Secondary battery diagnostic method, secondary battery diagnostic program, and secondary battery diagnostic device

Abstract

Provided is a secondary battery diagnostic method capable of evaluating a remaining service life of a secondary battery more accurately than in the past, without disassembling the secondary battery. This secondary battery diagnostic method includes: a parameter estimating step (step S1) for employing a predetermined model formula to estimate, on the basis of data obtained by measuring a load characteristic of a secondary battery being diagnosed, characteristic parameters, at a diagnosis time point, of the secondary battery being diagnosed, the characteristic parameters including an electrolytic solution diffusion coefficient Dn, at the diagnosis time point, of the secondary battery being diagnosed; a threshold estimating step (step S5) for employing a trained model to estimate, on the basis of input data, which are data including some of the characteristic parameters at the diagnosis time point, a threshold Dth of the electrolytic solution diffusion coefficient of the secondary battery being diagnosed; and a difference calculating step (step S4) for finding a difference ΔD between the threshold Dth and the electrolytic solution diffusion coefficient Dn at the diagnosis time point.

Description

[Technical Field]

[0001] The present invention relates to a diagnostic method for a secondary battery, a diagnostic program for a secondary battery, and a diagnostic device for a secondary battery. [Background technology]

[0002] The discharge capacity of secondary batteries such as lithium-ion batteries gradually decreases with repeated charging and discharging. Therefore, it is preferable to diagnose secondary batteries at appropriate times to evaluate their degree of deterioration and determine whether they can be reused or when they should be replaced. Furthermore, since secondary batteries can be reused after diagnosis, it is preferable that the diagnosis be performed non-destructively.

[0003] JP 2017-97997 A describes a characteristic analysis method for a secondary battery that uses a model equation with the characteristic values ​​of the components constituting the battery as parameters and estimates the characteristic values ​​of the components by fitting the battery voltage value expressed by the model equation to actual measurement data. This characteristic analysis method uses the actual measurement data obtained by applying to the battery under analysis a charge / discharge pattern including an operating period consisting of either a constant current discharge period or a constant current charge period, and a rest period provided following the operating period.

[0004] The publication also describes characteristic values ​​estimated by fitting to actual measurement data, such as the lithium ion diffusion coefficient in the positive electrode active material, the lithium ion diffusion coefficient in the negative electrode active material, the lithium ion diffusion coefficient in the electrolyte (electrolyte diffusion coefficient), the interface resistance in the positive electrode active material, the interface resistance in the negative electrode active material, and the lithium ion salt concentration in the electrolyte. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Publication No. 2017-97997 Summary of the Invention [Problem to be solved by the invention]

[0006] Conventionally, the degree of deterioration of a secondary battery has been evaluated based on the magnitude of discharge capacity and the magnitude of internal resistance at the time of diagnosis. However, according to research by the present inventors, even if the magnitude of discharge capacity and the magnitude of internal resistance at the time of diagnosis are similar, the number of times the secondary battery can be used thereafter is not necessarily the same. Specifically, even if the magnitude of discharge capacity and the magnitude of internal resistance at the time of diagnosis are similar, there are secondary batteries whose discharge capacity rapidly decreases after a small number of charge / discharge cycles, and those whose discharge capacity does not decrease significantly even after charge / discharge. Therefore, the remaining life of a secondary battery cannot be accurately evaluated simply by measuring the magnitude of discharge capacity and the magnitude of internal resistance at the time of diagnosis.

[0007] The aforementioned Japanese Patent Application Laid-Open No. 2017-97997 describes a method for estimating the characteristic values ​​of components constituting a secondary battery without disassembling the secondary battery, but does not describe a specific method for evaluating the remaining life of the secondary battery from these characteristic values.

[0008] An object of the present invention is to provide a secondary battery diagnostic method, a secondary battery diagnostic program, and a secondary battery diagnostic device that can evaluate the remaining life of a secondary battery more accurately than conventional methods without disassembling the secondary battery. [Means for solving the problem]

[0009] A secondary battery diagnosis method according to one embodiment of the present invention includes: a parameter estimation step of estimating characteristic parameters of a secondary battery to be diagnosed at the time of diagnosis, including an electrolyte diffusion coefficient Dn of the secondary battery to be diagnosed, using a predetermined model formula based on data obtained by measuring load characteristics of the secondary battery to be diagnosed; a threshold estimation step of estimating a threshold value Dth of the electrolyte diffusion coefficient of the secondary battery to be diagnosed, using a trained model, based on input data that is data including some of the characteristic parameters at the time of diagnosis; and a difference calculation step of calculating a difference ΔD between the threshold value Dth and the electrolyte diffusion coefficient Dn at the time of diagnosis, wherein the trained model is a plurality of learning a step of determining characteristic parameters of the learning secondary battery; a step of calculating the discharge capacity when the electrolyte diffusion coefficient of the learning secondary battery is changed based on the model formula and the characteristic parameters of the learning secondary battery, thereby determining the relationship between the electrolyte diffusion coefficient and the discharge capacity of the learning secondary battery; and a step of determining a threshold value of the electrolyte diffusion coefficient of the learning secondary battery based on the relationship between the electrolyte diffusion coefficient and the discharge capacity of the learning secondary battery, and the learning secondary battery is generated by machine learning using teacher data in which data including some of the characteristic parameters of the learning secondary battery is used as input data and the determined threshold value of the electrolyte diffusion coefficient of the learning secondary battery is output data.

[0010] A diagnostic program for a secondary battery according to one embodiment of the present invention causes a computer to execute the following steps: a parameter estimation step of estimating characteristic parameters of a secondary battery to be diagnosed at the time of diagnosis, including an electrolyte diffusion coefficient Dn of the secondary battery to be diagnosed, using a predetermined model formula based on data obtained by measuring load characteristics of the secondary battery to be diagnosed; a threshold estimation step of estimating a threshold value Dth of the electrolyte diffusion coefficient of the secondary battery to be diagnosed, using a trained model, based on input data that is data including some of the characteristic parameters at the time of diagnosis; and a difference calculation step of calculating a difference ΔD between the threshold value Dth and the electrolyte diffusion coefficient Dn at the time of diagnosis, wherein the trained model The learning secondary batteries are generated by machine learning using teacher data in which the following steps are performed for a plurality of learning secondary batteries: determining characteristic parameters of the learning secondary batteries; calculating the discharge capacity when the electrolyte diffusion coefficient of the learning secondary batteries is changed based on the model formula and the characteristic parameters of the learning secondary batteries, and determining the relationship between the electrolyte diffusion coefficient and the discharge capacity of the learning secondary batteries; and determining a threshold value of the electrolyte diffusion coefficient of the learning secondary batteries based on the relationship between the electrolyte diffusion coefficient and the discharge capacity of the learning secondary batteries. Data including some of the characteristic parameters of the learning secondary batteries is used as input data, and the determined threshold value of the electrolyte diffusion coefficient of the learning secondary batteries is used as output data.

