Battery forecasting method and system

Abstract

The electric vehicle includes an electric motor for propelling the electric vehicle and a battery adapted to provide electrical energy to the electric motor to propel the electric vehicle. The electric vehicle also includes one or more controllers that are co-programmed with instructions that: measure battery data of the battery; creating a model of a storage capacity of the battery on a driving distance of the electric vehicle using the battery data; evaluating the reliability of the model; if the model is evaluated as sufficiently reliable, performing a forecast on the battery using the model; and if the model is evaluated to be not sufficiently reliable, employing an alternative model of the storage capacity of the battery to the travel distance of the electric vehicle, the alternative model being based on battery data of one or more other vehicles, and performing a forecast on the battery using the alternative model.

Description

[0001] Introduction

[0002] This disclosure pertains to the field of battery prediction.

[0003] Batteries are the source of electrical energy used to propel electric vehicles. Predicting premature or excessive degradation of the battery is useful, including for notifying vehicle owners in advance of premature or excessive degradation of the vehicle's battery and providing feedback to engineering and manufacturing departments for future improvements to the battery. Summary of the Invention

[0004] An electric vehicle includes: an electric motor for propelling the electric vehicle; and a battery adapted to supply electrical energy to the electric motor to propel the electric vehicle. The electric vehicle further includes one or more controllers that are jointly programmed with the following instructions: measure battery data of the battery; use the battery data to create a model of the storage capacity of the battery versus the driving distance of the electric vehicle; evaluate the reliability of the model; if the model is evaluated as being sufficiently reliable, use the model to perform a prediction on the battery; and if the model is evaluated as being insufficiently reliable, adopt an alternative model of the storage capacity of the battery versus the driving distance of the electric vehicle, the alternative model being based on battery data of one or more other vehicles, and use the alternative model to perform a prediction on the battery.

[0005] If the battery data has an insufficient number of data measurements, the model can be evaluated as being insufficiently reliable. Additionally, if the model exhibits an increasing driving distance versus storage capacity, the model can be evaluated as being insufficiently reliable. Still additionally, if the model exhibits noise, the model can be evaluated as being insufficiently reliable.

[0006] The one or more other vehicles can be selected based on having operating characteristics similar to those of the electric vehicle. These operating characteristics can include the average distance traveled per day. Alternatively or additionally, these operating characteristics can include the average ambient temperature.

[0007] The prediction can include comparing the predicted storage capacities between components of the battery. The prediction can alternatively or additionally include comparing the predicted storage capacity loss between components of the battery to a threshold. The prediction can alternatively or additionally include using the model or alternative model in conjunction with a measurement of the internal resistance of the battery.

[0008] The second electric vehicle includes an electric motor for propelling the electric vehicle and a battery adapted to supply electrical energy to the electric motor to propel the electric vehicle. The electric vehicle further includes one or more controllers that are jointly programmed with the following instructions: measure battery data of the battery; use the battery data to create a model of the storage capacity of the battery versus the time of use of the battery in the electric vehicle; evaluate the reliability of the model; if the model is evaluated as being sufficiently reliable, use the model to perform a prediction on the battery; and if the model is evaluated as being insufficiently reliable, adopt an alternative model of the storage capacity of the battery versus the time of operation of the electric vehicle, the alternative model being based on battery data of one or more other vehicles, and use the alternative model to perform a prediction on the battery.

[0009] In the second electric vehicle, if the battery data has an insufficient number of data measurements, the model can be evaluated as being insufficiently reliable. Additionally, if the model exhibits an increasing storage capacity versus the time of use of the battery in the electric vehicle, the model can be evaluated as being insufficiently reliable. Still additionally, if the model exhibits noise, the model can be evaluated as being insufficiently reliable.

[0010] In the second electric vehicle, the one or more other vehicles can be selected based on the one or more other vehicles having operating characteristics similar to the operating characteristics of the electric vehicle. These operating characteristics can include the average distance traveled per day. Alternatively or additionally, these operating characteristics can include the average ambient temperature.

[0011] In the second electric vehicle, the prediction can include comparing the predicted storage capacities between components of the battery. The prediction can alternatively or additionally include comparing the predicted loss of storage capacity between components of the battery to a threshold. The prediction can alternatively or additionally include using the model or alternative model in combination with a measurement of the internal resistance of the battery.

