Method for creating inference of full charge capacity after battery degradation of electric vehicle

Abstract

Provided is a method for creating a highly accurate estimation formula for the amount of full charge after deterioration of a battery. Comprising the following treatment. A first estimation formula and a second estimation formula are prepared, the first estimation formula including a predetermined coefficient and a first battery variable related to the state of the battery and obtaining a first estimation value of the full charge capacity after the degradation of the battery, and the second estimation formula including an undetermined coefficient and a second battery variable related to the state of the battery and obtaining a second estimation value of the full charge capacity after the degradation of the battery. Actual performance data of the first battery variable and the second battery variable are collected from the L electric vehicles. A coefficient to be determined is determined by multiple regression analysis using K pieces of actual performance data relating to the first battery variable among the L pieces of actual performance data and the first inference equation. The determined coefficient to be determined is corrected using (L-K) pieces of actual data so that the correlation between the first estimated value and the second estimated value becomes strong. A second inference equation including the determined and corrected coefficient to be determined is a final inference equation.

Description

Technical Field

[0001] The technology disclosed in this specification relates to a method for creating an inference formula for the fully charged capacity after battery degradation of an electric vehicle. Background Art

[0002] The fully charged capacity of a battery that has been used for a long time decreases due to degradation from the fully charged capacity at the start of use (initial fully charged capacity). Even if left unused for a long time, the fully charged amount also becomes lower than the initial fully charged capacity due to degradation. In this specification, the fully charged capacity of a battery after a specified period is referred to as the "fully charged capacity after degradation". An example of a technology for inferring the fully charged capacity after degradation is disclosed in Japanese Patent Laid-Open Publication No. 2020-042036. The method disclosed in Japanese Patent Laid-Open Publication No. 2020-042036 uses the charge-discharge history of the battery to infer the fully charged capacity after degradation. Summary of the Invention

[0003] This specification provides a technology for inferring the fully charged capacity after battery degradation with higher accuracy than before. Here, in this specification, the fully charged capacity after degradation is expressed as a ratio when the initial fully charged capacity is set to 100 [%].

[0004] The method for creating the inference formula disclosed in this specification includes the following seven steps.

[0005] (Step 1) Prepare a first inference formula and a second inference formula. The first inference formula includes a fixed coefficient and a first battery variable related to the state of the battery, and obtain a first inference value of the fully charged capacity after battery degradation. The second inference formula includes undetermined coefficients and a second battery variable related to the state of the battery, and obtain a second inference value of the fully charged capacity after battery degradation. The specific structures of the first inference formula and the second inference formula will be shown in the "Detailed Description".

[0006] (Step 2) Collect the actual data of the first battery variable and the second battery variable from L electric vehicles. (Step 3) Substitute the K actual data related to the first battery variable among the L actual data into the first inference formula to obtain K first inference values. (Step 4) Determine the undetermined coefficients through multiple regression analysis according to the second inference formula into which the K actual data related to the second battery variable and the K first inference values among the L actual data are substituted. (Step 5) Substitute the (L - K) actual data related to the second battery variable among the actual data into the second inference formula including the determined undetermined coefficients to obtain (L - K) second inference values, and substitute the (L - K) actual data related to the first battery variable into the first inference formula to obtain new first inference values. (Step 6) Modify the determined undetermined coefficients in such a way that the correlation between the (L - K) first inference values and the second inference values obtained in Step 5 becomes stronger. (Step 7) Use the second inference formula including the determined and modified undetermined coefficients and the second battery variable as the inference formula for the fully charged capacity after deterioration of the battery.

[0007] Among them, the first battery variable and the second battery variable related to the state of the battery include the placement time, the energized power, etc. of each temperature range obtained by dividing the battery temperature into multiple temperature ranges.

[0008] Since the inference formula creation method disclosed in this specification adopts the battery variables defined for each temperature range of the battery, the fully charged capacity after deterioration can be inferred with high accuracy. In addition, K actual data among the L actual data are used in the determination of the undetermined coefficients, and the remaining (L - K) actual data are used in the modification of the determined undetermined coefficients. This also contributes to inferring the fully charged capacity after deterioration with high accuracy.

[0009] The detailed technology and further improvements disclosed in this specification will be described in the following "Detailed Description of the Embodiments". Description of the Drawings

[0010] Figure 1 is a flowchart of the steps for creating an inference formula.