[0011] A diagnostic device for a secondary battery according to one embodiment of the present invention includes: a parameter estimation device that estimates characteristic parameters of a secondary battery to be diagnosed at the time of diagnosis, including an electrolyte diffusion coefficient Dn of the secondary battery to be diagnosed, using a predetermined model formula based on data obtained by measuring load characteristics of the secondary battery to be diagnosed; a threshold estimation device that estimates a threshold Dth of the electrolyte diffusion coefficient of the secondary battery to be diagnosed based on input data that is data including some of the characteristic parameters at the time of diagnosis, using a trained model; and a difference calculation device that calculates a difference ΔD between the threshold Dth and the electrolyte diffusion coefficient Dn at the time of diagnosis, wherein the trained model is a plurality of learning a step of determining characteristic parameters of the learning secondary battery; a step of calculating the discharge capacity when the electrolyte diffusion coefficient of the learning secondary battery is changed based on the model formula and the characteristic parameters of the learning secondary battery, thereby determining the relationship between the electrolyte diffusion coefficient and the discharge capacity of the learning secondary battery; and a step of determining a threshold value of the electrolyte diffusion coefficient of the learning secondary battery based on the relationship between the electrolyte diffusion coefficient and the discharge capacity of the learning secondary battery, and the learning secondary battery is generated by machine learning using teacher data in which data including some of the characteristic parameters of the learning secondary battery is used as input data and the determined threshold value of the electrolyte diffusion coefficient of the learning secondary battery is output data.

[0012] A secondary battery diagnostic device according to one embodiment of the present invention includes a parameter estimation device that estimates characteristic parameters of a secondary battery to be diagnosed at the time of diagnosis, including an electrolyte diffusion coefficient Dn of the secondary battery to be diagnosed, using a predetermined model formula based on data obtained by measuring the load characteristics of the secondary battery to be diagnosed; a first data generation device that determines characteristic parameters of a plurality of learning secondary batteries; a second data generation device that calculates the discharge capacity when the electrolyte diffusion coefficient of the learning secondary battery is changed based on the model formula and the characteristic parameters of the learning secondary batteries, and calculates the relationship between the electrolyte diffusion coefficient and the discharge capacity of the learning secondary battery; The system includes: a third data generation device that determines a threshold value for the electrolyte diffusion coefficient of the training secondary battery based on the relationship between the electrolyte diffusion coefficient and the discharge capacity; a model generation device that generates a trained model by machine learning using teacher data in which data including some of the characteristic parameters of the training secondary battery is input data and the determined threshold value for the electrolyte diffusion coefficient of the training secondary battery is output data; a threshold estimation device that uses the trained model to estimate a threshold value Dth for the electrolyte diffusion coefficient of the secondary battery to be diagnosed based on input data that is data including some of the characteristic parameters at the time of diagnosis; and a difference calculation device that calculates the difference ΔD between the threshold value Dth and the electrolyte diffusion coefficient Dn at the time of diagnosis. [Effects of the Invention]

[0013] According to the present invention, the remaining life of a secondary battery can be evaluated more accurately than ever before without disassembling the secondary battery. [Brief explanation of the drawings]

[0014] [Figure 1] FIG. 1 is a flow diagram of a diagnostic method for a secondary battery according to one embodiment of the present invention. [Figure 2] FIG. 2 is a flowchart showing an example of a more specific procedure for the step of estimating the characteristic parameters. [Figure 3] FIG. 3 is a graph showing an example of the relationship between the electrolyte diffusion coefficient and the discharge capacity. [Figure 4] FIG. 4 is a diagram showing an example of how the threshold value Dth is determined. [Figure 5] FIG. 5 is a flow diagram of a diagnostic method for a secondary battery according to one embodiment of the present invention. [Figure 6] FIG. 6 is a flow diagram showing an example of a method for creating training data. [Figure 7] FIG. 7 is a diagram showing the relationship between ΔD obtained by diagnosis and the number of cycles at which the sudden drop begins. [Figure 8] FIG. 8 is a diagram showing the machine learning procedure. [Figure 9] FIG. 9 is a scatter plot in which the predicted values ​​of the electrolyte diffusion coefficient thresholds are plotted on the vertical axis and the true values ​​(values ​​obtained by analysis) are plotted on the horizontal axis for the training data. [Figure 10] FIG. 10 is a scatter plot of the validation data, with predicted values ​​of the electrolyte diffusion coefficient thresholds plotted on the vertical axis and true values ​​(values ​​determined by analysis) plotted on the horizontal axis. [Figure 11] FIG. 11 is a scatter plot in which the predicted values ​​of the electrolyte diffusion coefficient thresholds are plotted on the vertical axis and the true values ​​(values ​​obtained by analysis) are plotted on the horizontal axis for the training data. [Figure 12] FIG. 12 is a scatter plot of the validation data, with predicted values ​​of the electrolyte diffusion coefficient thresholds plotted on the vertical axis and true values ​​(values ​​determined by analysis) plotted on the horizontal axis. DETAILED DESCRIPTION OF THE INVENTION

[0015] The inventors focused on the electrolyte diffusion coefficient when evaluating the remaining life of secondary batteries. As secondary batteries undergo cycle degradation, the volume of electrolyte and the salt concentration in the electrolyte decrease, leading to a state known as "depletion of electrolyte." It is known that the value of the electrolyte diffusion coefficient decreases at this time. The value Dn of the electrolyte diffusion coefficient of a secondary battery at the time of diagnosis can be estimated nondestructively using measured data on load characteristics and a model equation well known in this field. Furthermore, the relationship between the electrolyte diffusion coefficient and discharge capacity can be determined by simulation using the aforementioned model equation.

[0016] From the relationship between the electrolyte diffusion coefficient and discharge capacity, the electrolyte diffusion coefficient value Dth at which the secondary battery becomes unsuitable for reuse is determined, and the difference between this and the electrolyte diffusion coefficient value Dn at the time of diagnosis, ΔD = Dn - Dth, is calculated. This ΔD can be used as an indicator of the remaining life of the secondary battery. In other words, the larger ΔD, the lower the possibility of early battery depletion and the higher the likelihood of long-term use. Conversely, the smaller ΔD, the higher the possibility of early battery depletion and the lower the likelihood of long-term use. Until now, there has been no method for predicting such a sudden drop in discharge capacity due to battery depletion. By using this method, the remaining life of a secondary battery can be evaluated more accurately than before.

[0017] The lifespan of a secondary battery varies greatly depending on the conditions of use (e.g., discharge rate, etc.). Therefore, "accurate evaluation of remaining lifespan" does not necessarily mean predicting the specific number of charge / discharge cycles until the battery becomes unsuitable for reuse.

[0018] In the above-described method for evaluating the remaining life of a secondary battery, in order to determine the threshold value Dth, a simulation must be performed for each secondary battery to be evaluated to determine the relationship between the electrolyte diffusion coefficient and the discharge capacity. This simulation requires a certain amount of time and requires mastery of the software used for the simulation. Therefore, in order to more easily evaluate the remaining life of a secondary battery, it is preferable to be able to determine the threshold value Dth without performing this simulation.

[0019] The inventors performed the above-described simulations to determine the threshold Dth for multiple secondary batteries with different types of positive and negative electrode active materials, different usage conditions, different degrees of degradation, etc. The inventors further constructed an estimation model (trained model) using machine learning, with relatively easily obtainable characteristic parameters of secondary batteries as input data (explanatory variables) and the determined threshold Dth as output data (objective variable), and successfully used this to estimate the threshold Dth of a secondary battery with an unknown threshold Dth. In this case, they found that the accuracy of the estimation could be improved by including the discharge capacity and electrolyte conductivity of the secondary battery in the input data.

