of 1 Hz, 4 Hz, 50.2 Hz, and 1,000 Hz. Data ".

  A degradation factor model (discriminant function) for evaluation by SVM is created with 5 samples of learning data of “25 ° C. cycle” and 12 samples of learning data of “50 ° C. cycle and 50 ° C. storage pair”. Deterioration factors were estimated using a degradation factor model created for a total of 29 samples of evaluation data 5 samples and a total of 24 samples of evaluation data of “50 ° C. cycle and 50 ° C. storage set”. As a result, one sample of the evaluation data was erroneously estimated, and the degradation factor estimation accuracy was 96.6%.

Step 152: “50 ° C. cycle” and “50 ° C. storage” are separated, and the feature quantity is “8-dimensional data consisting of complex impedances of four frequencies of 1 Hz, 4 Hz, 50.2 Hz, and 1,000 Hz”.
A deterioration factor model (discriminant function) for evaluation by SVM is created with 8 samples of learning data of “50 ° C. cycle” and 9 samples of learning data of “50 ° C. storage”, and 9 samples of evaluation data of “50 ° C. cycle” Deterioration factors were estimated using the degradation factor model created for a total of 19 samples of 10 samples for evaluation of “50 ° C. storage”. As a result, no sample was erroneously estimated in the evaluation data, and the degradation factor estimation accuracy was 100%.

<Capacity maintenance rate estimation by B battery complex impedance measurement>
FIG. 49 is a diagram (part 1) illustrating a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a capacity retention ratio (%). In the B battery, the capacity maintenance rate is 97% to 80% data (because there is only data up to about 80% capacity maintenance rate except for storage at 50 ° C). FIG. 49 shows a graph in which the real part of the complex impedance of 000 Hz is taken on the X axis and the capacity maintenance rate corresponding to the data of the real part of the complex impedance of 1,000 Hz is taken on the Y axis. The number of samples of actual measurement data and the range of the capacity maintenance ratio for each deterioration factor of the capacity maintenance ratio of 97% to 80% are as follows, and a total of 38 samples.

25 ° C. cycle: 10 samples (capacity retention: 97.2% to 84.2%)
50 ° C. cycle: 9 samples (capacity maintenance ratio: 97.1% to 81.0%)
Storage at 50 ° C .: 19 samples (capacity maintenance rate: 97.2% to 81.1%)
With these data, the estimation accuracy in the case of estimating the capacity maintenance rate from the real value of 1,000 Hz complex impedance after the discharge of 4.10V discharge is verified.

FIG. 49 shows a graph in which all data are plotted regardless of the deterioration factor and approximated by a cubic curve that minimizes the square error. By using this cubic approximate curve, the capacity maintenance rate of the battery can be estimated with a certain accuracy from the measurement of the 1,000 Hz impedance after the 4.10 V discharge is stopped.
This accuracy (accuracy retention capacity estimation accuracy of the approximate curve) is obtained by approximating the capacity maintenance rate of the sample measured in the same manner as in the case of the battery A and the real value of 1,000 Hz complex impedance after the 4.10V discharge stop of the sample. When the error from the capacity maintenance ratio obtained by substituting for is assumed to be ΔC, when assuming that ΔC is twice the standard deviation (2σΔC), that is, ΔC is normally distributed, a range of ± several percent from the estimated value is assumed In this case, it is evaluated whether or not the measured value of 95% falls within the error. The closer 2σΔC is to 0%, the higher the estimation accuracy.

The results of the evaluation are as follows.
All data: 38 samples (2σΔC = 6.79%)
25 ° C. cycle: 10 samples (2σΔC = 4.35%)
50 ° C. cycle: 9 samples (2σΔC = 8.05%)
Storage at 50 ° C .: 19 samples (2σΔC = 4.09%)
The accuracy evaluation value for each degradation factor is the result of calculating 2σΔC for the approximate curve using only the data for each degradation factor.

  FIG. 50 is a diagram (part 2) illustrating a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a capacity retention ratio (%). FIG. 50 is a graph that is the same graph as FIG. 49, and shows a classification display indicating which deterioration factor causes each point to deteriorate. As is apparent from FIG. 50, when the deterioration factors are different, the relationship between the 1,000 Hz complex impedance real part and the change in the capacity retention rate is different. In particular, the cycle deterioration at 25 ° C. and 50 ° C. and the storage deterioration at 50 ° C. are on completely different curves. Therefore, the capacity maintenance estimated from the measurement of the 1000 Hz complex impedance real part is obtained by obtaining a cubic curve that minimizes the square error of the relationship between the 1,000 Hz complex impedance real part and the capacity maintenance rate separately for each deterioration factor. The estimation accuracy of rate estimation is increased. The accuracy is specifically shown below.

