Method for predicting battery life

Abstract

The present disclosure provides "a method for predicting battery life". Methods and systems are provided for reliably providing a prognosis of the expected life of a vehicle battery. The degradation state of the battery is predicted based on the rate of convergence of a metric derived from sensed vehicle operating parameters towards a defined threshold, which is determined based on the past history of the metric. The predicted degradation state is then converted into an estimate of the remaining time or distance before a component needs to be serviced and displayed to the vehicle driver. Vehicle control and communication strategies can be defined with respect to the predicted degradation state.

Description

Technical Field

[0001] The present application relates to a method for estimating the remaining life of a battery that is commonly implemented to start a vehicle and support electrical loads when the ignition device is turned off, which is performed in conventional motor vehicles, hybrid vehicles, and electric vehicles. Such a battery is commonly referred to in the literature as a starting-lighting-ignition battery (or SLI-battery). The basic method described herein can also be applied to propulsion batteries used in hybrid or electric vehicles based on Li-ion or other technologies.

[0002] Background Art and Summary of the Invention

[0003] A vehicle engine includes an energy storage device, such as a lead-acid battery, to power a starter motor and support electrical load transients. Such a battery is commonly referred to in the literature as an SLI battery. A propulsion battery may also be included in a hybrid electric vehicle to power a motor coupled to the driveline. Batteries deteriorate over time and must be repaired or replaced. The rate of battery deterioration can be affected by multiple parameters, such as battery usage rate, battery life, temperature conditions, battery properties, etc.

[0004] Various methods have been developed to predict the health state of a vehicle system battery. An example method is shown by Uchida in US 8,676,4825. wherein the health of the battery of a hybrid vehicle is predicted based on a decrease in the fuel economy of the vehicle. Another example method is shown by Kozlowski et al. in US 20030184307. wherein the health state of the system battery is predicted based on the frequency of battery charging and discharging and its impact on battery parameters (such as impedance, electrolyte state, etc.). The battery health state is then indicated in terms of the remaining useful cycle count.

[0005] However, the present inventors have identified various problems with these methods. As an example, there may be various mechanisms that affect battery health, some of which are interdependent while others are independent of each other. The example methods discussed above may not take into account different battery characteristics that are affected at different rates based on vehicle operating conditions, which affect the battery health state. For example, during conditions that increase battery corrosion, the internal resistance of the battery may be more significantly affected, while during conditions that increase battery sulfation or loss of active material, the total capacity of the battery may be more significantly affected. Additionally, the effects of some of these conditions may be reversible. For example, the level of battery sulfation can be increased or decreased. Further, some characteristics may be more significantly affected by temperature while other characteristics are temperature independent. If the end-of-life of the battery is predicted without considering all of these effects, then it may be necessary to service the functional battery earlier than expected. Or, it may not be possible to send the degraded battery for service in a timely manner. In either case, driver satisfaction may be reduced due to the battery not starting, the resulting loss of mobility, and reduced electrical functions that affect driving performance (such as loss of power steering or electric assist for the electric power brakes).

[0006] These problems may be further exacerbated in autonomous vehicles, where the battery (e.g., a 12V SLI battery) supports the basic operating system and safety-critical systems in certain operating modes. Wherein if the battery is defective or approaching its end-of-life, then the autonomous functions of the vehicle can be disabled or scaled back. Suddenly disabling the autonomous functions can lead to customer dissatisfaction and inconvenience, especially in cases where the vehicle cannot be driven manually.

[0007] In one example, some of the above problems can be solved by a method for coupling to a vehicle's battery, the method comprising: predicting a vehicle degradation state of a plurality of battery metrics derived from sensed vehicle operating parameters relative to a corresponding threshold, the threshold being determined based on past driving history data, the past driving history data including the past history of each of the plurality of battery metrics; and converting the predicted degradation state into a remaining time or duration to be displayed to the vehicle driver based on a rate of convergence towards a threshold defining end-of-life. In this way, the remaining useful life of the vehicle battery can be predicted more accurately, and information can be communicated to the vehicle driver in a timely manner.