[0020] The present invention has been completed based on the above findings. Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

[0021] [Secondary battery diagnostic method] In the following explanation, first, a method for determining the thresholds Dth and ΔD without using an estimation model based on machine learning (hereinafter referred to as a "trained model") will be described. Then, a method for determining the thresholds Dth and ΔD using a trained model will be described. In addition, a method for generating a trained model (training method) will also be described.

[0022] [Method for determining the threshold Dth and ΔD without using a trained model] 1 is a flow diagram of a secondary battery diagnostic method (without using a trained model) according to one embodiment of the present invention. This diagnostic method includes the steps of: estimating characteristic parameters of a secondary battery to be diagnosed (hereinafter referred to as the "target battery") at the time of diagnosis (step S1); determining the relationship between the electrolyte diffusion coefficient and the discharge capacity (step S2); determining a threshold value Dth for the electrolyte diffusion coefficient (step S3); and determining the difference ΔD between the threshold value Dth and the electrolyte diffusion coefficient Dn at the time of diagnosis (step S4). Each step will be described in detail below.

[0023] [Process for estimating characteristic parameters] The characteristic parameters of the target battery at the time of diagnosis are estimated (step S1). More specifically, the characteristic parameters of the target battery at the time of diagnosis, including the electrolyte diffusion coefficient Dn of the target battery at the time of diagnosis, are estimated using a predetermined model formula based on data obtained by measuring the load characteristics of the target battery.

[0024] In this process, the characteristic parameters of the target battery at the time of diagnosis are estimated by fitting data obtained by measuring the load characteristics of the target battery using a predetermined model formula. This analysis (simulation) can be performed using a computer program capable of fluid analysis, such as the software Battery Design Studio manufactured by Siemens.

[0025] The model formula may be one well known in the art, such as the one described in Marc Doyle et al., "Modeling of Galvanostatic Charge and Discharge of the Lithium / Polymer / Insertion Cell," J. Electrochem. Soc., Vol. 140, No. 6, June (1993).

[0026] The target battery is, for example, a lithium ion battery.

[0027] The data obtained by measuring the load characteristics of the target battery may be, for example, a discharge curve obtained by measuring the target battery at multiple discharge rates. This data preferably includes a discharge curve measured at a very low discharge rate (e.g., 0.02 C). This data also preferably includes a discharge curve measured at a discharge rate of 1 C or higher. This data preferably includes discharge curves measured at three or more levels of discharge rate, and more preferably includes discharge curves measured at four or more levels of discharge rate. The data obtained by measuring the load characteristics of the target battery may also be a charge curve obtained by measuring the target battery at multiple charge rates.

[0028] The characteristic parameters estimated in this step (characteristic parameters at the time of diagnosis of the target battery) include at least the electrolyte diffusion coefficient Dn at the time of diagnosis of the target battery. The characteristic parameters may also include, for example, the solid phase diffusion coefficient of the positive and negative electrode active materials, the electrolyte conductivity, etc. Other specific examples of the characteristic parameters will be described later.

[0029] 2 is a flow chart showing an example of a more specific procedure for the step of estimating the characteristic parameters (step S1). In this example, the step of estimating the characteristic parameters (step S1) includes a step of inputting the basic specifications of the target battery (step S1-1), a step of inputting data obtained by measuring the load characteristics of the target battery (step S1-2), a step of estimating the static parameters of the target battery (step S1-3), and a step of estimating the dynamic parameters of the target battery (step S1-4).

[0030] The basic specifications of the secondary battery to be diagnosed are input into the analysis software (step S1-1). The basic specifications to be input are not limited to these, but may include, for example, the following: Positive and negative electrode composition (component materials, content, particle size, etc.) · Thickness, density, and bending rate of the positive and negative electrodes (= about 1.5 in most cases) ·Material, thickness, and electrical conductivity of positive and negative electrode current collector foils Separator thickness and porosity Electrolyte composition (component materials, content) -Thermal conductivity and heat capacity of the above constituent materials (basic physical properties specific to the material) ·Electrode area

[0031] Since the diagnosis is basically non-destructive, accurate information on the composition of the electrolyte at the time of diagnosis is not available. Therefore, general information on the secondary battery to be diagnosed (or specification information for a new battery) is obtained and input as a parameter. Although some values ​​must be input when actually performing a simulation, the composition of the electrolyte itself does not have a significant impact on the results of the simulation. In the diagnosis method of this embodiment, the composition information on the electrolyte is only used as a reference.

[0032] The density of the positive and negative electrodes is also expected to have changed due to expansion from the initial state, but the exact value cannot be measured at the time of diagnosis. Therefore, the initial value (standard value, etc.) or a value predicted from the initial value is entered. If it is completely unknown, a general value may be entered. If necessary, fine adjustments may be made in step S1-4.

[0033] The data obtained by measuring the load characteristics of the target battery is input into the analysis software (step S1-2). As described above, the data obtained by measuring the load characteristics of the target battery is a discharge curve obtained by measuring the target battery at multiple discharge rates. Hereinafter, "data obtained by measuring the load characteristics of the target battery" may be referred to as "actual measurement data."

[0034] The static parameters of the target battery are estimated from the actual measurement data and the model formula (step S1-3). For example, the static parameters of the target battery are adjusted to fit the shape of the discharge curve measured at a very low discharge rate. The discharge curve measured at a very low discharge rate (e.g., 0.02 C) can be considered to roughly match the voltage curve (OCV curve) when no load is connected. The static parameters can include, but are not limited to, the following: Capacity per unit weight of positive and negative electrode active material (discharge capacity of a battery decreases after use) Utilization rate of each positive and negative electrode active material (not all of the active materials are used) - Maximum and minimum voltages for the target battery range

[0035] The dynamic parameters of the target battery are estimated from the actual measurement data and the model formula (step S1-4). For example, a simulation is performed in which the target battery is discharged at a current value equivalent to the measurement conditions of the actual measurement data, and the results of this simulation are compared with the actual measurement data, and the dynamic parameters are adjusted so that the two match. This simulation can be performed, for example, using the discharge curve prediction function of Battery Design Studio mentioned above. The dynamic parameters can include, but are not limited to, the following: Electrolyte conductivity Electrolyte diffusion coefficient Diffusion coefficient in the solid phase of positive and negative electrode active materials -Heat capacity of the target battery

[0036] It is preferable to set the environmental temperature during the simulation to match the environmental temperature when the actual measurement data was obtained. For medium-sized or larger product batteries, especially those expected to be used at high rates, it is preferable to consider the effects of heat generation. To do so, it is preferable to perform measurements at least at 1C and perform fitting with the actual measurement data affected by heat generation. On the other hand, if the target battery is a small cell for desktop testing, it is not necessary to consider the effects of heat generation.

[0037] Through the above steps, it is possible to estimate the characteristic parameters of the target battery at the time of diagnosis, including the electrolyte diffusion coefficient Dn of the target battery at the time of diagnosis.