[0011] Figure 2 is a diagram showing an example of the placement time of each temperature range.

[0012] Figure 3 is an explanatory diagram of the function y1(X1, Z1).

[0013] Figure 4 is a diagram showing an example of the placement time of vehicle A in each temperature range.

[0014] Figure 5 is a diagram showing the battery placement time x1 of each driving rangei A diagram of an example.

[0015] Figure 6 A diagram showing an example of the amount of electricity supplied in each temperature range.

[0016] Figure 7 An explanatory diagram of the function y2(X2, Z2).

[0017] Figure 8 A diagram showing an example of the driving time of vehicle A in each temperature range.

[0018] Figure 9 A diagram showing an example of the amount of electricity supplied x2 in each temperature range i A diagram of an example.

[0019] Figure 10 A diagram showing an example of the small - power charging time in each temperature range.

[0020] Figure 11 A diagram showing an example of the correlation between the first inferred value and the second inferred value.

[0021] Figure 12 A diagram showing an example of the correlation between the target variable and the second inferred value.

[0022] Figure 13 A diagram showing an example of the result of repeatedly optimizing the undetermined coefficients. Detailed implementation mode

[0023] As described above, in this specification, the fully - charged capacity after deterioration is represented by the ratio with respect to the initial fully - charged capacity. That is, the initial fully - charged capacity is set to 100.

[0024] The inference formula creation method disclosed in this specification includes 7 steps. Figure 1 A flowchart showing the steps of the inference formula creation method is shown. Each step will be specifically described in detail.

[0025] (The first step) In the first step, a first inference formula and a second inference formula for calculating the inferred value of the fully - charged capacity after deterioration are prepared (step S1). The first inference formula includes established coefficients and a first battery variable related to the state of the battery. The inferred value of the fully - charged capacity after deterioration obtained by the first inference formula is called the first inferred value. The second inference formula includes undetermined coefficients and a second battery variable related to the state of the battery. The inferred formula of the fully - charged capacity after deterioration obtained by the second inference formula is called the second inferred value.

[0026] The ultimate goal of the inferential creation method of this embodiment is to collect measurement values (i.e., actual performance data) corresponding to the first battery variable and the second battery variable from multiple electric vehicles actually being used by the user, and use these actual performance data and the first inference formula to determine the undetermined coefficients for optimization. By obtaining the optimized values of the undetermined coefficients, the second inference formula can infer the fully charged capacity after deterioration with high accuracy. The second inference formula including the determined and optimized undetermined coefficients is the desired inference formula.

[0027] The first inference formula is given by the following (Equation 1). Among them, there may be cases where the "coefficients" in this embodiment include constants.

[0028]

[0029] In (Equation 1), y on the left side is the first inferred value. "100" on the right side refers to the initial fully charged capacity.

[0030] The first battery variable z1 represents one of the multiple temperature ranges that the battery can achieve when not in use. The first battery variable x1 represents the time the battery is placed in the temperature range z1. Hereinafter, the time the battery is placed is sometimes simply referred to as the placement time. The function y1(x1, z1) on the right side represents the inferred value of the fully charged capacity after deterioration caused by the placement time. The function y1(x1, z1) includes predetermined coefficients (i.e., established coefficients), and the inferred value can be obtained by substituting the first battery variables x1 and z1. The established coefficients will be described later.

[0031] An example of the temperature range and the placement time is as Figure 2 shown. In the example of Figure 2 , the temperature range is divided into 7 (Ts1 - Ts7). In addition, the function y1(x1, z1) is as Figure 3 shown. The fully charged capacity after deterioration decreases according to the placement time. The function y1(x1, z1) becomes a linear function proportional to the square root of the placement time x1. Its slope Ra takes a negative value. Figure 3 The (Equation 1a) of Figure 3 corresponds to the chart of Figure 2 . The constant Ca and the slope Ra are determined in advance through experiments / simulations / analyses, etc. The fact that y1 is proportional to the square root of the placement time x1 is also an insight obtained through experiments and analyses. The constant Ca and the slope Ra are established coefficients. An example of the constant Ca is 100. In the example of

[0032] , the temperature that the battery can achieve is divided into 7. Therefore, the first battery variable x1 (placement time x1) is assigned to each of the 7 temperature ranges. Figure 3obtained by (Equation 1b). In (Equation 1b), "n" refers to the number of temperature ranges when the battery is placed. In Figure 1 the example, n = 7. In addition, the slope Rb of (Equation 1b) can use the battery characteristic values obtained through experiments.