[0008] As an example, a vehicle system can include a battery that uses statistical and experimental methods to predict its end of life. Based on the nature of the battery (e.g., based on the battery's chemical composition), a plurality of battery attributes (such as a subset of all battery attributes) that can be monitored to measure battery degradation can be identified. Additionally, a method for online measuring selected battery attributes during vehicle operation can be determined, the method including: identifying the required battery sensors; sampling the frequency; and calculating the algorithms required for the signals or metrics. For example, in the case of a lead-acid battery, at least the battery internal resistance and the battery capacity can be monitored, the battery internal resistance being measured based on the changes in the battery voltage and current during vehicle operation, and the battery capacity being based on the battery internal resistance at a low state of charge and / or the changes in the minimum achievable open circuit voltage (OCV) and the maximum achievable open circuit voltage calculated when the battery is fully discharged and fully charged. The measured characteristics can be normalized to temperature and the state of charge of the battery (SOC) to account for the different effects of temperature on each battery characteristic at a given state of charge. Then, thresholds can be defined offline for each of the selected battery attributes based on statistical parameterization methods, the past history of the given battery, the vehicle driving history, the battery repair history, fleet data, etc. The thresholds can be temperature-independent calibrated thresholds that are also normalized to a predefined temperature and state of charge (e.g., normalized to 25 °C and 100% SOC). Based on the rate of convergence of the measured battery attributes towards their respective thresholds, the end of life of the battery can be predicted. The remaining battery life can then be displayed to the vehicle driver as the remaining number of vehicle miles driven, the remaining number of vehicle starts, the remaining number of fuel tank refueling events, etc.

[0009] In this way, the remaining life of a vehicle battery can be accurately predicted without relying on computationally intensive algorithms. By using the data sensed on the vehicle, associated with vehicle and fleet driving statistics, the state of health of the battery can be calculated more accurately. For example, the internal resistance and capacitance of the battery can be better determined by considering the temperature effect. The technical effect of estimating the end of battery life based on thresholds defined for each battery characteristic and the trajectory of each battery characteristic towards the corresponding threshold, taking into account different mechanisms of battery degradation, is considered. For example, it is considered that battery degradation due to corrosion effects is different from battery degradation due to sulfation effects, and the overall battery health can be calculated more reliably. By converting the sensed state of health into an estimate of the remaining time or duration of vehicle operation before component repair is required, the vehicle driver can be better informed about the condition of the component. Thus, timely component repair can be ensured, thereby improving vehicle performance. Additionally, in cases where battery characteristics are affected by operating driving behavior, timely notification can enable the driver to adjust their driving behavior to extend battery life. By predicting the remaining life of vehicle components through recursive estimation of statistical features, the remaining life of components can be predicted with a lower computational intensity without sacrificing the accuracy of the prediction. This enables a margin to be provided to better ensure the healthy operation of components within the estimated remaining life. The prognostic feature can provide an early indication of the remaining life of the battery to help customers plan maintenance in advance and avoid component failures. Additionally, the convenience of online estimation can be provided in an easily implementable package.

[0010] It should be understood that the above invention content provides a selection of some concepts further described in the detailed implementation in a simplified form. This does not mean identifying the key or essential features of the claimed subject matter, the scope of which is uniquely defined by the claims after the detailed description. Furthermore, the claimed subject matter is not limited to implementations that solve any disadvantages noted above or in any part of this disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1A An example embodiment of a vehicle low voltage power supply is schematically shown.

[0012] Figure 1B An example embodiment of a power supply system is shown, the power supply system including a conventional 12V-based alternator with a single lead acid battery and an associated battery monitoring sensor.

[0013] Figure 2 An example battery life end estimation curve for a vehicle battery is shown.

[0014] Figure 3 A high-level flowchart for performing prognostics and diagnostics of a vehicle battery using sensed data and statistical estimation is shown.

[0015] Figure 4 A high - level flowchart showing an example method that can be used to predict the remaining life of a vehicle battery.

[0016] Figure 5 An example adjustment of the EOL threshold for battery resistance during battery life - end prediction is shown.