[0038] [Step of determining the relationship between electrolyte diffusion coefficient and discharge capacity] Based on the model formula used in step S1 and the characteristic parameters estimated in step S1, the relationship between the electrolyte diffusion coefficient and the discharge capacity is determined (step S2). More specifically, among the characteristic parameters estimated in step S1, a discharge simulation is performed by varying only the electrolyte diffusion coefficient while keeping the other characteristic parameters constant, and the discharge capacity is determined. The discharge rate and environmental temperature during the discharge simulation are preferably set according to the intended reuse application. For example, if the intended application is one in which the battery will be used at an average rate of about 1 C, the discharge rate used to determine the relationship between the electrolyte diffusion coefficient and the discharge capacity is also 1 C. Since it is difficult to precisely match all environments, the simulation may be performed using average values.

[0039] 3 is a graph showing an example of the relationship between the electrolyte diffusion coefficient and the discharge capacity. In this example, the discharge capacity when the environmental temperature is 45°C and the discharge rate is 0.5 C is calculated by dividing the discharge capacity by the electrolyte diffusion coefficient of 4.8×10 -6 , 4.0×10 -6 , 2.95×10 -6 , 1.85×10 -6 , 1.48×10 -6 , and 1.1 × 10 -6 cm 2 / sec.

[0040] As shown in this example, the smaller the electrolyte diffusion coefficient, the smaller the discharge capacity. Furthermore, the relationship between the electrolyte diffusion coefficient and the discharge capacity is not linear, but tends to show a curve in which the smaller the electrolyte diffusion coefficient, the greater the decrease in discharge capacity.

[0041] [Step of determining the threshold value Dth of the electrolyte diffusion coefficient] Based on the relationship between the electrolyte diffusion coefficient and the discharge capacity obtained in step S2, a threshold Dth for the electrolyte diffusion coefficient is determined (step S3). More specifically, the value of the electrolyte diffusion coefficient at which the target battery becomes unsuitable for reuse is determined as the threshold Dth, with reference to the relationship between the electrolyte diffusion coefficient and the discharge capacity obtained in step S2. The circumstances under which a battery is determined to be "unsuitable for reuse" vary depending on the reuse application of the target battery. Therefore, a criterion for determining that a battery is "unsuitable for reuse" is set according to the application.

[0042] For example, when the discharge capacity falls below a predetermined allowable value, the battery may be determined to be unsuitable for reuse. In this case, the electrolyte diffusion coefficient when the discharge capacity falls below the predetermined allowable value is determined as the threshold value Dth. For example, in the example of FIG. 3, if the allowable value of the discharge capacity is 36.02 mAh, the threshold value Dth is 1.40×10 -6 cm 2 / sec.

[0043] Alternatively, the battery may be determined to be unsuitable for reuse when the discharge capacity begins to drop sharply. In this case, the electrolyte diffusion coefficient when the discharge capacity begins to drop sharply is determined as the threshold value Dth. For example, the electrolyte diffusion coefficient when the slope of the discharge capacity becomes equal to or greater than a predetermined magnitude may be set as the threshold value Dth. Furthermore, as shown in FIG. 4, the point where the tangents to the curves before and after the start of the drop in discharge capacity intersect may be set as the threshold value Dth.

[0044] [Step of calculating the difference ΔD between the threshold value Dth and the electrolyte diffusion coefficient Dn at the time of diagnosis] The difference ΔD between the threshold value Dth determined in step S3 and the electrolyte diffusion coefficient Dn estimated in step S1 at the time of diagnosis is calculated (step S4). For example, Dn=2.22×10 -6 cm 2 / sec, Dth=1.40×10 -6 cm 2 / sec, ΔD=Dn-Dth=0.82×10 -6 cm 2 / sec.

[0045] This ΔD can be used as an indicator of the remaining lifespan of the target battery. In other words, the larger ΔD, the higher the possibility that the target battery can be used for a long period of time, and the smaller ΔD, the lower the possibility that the target battery can be used for a long period of time. Even if the discharge capacity at the time of diagnosis is approximately the same, ΔD may differ. By using ΔD, the remaining lifespan of the target battery can be evaluated more accurately than the conventional method of evaluating the remaining lifespan based on the magnitude of the discharge capacity at the time of diagnosis.

[0046] As mentioned above, the lifespan of a secondary battery varies greatly depending on the conditions of use, so "accurate evaluation of remaining life" does not necessarily mean predicting the specific number of charge / discharge cycles until the battery becomes unsuitable for reuse. However, if it is assumed that the target battery will continue to be used under certain conditions, it is also possible to predict the remaining lifespan (number of charge / discharge cycles) of the target battery from ΔD. For example, the relationship between ΔD and remaining lifespan may be measured in advance, and the remaining lifespan of the target battery may be predicted based on this relationship between ΔD and remaining lifespan.

[0047] [Method for determining the threshold Dth and ΔD using a trained model] 5 is a flow diagram of a secondary battery diagnostic method (using a trained model) according to one embodiment of the present invention. In this diagnostic method, the step of determining the relationship between the electrolyte diffusion coefficient and the discharge capacity (step S2) and the step of determining the threshold value Dth of the electrolyte diffusion coefficient (step S3) in FIG. 1 are replaced with a step of estimating the threshold value Dth of the electrolyte diffusion coefficient (step S5, hereinafter referred to as the "threshold estimation step"). The threshold estimation step (step S5) is a step of estimating the threshold value Dth of the electrolyte diffusion coefficient of the target battery using the "trained model" described below, based on input data that is data including some of the characteristic parameters of the target battery at the time of diagnosis estimated in step S1.

[0048] [Pre-trained model] The trained model is an estimation model that estimates output data from input data (explanatory variables) using data including some of the characteristic parameters of a secondary battery and the threshold value of the electrolyte diffusion coefficient of that secondary battery as output data (objective variable), and is obtained by machine learning using training data created in advance for multiple secondary batteries (hereinafter referred to as "training secondary batteries").

[0049] [Creating training data] 6 is a flow diagram showing an example of a method for creating training data. This method for creating training data includes a step of determining characteristic parameters of a training secondary battery (step SA-1), a step of determining the relationship between the electrolyte diffusion coefficient and the discharge capacity of the training secondary battery (step SA-2), and a step of determining a threshold value of the electrolyte diffusion coefficient of the training secondary battery (step SA-3).

[0050] It is preferable to create training data for multiple secondary batteries that differ in the type of positive and negative electrode active material, usage conditions, degree of deterioration, etc. The greater the number of training secondary batteries (i.e., the greater the amount of training data used for machine learning), the more accurate the trained model that can be obtained.

[0051] The step of determining the characteristic parameters of the learning secondary battery (step SA-1) is, more specifically, a step of estimating the characteristic parameters of the real battery using a model formula based on data obtained by measuring the load characteristics of the real battery. This estimation can be performed in the same manner as the step of estimating the characteristic parameters of the target battery at the time of diagnosis (step S1 in FIG. 1). Note that in the step of determining the characteristic parameters of the learning secondary battery (step SA-1), it is not necessary to perform analysis using data obtained by measuring the load characteristics of the real battery (actual measurement data) for all of the learning secondary batteries to be analyzed. For learning secondary batteries similar to analyzed real batteries, the characteristic parameters may be determined based on data obtained by analyzing the real battery (for example, by changing some of the characteristic parameters obtained by analyzing the real battery). However, for learning secondary batteries that are significantly different in shape, etc., it is preferable to estimate the characteristic parameters by performing analysis using actual measurement data.