FIG. 51 is a diagram (part 3) illustrating a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a capacity retention ratio (%).
FIG. 51 shows a graph obtained by plotting only a sample with a cycle of 25 ° C. and approximating with a cubic curve that minimizes the square error. At this time, 2σΔC is 2.46%. This accuracy is higher than the estimated accuracy of 6.79% for all data, and the approximate accuracy using all data is higher than the estimated accuracy of 4.35% for the 25 ° C. cycle.

FIG. 52 is a diagram (part 4) illustrating a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a capacity retention ratio (%).
FIG. 52 shows a graph in which only a sample at 50 ° C. is plotted and approximated by a cubic curve that minimizes the square error. At this time, 2σΔC is 1.36%. This accuracy is higher than the estimated accuracy of 6.79% for all data, and the approximated curve using all data is higher than the estimated accuracy of 8.05% for the 50 ° C. cycle.

FIG. 53 is a diagram (No. 5) showing a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a capacity retention ratio (%).
FIG. 53 shows a graph in which only a sample stored at 50 ° C. is plotted and approximated by a cubic curve that minimizes the square error. At this time, 2σΔC is 1.54%. This accuracy is higher than the estimated accuracy of 6.79% for all data, and the approximate accuracy using all data is higher than the estimated accuracy of 4.09% stored at 50 ° C.
From the above results, it can be seen that the estimation accuracy of the capacity maintenance rate is improved by estimating and identifying the deterioration factor.

<Estimation of power capacity maintenance rate by B battery complex impedance measurement>
FIG. 54 is a diagram (part 1) illustrating a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a power capacity retention rate (%). In battery B, data with a power capacity maintenance rate of 97% to 80% (because there is only data up to about 80% of the power capacity maintenance rate except for storage at 50 ° C.), it is included in the 8-dimensional feature amount after 4.10V pause FIG. 54 shows a graph in which the real part of the complex impedance of 1,000 Hz is taken on the X axis and the power capacity maintenance rate corresponding to the data of the real part of the complex impedance of 1,000 Hz is taken on the Y axis. The actual number of measurement data samples and the range of the power capacity maintenance rate for each deterioration factor are as follows, which is a total of 38 samples.

25 ° C. cycle: 10 samples (power capacity retention rate: 96.7% to 83.1%)
50 ° C. cycle: 9 samples (power capacity retention rate: 97.0% to 80.6%)
Storage at 50 ° C .: 19 samples (power capacity maintenance rate: 97.1% to 81.2%)
With these data, the estimation accuracy in the case of estimating the power capacity maintenance rate from the real value of 1,000 Hz complex impedance after the 4.10 V discharge pause is verified. FIG. 54 shows a graph in which all data are plotted regardless of deterioration factors and approximated by a cubic curve that minimizes the square error. By using this cubic approximate curve, it is possible to estimate the power capacity maintenance rate of the battery with a certain accuracy from the measurement of the 1,000 Hz impedance after the 4.10 V discharge is stopped.

  This accuracy (estimated curve power capacity maintenance rate estimation accuracy) was measured in the same manner as the capacity maintenance rate, and the sample power capacity maintenance rate and the real value of 1,000 Hz complex impedance after the 4.10V discharge pause of the sample. Is assumed to be twice the standard deviation of ΔC (2σΔC), that is, assuming that ΔC is normally distributed, ±± If a percent width is assumed, it is evaluated whether the measured value of 95% falls within the error. The closer 2σΔC is to 0%, the higher the estimation accuracy.

The results of the evaluation are as follows.
All data: 38 samples (2σΔC = 7.19%)
25 ° C. cycle: 10 samples (2σΔC = 4.79%)
50 ° C. cycle: 9 samples (2σΔC = 8.36%)
Storage at 50 ° C .: 19 samples (2σΔC = 4.92%)
The accuracy evaluation value for each degradation factor is the result of calculating 2σΔC for the approximate curve using only the data for each degradation factor.

  FIG. 55 is a diagram (part 2) illustrating a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a power capacity retention rate (%). FIG. 55 is a graph that is the same as FIG. 54 and shows a classification display indicating which deterioration factor causes each point to deteriorate. As is clear from FIG. 55, when the deterioration factors are different, the relationship between the 1,000 Hz complex impedance real part and the change in the power capacity maintenance rate is different. In particular, the cycle deterioration at 25 ° C. and 50 ° C. and the storage deterioration at 50 ° C. are on completely different curves. Therefore, the power estimated from the measurement of the 1000 Hz complex impedance real part is obtained separately by obtaining a cubic curve that minimizes the square error of the relationship between the real part of the 1,000 Hz complex impedance and the power capacity maintenance ratio for each degradation factor. The estimation accuracy of capacity maintenance rate estimation increases. The accuracy is specifically shown below.