[0017] Figure 6 An example determination of the convergence rate of battery characteristics towards their thresholds is shown.

[0018] Figure 7 A high - level flowchart showing an example method that can be used to update the EOL threshold based on vehicle performance metrics.

[0019] Figure 8 An example intermediate EOL threshold that can be implemented during battery life - end prediction is shown.

[0020] Figure 9 A high - level flowchart of an exemplary method that can be used to update a notification provided to a vehicle driver and limit autonomous vehicle functions when an intermediate EOL threshold is exceeded.

[0021] Figure 10 An example adjustment of the EOL threshold for battery capacity during battery life - end prediction is shown.

[0022] Figure 11 A high - level flowchart showing an example method that can be used to update the battery charge voltage in response to exceeding an intermediate EOL threshold.

[0023] Figure 12 An example pre - diagnosis architecture that can be used for data processing and threshold calculation is shown.

[0024] Figure 13 An example communication strategy for a faulty battery is shown.

[0025] Figure 14 A general communication strategy for a faulty battery is shown.

[0026] Figure 15 An example mode - limiting control strategy for an autonomous vehicle is shown.

[0027] Figure 16 A control strategy that uses the EOL criteria of a battery to limit the electro - hydraulic roll control function of a vehicle is shown. Detailed Description

[0028] The following description relates to systems and methods for predicting the remaining life of a battery of a vehicle system, such as including Figure 1B of the battery architecture Figure 1AAn example vehicle system. The vehicle controller may be configured to execute control programs, such as Figures 3 to 4 an example program, to use statistical and measurement data to predict the remaining life of a vehicle battery. The controller may compare the estimated battery characteristics with corresponding thresholds, as Figure 2 shown. Additionally, the controller may estimate the rate of convergence of the estimated battery characteristics towards the corresponding thresholds, as Figure 6 shown, to predict the end of life of the battery. The thresholds may be calibrated based on data received from a given vehicle and other vehicle data, as Figure 5 、 Figure 7 、 Figure 10 and Figure 12 explained. Additionally, multiple intermediate thresholds may be selected before the EOL. Various notifications may be sent, and control actions may be taken when each intermediate threshold is exceeded, as Figures 8 to 9 、 Figure 11 and Figures 13 to 16 explained. In this way, regular battery maintenance can be better ensured, and battery warranty issues can be reduced.

[0029] Figure 1A FIG. shows an example of a combustion chamber or cylinder of an internal combustion engine 10. The engine 10 may be coupled in a propulsion system (such as vehicle system 5) for on-road driving. In one example, the vehicle system 5 may be a hybrid electric vehicle system.

[0030] The engine 10 may be at least partially controlled by a control system including a controller 12 and an input from a vehicle driver 130 via an input device 132. In this example, the input device 132 includes an accelerator pedal and a pedal position sensor 134 for generating a proportional pedal position signal PP. A cylinder 14 (also referred to herein as a "combustion chamber") of the engine 10 may include a combustion chamber wall 136 within which a piston 138 is located. The piston 138 may be coupled to a crankshaft 140 such that the reciprocating motion of the piston is transformed into rotational motion of the crankshaft. The crankshaft 140 may be coupled via a transmission system to at least one drive wheel of a passenger vehicle. Additionally, a starter motor (not shown) may be coupled to the crankshaft 140 via a flywheel to effect a starting operation of the engine 10.

[0031] The cylinder 14 may receive intake air via a series of intake passages 142, 144, and 146. The air received via the intake passage 142 may be filtered via an air filter 135 before the air moves into the air passages 144, 146. In addition to the cylinder 14, the intake passage 146 may also communicate with other cylinders of the engine 10. In some examples, one or more of the intake passages may include a boosting device, such as a turbocharger or a supercharger. For example, Figure 1AAn engine 10 configured with a turbocharger is shown. The turbocharger includes a compressor 174 disposed between intake passages 142 and 144, and an exhaust turbine 176 disposed along an exhaust passage 148. The compressor 174 can be powered at least in part by the exhaust turbine 176 via a shaft 180, where the supercharging device is configured as a turbocharger. However, in other examples, such as when the engine 10 is provided with a supercharger, the exhaust turbine 176 can optionally be omitted, where the compressor 174 can be powered by a mechanical input from a motor or an engine. A throttle valve 162 including a throttle plate 164 can be disposed along the intake passage of the engine to vary the flow rate and / or pressure of the intake air provided to the engine cylinders. For example, the throttle valve 162 can be positioned downstream of the compressor 174, as Figure 1A shown, or alternatively, can be provided upstream of the compressor 174.