[0052] The step of determining the relationship between the electrolyte diffusion coefficient and discharge capacity of the learning secondary battery (step SA-2) is, more specifically, a step of determining the discharge capacity when the electrolyte diffusion coefficient of the learning secondary battery is changed, based on the model formula used in step SA-1 and the characteristic parameters of the learning secondary battery determined in step SA-1, and determining the relationship between the electrolyte diffusion coefficient and discharge capacity of the learning secondary battery. The step of determining the relationship between the electrolyte diffusion coefficient and discharge capacity of the learning secondary battery (step SA-2) can be performed in the same manner as the step of determining the relationship between the electrolyte diffusion coefficient and discharge capacity of the target battery (step S2 in FIG. 1).

[0053] The step of determining the threshold value of the electrolyte diffusion coefficient of the learning secondary battery (step SA-3) is, more specifically, a step of determining the threshold value of the electrolyte diffusion coefficient of the learning secondary battery based on the relationship between the electrolyte diffusion coefficient and the discharge capacity of the learning secondary battery obtained in step SA-2. The step of determining the threshold value of the electrolyte diffusion coefficient of the learning secondary battery (step SA-3) can be performed in the same manner as the step of determining the threshold value Dth of the electrolyte diffusion coefficient of the target battery (step S3 in FIG. 1).

[0054] The above steps (steps SA-1 to SA-3) are performed for a plurality of learning secondary batteries, and the threshold value of the electrolyte diffusion coefficient is determined for the plurality of learning secondary batteries. Several parameters selected from the parameters used in the step (step SA-2) of determining the relationship between the electrolyte diffusion coefficient and the discharge capacity of the learning secondary battery are selected as input data (explanatory variables), and the threshold value of the electrolyte diffusion coefficient is used as output data to create training data.

[0055] The input data includes some of the characteristic parameters of the learning secondary battery determined in step SA-1. The input data preferably includes the discharge capacity and electrolyte conductivity of the learning secondary battery. Including these parameters can increase the accuracy of the estimation.

[0056] Since the ohmic resistance obtained by measuring the impedance of a secondary battery is mostly due to the resistance of the electrolyte, the "ohmic resistance converted to a value per unit area of ​​the electrode" can be used instead of the electrolyte conductivity. While the ohmic resistance is easy to obtain, it may be less accurate than the electrolyte conductivity because it includes the resistance of the electrode and the contact resistance of components such as tabs.

[0057] The input data preferably further includes the magnitude of the current when the secondary battery is used (discharge rate). Here, "the magnitude of the current when the secondary battery is used" means the magnitude of the current when the secondary battery to be diagnosed is used after diagnosis. In other words, the magnitude of the current when the secondary battery is used is not a value determined in step SA-1 (FIG. 6) or a value estimated in step S1 (FIG. 5), but a value determined by the use of the secondary battery to be diagnosed, etc.

[0058] The input data may further include the temperature at which the secondary battery is used. Here, "the temperature at which the secondary battery is used" means the temperature at which the secondary battery to be diagnosed is used after the diagnosis.

[0059] The input data preferably further includes one or more parameters selected from the group consisting of the electrode area, the positive electrode coating amount, the positive electrode active material type, the positive electrode porosity, the negative electrode porosity, and the heat capacity of the learning secondary battery. All of these parameters are relatively easy to obtain and have a relatively large effect on the accuracy of the estimation.

[0060] Of the parameters listed above, the electrode area, the amount of positive electrode coating, and the type of positive electrode active material do not change with use, so data from when a new battery was manufactured can be used. For the positive electrode porosity and the negative electrode porosity, data from when a new battery was manufactured can be used, or, since the amount of expansion is generally known depending on the type of active material, the maximum value expected after degradation can be used. The heat capacity of the secondary battery can be estimated, for example, in step SA-1. Since the heat capacity is the same for the same model of secondary battery, the heat capacity can be estimated only once for the same model of secondary battery, and thereafter can be treated as a constant.

[0061] In addition to these, the input data may also include the positive electrode active material capacity, the positive electrode active material solid phase diffusion coefficient, the negative electrode active material capacity, the negative electrode utilization rate, the negative electrode active material solid phase diffusion coefficient, and the separator thickness.

[0062] [Generate a trained model] Using the created training data, an estimation model (trained model) that estimates output data (objective variables) from input data (explanatory variables) is generated by machine learning. There are no particular limitations on the machine learning algorithm, but nonlinear support vector regression, for example, can be used.

[0063] In the secondary battery diagnostic method according to this embodiment (FIG. 5), this trained model is used to estimate the threshold value Dth of the electrolyte diffusion coefficient of the target battery based on input data that includes some of the characteristic parameters of the target battery at the time of diagnosis estimated in step S1 (step S5).

[0064] The input data used in the threshold estimation step (step S5) corresponds to the input data used in the machine learning. For example, if the discharge capacity and electrolyte conductivity of the training secondary battery are used as input data in the machine learning, the discharge capacity and electrolyte conductivity of the target battery are used as input data in the threshold estimation step.

[0065] According to this embodiment, once a trained model is generated, it is not necessary to perform the process of determining the relationship between the electrolyte diffusion coefficient and the discharge capacity (step S2 in Figure 1) each time a diagnosis is performed, making it possible to more easily evaluate the remaining life of a secondary battery.

[0066] [Diagnostic program for secondary batteries, etc.] The above-described secondary battery diagnostic method can also be realized as a computer program. A secondary battery diagnostic program according to one embodiment of the present invention causes a computer to execute the following steps: a parameter estimation step of estimating characteristic parameters of the secondary battery at the time of diagnosis, including the electrolyte diffusion coefficient Dn of the secondary battery at the time of diagnosis, using a predetermined model formula based on data obtained by measuring the load characteristics of the secondary battery; a threshold estimation step of estimating a threshold value Dth of the electrolyte diffusion coefficient of the secondary battery at the time of diagnosis, using a trained model, based on input data that includes some of the characteristic parameters at the time of diagnosis; and a difference calculation step of calculating the difference ΔD between the threshold value Dth and the electrolyte diffusion coefficient Dn at the time of diagnosis. This computer program may be recorded on a computer-readable recording medium. This embodiment also enables a more accurate evaluation of the remaining life of the target battery compared to conventional methods that evaluate the remaining life based on the magnitude of the discharge capacity at the time of diagnosis.

[0067] The above-described secondary battery diagnostic method can also be realized as a computer system. A secondary battery diagnostic system according to one embodiment of the present invention includes a memory and a processor, and the processor, in accordance with a program stored in the memory, executes the following steps: a parameter estimation step of estimating characteristic parameters of the secondary battery to be diagnosed at the time of diagnosis, including an electrolyte diffusion coefficient Dn of the secondary battery to be diagnosed, using a predetermined model formula based on data obtained by measuring the load characteristics of the secondary battery to be diagnosed; a threshold estimation step of estimating a threshold value Dth of the electrolyte diffusion coefficient of the secondary battery to be diagnosed, using the trained model, based on input data that includes some of the characteristic parameters at the time of diagnosis; and a difference calculation step of calculating a difference ΔD between the threshold value Dth and the electrolyte diffusion coefficient Dn at the time of diagnosis.