[0012] A method for predicting a battery for an electric vehicle includes measuring battery data of the battery by one or more controllers. The method further includes creating, by the one or more controllers, a model of the storage capacity of the battery versus the driving distance of the electric vehicle or a model of the storage capacity of the battery versus the usage time of the battery in the electric vehicle using the battery data. Additionally, the method includes: selecting the one or more other vehicles based on the one or more other vehicles having operating characteristics similar to the operating characteristics of the electric vehicle. Still additionally, the method includes: creating an alternative model of the storage capacity of the battery versus the driving distance of the electric vehicle or an alternative model of the storage capacity of the battery versus the usage time of the battery in the electric vehicle based on the battery data of the one or more other vehicles. The method further includes: using the model or the alternative model in combination with a measured value of the internal resistance of the battery as a tool for predicting the battery; and identifying one or more root causes of predicted degradation of the battery.

[0013] The present disclosure provides the following technical solutions:

[0014] Technical solution 1. An electric vehicle, comprising:

[0015] A motor for propelling the electric vehicle;

[0016] A battery adapted to supply electrical energy to the motor to propel the electric vehicle; and

[0017] One or more controllers jointly programmed with the following instructions:

[0018] Measure the battery data of the battery;

[0019] Create a model of the storage capacity of the battery versus the driving distance of the electric vehicle using the battery data;

[0020] Evaluate the reliability of the model;

[0021] If the model is evaluated as being sufficiently reliable, use the model to perform a prediction on the battery; and

[0022] If the model is evaluated as being insufficiently reliable, adopt an alternative model of the storage capacity of the battery versus the driving distance of the electric vehicle, the alternative model being based on the battery data of one or more other vehicles, and use the alternative model to perform a prediction on the battery.

[0023] Technical solution 2. The electric vehicle according to technical solution 1, further comprising selecting the one or more other vehicles based on the one or more other vehicles having operating characteristics similar to the operating characteristics of the electric vehicle.

[0024] Solution 3. The electric vehicle according to Solution 2, wherein the operating characteristic includes the average distance traveled per day.

[0025] Solution 4. The electric vehicle according to Solution 2, wherein the operating characteristic includes the average ambient temperature.

[0026] Solution 5. The electric vehicle according to Solution 1, wherein if the battery data has an insufficient number of data measurements, the model is evaluated as not reliable enough.

[0027] Solution 6. The electric vehicle according to Solution 1, wherein if the model exhibits an increased storage capacity for the driving distance or if the model exhibits noise, the model is evaluated as not reliable enough.

[0028] Solution 7. The electric vehicle according to Solution 1, wherein the prediction includes comparing the predicted storage capacities between the components of the battery.

[0029] Solution 8. The electric vehicle according to Solution 1, wherein the prediction includes comparing the predicted storage capacity loss between the components of the battery with a threshold value.

[0030] Solution 9. The electric vehicle according to Solution 1, wherein the prediction includes using the model or the alternative model in combination with the measured value of the internal resistance of the battery.

[0031] Solution 10. An electric vehicle, comprising:

[0032] An electric motor for propelling the electric vehicle;

[0033] A battery adapted to supply electrical energy to the electric motor to propel the electric vehicle; and

[0034] One or more controllers jointly programmed with the following instructions:

[0035] Measure the battery data of the battery;

[0036] Create a model of the storage capacity of the battery versus the usage time of the battery in the electric vehicle using the battery data;

[0037] Evaluate the reliability of the model;

[0038] If the model is evaluated as reliable enough, use the model to perform a prediction on the battery; and

[0039] If the model is evaluated as not being reliable enough, an alternative model of the usage time of the battery in the electric vehicle using the storage capacity of the battery is adopted. The alternative model is based on battery data of one or more other vehicles, and the alternative model is used to perform a prediction on the battery.

[0040] Technical solution 11. The electric vehicle according to technical solution 10 further includes selecting the one or more other vehicles based on the one or more other vehicles having operating characteristics similar to the operating characteristics of the electric vehicle.

[0041] Technical solution 12. The electric vehicle according to technical solution 11, wherein the operating characteristics include the average distance traveled per day.

[0042] Technical solution 13. The electric vehicle according to technical solution 11, wherein the operating characteristics include the average ambient temperature.

[0043] Technical solution 14. The electric vehicle according to technical solution 10, wherein if the battery data has an insufficient number of data measurement values, the model is evaluated as not being reliable enough.

[0044] Technical solution 15. The electric vehicle according to technical solution 10, wherein if the model shows an increase in the storage capacity for the usage time of the battery in the electric vehicle or if the model shows noise, the model is evaluated as not being reliable enough.

[0045] Technical solution 16. The electric vehicle according to technical solution 10, wherein the prediction includes:

[0046] comparing the storage capacities between components of the battery; and

[0047] identifying the battery as faulty based on the comparison.

[0048] Technical solution 17. The electric vehicle according to technical solution 10, wherein the prediction includes:

[0049] comparing the storage capacity losses between components of the battery; and

[0050] identifying the battery or its components as faulty based on the comparison.

