Battery degradation prediction system, battery degradation prediction method, and battery degradation prediction program

Abstract

This disclosure relates to a battery degradation prediction system, a battery degradation prediction method, and a battery degradation prediction procedure. The battery degradation prediction system using a prediction model includes: a reliability calculation unit configured to calculate prediction reliability using the difference between a predicted SOH value and an actual measured SOH value of a target battery, the predicted SOH value being calculated using the prediction model; and a model adjustment unit configured to adjust the prediction model when the prediction reliability exceeds a predetermined threshold. Therefore, the model to be used is adjusted by providing a quantitative prediction reliability to the output.

Description

Technical Field

[0001] This disclosure relates to a battery degradation prediction system, a battery degradation prediction method, and a battery degradation prediction program. Background Technology

[0002] In recent years, technologies have been used to diagnose and predict battery degradation by taking into account the battery’s physical or chemical degradation characteristics and usage history (such as elapsed time, charge, state of charge (SOC), and temperature).

[0003] Japanese Unexamined Patent Application Publication No. 2013-089424 (JP 2013-089424 A) discloses a method for constructing an objective function by combining a model that separates the aging part and the current-carrying part of the battery with a calculation model such as the root law, and using a solver to create a table of discharge coefficient ha(T,S) and current-carrying coefficient ac(T,S), where T is the temperature and S is the state of charge (SOC), and predicting battery degradation based on this table. Summary of the Invention

[0004] In existing technologies, it is not possible to adjust the adaptability and robustness of predicted outputs based on the physical and chemical degradation characteristics of batteries. Therefore, there is a risk that the model may have good robustness but poor adaptability to a particular individual, or the model may have good adaptability to a specific individual but poor robustness and generalization performance.

[0005] A technique is also known to convert time-series data such as battery voltage or SOC, temperature, and current into intermediate data, thereby using this data as input for potential features without explicitly indicating the physical and chemical degradation characteristics of the data. However, in this method, physical and chemical degradation characteristics cannot be guaranteed. Therefore, reasonable predictive reliability cannot be calculated, and even feedback cannot be provided. Consequently, the risk that the model may have poor individual adaptability or poor generalization performance cannot be eliminated.

[0006] Furthermore, a technique is known to correct the prediction formula by evaluating the deviation from the predicted value when additional measurement data is input. However, in this correction method, the moderating of the contributions of physical and chemical degradation properties cannot be reflected in the strength of the correction. Therefore, the risk that the model may have poor individual adaptability or poor generalization performance cannot be eliminated.

[0007] This disclosure relates to a battery degradation prediction system, battery degradation prediction method, and battery degradation prediction program with adjustable adaptability and robustness.

[0008] The battery degradation prediction system according to this disclosure is a battery degradation prediction system using a prediction model. The battery degradation prediction system includes: a reliability calculation unit configured to calculate predicted reliability using the difference between a predicted state of health value and an actual measured state of health value of a target battery, the predicted state of health value being calculated using the prediction model; and a model adjustment unit configured to adjust the prediction model when the predicted reliability exceeds a predetermined threshold.

[0009] Therefore, the model to be used is tuned by providing a quantitative prediction reliability to the output.

[0010] The battery degradation prediction method according to this disclosure is a battery degradation prediction method using a prediction model. The battery degradation prediction method includes: calculating prediction reliability using the difference between a predicted state of health value and an actual measured state of health value of a target battery, wherein the predicted state of health value is calculated using the prediction model; and adjusting the prediction model when the prediction reliability exceeds a predetermined threshold.

[0011] Therefore, the model to be used is tuned by providing a quantitative prediction reliability to the output.

[0012] The battery degradation prediction program according to this disclosure is a battery degradation prediction program that uses a prediction model. The battery degradation prediction program enables a computer to: calculate prediction reliability using the difference between a predicted state of health value and an actual measured state of health value of a target battery, the predicted state of health value being calculated using the prediction model; and adjust the prediction model when the prediction reliability exceeds a predetermined threshold.

[0013] Therefore, the model to be used is tuned by providing a quantitative prediction reliability to the output.