[0068] [Secondary battery diagnostic device] A secondary battery diagnostic device according to one embodiment of the present invention includes a parameter estimation device that estimates characteristic parameters of the secondary battery to be diagnosed at the time of diagnosis, including an electrolyte diffusion coefficient Dn of the secondary battery to be diagnosed, using a predetermined model formula based on data obtained by measuring the load characteristics of the secondary battery to be diagnosed; a threshold estimation device that estimates a threshold Dth of the electrolyte diffusion coefficient of the secondary battery to be diagnosed based on input data that includes some of the characteristic parameters at the time of diagnosis, using a trained model; and a difference calculation device that calculates the difference ΔD between the threshold Dth and the electrolyte diffusion coefficient Dn at the time of diagnosis.

[0069] A secondary battery diagnostic device according to another embodiment of the present invention includes a parameter estimation device that estimates characteristic parameters of a secondary battery to be diagnosed at the time of diagnosis, including an electrolyte diffusion coefficient Dn of the secondary battery to be diagnosed, using a predetermined model formula based on data obtained by measuring the load characteristics of the secondary battery to be diagnosed; a first data generation device that determines characteristic parameters of the learning secondary batteries for a plurality of learning secondary batteries; a second data generation device that determines the discharge capacity when the electrolyte diffusion coefficient of the learning secondary battery is changed based on the model formula and the characteristic parameters of the learning secondary batteries, and determines the relationship between the electrolyte diffusion coefficient and the discharge capacity of the learning secondary batteries; The system includes a third data generation device that determines a threshold value for the electrolyte diffusion coefficient of the training secondary battery based on the relationship between the electrolyte diffusion coefficient and the discharge capacity; a model generation device that generates a trained model by machine learning using teacher data in which data including some of the characteristic parameters of the training secondary battery is used as input data and the determined threshold value for the electrolyte diffusion coefficient of the training secondary battery is used as output data; a threshold estimation device that uses the trained model to estimate a threshold value Dth for the electrolyte diffusion coefficient of the secondary battery to be diagnosed based on input data that is data including some of the characteristic parameters at the time of diagnosis; and a difference calculation device that calculates the difference ΔD between the threshold value Dth and the electrolyte diffusion coefficient Dn at the time of diagnosis.

[0070] These embodiments also enable a more accurate evaluation of the remaining life of the target battery compared to the conventional method of evaluating the remaining life based on the magnitude of the discharge capacity at the time of diagnosis. [Example]

[0071] The present invention will be described in more detail below with reference to examples, but the present invention is not limited to these examples.

[0072] We fabricated multiple medium-sized laminated cells with a rated capacity of 5 Ah and multiple small-sized laminated cells with a rated capacity of 36 mAh.

[0073] [Medium-sized laminated cell] <Preparation of positive electrode> A positive electrode mixture-containing slurry was prepared by uniformly mixing 93 parts by weight of LiCoO2 (positive electrode active material), 3 parts by weight of carbon black (conductive additive), and 4 parts by weight of PVDF (binder) using NMP (solvent). This positive electrode mixture-containing slurry was applied to both sides of a positive electrode current collector made of aluminum foil with a thickness of 15 μm, dried, and then pressure-molded using a roller press. A positive electrode was fabricated by punching out a portion of the positive electrode current collector not coated with the positive electrode mixture-containing slurry to form a tab.

[0074] <Preparation of negative electrode> A negative electrode mixture slurry was prepared by mixing 97.5 parts by mass of graphite (a negative electrode active material), 1.5 parts by mass of carboxymethyl cellulose (a binder), and 1 part by mass of styrene-butadiene rubber, adding an appropriate amount of water and thoroughly mixing. This negative electrode mixture slurry was applied to both sides of a negative electrode current collector made of copper foil with a thickness of 10 μm, dried, and then pressure-molded using a roller press. A negative electrode was fabricated by punching out a portion of the negative electrode current collector not coated with the negative electrode mixture slurry to form a tab.

[0075] <Battery construction> The seven positive electrodes and the eight negative electrodes were alternately stacked with an 18 μm-thick polyolefin microporous film separator having a three-layer structure with a polyethylene layer as the middle layer and two polypropylene layers as the outer layers interposed therebetween to form a laminated electrode body.

[0076] Next, the positive electrode tabs of the laminated electrode body were welded together, and the negative electrode tabs were welded together, and leads were connected to each of them. After that, LiPF was dissolved at a concentration of 1 mol / L in a solution obtained by mixing ethylene carbonate, diethyl carbonate, and methyl ethyl carbonate in a volume ratio of 1:1:1, and then vinylene carbonate was further dissolved in an amount to make 1 mass % to prepare a non-aqueous electrolyte solution. This was then sealed in an exterior body made of an aluminum laminate film, together with the non-aqueous electrolyte solution, to produce a non-aqueous electrolyte secondary battery with a rated capacity of 5 Ah.

[0077] [Small laminated cell] <Preparation of positive electrode> A positive electrode mixture-containing slurry was prepared by uniformly mixing 94 parts by weight of LiCoO2 (positive electrode active material), 4 parts by weight of carbon black (conductive additive), and 2 parts by weight of PVDF (binder) using NMP (solvent). This positive electrode mixture-containing slurry was applied to both sides of a positive electrode current collector made of aluminum foil with a thickness of 15 μm, dried, and then pressure-molded using a roller press. A positive electrode was fabricated by punching out a portion of the positive electrode current collector not coated with the positive electrode mixture-containing slurry to form a tab.

[0078] <Preparation of negative electrode> A negative electrode mixture slurry was prepared by mixing 94.5 parts by mass of graphite (negative electrode active material), 3 parts by mass of carbon-coated SiO particles (D50: 5.0 μm), 1.5 parts by mass of carboxymethyl cellulose (binder), and 1 part by mass of styrene-butadiene rubber, and adding an appropriate amount of water and thoroughly mixing. This negative electrode mixture slurry was applied to both sides of a negative electrode current collector made of copper foil with a thickness of 10 μm, dried, and then pressure-molded using a roller press. A negative electrode was fabricated by punching out a portion of the negative electrode current collector not coated with the negative electrode mixture slurry to form a tab.

[0079] <Battery construction> The positive electrode and the negative electrode were stacked via a 12 μm thick polyolefin microporous film separator with a three-layer structure consisting of a polyethylene layer as an intermediate layer and two polypropylene layers as outer layers to form a laminated electrode body.

[0080] Next, leads were connected to the positive electrode tab and the negative electrode tab of the laminated electrode body, respectively. LiPF was then dissolved at a concentration of 1 mol / L in a solution prepared by mixing ethylene carbonate and diethyl carbonate in a volume ratio of 3:7, and then vinylene carbonate was further dissolved in an amount to give a concentration of 1 mass % to prepare a non-aqueous electrolyte solution. The resultant was then enclosed in an exterior body made of an aluminum laminate film, thereby producing a non-aqueous electrolyte secondary battery with a rated capacity of 36 mAh.