[0033] The placement time x1 of (Equation 1b) i will be described. Among them, the subscript "i" refers to the i-th temperature range. That is, "x1 i " refers to the placement time of the battery in the i-th temperature range. The values of the battery placement time recorded for each temperature range are directly used as "x1 i ". An example of "x1 i " is as Figure 4 , Figure 5 shown. Figure 4 is an example of the battery placement time of vehicle A in each temperature range. For example, the battery placement time x12 of vehicle A in the temperature range of "16 - 20 °C" (the second temperature range) is "2400". Figure 5 represents an example of the battery placement time "x1 i " for each "i". Figure 4 The battery placement time x1 in each temperature range of i is substituted into each segment of Figure 5 . Hereinafter, "battery placement time" will sometimes be abbreviated as "placement time".

[0034] The function y2(x2, z2) on the right side of (Equation 1) will be described. The first battery variable z2 represents one of the temperature ranges obtained by dividing the temperature that the battery can reach during the driving of the electric vehicle into multiple temperature ranges. The temperature range of the first battery variable z2 can be the same as or different from the previous first battery variable z1. The first battery variable x2 represents the amount of electricity flowing in and out of the battery in the temperature range z2. Hereinafter, the amount of electricity flowing in and out of the battery will be referred to as the energized electricity. The function y2(x2, z2) represents the estimated value of the fully charged capacity after deterioration caused by the energized electricity during driving. The function y2(x2, z2) includes predetermined coefficients (i.e., established coefficients), and the estimated value can be obtained by substituting the first battery variables x2 and z2. The established coefficients will be described later.

[0035] An example of the temperature range and the energized electricity is as Figure 6 shown. In this example, the temperature range of the battery is the same as that in Figure 2 , and is divided into 7 (Ts1 - Ts7). The function y2(x2, z2) is as Figure 7As shown, the function y2(x2, z2) becomes a first-order function proportional to the energized charge x2. Its slope Rb takes a negative value. That is, the larger the energized charge, the smaller the fully charged capacity after deterioration. Figure 7 of (Equation 1c) corresponds to Figure 7 the graph of. The constants Cb and slope Rb are determined in advance through experiments / simulations / analyses, etc. That y2 is proportional to the energized charge x2 is also an insight obtained through experiments, analyses, etc. The constants Cb and slope Rb are established coefficients. In Figure 6 the example of, the temperatures that the battery can achieve are divided into 7. Therefore, the energized charge x2 (the first battery variable x2) is assigned to each of the 7 temperature ranges.

[0036] The inferred value y2 due to the energized charge is obtained through Figure 7 of (Equation 1d). In (Equation 1d), "n" refers to the number of temperature ranges during driving. In Figure 6 the example of, n = 7. Additionally, an example of the constant Ca in (Equation 1d) is 100. The slope Ra in (Equation 1d) uses the battery characteristic values obtained through experiments.

[0037] The method for obtaining the energized charge x2 in (Equation 1d) i is explained. Among them, the subscript "i" refers to the i-th temperature range. That is, "x2 i " refers to the energized charge in the i-th temperature range. The energized charge x2 i is obtained through the following two steps.

[0038] (Step 1) Frequencyize the driving time of each temperature range during driving. Figure 8 The upper part of shows an example of the driving time of vehicle A in each temperature range. For example, the driving time of vehicle A in the temperature range "16 - 20°C" (the second temperature range) is "100". In Figure 8 the example of, the total driving time is 1540. The value obtained by dividing the driving time of each temperature range by the total driving time is the frequencyized driving time, that is, the driving frequency. In Figure 8 the lower part of shows the driving frequencies of each temperature range. The sum of the driving frequencies of all temperature ranges is 1.0.

[0039] (Step 2) Multiply the driving frequency of each temperature range by the total energized charge of vehicle A. Figure 9 shows an example of the energized charge x2 of each temperature range obtained through Step 2. For example, the energized charge x25 in the fifth temperature range (i = 5) is x25 = 0.227 (the driving frequency in the fifth temperature range) × 8500 (the total energized charge) = 1932.

[0040] (Equation 1) is a formula for obtaining the first inferred value y of the fully charged capacity after degradation based on the inferred value y1 caused by the battery storage time and the inferred value y2 caused by the charged power. (Equation 1) includes established coefficients and does not include undetermined coefficients. If the actual data of the battery storage time (the first battery variable x1) and the charged power (the first battery variable x2) in each temperature range are substituted into (Equation 1), the first inferred value y can be obtained.