  FIG. 56 is a diagram (part 3) illustrating a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a power capacity retention rate (%). FIG. 56 shows a graph obtained by plotting only the sample of the 25 ° C. cycle and approximating with a cubic curve that minimizes the square error. At this time, 2σΔC is 2.60%. This accuracy is higher than the estimated accuracy of 7.19% for all data, and the approximated curve using all data is higher than the estimated accuracy of 4.79% for the 25 ° C. cycle.

  FIG. 57 is a diagram (part 4) illustrating a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a power capacity retention rate (%). FIG. 57 shows a graph obtained by plotting only the sample of the 50 ° C. cycle and approximating with a cubic curve that minimizes the square error. At this time, 2σΔC is 1.39%. This accuracy is higher than the estimated accuracy of 7.19% for all data, and the estimated accuracy of 8.36% for the 50 ° C. cycle in the approximate curve using all data.

FIG. 58 is a diagram (No. 5) showing a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a power capacity retention rate (%). FIG. 58 shows a graph in which only a sample stored at 50 ° C. is plotted and approximated by a cubic curve that minimizes the square error. At this time, 2σΔC is 1.51%. This accuracy is higher than the estimated accuracy of 7.19% for all data, and the estimated accuracy of 4.92% stored at 50 ° C. for the approximate curve using all data.
From the above results, it can be seen that the estimation accuracy of the power capacity maintenance rate is improved by estimating and identifying the deterioration factor.

<Residual current capacity estimation by B battery complex impedance measurement>
FIG. 59 is a diagram (part 1) illustrating a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a remaining current capacity (Ah) at 4.0V. The “residual current capacity at 4.0 V” refers to the discharge current capacity when discharged at 0.5 C from the 4.0 V state to the fully discharged state.

  In B battery, it is data of capacity maintenance rate 97% to 80% (because there is only data up to about 80% of power capacity maintenance rate except for storage at 50 ° C). FIG. 59 is a graph in which the real part of the complex impedance of 1,000 Hz is taken as the X axis, and the residual current capacity at 4.0 V corresponding to the data of the real part of the complex impedance of 1,000 Hz is taken as the Y axis. The range of the number of actual measurement data samples and the capacity maintenance ratio for each deterioration factor is as follows, which is a total of 38 samples.

25 ° C. cycle: 11 samples (power capacity maintenance rate: 96.9% to 83.2%)
50 ° C. cycle: 9 samples (power capacity maintenance rate: 96.0% to 80.2%)
Storage at 50 ° C .: 18 samples (power capacity maintenance rate: 96.7% to 81.3%)
With these data, the estimation accuracy in the case of estimating the 4.0 V residual current capacity from the real value of 1,000 Hz complex impedance after the pause of the 4.10 V discharge is verified. FIG. 59 shows a graph in which all data are plotted regardless of deterioration factors and approximated by a cubic curve that minimizes the square error. By using this cubic approximation curve, the 4.0 V residual current capacity of the battery can be estimated with certain accuracy from the measurement of the 1,000 Hz impedance after the 4.10 V discharge is stopped. Since the battery voltage can be easily measured, it is practically important to accurately estimate the remaining current capacity from the battery voltage.

  This accuracy (4.0V residual current capacity estimation accuracy of the approximate curve) was measured in the same manner as in the case of the capacity maintenance rate, and the sample's 4.0V residual current capacity and the sample's 1,000 Hz after 4.0V discharge pause. When the difference from the 4.0V residual current capacity obtained by substituting the real value of complex impedance into the approximate curve is ΔC, it is twice the standard deviation of ΔC (2σΔC), that is, when ΔC is normally distributed Then, it is evaluated whether 95% of the actually measured value falls between the errors if a range of ±% from the estimated value is assumed. The closer 2σΔC is to 0%, the higher the estimation accuracy.

The results of the evaluation are as follows.
All data: 38 samples (2σΔC = 8.15%)
25 ° C. cycle: 11 samples (2σΔC = 5.14%)
50 ° C. cycle: 9 samples (2σΔC = 8.01%)
Storage at 50 ° C .: 18 samples (2σΔC = 5.57%)
The accuracy evaluation value for each degradation factor is the result of calculating 2σΔC for the approximate curve using only the data for each degradation factor.

  FIG. 60 is a diagram (part 2) illustrating a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a remaining current capacity (Ah) at 4.0 V. FIG. 60 is a graph that is the same graph as FIG. 59 and shows a classification of the deterioration factor for each point. As is clear from FIG. 60, when each deterioration factor is different, the relationship between the 1,000 Hz complex impedance real part and the change in the 4.0 V residual current capacity is different. In particular, the cycle deterioration at 25 ° C. and 50 ° C. and the storage deterioration at 50 ° C. are on completely different curves. Therefore, it is estimated from the measurement of the 1,000 Hz complex impedance real part by obtaining a cubic curve that minimizes the square error of the relationship between the 1,000 Hz complex impedance real part and the 4.0 V residual current capacity separately for each deterioration factor. The estimated accuracy of the 4.0V residual current capacity is increased. The accuracy is specifically shown below.