[0051] Technical solution 18. The electric vehicle according to technical solution 10, wherein the prediction includes using the model or the alternative model in combination with the internal resistance measurement value of the battery.

[0052] Technical solution 19. A method for predicting a battery for an electric vehicle, the method comprising:

[0053] Measure battery data of the battery via one or more controllers;

[0054] Via one or more controllers, create a model of the storage capacity of the battery versus the driving distance of the electric vehicle or a model of the storage capacity of the battery versus the usage time of the battery in the electric vehicle using the battery data;

[0055] Select the one or more other vehicles based on the one or more other vehicles having operating characteristics similar to the operating characteristics of the electric vehicle;

[0056] Based on the battery data of the one or more other vehicles, create an alternative model of the storage capacity of the battery versus the driving distance of the electric vehicle or an alternative model of the storage capacity of the battery versus the usage time of the battery in the electric vehicle;

[0057] Use the model or the alternative model in combination with the measured internal resistance value of the battery for prediction of the battery; and

[0058] Identify one or more root causes of battery degradation predicted by the prediction.

[0059] Technical solution 20. The method according to technical solution 19, wherein the prediction includes:

[0060] Compare the storage capacity or storage capacity loss between components of the battery; and

[0061] Identify the battery as faulty based on the comparison.

[0062] The above summary of the invention does not represent every embodiment or every aspect of the present disclosure. When understood in conjunction with the drawings and the claims, the above features and advantages of the present disclosure, as well as other possible features and advantages, will be apparent from the following detailed description of the embodiments and best modes for carrying out the present disclosure. Additionally, the present disclosure expressly includes combinations and sub - combinations of the elements and features presented above and below. Brief Description of the Drawings

[0063] Figure 1 Illustrates an electric vehicle and a backend associated with the electric vehicle.

[0064] Figure 2 Illustrates Figure 1 the electrical system of a motor vehicle.

[0065] Figure 3 Illustrates a battery pack of an electric vehicle and a model of the capacity of the battery pack of a representative electric vehicle fleet versus vehicle mileage.

[0066] Figure 4Illustrated is a method for modeling battery capacity.

[0067] Figure 5 Illustrated are clusters into which a fleet of electric vehicles can be grouped based on operating characteristics of the electric vehicles.

[0068] Figure 6 Illustrated is a method for monitoring the internal resistance and capacity of batteries of multiple electric vehicles over time.

[0069] Figure 7 Illustrated are various root causes of battery failures. Detailed Description

[0070] The present disclosure may take many different forms of embodiments. Representative examples of the present disclosure are shown in the drawings and are described herein in detail as non-limiting examples of the disclosed principles. To this end, elements and limitations described in the abstract, introduction, summary of the invention, and detailed description sections, but not explicitly set forth in the claims, should not be incorporated into the claims singly or jointly by implication, inference, or otherwise.

[0071] For the purposes of this description, unless explicitly disclaimed, the use of the singular includes the plural and vice versa, the terms "and" and "or" shall be both conjunctive and disjunctive, "any" and "all" shall mean "any and all", and the words "comprising", "having", "including", "containing", etc. shall mean "including but not limited to". In addition, approximating words such as "about", "almost", "substantially", "generally", "approximate", etc. may be used herein in the sense of "at, close to, or almost at" or "within 0 - 5%", or "within acceptable manufacturing tolerances", or a logical combination of the foregoing.

[0072] First refer to Figure 1 , in which an electric vehicle 10 is illustrated. Also refer to Figure 2 , the electric vehicle 10 has an electrical system 12. The electrical system 12 includes a powertrain for the electric vehicle 10. The powertrain for the electric vehicle 10 includes an electric motor 14 for propelling the electric vehicle 10. The electric motor 14 may be an alternating current (AC) motor. The electrical system 12 also includes a battery pack 16. The battery pack 16 is coupled to a power inverter module (PIM) 18. The battery pack 16 provides propulsion energy to the electric motor 14 via switching performed within the PIM 18 controlled by a powertrain control unit (PCU) 20. The PCU 20 may be an independent controller, may be integrated with the PIM 18, or may be integrated with other controllers in the electrical system 12 of the electric vehicle 10.

[0073] The electric vehicle 10 can be any vehicle that uses electric energy for all parts of propulsion and can include a pure electric vehicle and a hybrid electric vehicle. In addition, the electric vehicle 10 can be any type of vehicle, such as a passenger car, a truck, a van, a sport utility vehicle, a motorcycle, a bicycle, a scooter, a boat, an airplane, etc.