[0041] The second inference formula is given by the following (Equation 2).

[0042] Qdeg = 100(Dpark + Dlow + Dhigh + Drun) (Equation 2)

[0043] Qdeg on the left side of (Equation 2) also refers to the inferred value of the fully charged capacity after degradation. Qdeg in (Equation 2) corresponds to the second inferred value.

[0044] The battery mounted on the electric vehicle deteriorates due to the electric vehicle and various variables related to the state of the battery. The main factors affecting the deterioration of the battery are as follows: (1) The storage deterioration amount Dpark representing the deterioration amount caused by the time when the battery is not used (battery storage time). (2) The small-power charging deterioration amount Dlow representing the deterioration amount generated when the battery is charged with a small power smaller than the specified power threshold. (3) The large-power charging deterioration amount Dhigh representing the deterioration amount generated when the battery is charged with a large power larger than the specified power threshold. (4) The driving deterioration amount Drun representing the deterioration amount generated during the driving of the electric vehicle. Among them, the power threshold for distinguishing small-power charging and large-power charging is set to several tens of kilowatts (for example, 20 kilowatts). The power threshold is determined in advance and stored in the controller of the electric vehicle.

[0045] (Equation 1) means that the reduction in the fully charged capacity due to deterioration is expressed by the sum of the storage deterioration amount Dpark, the small-power charging deterioration amount Dlow, the large-power charging deterioration amount Dhigh, and the driving deterioration amount Drun. The inference method disclosed in this specification independently obtains the main four factors (Dpark, Dlow, Dhigh, Drun) affecting the deterioration, and infers the total deterioration amount by summing them. Since (Equation 2) infers the deterioration amount according to each factor of deterioration, the fully charged capacity after deterioration can be accurately inferred. (Equation 2) includes multiple battery variables. The battery variables included in (Equation 2) are called the second battery variables, which are distinguished from the first battery variables included in (Equation 1).

[0046] The storage deterioration amount Dpark will be specifically described. The storage deterioration amount Dpark can be inferred by the following (Equation 2a).

[0047]

[0048] The meanings of the symbols in (Formula 2a) are as described below.

[0049] a, b: coefficients to be determined

[0050] X i : the second battery variable representing the time (placement time) when the battery temperature is divided into n temperature intervals and the battery is placed in the i-th temperature interval

[0051] r i : the coefficient to be determined representing the contribution degree of the placement time in the i-th temperature interval to deterioration

[0052] C: the second battery variable representing the time from the factory of the electric vehicle until it reaches the user

[0053] Y1: the second battery variable representing the time (total high-charge time) when the battery remaining capacity is 80% or more of the initial full charge during placement

[0054] Coefficients to be determined a, b, r i Are determined based on the first inference formula and the actual performance data of multiple electric vehicles. The determination steps of the coefficients to be determined will be described later.

[0055] The time C from the factory of the electric vehicle until it reaches the user is uniquely determined at the moment when the electric vehicle reaches the user. That is, C is a given constant. The second battery variable, that is, the placement time X in each temperature interval i And the total high-charge time Y1 are regularly measured and accumulated by the controller of the electric vehicle. The accumulated values correspond to the actual performance data of the battery variables.

[0056] The small-power charging deterioration amount Dlow will be specifically described. The small-power charging deterioration amount Dlow can be inferred by the following (Formula 2b).

[0057]

[0058] The meanings of the symbols in (Formula 2b) are as described below.

[0059] c k : the coefficient to be determined for the k-th power amount interval when the power amount charged by one small-power charging is divided into m power amount intervals

[0060] Tempa i : the second battery variable representing the value (small-power charging time) obtained by dividing the time required for small-power charging in the i-th temperature interval by the total time of small-power charging when the battery temperature is divided into n temperature intervals

[0061] ra i: Coefficient to be determined representing the contribution of battery usage in the \(i\)-th temperature range to degradation

[0062] Z k : Second battery variable representing the number of charging times (number of small-power charging times) that the charge amount charged by one small-power charging is divided into \(m\) charge amount ranges and belongs to the \(k\)-th charge amount range

[0063] Coefficient to be determined \(r_a\) i Refers to the coefficient that converts the usage of the battery per unit time into the amount of degradation. "Usage of the battery" means current flowing in and out of the battery. The coefficient to be determined \(c\) is optimized based on the actual data of the usage status of multiple electric vehicles used by the user k Coefficient to be determined \(c\) k 、\(r_a\) i Are also determined according to the first inference formula and the actual data of multiple electric vehicles. The determination steps of the coefficients to be determined will be described later.