[0014] According to this disclosure, a battery degradation prediction system, a battery degradation prediction method, and a battery degradation prediction program with adjustable adaptability and robustness can be provided. Attached Figure Description

[0015] The features, advantages, and technical and industrial significance of exemplary embodiments of the present invention will now be described with reference to the accompanying drawings, wherein like symbols denote like elements, and wherein: Figure 1 This is a block diagram illustrating the configuration of the degradation prediction system according to the first embodiment; Figure 2 This is a flowchart of the operation of the degradation prediction system according to the first embodiment; Figure 3 This is a diagram illustrating an example of a prediction model f for SOH at a future time t according to the first embodiment; Figure 4It is a diagram showing the overall calculation process of prediction reliability according to the first embodiment; Figure 5 It is a diagram showing an example of the prediction reliability system according to the first embodiment; and Figure 6 It is a diagram showing another example of the prediction reliability system according to the first embodiment. Detailed Description of the Invention

[0016] First Embodiment Hereinafter, a battery degradation prediction system according to an embodiment will be described with reference to the accompanying drawings. Figure 1 It is a block diagram showing an example of the configuration of the degradation prediction system 1. The degradation prediction system 1 includes a State of Health (SOH) prediction unit 11, a reliability calculation unit 12, an adjustment determination unit 13, a model adjustment unit 14, and an additional prediction execution determination unit 15. The battery is a secondary battery installed in a vehicle, and the description will be given assuming that multiple batteries are usually installed in the vehicle.

[0017] The SOH prediction unit 11 predicts values by evaluating the state of the battery compared with the ideal state. In other words, the SOH prediction unit 11 can predict the health state and degradation of the battery.

[0018] More specifically, the SOH prediction unit 11 receives the SOH history x = [x(0),..., x(t')] of the target battery that satisfies t' < t at an arbitrary future time t as input. Then, the SOH prediction unit 11 predicts the future SOH y(t) at time t using the future SOH prediction model f as y(t) = f(x|θ). In this equation, θ is the parameter of the model f.

[0019] The reliability calculation unit 12 calculates the prediction reliability C(t) of the output y(t) of the SOH prediction model output by the SOH prediction unit 11 through the calculation result g(x, f) of the reliability calculation model g.

[0020] The adjustment determination unit 13 determines whether to perform model adjustment of the SOH prediction model f used in the SOH prediction unit 11 based on the result of the prediction reliability C(t) calculated by the reliability calculation unit 12.

[0021] The model adjustment unit 14 adjusts the balance between the adaptability and robustness of the SOH model f by changing the parameter θ.

[0022] The additional prediction execution determination unit 15 determines whether to perform further prediction using the new input x'.

[0023] Next, reference will be made to Figure 2 Describe the operation process of the degradation prediction system 1.

[0024] The SOH prediction unit 11 predicts a value by evaluating the state of the battery compared to an ideal state (step S1).

[0025] The reliability calculation unit 12 calculates the prediction reliability C(t) of the output y(t) of the SOH prediction model through the calculation result g(x,f) of the reliability calculation model g (step S2).

[0026] The adjustment determination unit 13 determines whether to perform model adjustment of the SOH prediction model f based on the result of the prediction reliability C(t) calculated by the reliability calculation unit 12 (step S3). When the adjustment determination unit 13 determines to perform adjustment (yes in step S3), the process proceeds to step S4. When the adjustment determination unit 13 determines not to perform adjustment (no in step S3), the process proceeds to step S5.

[0027] The model adjustment unit 14 adjusts the SOH model f (step S4). Then, the process proceeds to step S5.

[0028] The additional prediction execution determination unit 15 determines whether to perform an additional prediction using the new input x' (step S5). When it is determined to perform an additional prediction (yes in step S5), the process returns to step S1. When it is determined not to perform an additional prediction (no in step S5), the process ends.

[0029] Next, detailed operation examples in each of the steps shown will be described Figure 2 First, the calculation of the future SOH y(t) at time t by the SOH prediction unit 11 in step S1 will be described.

[0030] The SOH prediction unit 11 uses a secondary battery SOH prediction model based on the physical and chemical characteristics of battery degradation to predict the future SOH of the target battery. Figure 3 FIG. is an example of the prediction model f showing the SOH at a future time t.

[0031] As described above, the SOH prediction unit 11 receives, as input, the SOH history x = [x(0),..., x(t')] of the target battery satisfying t' < t at any future time t. Then, the SOH prediction unit 11 calculates the future SOH y(t) at time t through y(t) = f(x|θ) using the prediction model f. In this equation, θ is a parameter of the model f. At this time, the SOH prediction unit 11 can include battery-related information in the SOH history x as information other than the time series information of the actual values of the SOH itself, and this battery-related information is information on battery-related conditions that affect SOH degradation.