[0032] In addition to cylinder 14, the exhaust passage 148 can also receive exhaust from other cylinders of the engine 10. An exhaust sensor 128 is shown coupled to the exhaust passage 148 upstream of an emissions control device 178. The sensor 128 can be selected from a variety of suitable sensors for providing an indication of the exhaust air-fuel ratio, such as a linear oxygen sensor or UEGO (universal or wide-range exhaust gas oxygen), a two-state oxygen sensor or EGO (as shown), a HEGO (heated EGO), a NOx, HC, or CO sensor. The emissions control device 178 can be a three-way catalyst (TWC), a NOx trap, various other emissions control devices, or a combination thereof.

[0033] Each cylinder of the engine 10 can include one or more intake valves and one or more exhaust valves. For example, cylinder 14 is shown as including at least one intake lift valve 150 and at least one exhaust lift valve 156 located in the upper region of cylinder 14. In some examples, each cylinder of the engine 10 (including cylinder 14) can include at least two intake lift valves and at least two exhaust lift valves located in the upper region of the cylinder.

[0034] The intake valve 150 can be controlled by the controller 12 via the actuator 152. Similarly, the exhaust valve 156 can be controlled by the controller 12 via the actuator 154. In some cases, the controller 12 can change the signals provided to the actuators 152 and 154 to control the opening and closing of the corresponding intake and exhaust valves. The positions of the intake valve 150 and the exhaust valve 156 can be determined by respective valve position sensors (not shown). The valve actuators can be of the electric valve actuation type or the cam actuation type, or a combination thereof. The intake and exhaust valve timings can be controlled simultaneously, or any possibility of variable intake cam timing, variable exhaust cam timing, dual independent variable cam timing, or fixed cam timing can be used. Each cam actuation system can include one or more cams and can utilize one or more of a cam profile switching (CPS), variable cam timing (VCT), variable valve timing (VVT), and / or variable valve lift (VVL) system operable by the controller 12 to change valve operation. For example, cylinder 14 can optionally include an intake valve controlled by electric valve actuation and an exhaust valve controlled by cam actuation including CPS and / or VCT. In other examples, the intake and exhaust valves can be controlled by a common valve actuator or actuation system, or a variable valve timing actuator or actuation system.

[0035] Cylinder 14 can have a compression ratio that is the volume ratio when the piston 138 is at bottom center to top center. In one example, the compression ratio is in the range of 9:1 to 10:1. However, in some examples using different fuels, the compression ratio can be increased. For example, this can occur when using a fuel with a higher octane number or a fuel with a higher latent heat of vaporization. If direct injection is used due to its effect on engine knock, the compression ratio may also increase.

[0036] In some examples, each cylinder of the engine 10 can include a spark plug 192 for initiating combustion. In a selected operating mode, the ignition system 190 can provide an ignition spark to the combustion chamber 14 via the spark plug 192 in response to a spark advance signal SA from the controller 12. However, in some embodiments, the spark plug 192 can be omitted, such as in the case where the engine 10 can initiate combustion by auto-ignition or by injecting fuel, as in the case of certain diesel engines.