[0074] The battery pack 16 can include a plurality of battery cells arranged in battery cell groups. Four such battery cell groups are illustrated, namely battery cell group 30, battery cell group 32, battery cell group 34, and battery cell group 36. The battery pack 16 can have any number of battery cells and battery cell groups. The battery cell groups can be grouped into battery modules, such as battery module 38 that includes battery cell group 30 and battery cell group 32, and battery module 40 that includes battery cell group 34 and battery cell group 36. Although two battery cell groups are illustrated in each of battery module 38 and battery module 40, this is for convenience only. A battery module can include any number of battery cell groups, and the battery pack 16 can have any number of battery modules. The battery cells, battery cell groups, and battery modules mentioned herein can be referred to as components of the battery pack 16.

[0075] The electrical system 12 can also include a battery management system (BMS) controller 22 for monitoring the battery pack 16. Specifically, the BMS controller 22 can monitor the state of charge of the battery pack 16 and the charging and discharging of the battery pack 16. Additionally, the BMS controller 22 can monitor the state of charge of each battery module (such as battery module 38) and its charging and discharging. Further additionally, the BMS controller 22 can monitor the state of charge of each battery cell group (such as battery cell group 30) and its charging and discharging. The BMS controller 22 can also monitor the internal resistance of the battery pack 16 and / or its battery cell groups and / or battery modules, such as by monitoring the charging and discharging current and voltage of the battery pack 16 and its battery cell groups and battery modules, where the internal resistance is the quotient of voltage divided by current.

[0076] The BMS controller 22 can be integrated within the battery pack 16 or can be separated from the battery pack 16. The BMS controller 22 is understood to have the connections required for monitoring the entire battery pack 16 and for monitoring each battery cell group and each battery module that the BMS controller 22 monitors. The BMS controller 22 is understood to have sufficient resources (e.g., a microcontroller, software, inputs, outputs, memory, etc.) to perform the tasks ascribed to it herein.

[0077] The electrical system 12 can have a temperature sensor 23 for measuring the ambient temperature near the electric vehicle 10. The temperature sensor 23 can be connected to the BMS controller 22 or to another controller in the electrical system 12.

[0078] In addition to the powertrain control unit 20 and the BMS controller 22, the electric vehicle 10 may also include several other controllers in the electrical system 12. For example, the electric vehicle 10 may include a telematics control unit (TCU) 26, which allows the electric vehicle 10 to participate in cellular remote communication, including cellular data communication. Such remote communication may include remote communication with a "back end" 11 operated by the manufacturer of the electric vehicle 10 (operated by General Motors is an example of such a back end). The TCU 26 may have or may be connected to a suitable antenna 27 that facilitates such communication. The electric vehicle 10 may also have any number of other controllers (generally illustrated as controller 28a, controller 28b, and controller 28n in Figure 2 ) for controlling other functions of the electric vehicle 10, such as braking, infotainment, steering, lighting, etc. The various controllers on the electric vehicle 10 may be networked together via one or more data buses 29 or via individual circuits, and the controllers can thus share data when performing their various tasks. In addition, the electrical system 12 of the electric vehicle 10 may be partitioned differently than shown in Figure 2 , where various tasks are performed by different or other controllers or shared with different or other controllers.

[0079] The various calculations, comparisons, and other tasks described hereinafter in the present disclosure may be performed by one controller or jointly by several controllers that may be networked together in the electrical system 12 of the electric vehicle 10. Each of these controllers is individually and / or jointly understood to have sufficient electronic resources (microprocessor, memory, input, output, software, cellular network access device (NAD)) to perform the functions described in the present disclosure. Some of the functions described herein may be performed by a computer outside the electric vehicle 10, such as a computer in the back end 11 or in the cloud.

[0080] One or several controllers may respond to an "instruction", which may include one or more software commands. In addition, an instruction may include one or more additional instructions.

[0081] Now refer to Figure 3 and Figure 4 , which illustrate a battery capacity modeling algorithm. The function of this algorithm may be to generate a model of the battery storage capacity of the battery pack 16 versus the driving distance (i.e., the odometer reading in, for example, miles or kilometers), represented by the curve 200.

[0082] At block 300, battery data, i.e., the storage capacity of the battery pack 16, is measured or calculated. The BMS controller 22 that monitors the charging and discharging of the battery pack 16 can know the current storage capacity of the battery pack 16, in ampere-hours, kilowatt-hours, or other suitable storage capacity units. This capacity can be continuously and periodically calculated. The storage capacities of the battery cell groups and battery modules within the battery pack 16 can also be continuously measured or calculated.