[0032] Battery-related information includes, but is not limited to, battery characteristics, characteristics of the vehicle containing the battery, driving environment and vehicle condition history of the vehicle containing the battery, and driving and charging operation history of the vehicle containing the battery.

[0033] Examples of battery characteristics include the battery's full-charge capacity, cathode material, anode material, and electrolyte material, as well as the manufacturer. Examples of characteristics of vehicles including batteries include vehicle model, performance such as average fuel efficiency, and vehicle weight. Examples of driving environment and vehicle condition history for vehicles including batteries include the frequency distribution of driving areas, state of charge (SOC) during driving, and battery temperature. Examples of driving and charging operation history for vehicles including batteries include the number of sudden accelerations and decelerations, and the ratio of fast charging to normal charging.

[0034] SOH prediction models f can be SOH prediction models based on electrode degradation behavior according to Arrhenius equations or the root law of degradation over time. This is a known technique for predicting SOH based on the physical and chemical characteristics of battery degradation.

[0035] The parameter θ used in model f can be pre-calculated or recalculated using SOH history x.

[0036] Next, an example of the reliability prediction calculated by the reliability calculation unit 12 in step S2 will be described. The following two cases will be described: the case where the degradation prediction system 1 can obtain the SOH history x=[x(0),…,x(t')] of multiple batteries, and the case where the degradation prediction system 1 cannot obtain the SOH history x=[x(0),…,x(t')] of multiple batteries.

[0037] First, a description will be given of the situation where the degradation prediction system 1 cannot obtain the SOH history of multiple cells (i.e., it can only obtain the SOH history of a single cell).

[0038] First, the reliability calculation unit 12 performs calculations on the inputs of the reliability system. Figure 4 This is a diagram illustrating an example of the overall calculation process for predicting reliability. (See diagram for example.) Figure 4 As shown, the reliability calculation unit 12 treats the past SOH history x=[x(0),…,x(t')] as a sequential input in order starting from the earliest time, and uses x up to a specific time (such as [x(0)], [x(0),x(1)], [x(0),x(1),x(2)],…,[x(0),…,x(t')]) to calculate the predicted output y_0(t),…,y_t'(t) of the prediction system at each time, as well as the deviation y_0(t)-x(t),…,y_t'(t)-x(t) of the model's predicted output at each time.

[0039] Next, the reliability calculation unit 12 calculates and predicts the reliability through the reliability system. Figure 5 This is a diagram illustrating an example of a predictive reliability system. (For example...) Figure 5 As shown, the reliability calculation unit 12 uses the deviations of the model's predicted output at each time, y_0(t)-x(t), ..., y_t'(t)-x(t), to calculate the expected degree of the prediction deviation between the model creation time and the future time used for prediction through methods such as regression, and sets it as the prediction reliability C(t) of the final output y(t) of the future SOH prediction.

[0040] Next, we will describe the situation where the degradation prediction system 1 can obtain the SOH history x=[x(0),…,x(t')] of multiple batteries.

[0041] First, the reliability calculation unit 12 performs calculations on the inputs of the reliability system. Specifically, the reliability calculation unit 12 performs the aforementioned calculations on the deviations of the model's predicted outputs at various times for each of the multiple batteries, as the reliability of a single battery.

[0042] That is, the reliability calculation unit 12 calculates the predicted output y_i=[y_0(t),…,y_t'(t)] of the prediction system for battery i and the deviation e_i=[y_0(t)-x(t),…,y_t'(t)-x(t)] of the predicted output of the model at each time for multiple batteries. When there is historical SOH data for n batteries, i=0,…,n-1.

[0043] Next, the reliability calculation unit 12 calculates the predicted reliability for n batteries using the predicted output y_i=[y_0(t),…,y_t'(t)] of the calculated prediction system and the deviation e_i=[y_0(t)-x(t),…,y_t'(t)-x(t)] of the predicted output of the calculated model at each time. Figure 6 This is a diagram illustrating another example of a predictive reliability system. (See diagram for example.) Figure 4 and Figure 6 As shown, the reliability calculation unit 12 receives n pairs (y_i, e_i) as input and calculates the expected degree of prediction deviation between the model creation time and the predicted future time using methods such as regression. Then, the reliability calculation unit 12 sets this as the prediction reliability C_i(t) of the final output y_i(t) of the future SOH prediction.