[0064] Second battery variable \(Tempa\) representing the small-power charging time in each temperature range i And second battery variable \(Z\) representing the number of small-power charging times k Are regularly measured and stored by the controller of the electric vehicle. The stored values correspond to the actual data of the second battery variable.

[0065] Small-power charging time \(Tempa\) in each temperature range i An example of Figure 10 Is shown as follows. In the example of Figure 10 , the battery temperature is divided into 7 temperature ranges. That is, in (Equation 2b), \(n = 7\). Figure 10 The upper part shows the small-power charging time in each temperature range. A temperature sensor for measuring the temperature of the battery is installed in the electric vehicle, and the controller accumulates and stores each data in the table in the upper part of Figure 10 According to the measured value of the temperature sensor and the time required for small-power charging. In the rightmost column of the table in the upper part of Figure 10 , the total time required for small-power charging is recorded. In the lower table of Figure 10 , the value obtained by dividing the charging time in each temperature range by the total time is substituted. Figure 10 The values in the lower table of i Are an example of the actual data of \(Tempa\) in (Equation 2b).

[0066] The large-power charging degradation amount \(D_{high}\) is specifically described. The large-power charging degradation amount \(D_{high}\) can be inferred by the following (Equation 2c).

[0067]

[0068] The meanings of the symbols in (Equation 2c) are described as follows.

[0069] e j : The electric quantity charged by one large-power charge is divided into p electric quantity intervals, and it is the undetermined coefficient for the j-th electric quantity interval

[0070] Tempb i : It represents the second battery variable that divides the battery temperature into n temperature intervals and divides the time required for large-power charging in the i-th temperature interval by the total large-power charging time (large-power charging time)

[0071] ra i : It represents the undetermined coefficient of the contribution of battery usage in the i-th temperature interval to degradation

[0072] W j : It represents the second battery variable that divides the electric quantity charged by one large-power charge into p electric quantity intervals and represents the number of large-power charging times (large-power charging times) belonging to the j-th electric quantity interval

[0073] The undetermined coefficient ra representing the contribution degree i is the same as the undetermined coefficient used in (Equation 2b). The undetermined coefficient e k , ra i are also determined as the optimal values based on the first inference formula and the actual performance values of multiple electric vehicles. The determination steps of the undetermined coefficients will be described later. The battery variable Tempb i is the same as the battery variable Tempa in (Equation 2b) i

[0074] The second battery variable Tempb of the battery representing the large-power charging time of each temperature interval i and the second battery variable W representing the number of large-power charging times k are regularly measured and accumulated by the controller of the electric vehicle. The accumulated values are equivalent to the actual performance data of the second battery variable.

[0075] The driving degradation amount Drun will be specifically described. The driving degradation amount Drun can be inferred by the following (Equation 2d).

[0076]

[0077] The meanings of the symbols in (Equation 2d) are described as follows.

[0078] d: undetermined coefficient

[0079] Tempc i : It represents the second battery variable that divides the battery temperature into n temperature intervals and divides the time traveled in the i-th temperature interval by the total driving time (driving time)​

[0080] ra i : The undetermined coefficient representing the contribution of battery usage in the i-th temperature range to degradation

[0081] I EV : The second battery variable representing the total charge flowing in and out of the battery during driving

[0082] The second battery variable Tempc representing the driving time of each temperature range of the battery i and the second battery variable Tempa in (Equation 2b) i are the same. The undetermined coefficient ra i is the same as the undetermined coefficient used in (Equation 2b). The undetermined coefficients d, ra i are determined as optimal values based on the first inference formula and the actual data of the usage status of multiple electric vehicles being used by the user.

[0083] The driving time Tempc of each temperature range i and the total charge I EV are periodically measured and stored by the controller of the electric vehicle. The stored values correspond to the actual data of the second battery variable.