[0083] At block 301, telematics data can be collected via the TCU 26. The telematics data can be provided to the back-end 11. The telematics data can include the measured capacity of the battery pack 16 or its components (battery cell groups, battery modules). The telematics data can also include "good fit" data (from block 320). At block 302, it is determined whether the battery pack 16 has been replaced. At block 304, if the battery pack 16 has been replaced with a new battery pack, the number of months of use of the battery pack 16 and the driving distance of the electric vehicle 10 are updated. At block 306, any deviation generated by replacing the battery pack 16 with a new battery pack (the new battery pack has a known beginning-of-life (BOL) capacity within strict tolerances) is eliminated, and the algorithms described herein are reset / re-learned to reflect the replacement of the battery pack.

[0084] Then, at the start of the data classification section 307 of the algorithm, the measured or calculated storage capacity is used, where data regarding the capacity of the battery pack 16 can be processed, filtered, and classified. At block 308, the calculated or measured capacity data of the battery pack 16 is fitted to a regression model of storage capacity versus driving distance of the electric vehicle 10. At block 310, statistics of the regression model are calculated. These statistics can include slope, root mean square error, and r-squared statistic. Then, at block 312, it is determined whether the number of data points in the regression model is sufficient (i.e., above a predetermined threshold) for inferring that the model of the capacity of the battery pack 16 versus odometer reading can be expected to be reliable for characterizing the battery pack 16. As a non-limiting example, the number of data points judged to be "sufficient" can be two to three data points per month over a period of five or six months. If the inference at block 312 is "no" (i.e., the number of data points is insufficient), then the model of the capacity of the battery pack 16 versus odometer reading is determined at block 314 to be based on insufficient data and is labeled as a "premature fit data" model. If the inference at block 312 is that there are sufficient data points in the regression model, then the method proceeds to block 316. At block 316, the statistics from the model are tested to determine the reliability of the model, such as by comparing with a threshold suitable for the particular statistic being tested. For example, the r-squared of the model can be compared with a threshold, which can be 95%. Additionally, at block 316, it can be determined whether the slope of the modeling curve is negative or whether the slope of the modeling curve or a portion thereof is positive; the capacity of the battery decreases as the mileage increases, so if the data is "good", then the curve is expected to have a negative slope. If the statistic being tested for reliability does not pass its corresponding test (e.g., the r-squared value of the data generating the model is not higher than the threshold or the slope of the model curve is not negative or the slope of the modeling curve or a portion thereof is positive), then it can be inferred that the data generating the model is a poor fit (e.g., the data can be considered "noisy"). Thus, at block 318, the data can be labeled as "poor fit data". Clearly, the tests at blocks 312 and 316 can determine whether the battery data can be used to generate a sufficiently reliable model.

[0085] However, at block 316, if it is determined that the statistics being tested pass their respective tests (e.g., the r-squared of the data is above a threshold and the slope of the curve representing the model is reliably negative), then it can be inferred that the data used to generate the model is "good fit data" (i.e., the data is sufficient in quantity and quality to generate a reliable model of the battery pack 16), and the data can be so marked at block 320. Thus, at block 322, a final prediction of the storage capacity of the battery pack 16 with respect to the driving distance during the life of the battery pack 16 can be identified. This final prediction can be used for prognostics of the battery pack 16 (block 324). This can include prognostics regarding whether the battery pack 16 is likely to have a capacity less than expected during its life (block 326), and predicting the likelihood that the battery pack 16 is likely to be the subject of a warranty claim (block 328). This information can be sent to the backend 11 and can be used to communicate to the owner of the electric vehicle 10 that the battery pack 16 may have a shorter life than expected.

[0086] If the answer at block 316 is no, then at block 318 the data can be characterized as "poor fit" data.

[0087] Once the data from the electric vehicle 10 has been characterized as "good fit data", then the data from block 320 can be passed to block 332.

[0088] Further reference Figure 4 , and then the aggregation and regression prediction section 330 of the algorithm disclosed herein can be entered. There, starting at block 332, the "good fit data" from the vehicle fleet can be aggregated (here, an automaker that can access data from the vehicle fleet via the backend 11 may be able to use this data to significant advantage). Then the "good fit data" can be aggregated based on the average distance traveled per day by the vehicles in the fleet on the x-axis (e.g., in miles or kilometers) and the average ambient temperature at which the vehicles in the fleet operate on the y-axis. Figure 5 The aggregation of the vehicle population is illustrated at, where clusters 350, 352, 354, and 356 are illustrated. The "regression" used herein can be linear regression or other regression analysis.