[0044] Therefore, the reliability calculation unit 12 can use the difference between the predicted SOH value of the target battery calculated using the prediction model and the actual measured SOH value to calculate the predicted reliability.

[0045] Next, an example of whether to perform model adjustment is determined by the adjustment determination unit 13 in step S3 will be described.

[0046] The adjustment decision unit 13 determines whether to perform model adjustment based on the predicted reliability C(t) output by the reliability calculation unit 12.

[0047] For example, when C_i(t) exceeds a preset threshold relative to y(t) (i.e., it is a poor state), the adjustment determination unit 13 determines to perform adjustment on the prediction model, regardless of whether the SOH history is obtained for a single cell or multiple cells in the processing of step S2.

[0048] For example, in the processing of step S2, if the SOH history can be obtained for multiple batteries, when C_i(t) is relatively large (i.e. relatively poor) compared to other batteries, the adjustment determination unit 13 can determine to perform adjustment on the battery prediction model. That is, the adjustment determination unit 13 can use the values ​​of C_i(t) of other batteries as thresholds for the target battery and make a determination based on whether these thresholds are exceeded.

[0049] Based on the above, in the adjustment determination unit 13, the calculation regarding whether the predicted reliability exceeds a predetermined threshold can be changed based on whether the actual measured SOH value of the battery has been obtained from multiple batteries.

[0050] Next, the model adjustment performed by the model adjustment unit 14 in step S4 will be described.

[0051] Model tuning unit 14 adjusts the balance between the fitness and robustness of model f by changing the parameter θ. Model tuning unit 14 uses knowledge of the physics of battery degradation (such as degradation patterns or root square laws) to adjust the trade-off between the fitness and robustness of the prediction model. The current time is represented by t_{now}, and the future time predicted by SOH is represented by t_{future}. The following feedback on the prediction reliability C(t) is then given as an example.

[0052] During parameter learning, model tuning unit 14 adjusts the weights of the learning data for each battery based on the battery degradation mode (early degradation, mid-term degradation, or late degradation) determined by t_{now}. For example, to improve adaptability, model tuning unit 14 increases the weight of battery data in degradation modes close to the prediction target (close to t_{now}) during parameter learning. To improve robustness, model tuning unit 14 decreases the weight of battery data in degradation modes close to the prediction target (close to t_{now}).

[0053] During parameter learning, for a range of t_{future} where the prediction reliability C(t) is large, the model conditioning unit 14 can improve robustness by fitting the prediction model to a general degradation curve (such as the root law) that does not depend on individual cells.

[0054] Therefore, when the prediction reliability exceeds a predetermined threshold in the judgment made by the adjustment judgment unit 13, the model adjustment unit 14 can adjust the prediction model.

[0055] Next, an example will be described of whether to perform additional prediction by the additional prediction execution determination unit 15 in step S5.

[0056] For example, as shown below, when a new input x' exists, the additional prediction execution decision unit 15 decides to execute additional prediction.

[0057] When the latest data is added to the original input x=[x(0),…,x(t')], the additional prediction execution decision unit 15 determines to execute the additional prediction, resulting in x'=[x(0),…,x(t'),x(t'+1)]. This is an assumption about the data accumulating over time.

[0058] When a new feature z is added to the original input x=[x(0),…,x(t')], the additional prediction execution decision unit 15 decides to perform additional prediction, resulting in x'=[x(0),…,x(t'),x(t'+1)]. This is an assumption about adding information from another database.

[0059] When a new battery SOH history is added to the original input x=[x(0),…,x(t')], the additional prediction execution decision unit 15 determines to execute the additional prediction, obtaining x'=[x,x_additional]. For example, this is an assumption for the case where later battery data is added.

[0060] Even if the original input x=[x(0),…,x(t')] is directly used as x'=x, the additional prediction execution decision unit 15 can still decide to execute additional prediction. This means that the input remains unchanged, and the adjusted output can be obtained using the adjusted model.

[0061] In this way, when the prediction reliability does not exceed a predetermined threshold in the determination made by the adjustment determination unit 13, or when the prediction reliability exceeds a predetermined threshold in the determination made by the adjustment determination unit 13 and the model adjustment unit 14 has adjusted the prediction model, the additional prediction execution determination unit 15 can determine whether to execute additional prediction.