[0084] The first inference formula (Equations 1, 1a - 1d) includes the first battery variables x1, x2, z1, z2 and the established coefficients Ca, Cb, Ra i , Rb i (i is the index representing the temperature range). The established coefficients Ca, Ra i determine the relationship between the battery's placement time and the fully charged capacity after degradation. The established coefficients Cb, Rb i determine the relationship between the battery's charged power and the fully charged capacity after degradation. The established coefficients of the first inference formula are determined in advance through experiments / simulations / theoretical analyses, etc.

[0085] The second inference formula (Equation 2) includes the second battery variables X i , C, Y1, Tempa i , Z k , Tempb i , W j , Tempc i , I EV . Additionally, the second inference formula (Equation 2) includes the undetermined coefficients a, b, r i , c k , ra i , ei, ra i , d. Hereinafter, the first battery variables and the second battery variables are collectively referred to as "battery variables".

[0086] (The second step) Collect the actual data of battery variables from L electric vehicles ( Figure 1 , step S2). The number L is a number much larger than the number of undetermined coefficients. A communication device is installed in the electric vehicle that is the object of the inference formula creation method. In addition, the electric vehicle regularly stores measurement values corresponding to the battery variables. The measurement values stored in the electric vehicle correspond to the actual data of the battery variables. The electric vehicle regularly sends the actual data of the battery variables to the management center. The computer of the management center collects the actual data of the battery variables of L electric vehicles.

[0087] (The third step) Substitute the K actual data related to the first battery variable in the L actual data into the first inference formula to obtain K first inference values of the fully charged capacity after battery degradation (step S3).

[0088] (The fourth step) Substitute the K actual data related to the second battery variable and the K first inference values in the L actual data into the second inference formula, and determine the undetermined coefficients through multiple regression analysis based on the obtained K second inference formulas ( Figure 1 , step S4). The actual data of the first battery variable and the actual data of the second battery variable in the actual data of one electric vehicle both correspond to the fully charged capacity after battery degradation of the electric vehicle. The first inference value obtained by substituting the actual data of the first battery variable into the first inference formula should be equal to the left side of the second inference formula (i.e., the second inference value) when the actual data of the second battery variable is input. That is, the first inference value can be substituted into the left side of the second inference formula. At this time, the K second inference formulas with the actual data of the second battery variable substituted into the right side of the second inference formula can be regarded as a multiple regression formula with the first inference value on the left side as the target variable and the undetermined coefficients as the explanatory variables. Therefore, the undetermined coefficients can be determined through multiple regression analysis based on the K second inference formulas.

[0089] (The fifth step) Substitute the (L - K) actual data related to the second battery variable in the actual data into the second inference formula including the determined undetermined coefficients to obtain (L - K) second inference values, and substitute the (L - K) actual data related to the first battery variable into the first inference formula to obtain new first inference values ( Figure 1 , step S5).

[0090] (Step 6) Correct the determined undetermined coefficients in such a way that the correlation between the (L−K)-th first inference value and the second inference value obtained in Step 5 becomes stronger. The (L−K) actual data are not used in the determination of the undetermined coefficients. The computer corrects the coefficients by using the (L−K) actual data and the corresponding first inference values. The computer substitutes the (L−K) actual data related to the second battery variable into (Equation 2) to obtain the second inference value. The undetermined coefficients of (Equation 2) at this time use the values determined in Step S5. In addition, the computer substitutes the (L−K) actual data related to the first battery variable into (Equation 1) to re-obtain the first inference value.

[0091] Figure 11 An example showing the correlation between the first inference value and the second inference value. Figure 11 is a graph depicting the set of the first inference value y and the second inference value Qdeg calculated using the (L−K) actual data. Group A is the result of the set of inference values when the coefficients before correction are used. Preferably, the first inference value is equal to the second inference value. If the first inference value is equal to the second inference value, the set of the first inference value and the second inference value lies on a 45-degree straight line. For the case before correction (Group A), the determined undetermined coefficients are inaccurate, and the set of the first correction value and the second correction value deviates from the 45-degree straight line.

[0092] In Step 6, correct the determined undetermined coefficients in such a way that the set of the first correction value and the second correction value is distributed along the 45-degree straight line. Figure 11 Group B shows the distribution of the set of the first inference value and the second inference value when the corrected coefficients are used. By optimizing the determined undetermined coefficients based on the correction, the accuracy of (Equation 2) is improved.

[0093] (Step 7) Use the second inference formula including the determined and corrected undetermined coefficients and the second battery variable as the inference formula for the fully charged capacity after battery degradation. As described above, by using the (L−K) actual data and (Equation 1) to optimize the coefficients (the determined undetermined coefficients), an inference formula for the fully charged capacity after degradation with high accuracy can be obtained.