[0089] At block 334, a regression model is generated for the "well - fit data" of each cluster. At block 336, if the battery pack 16 has an "early - fit data" or "poor - fit data" model, a model from the appropriate cluster can be selected as a "placeholder" or "substitute" for the actual model of the battery pack 16. If the battery pack 16 has an "early - fit data" model, the method proceeds to block 338, and the regression model for the appropriate cluster is temporarily used as the capacity - to - mileage model of the battery pack 16. However, if the battery pack 16 has a "poor - fit data" model, the method proceeds from block 336 to block 340 and the regression model for the appropriate cluster is temporarily used as the predicted - capacity - to - distance model of the battery pack 16. Thus, it may be temporarily assumed that since the battery pack 16 does not yet have its own "well - fit data" model, the model of the battery pack 16 can be comparable to the models of other vehicles in the same cluster. That is, it can be temporarily assumed that the capacity - to - miles - traveled model of the battery pack 16 of the electric vehicle 10 will be generally consistent with the "well - fit data" models of other vehicles operating under similar conditions.

[0090] Reference Figure 3 , illustrates examples of "poor - fit data" or "early - fit data" models and their correction ([ Figure 3 The x - axis of the graph of [ Figure 3 can be the distance traveled by the electric vehicle 10, as represented by miles or kilometers recorded on the odometer of the electric vehicle 10. Figure 3 The y - axis of the graph of [ Figure 3 can be the electrical storage capacity of the battery pack 16 in appropriate units (such as amp - hours, kilowatt - hours, or other appropriate units). First, it can be assumed that at the beginning of life ("BOL"), marked as "0K" (or zero thousand) miles or kilometers on the x - axis of [ Figure 3 , the capacity of the new battery pack is known within strict tolerances because the battery has just been manufactured and has not experienced any use or aging. As an example, the BOL capacity 210 of the battery pack 16 can be approximately 181 amp - hours, as illustrated in the exemplary example of [

[0091] In addition, the curve segment 214 (a segment of the "original" capacity model of the battery pack 16 before bias correction) and the curve segment 216 (a segment of the "original" capacity model of the battery pack 16 after bias correction) may exhibit noise. The battery storage capacity is expected to decrease over time by nature. However, the portions of the curve 200 illustrated by the portions 217a and 217b showing an upward (or positive) slope demonstrate noise in the capacity data of the battery pack 16. Such noise can be the basis for inferring that the model of the battery pack 16 is based on "poorly fitting data" (see also Figure 4 box 316).

[0092] Calculating the capacity-versus-mileage model based on the "poorly fitting data" or "prematurely fitting data" of the battery pack 16 of the electric vehicle 10 may result in the curve 200. However, substituting an alternative "well-fitting" data model based on a cluster of vehicles in a similar scenario may result in the curve 212. Using the curve 200 may lead to a premature prediction that the life of the battery pack 16 will be shortened (i.e., the life error amount 230 of the battery pack 16, e.g., because the long-term capacity degradation predicted according to the curve 200 is greater than that for the case of the curve 212). Alternatively, using the curve 200 may result in a premature prediction that the battery pack 16 will have more capacity degradation (i.e., the error amount 232 regarding the degradation amount, e.g., because the long-term capacity degradation predicted according to the curve 200 is greater than that for the case of the curve 212). Therefore, for the purpose of battery prediction, the model that results in the curve 212 can be adopted as an alternative model to the actual model of the battery pack 16 until the battery pack 16 has a model generated using "well-fitting data".

[0093] Although Figure 3 the graph is illustrated as storage capacity (y-axis) versus driving distance (x-axis), the model can alternatively or additionally be generated based on storage capacity versus the usage time of the battery pack 16 in the electric vehicle 10 (such as the number of months the battery pack 16 has been installed in the electric vehicle 10). Alternatively or additionally, the model can be generated based on storage capacity versus the operation time of the electric vehicle 10 in which the battery pack 16 is installed (i.e., the cumulative elapsed time during which the electric vehicle 10 has actually used the battery pack 16 for propulsion). The discussions in the present disclosure should be understood accordingly. In addition, the predictions of the model based on storage capacity versus driving distance described herein and the predictions of the model based on storage capacity versus one or both of the time metrics discussed in this paragraph can be used in combination; they are not mutually exclusive.

[0094] Once a model of the battery pack 16 has been formed using "well - fit data" or a temporary or alternative model (in the case of "over - fit data" or "poor - fit data"), it can be used for prognostics of the battery pack 16 and / or its battery cell groups and / or battery modules. Here, "prognostics" is used to refer to the prediction of the future performance or future condition of the battery pack 16 and / or its battery cell groups and / or battery modules. Of course, the first prognostics is provided by the capacity - versus - odometer - reading model itself (whether it is a model represented by curve 200 as in the case where it is derived from "well - fit data" or an alternative model such as the model represented by curve 212). This model predicts the storage capacity of the battery pack 16 with respect to the driving distance of the electric vehicle 10.