[0081] [Preparation of degradation cell] Charge-discharge cycle tests were conducted on a medium-sized laminated cell with a rated capacity of 5 Ah under multiple conditions, including different charge-discharge rates and ambient temperatures, to produce several degraded cells whose discharge capacity had dropped to 4.8 Ah. Similarly, charge-discharge cycle tests were conducted on a small laminated cell with a rated capacity of 36 mAh under multiple conditions, including different charge-discharge rates and ambient temperatures, to produce several degraded cells whose discharge capacity had dropped to 35 mAh.

[0082] [Load characteristic measurement] The load characteristics of these deteriorated cells were measured, specifically, the discharge curves were measured at discharge rates of 0.02C, 0.2C, 0.5C, and 1C.

[0083] [Secondary battery diagnosis] The secondary battery diagnostic method described in the embodiment was carried out using these deteriorated cells as target batteries. The analysis (simulation) was carried out using the software Battery Design Studio manufactured by Siemens. Among the basic specifications, the solvent ratio and salt concentration were input as the same values ​​as at the time of production (because the values ​​at the time of diagnosis cannot be measured).

[0084] After estimating the characteristic parameters at the time of diagnosis, the discharge capacity was calculated at an ambient temperature of 45°C and a discharge rate of 0.5C while changing the electrolyte diffusion coefficient based on the estimated characteristic parameters, and the relationship between the electrolyte diffusion coefficient and discharge capacity was obtained. The point where the tangents to each curve before and after the discharge capacity began to drop sharply was set as the threshold Dth, and the difference ΔD between the threshold Dth and the electrolyte diffusion coefficient Dn at the time of diagnosis was calculated.

[0085] [Measurement of remaining life] Charge-discharge cycle tests were performed on these deteriorated cells under the same conditions, and the number of charge-discharge cycles until the discharge capacity suddenly dropped (number of cycles until the sudden drop began) was measured. The relationship between ΔD obtained by diagnosis and the number of cycles until the sudden drop began is shown in Figure 7.

[0086] As shown in Figure 7, even when the discharge capacity at the time of diagnosis was the same (4.8 Ah or 35 mAh), differences in ΔD were observed. It was also found that the number of cycles at which the sharp drop began changed depending on ΔD, confirming the validity of this diagnostic method.

[0087] [Generate a trained model] Next, we prepared secondary batteries using LCO or NCM as the positive electrode active material and graphite or SiO as the negative electrode active material. The load characteristics of these secondary batteries were measured, and the threshold electrolyte diffusion coefficient was determined using the method described above. These data were used as training data (however, some data was used as validation data rather than training data), and machine learning was performed to generate a trained model.

[0088] The following nine variables were used as input data (explanatory variables). 1.Discharge capacity 2. Electrode area 3. Current magnitude when using secondary batteries 4. Amount of coating on the positive electrode 5.Cathode active material species 6. Positive electrode porosity 7.Negative electrode porosity 8.Electrolyte Conductivity 9. Heat capacity of secondary batteries

[0089] Machine learning was performed using the open source library scikit-learn. The machine learning procedure is shown in Figure 8. The machine learning algorithm used was nonlinear support vector regression (nonlinear SVR).

[0090] First, all data was normalized using the following formula: Xn=(X-Xa) / Xd X: original value, Xn: normalized value, Xa: mean value of variable X, Xd: standard deviation

[0091] In the nonlinear SVR, the weighting coefficient w was calculated so that the following function was minimized:

number

[0092] h(y (i) -f(x (i) )) is a function expressed by the following formula: h(y (i) -f(x (i) ))=max(0,|y (i) -f(x (i) )|-ε) C and ε are so-called hyperparameters.

[0093] Figures 9 and 10 are scatter plots with the predicted values ​​of the electrolyte diffusion coefficient threshold on the vertical axis and the true values ​​(values ​​obtained by analysis) on the horizontal axis. Figure 9 is for training data (100 pieces of data used), and Figure 10 is for validation data (data not used for learning). As shown in Figure 10, the values ​​obtained by analysis can be predicted with good accuracy even for unlearned data.

[0094] <Contribution of electrolyte conductivity to Dth> Using the trained model, we investigated the effect of electrolyte conductivity on Dth.

[0095] For a certain secondary battery, if the electrolyte conductivity is reduced from 8.90 (mS / cm) to 1.75 (mS / cm) without changing other parameters, Dth will be 2.705×10 -6 (cm 2 / sec) to 2.758×10 -6 (cm 2 / sec).

[0096] For the same secondary battery, if the heat capacity is halved without changing the electrolyte conductivity, Dth becomes 2.705 × 10 -6 (cm 2 / sec) to 2.643 × 10 -6 (cm 2 / sec).

[0097] Next, consider the case where the heat capacity is halved and the electrolyte conductivity is reduced to 1.75 (mS / cm). From the above results, Dth is 2.643 × 10 -6 (cm 2 / sec), which is expected to increase as the electrolyte conductivity decreases. However, according to the results of the trained model, Dth is 2.643×10 -6 (cm 2 / sec) to 2.575×10 -6 (cm 2 / sec).

[0098] As described above, the effect of electrolyte conductivity on Dth is not constant depending on the amount of change in other parameters, and there is a correlation between the parameters. Therefore, it is clear that simple prediction is difficult.

[0099] For comparison, machine learning was performed excluding electrolyte conductivity from the input data (explanatory variables), and a trained model was created. Figures 11 and 12 are scatter plots with the predicted value of the "electrolyte diffusion coefficient threshold" on the vertical axis and the true value (value obtained by analysis) on the horizontal axis. Figure 11 is for the training data, and Figure 12 is for the validation data (data not used for training). As shown in Figure 12, this trained model had inferior prediction accuracy compared to the results of a trained model that included electrolyte conductivity in the input data (Figure 10).

[0100] This shows that the accuracy of prediction can be improved by including the electrolyte conductivity in the input data.

[0101] Although the embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments, and various modifications are possible within the scope of the invention.

Claims

1. a parameter estimation step of estimating characteristic parameters of the secondary battery to be diagnosed at the time of diagnosis, including an electrolyte diffusion coefficient Dn of the secondary battery to be diagnosed, using a predetermined model formula based on data obtained by measuring load characteristics of the secondary battery to be diagnosed; a threshold estimation step of estimating a threshold Dth of an electrolyte diffusion coefficient of the secondary battery to be diagnosed based on input data that is data including a part of characteristic parameters at the time of diagnosis, using a trained model; a difference calculation step of calculating a difference ΔD between the threshold value Dth and the electrolyte diffusion coefficient Dn at the time of diagnosis, The trained model is Regarding multiple secondary batteries for learning, determining characteristic parameters of the learning secondary battery; determining a discharge capacity when the electrolyte diffusion coefficient of the learning secondary battery is changed based on the model formula and characteristic parameters of the learning secondary battery, and determining a relationship between the electrolyte diffusion coefficient and the discharge capacity of the learning secondary battery; determining a threshold value of the electrolyte diffusion coefficient of the learning secondary battery based on a relationship between the electrolyte diffusion coefficient of the learning secondary battery and a discharge capacity; A secondary battery diagnostic method, wherein the secondary battery diagnostic data is generated by machine learning using training data in which data including some of the characteristic parameters of the learning secondary battery is used as input data and the determined threshold value of the electrolyte diffusion coefficient of the learning secondary battery is used as output data.