[0094] Next, an example of the optimization of the undetermined coefficients is shown. The first inference formula (Equation 1) and the second inference formula (Equation 2) are the same as above (Step 1).

[0095] (Step 2) Collect the actual data of the first battery variable and the second battery variable from L electric vehicles. The first battery variable is x1, x2, z1, z2 in (Equation 1). x1 uses the battery placement time. z1 uses each temperature range obtained by dividing the temperature range of the battery when it is placed into multiple temperature intervals. x2 uses the total power consumption. z2 uses each temperature range obtained by dividing the temperature range during the driving of the electric vehicle into multiple temperature intervals. The second battery variable is X, C, Y1, Tempa in (Equation 2a), (Equation 2b), (Equation 2c), (Equation 2d). i , C, Y1, Tempa i , Z k , Tempb i , W j , Tempc i , I EV . The meanings of the respective symbols are as described above. The first battery variable and the second battery variable are always measured and stored in the actually running electric vehicle.

[0096] (Step 3a) Substitute the L first battery variables included in the L actual data into the first inference formula (Equation 1) to obtain L first inference values y. For the convenience of explanation, the following L first inference values y are referred to as L target variables y.

[0097] (Step 3b) Divide the L target variables y into K learning data and (L - K) test data. The selection of the learning data and the test data can be random. The ratio of K and (L - K) is preferably 8:2.

[0098] For the undetermined coefficients a, b, r of (Equation 2a), (Equation 2b), (Equation 2c), (Equation 2d) i , c k , ra i , e j , ra i , d determine some initial values.

[0099] (Step 4) Substitute the initial values of the undetermined coefficients and the K battery variables into the second inference formula (Equation 2) to obtain K second inference values Qdeg. Optimize the undetermined coefficients a, b, r, c, ra, e, ra, d in such a way that the absolute value of the difference (|y - Qdeg|) between each of the K learning data y and each of the K inference values Qdeg is minimized. The genetic algorithm is suitable for the optimization algorithm. i , c k , ra i , e j , ra i , d.

[0100] (Step 5) Use (L - K) test data y to verify the accuracy of the second inference formula (Formula 2) (the accuracy of the undetermined coefficients determined by optimization). Specifically, perform the following processing. Substitute (L - K) second battery variables into the second inference formula after optimizing the undetermined coefficients to obtain (L - K) second inference values Qdeg. Plot the group (y, Qdeg) of (L - K) test data y and the corresponding second inference value Qdeg for each test data y on a graph. An example of the plot is as Figure 12 shown. Calculate the average error and the maximum error of the (L - K) groups (y, Qdeg).

[0101] (Step 6) Repeat the processing from Step 3b to Step 5 ten times. An example of the results (average error and maximum error) when the processing is repeated ten times is as Figure 13 shown.

[0102] (Step 7) As the accuracy of the second inference formula, adopt the average value of the average error and the maximum error when the processing from Step 3b to Step 5 is repeated ten times.

[0103] Describe the points to note regarding the technology described in the embodiments. The structure of the formulas introduced in the embodiments is based on insights obtained from experiments / simulations / analyses, etc.

[0104] The method for creating the inference formula disclosed in this specification can be performed by a computer or by a person.

[0105] As described above, specific examples of the present invention have been described in detail, but these are merely examples and do not limit the scope of protection claimed in this application. The technology described in the scope of protection claimed in this application includes technologies obtained by various deformations and changes of the above - illustrated specific examples. The technical elements described in this specification or the drawings exhibit technical usefulness alone or through various combinations, and are not limited to the combinations described in the technical solution at the time of application. In addition, the technologies illustrated in this specification or the drawings can achieve multiple purposes simultaneously, and achieving one of the purposes itself has technical usefulness.