[0095] In the present disclosure, "prognostics" can also refer to predicting the capacity differences between the battery cell groups of the battery pack 16 and / or between the battery modules of the battery pack 16; for example, this can be used to isolate one or more battery cells or battery modules that are predicted to behave differently from each other due to one or more expected failure modes or failure mechanisms. "Prognostics" in the present disclosure can also refer to predicting which failure modes or failure mechanisms may be forming or may form in the battery pack 16 and / or its battery cell groups and / or battery modules.

[0096] Prognostics can be performed via one or more capacity - monitoring algorithms. One such algorithm can compare the capacity models of the individual battery cell groups within the battery pack 16 at any particular odometer reading. Various comparisons can be performed, such as the following:

[0097]

[0098]

[0099] where CG iis the battery cell group (“CG”) number of the battery pack 16, n is the total number of battery cell groups in the battery pack 16, “max” is the maximum capacity among the “n” battery cell groups, “min” is the minimum capacity among the “n” battery cell groups, “avg” is the average capacity of all “n” battery cell groups, and “median” is the median capacity of all “n” battery cell groups. If one or more of the values in the above table are higher than the threshold, a flag for the predicted loss of storage capacity for the battery pack 16 can be set, and the failure of the (one or more) suspect battery cell groups can be predicted (e.g., over time, having a shorter life than the designed life or a greater loss of storage capacity than the designed storage capacity loss). Additionally, for the purpose of forecasting the (one or more) root causes related to the predicted storage capacity loss, the (one or more) suspect battery cell groups can be identified. The algorithm can be used with reference to any specific future odometer reading. As an illustration, row (a) in Table 1 illustrates the calculation of the difference between the maximum estimated capacity of a battery cell group and the minimum estimated capacity of the battery cell group, and compares this difference with the threshold. Row (b) in Table 1 illustrates the calculation of the difference between the maximum estimated capacity of a battery cell group and the average or median of the estimated capacities of all “n” battery cell groups, and compares this difference with the threshold. Row (c) in Table 1 illustrates the calculation of the difference between the average or median of the estimated capacities of all “n” battery cell groups and the minimum estimated capacity of the battery cell group, and compares this difference with the threshold.

[0100] Instead of performing the aforementioned forecasting at the battery cell group level, the aforementioned forecasting can be performed at the battery module level (e.g., between battery module 38 and battery module 40 and any other battery modules within the battery pack 16). For the purpose of forecasting at the battery cell group level, the BMS controller 22 should be understood to have an electrical connection within the battery pack 16 at the battery cell group level. For the purpose of forecasting at the module level, the BMS controller 22 should be understood to have an electrical connection within the battery pack 16 at the module level.

[0101] Another algorithm that can be employed can use the predicted estimated capacity loss (which can also be referred to as estimated capacity fade) for an individual battery cell group. In this case, the percentage of capacity fade over time can be calculated as follows:

[0102]

[0103] where CG i _Capacity(t) is the capacity of the battery cell group CG i at any selected time t, and CG i _Capacity(0) is the capacity of the battery cell group CGi Capacity at time 0. If any capacity degradation calculation result is higher than the threshold, then one or more flags can be set for the suspect battery cell group(s) and the suspect battery cell group(s) is / are identified. Such battery cell group(s) may be expected to fail (e.g., over time, having a shorter life than the design-specified life or a greater storage capacity loss than the design-specified storage capacity loss).

[0104] Instead of making the above prediction at the battery cell group level, the above prediction can be made at the module level (e.g., between battery module 38 and battery module 40 and any other battery modules within battery pack 16). The BMS controller 22 should be understood to have connections within battery pack 16 to have electrical connections at the module level within battery pack 16.