2. The diagnostic method for a secondary battery according to claim 1, the input data used in the machine learning includes a discharge capacity and an electrolyte conductivity of the learning secondary battery, A secondary battery diagnostic method, wherein input data used in the threshold estimation step includes a discharge capacity and an electrolyte conductivity of the secondary battery to be diagnosed at the time of diagnosis.

3. The diagnostic method for a secondary battery according to claim 1, the input data used in the machine learning includes a discharge capacity of the learning secondary battery and an ohmic resistance converted per electrode unit area, A secondary battery diagnostic method, wherein input data used in the threshold estimation step includes a discharge capacity and an ohmic resistance converted per unit electrode area of ​​the secondary battery to be diagnosed at the time of diagnosis.

4. The diagnostic method for a secondary battery according to claim 2 or 3, A diagnostic method for a secondary battery, wherein the input data used in the machine learning and threshold estimation processes further includes the magnitude of the current when the secondary battery is used.

5. The diagnostic method for a secondary battery according to claim 2 or 3, The input data used in the machine learning further includes one or more selected from the group consisting of an electrode area, a positive electrode coating amount, a positive electrode active material type, a positive electrode porosity, a negative electrode porosity, and a heat capacity of the learning secondary battery, The method for diagnosing a secondary battery, wherein the input data used in the threshold estimation step further includes one or more selected from the group consisting of an electrode area, a positive electrode coating amount, a positive electrode active material type, a positive electrode porosity, a negative electrode porosity, and a heat capacity of the secondary battery to be diagnosed.

6. A diagnostic method for a secondary battery according to any one of claims 1 to 3, comprising: The method for diagnosing a secondary battery, wherein the step of determining the threshold value of the electrolyte diffusion coefficient of the learning secondary battery is a step of determining the electrolyte diffusion coefficient when the discharge capacity is equal to or less than a predetermined allowable value as the threshold value.

7. A diagnostic method for a secondary battery according to any one of claims 1 to 3, comprising: The method for diagnosing a secondary battery, wherein the step of determining the threshold value of the electrolyte diffusion coefficient of the learning secondary battery is a step of determining the electrolyte diffusion coefficient at which the discharge capacity starts to drop rapidly as the threshold value.

8. A diagnostic method for a secondary battery according to any one of claims 1 to 3, comprising: A method for diagnosing a secondary battery, wherein the data obtained by measuring the load characteristics includes discharge curves measured at a plurality of discharge rates of the secondary battery to be diagnosed.

9. A diagnostic method for a secondary battery according to any one of claims 1 to 3, comprising: A method for diagnosing a secondary battery, wherein the secondary battery to be diagnosed and each of the plurality of learning secondary batteries are lithium ion batteries.

10. A diagnostic program for a secondary battery, comprising: a parameter estimation step of estimating characteristic parameters of the secondary battery to be diagnosed at the time of diagnosis, including an electrolyte diffusion coefficient Dn of the secondary battery to be diagnosed, using a predetermined model formula based on data obtained by measuring load characteristics of the secondary battery to be diagnosed; a threshold estimation step of estimating a threshold Dth of an electrolyte diffusion coefficient of the secondary battery to be diagnosed based on input data that is data including a part of characteristic parameters at the time of diagnosis, using a trained model; a difference calculation step of calculating a difference ΔD between the threshold value Dth and the electrolyte diffusion coefficient Dn at the time of diagnosis, The trained model is a diagnostic program for the secondary battery, Regarding multiple secondary batteries for learning, determining characteristic parameters of the learning secondary battery; determining a discharge capacity when the electrolyte diffusion coefficient of the learning secondary battery is changed based on the model formula and characteristic parameters of the learning secondary battery, and determining a relationship between the electrolyte diffusion coefficient and the discharge capacity of the learning secondary battery; determining a threshold value of the electrolyte diffusion coefficient of the learning secondary battery based on a relationship between the electrolyte diffusion coefficient of the learning secondary battery and a discharge capacity; A secondary battery diagnostic program generated by machine learning using training data in which data including some of the characteristic parameters of the learning secondary battery is used as input data and the determined threshold value of the electrolyte diffusion coefficient of the learning secondary battery is used as output data.

11. A diagnostic device for a secondary battery, comprising: a parameter estimation device that estimates characteristic parameters of a secondary battery to be diagnosed at the time of diagnosis, including an electrolyte diffusion coefficient Dn of the secondary battery to be diagnosed, using a predetermined model equation based on data obtained by measuring load characteristics of the secondary battery to be diagnosed; a threshold estimation device that estimates a threshold Dth of an electrolyte diffusion coefficient of the secondary battery to be diagnosed based on input data that is data including a portion of the characteristic parameters at the time of the diagnosis, using a trained model; a difference calculation device for calculating a difference ΔD between the threshold value Dth and the electrolyte diffusion coefficient Dn at the time of diagnosis, The trained model is configured by the secondary battery diagnostic device: Regarding multiple secondary batteries for learning, determining characteristic parameters of the learning secondary battery; determining a discharge capacity when the electrolyte diffusion coefficient of the learning secondary battery is changed based on the model formula and characteristic parameters of the learning secondary battery, and determining a relationship between the electrolyte diffusion coefficient and the discharge capacity of the learning secondary battery; determining a threshold value of the electrolyte diffusion coefficient of the learning secondary battery based on a relationship between the electrolyte diffusion coefficient of the learning secondary battery and a discharge capacity; A secondary battery diagnostic device generated by machine learning using training data in which data including some of the characteristic parameters of the training secondary battery is used as input data and the determined threshold value of the electrolyte diffusion coefficient of the training secondary battery is used as output data.

12. a parameter estimation device that estimates characteristic parameters of a secondary battery to be diagnosed at the time of diagnosis, including an electrolyte diffusion coefficient Dn of the secondary battery to be diagnosed, using a predetermined model equation based on data obtained by measuring load characteristics of the secondary battery to be diagnosed; a first data generating device that determines characteristic parameters of a plurality of learning secondary batteries; a second data generating device that calculates a discharge capacity when an electrolyte diffusion coefficient of the learning secondary battery is changed based on the model formula and characteristic parameters of the learning secondary battery, and calculates a relationship between the electrolyte diffusion coefficient and the discharge capacity of the learning secondary battery; a third data generating device that determines a threshold value of the electrolyte diffusion coefficient of the learning secondary battery based on a relationship between the electrolyte diffusion coefficient of the learning secondary battery and a discharge capacity; a model generation device that generates a trained model by machine learning using training data in which data including a portion of the characteristic parameters of the training secondary battery is used as input data and the determined threshold value of the electrolyte diffusion coefficient of the training secondary battery is used as output data; and a threshold estimation device that estimates a threshold Dth of an electrolyte diffusion coefficient of the secondary battery to be diagnosed based on input data that includes a portion of the characteristic parameters at the time of diagnosis, using the trained model; a difference calculation device that calculates a difference ΔD between the threshold value Dth and the electrolyte diffusion coefficient Dn at the time of diagnosis.

Images