FIG. 61 is a diagram (part 3) illustrating a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a remaining current capacity (Ah) at 4.0 V.
FIG. 61 shows a graph in which only a sample at 25 ° C. is plotted and approximated by a cubic curve that minimizes the square error. At this time, 2σΔC is 3.71%. This accuracy is higher than the estimation accuracy of 8.15% for all data, and the approximation accuracy using all data is higher than the estimation accuracy of 5.14% for the 25 ° C. cycle.

FIG. 62 is a diagram (part 4) illustrating a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a residual current capacity (Ah) at 4.0 V.
FIG. 62 shows a graph obtained by plotting only the sample of the 50 ° C. cycle and approximating it with a cubic curve that minimizes the square error. At this time, 2σΔC is 1.83%. This accuracy is higher than the estimation accuracy of 8.15% for all data, and the approximation accuracy using all data is higher than the estimation accuracy of 8.01% for the 50 ° C. cycle.

FIG. 63 is a diagram (No. 5) illustrating a relationship between a real part (Ω) of a complex impedance of 1,000 Hz and a remaining current capacity (Ah) at 4.0 V.
FIG. 63 shows a graph in which only a sample stored at 50 ° C. is plotted and approximated by a cubic curve that minimizes the square error. At this time, 2σΔC is 0.88%. This accuracy is higher than the estimated accuracy of 8.15% for all data, and the estimated accuracy stored at 50 ° C. is 5.57% for the approximate curve using all data.

  From the above results, it is understood that the estimation accuracy of the 4.0 V residual current capacity is improved by estimating and identifying the deterioration factor. Here, the case where the battery voltage is 4.0 V is specifically seen, but it is possible to improve the estimation accuracy of the remaining current capacity at the battery voltage by estimating and identifying the deterioration factor also at other voltages.

  According to the present invention, it is possible to provide a secondary battery diagnostic device and a secondary battery diagnostic method for determining a battery state such as a deterioration factor and a capacity based on complex impedance measurement data obtained by an AC impedance method.

DESCRIPTION OF SYMBOLS 1 Secondary battery 2 to be measured 2 Reference data holding unit 3 Impedance calculation unit 4 Degradation factor estimation unit 5 Diagnosis determination unit 6 Capacity estimation unit

Claims (8)

In the secondary battery diagnostic device for determining the battery state of the secondary battery,
A reference data holding unit for storing reference data associated with information on deterioration factors of the data collection measurement secondary battery and information on the value of the complex impedance of the data collection measurement secondary battery itself ;
An impedance calculation unit for calculating a complex impedance of the secondary battery to be measured which is a target for determining a battery state;
A deterioration factor estimation unit that estimates a deterioration factor of the secondary battery to be measured based on the reference data stored in the reference data holding unit and information on the complex impedance value itself calculated by the impedance calculation unit. When,
A secondary battery diagnosis apparatus comprising: a diagnosis determination unit that performs diagnosis determination based on an estimation result of the deterioration factor estimation unit and outputs the diagnosis result.   The secondary battery diagnostic apparatus according to claim 1, wherein the complex impedance is at least two types of complex impedances.   The secondary battery diagnostic apparatus according to claim 2, wherein the complex impedance is a complex impedance calculated using different frequencies.   The complex impedance is a real part or an imaginary part of the complex impedance when measured at a specific frequency, and a real part or an imaginary part of the complex impedance when measured at a frequency different from the specific frequency, The secondary battery diagnostic device according to claim 3. In the secondary battery diagnostic method for determining the battery state of the secondary battery,
Storing the reference data associated with the information on the deterioration factor of the data collection measurement secondary battery and the information of the complex impedance value itself of the data collection measurement secondary battery in the reference data holding unit;
A step of calculating a complex impedance of a measured secondary battery, which is a target for determining a battery state, by an impedance calculation unit;
Based on the reference data stored in the reference data holding unit and information on the complex impedance value itself calculated by the impedance calculation unit, the deterioration factor estimation unit estimates the deterioration factor of the secondary battery to be measured. And steps to
And a step of performing a diagnostic determination by a diagnosis determination unit based on an estimation result of the deterioration factor estimation unit and outputting the diagnosis result.   The secondary battery diagnosis method according to claim 5, wherein the complex impedance is at least two types of complex impedances.   The secondary battery diagnosis method according to claim 6, wherein the complex impedance is a complex impedance calculated using different frequencies.   The complex impedance is a real part or an imaginary part of the complex impedance when measured at a specific frequency, and a real part or an imaginary part of the complex impedance when measured at a frequency different from the specific frequency, The secondary battery diagnostic method according to claim 5, 6 or 7.

Images