[0105] Monitoring the capacity of battery pack 16 as described above herein can also be combined with monitoring the resistance of battery pack 16 over time, which is an additional prediction tool (such resistance monitoring can be performed at the battery cell group, module, or battery pack level, can be performed periodically, and extrapolate the data into the future to create a resistance vs. odometer reading model or a resistance vs. time model). For example, Figure 6 illustrates the use of three vehicles, namely vehicle 400, vehicle 402, and vehicle 404. The prediction made via the combined algorithm 410 of the predicted battery cell group resistance over time and the predicted battery cell group capacity over time can show that, over a comparable time range, one or more battery cell groups of the battery pack of electric vehicle 400 can show a more significant increase in resistance over time and / or a more significant decrease in capacity over time (see the graphs 450 and 452 of resistance (R) vs. time and storage capacity (C) vs. time, as compared to the battery pack of vehicle 402 (graphs 454 and 456) and the battery pack of vehicle 404 (graphs 458 and 460)). This can be used as a prediction tool, where the Figure 6 curves shown in are used to isolate the root cause of the predicted battery degradation through physics-based analysis or machine learning. Figure 6 The curves shown in can be compared with Figure 7 the curves of Figure 7 which may have been generated based on actual laboratory diagnoses of the root causes of failures in the batteries of a vehicle fleet. Such root causes may, for example, be a high moisture content in the battery electrolyte ( Figure 7 curves 530a and 530b in) as compared to the baseline for the vehicle and battery population, Figure 7 curves 500a and 500b in), metal contamination of the battery ( Figure 7 curves 510a and 510b in), loss of battery electrolyte (Figure 7 the curves 520a and 520b) in. The lithium plating of the battery electrodes can also be a root cause of battery degradation. Recall that a vehicle manufacturer that has a large number of its customer vehicles on the road and can access data about the vehicles via a telecommunication network including the backend 11 may be able to access a large amount of data. These analyses of the root causes of battery degradation can be fed back to the engineering and manufacturing departments responsible for the battery pack 16 for future design and manufacturing improvements as needed.

[0106] Embodiments of the present disclosure are described herein. However, it should be understood that the disclosed embodiments are merely examples, and other embodiments may take various and alternative forms. The drawings are not necessarily to scale; some features may be enlarged or minimized to show details of particular components. Accordingly, the specific structural and functional details disclosed herein should not be construed as limiting, but merely as a representative basis for teaching one skilled in the art to employ the present disclosure in various ways.

[0107] In addition, the features of the embodiments shown in the drawings or various embodiments mentioned in this specification need not be understood as independent embodiments of each other. On the contrary, it is possible that each feature described in one of the examples of an embodiment can be combined with one or more other desired features from other embodiments, resulting in other embodiments not described in words or with reference to the drawings. Accordingly, such other embodiments fall within the framework of the scope of the appended claims. In addition, the present disclosure expressly includes combinations and sub - combinations of the elements and features presented above and below.

Claims

1. An electric vehicle, comprising: an electric motor for propelling the electric vehicle; a battery adapted to provide electrical energy to the electric motor to propel the electric vehicle; and One or more controllers are programmed together using the following instructions: measuring battery data of the battery; creating a model of storage capacity of the battery versus driving range of the electric vehicle using the battery data; Assess the reliability of the model; If the model is assessed to be sufficiently reliable, performing a forecast on the battery using the model; and If the model is assessed to be insufficiently reliable, a surrogate model of the storage capacity of the battery versus the driving range of the electric vehicle is employed, the surrogate model being based on battery data of one or more other vehicles, and a forecast is performed for the battery using the surrogate model. 2 . The electric vehicle of claim 1 , further comprising selecting the one or more other vehicles based on the one or more other vehicles having operating characteristics similar to operating characteristics of the electric vehicle.

3. The electric vehicle of claim 2, wherein the operating characteristic comprises an average distance driven per day. The electric vehicle of claim 2 , wherein the operating characteristic comprises an average ambient temperature. 5 . The electric vehicle of claim 1 , wherein the model is evaluated as not being sufficiently reliable if the battery data has an insufficient number of data measurements.

6. The electric vehicle of claim 1, wherein the model is evaluated as not sufficiently reliable if the model exhibits increased storage capacity versus driving range or if the model exhibits noise.

7. The electric vehicle of claim 1, wherein the forecasting includes comparing projected storage capacities between component parts of the battery.

8. The electric vehicle of claim 1, wherein the forecasting comprises comparing a predicted loss of storage capacity between components of the battery to a threshold value.

9. The electric vehicle of claim 1, wherein the predicting comprises using the model or the surrogate model in conjunction with a measurement of an internal resistance of the battery.

10. An electric vehicle comprising: an electric motor for propelling the electric vehicle; a battery adapted to provide electrical energy to the electric motor to propel the electric vehicle; and One or more controllers are programmed together using the following instructions: measuring battery data of the battery; using the battery data to create a model of the storage capacity of the battery versus the time the battery is used in the electric vehicle; Assess the reliability of the model; If the model is assessed to be sufficiently reliable, performing a forecast on the battery using the model; and If the model is assessed to be insufficiently reliable, a surrogate model of the storage capacity of the battery versus the usage time of the battery in the electric vehicle is employed, the surrogate model being based on battery data of one or more other vehicles, and a forecast is performed for the battery using the surrogate model.

Images