Claims

1. A method for creating an inference formula for the fully charged capacity after deterioration of a battery of an electric vehicle, wherein, Comprising: The first step of preparing a first inference formula and a second inference formula, where the first inference formula includes established coefficients and a first battery variable related to the state of the battery, and obtaining a first inference value of the fully charged capacity after the battery deteriorates; the second inference formula includes undetermined coefficients and a second battery variable related to the state of the battery, and obtaining a second inference value of the fully charged capacity after the battery deteriorates. The second step of collecting actual data of the first battery variable and the second battery variable from L electric vehicles. The third step of substituting K actual data related to the first battery variable among the L actual data into the first inference formula to obtain K first inference values. The fourth step of determining the undetermined coefficients by multiple regression analysis according to the second inference formula into which K actual data related to the second battery variable among the L actual data and the K first inference values are substituted. The fifth step of substituting (L - K) actual data related to the second battery variable among the actual data into the second inference formula including the determined undetermined coefficients to obtain (L - K) second inference values, and substituting (L - K) actual data related to the first battery variable into the first inference formula to obtain new first inference values. The sixth step of correcting the determined undetermined coefficients in such a way that the correlation between the (L - K) first inference values and the second inference values obtained in the fifth step becomes stronger. And The seventh step of using the second inference formula including the determined and corrected undetermined coefficients and the second battery variable as the inference formula for the fully charged capacity after the battery deteriorates. The first inference formula is expressed by the following (Equation 1). Here, y is the first inference value of the fully charged capacity after the battery deteriorates. y1(x1, z1) is a function including established coefficients with the first battery variables x1 and z1 as independent variables. z1 is the first battery variable representing one of the multiple temperature intervals obtained by dividing the temperature that the battery can achieve when not in use. x1 is the first battery variable representing the time the battery is placed in the temperature interval z1. y2(x2, z2) is a function including established coefficients with the first battery variables x2 and z2 as independent variables. z2 is the first battery variable representing one of the multiple temperature intervals obtained by dividing the temperature that the battery can achieve during the driving of the electric vehicle. x2 is the first battery variable representing the amount of electricity flowing in and out of the battery in the temperature interval z2. The second inference formula is expressed by the following (Equation 2). Qdeg = 100 - (Dpark + Dlow + Dhigh + Drun) (Equation 2) Here, Qdeg is the second inference value of the fully charged capacity after the battery deteriorates. Dpark is the placement deterioration amount representing the deterioration amount caused by the placement time of the battery. Dlow is the small-power charging deterioration amount representing the deterioration amount generated when the battery is charged with a small power smaller than the specified power threshold. Dhigh is the large-power charging deterioration amount representing the deterioration amount generated when the battery is charged with a large power larger than the specified power threshold. Drun is the driving deterioration amount representing the deterioration amount generated during the driving of the electric vehicle. The storage degradation amount Dpark is as follows (Equation 2a), where a and b are undetermined coefficients, X i is the second battery variable representing the time the battery is placed in the i-th temperature range when the battery temperature is divided into n temperature ranges. r i is an undetermined coefficient representing the contribution of the placement time in the i-th temperature range to deterioration C is the second battery variable representing the time from the factory of the electric vehicle until it reaches the sales target, Y1 is the second battery variable representing the time during storage when the battery remaining capacity is 80% or more of the initial full charge amount, The small-power charging degradation amount Dlow is as follows (Equation 2b), where c k is the undetermined coefficient for the k-th power range obtained by dividing the power charged by a small power charge once into m power ranges Tempa i is the second battery variable representing a value obtained by dividing the battery temperature into n temperature ranges and dividing the time required for small-power charging in the i-th temperature range by the total time of small-power charging. ra i is a undetermined coefficient representing the contribution of battery usage in the i-th temperature range to deterioration Z k is the second battery variable representing the number of times of charging belonging to the k-th power range when the power charged by one small-power charging is divided into m power ranges The large-power charging degradation amount Dhigh is as follows (Equation 2c), where e j is the undetermined coefficient for the jth power range obtained by dividing the power charged by one large-power charge into p power ranges. Tempb i is the second battery variable representing a value obtained by dividing the time required for high-power charging in the i-th temperature range by the total time of high-power charging after dividing the battery temperature into n temperature ranges. ra i is a coefficient to be determined representing the contribution of battery usage in the i-th temperature range to degradation. W j is the second battery variable representing the number of charging times that belong to the j-th power range when the power charged by one-time large-power charging is divided into p power ranges. The driving degradation amount Drun is as follows (Equation 2d), where d is an undetermined coefficient, Tempc i is the second battery variable representing a value obtained by dividing the time traveled in the i-th temperature range by the total travel time after dividing the battery temperature into n temperature ranges. ra i is a undetermined coefficient representing the contribution of battery usage in the i-th temperature range to degradation. I EV is the second battery variable representing the total amount of electricity flowing in and out of the battery during driving.

Images