[0062] Based on the above, degradation prediction system 1 can calculate prediction reliability based on SOH history and perform model adjustment based on prediction reliability. Degradation prediction system 1 can adjust the balance between model adaptability and robustness by changing the parameters used in the model.

[0063] This invention is not limited to the embodiments described above, and can be appropriately modified without departing from the spirit and scope of the invention. That is, for clarity, the above description has been appropriately omitted or simplified, and those skilled in the art can easily modify, add, or transform the various elements of the embodiments within the scope of this invention.

[0064] The embodiments of this disclosure can be implemented by hardware or dedicated circuitry, software, logic, or any combination thereof. Some solutions can be implemented by hardware, while others can be implemented by firmware or software, which can be executed by a controller, microprocessor, or any other computing device.

[0065] This disclosure also provides at least one computer program product tangibly stored in a non-transitory computer-readable storage medium. The computer program product includes computer-executable instructions (such as instructions contained in a program module that executes on a target real or virtual processor device) to perform the processes or methods of this disclosure. Program modules include routines, programs, libraries, objects, classes, components, data structures, etc., that perform specific tasks or implement specific abstract data types. In various embodiments, the functionality of program modules can be combined or split among program modules as needed. The machine-executable instructions in the program modules can execute in a local or distributed device. In a distributed device, the program modules can reside in both local and remote storage media.

[0066] Program code used to perform the methods of this disclosure may be written in any combination of more than one programming language. This program code is provided to a processor or controller of a general-purpose computer, a special-purpose computer, or any other programmable data processing device. When executed by the processor or controller, the functions and operations in the flowchart and / or implementation block diagram are performed. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0067] Various types of non-transitory computer-readable media (storage media) can be used to store programs and provide them to a computer. Non-transitory computer-readable media include various types of tangible recording media. Examples of non-transitory computer-readable media include magnetic recording media, magneto-optical recording media, optical disc media, and semiconductor memory. Examples of magnetic recording media include flexible discs, magnetic tapes, and hard disk drives. Examples of magneto-optical recording media include magneto-optical discs. Examples of optical disc media include Blu-ray discs, optical discs (CDs) - read-only memory (ROM), CD-Rs (recordable), and CD-R Ws (rewritable). Examples of semiconductor memory include solid-state drives, mask ROMs, programmable ROMs (PROMs), erasable PROMs (EPROMs), flash memory ROMs, and random access memory (RAM). Programs can also be provided to a computer via various types of transient computer-readable media. Examples of transient computer-readable media include electrical signals, optical signals, and electromagnetic waves. Transient computer-readable media can provide programs to a computer via wired communication paths (such as wires or optical fibers) or via wireless communication paths.

Claims

1. A battery degradation prediction system using a predictive model, the battery degradation prediction system comprising: A reliability calculation unit is configured to calculate predicted reliability using the difference between a predicted state of health value and an actual measured state of health value of the target battery, the predicted state of health value being calculated using the prediction model. as well as A model adjustment unit is configured to adjust the prediction model when the prediction reliability exceeds a predetermined threshold.

2. The battery degradation prediction system according to claim 1 further includes an additional prediction execution determination unit, which is configured to determine whether to execute additional prediction when the prediction reliability does not exceed the predetermined threshold, or when the prediction reliability exceeds the predetermined threshold and the prediction model has been adjusted.

3. The battery degradation prediction system according to claim 1 or 2 further includes an adjustment determination unit configured to perform a calculation regarding whether the predicted reliability exceeds the predetermined threshold, wherein... The adjustment determination unit is configured to change the calculation regarding whether the predicted reliability exceeds the predetermined threshold based on whether the actual measured health status value of the battery has been obtained from multiple batteries.

4. A battery degradation prediction method using a predictive model, the battery degradation prediction method comprising: The prediction reliability is calculated using the difference between the predicted state of health (SHS) value and the actual measured SHS value of the target battery, wherein the predicted SHS value is calculated using the prediction model; and When the prediction reliability exceeds a predetermined threshold, the prediction model is adjusted.

5. A battery degradation prediction program using a predictive model, the battery degradation prediction program enabling a computer to: The prediction reliability is calculated using the difference between the predicted state of health (SHS) value and the actual measured SHS value of the target battery, wherein the predicted SHS value is calculated using the prediction model; and When the prediction reliability exceeds a predetermined threshold, the prediction model is adjusted.

Images