[0037] In some examples, each cylinder of engine 10 may be configured with one or more fuel injectors for supplying fuel thereto. As a non-limiting example, cylinder 14 is shown to include two fuel injectors 166 and 170. Fuel injectors 166 and 170 may be configured to deliver fuel received from fuel system 8. Fuel system 8 may include one or more fuel tanks, a fuel pump, and a fuel rail. Fuel injector 166 is shown to be directly coupled to cylinder 14 for directly injecting fuel therein in proportion to the pulse width of signal FPW-1 received from controller 12 via electronic driver 168. In this manner, fuel injector 166 provides so-called direct injection (hereinafter referred to as "DI") fuel into combustion cylinder 14. Although Figure 1A injector 166 is shown positioned to one side of cylinder 14, it may alternatively be located at the top of the piston, such as in a location near spark plug 192. Due to the lower volatility of some alcohol-based fuels, such a location may improve mixing and combustion when operating the engine with an alcohol-based fuel. Alternatively, the injector may be located at the top of the intake valve and near the intake valve to improve mixing. Fuel may be delivered from the fuel tank of fuel system 8 to fuel injector 166 via a high-pressure fuel pump and a fuel rail. Additionally, the fuel tank may have a pressure sensor that provides a signal to controller 12.

[0038] Fuel injector 170 is shown disposed in intake passage 146 rather than in cylinder 14, and is configured to provide so-called port fuel injection (hereinafter referred to as "PFI") into the intake passage upstream of cylinder 14. Fuel injector 170 may inject fuel received from fuel system 8 in proportion to the pulse width of signal FPW-2 received from controller 12 via electronic driver 171. Note that a single driver 168 or 171 may be used for both fuel injection systems, or multiple drivers may be used, such as driver 168 for fuel injector 166 and driver 171 for fuel injector 170, as shown.

[0039] In an alternative example, each of fuel injectors 166 and 170 may be configured as a direct fuel injector for directly injecting fuel into cylinder 14. In yet another example, each of fuel injectors 166 and 170 may be configured as an intake port fuel injector for injecting fuel upstream of intake valve 150. In other examples, cylinder 14 may include only a single fuel injector that is configured to receive different fuels from the fuel system as a fuel mixture in different relative amounts and is also configured to directly inject the fuel mixture into the cylinder as a direct fuel injector or directly inject the fuel mixture upstream of the intake valve as an intake port fuel injector. Thus, it should be understood that the fuel system described herein should not be limited to the specific fuel injector configurations described herein by way of example.

[0040] During a single cycle of the cylinder, fuel may be delivered to the cylinder by two injectors. For example, each injector may deliver a portion of the total fuel injection that is combusted in cylinder 14. Additionally, the distribution and / or relative amount of fuel delivered from each injector may vary with operating conditions such as engine load, knock, and exhaust temperature, as described below. Fuel injected at the intake port may be delivered during an open intake valve event, a closed intake valve event (e.g., substantially prior to the intake stroke), and during open and closed intake valve operations. Similarly, for example, fuel injected directly may be delivered during the intake stroke, may also be delivered in part during a previous exhaust stroke, during the intake stroke, and in part during the compression stroke. Thus, even for a single combustion event, fuel injected from the intake port and the direct injector may be injected at different timings. Additionally, for a single combustion event, multiple injections of the fuel delivered may be performed per cycle. The multiple injections may be performed during the compression stroke, the intake stroke, or any suitable combination thereof.

[0041] Fuel injectors 166 and 170 may have different characteristics. These include size differences, e.g., one injector may have larger injection holes than the other. Other differences include but are not limited to different spray angles, different operating temperatures, different aimings, different injection times, different spray characteristics, different positions, etc. Additionally, different effects may be achieved depending on the distribution ratio of the fuel injected between injectors 170 and 166.

[0042] The fuel tank in the fuel system 8 can hold fuels of different fuel types, such as fuels with different fuel qualities and different fuel compositions. The differences can include different alcohol contents, different water contents, different octane numbers, different heat of vaporization, different fuel mixtures, and / or combinations thereof, etc. An example of a fuel with a different heat of vaporization can include gasoline as a first fuel type with a lower heat of vaporization and ethanol as a second fuel type with a higher heat of vaporization. In another example, the engine can use gasoline as the first fuel type and an alcohol-containing fuel mixture, such as E85 (which is approximately 85% ethanol and 15% gasoline) or M85 (which is approximately 85% methanol and 15% gasoline) as the second fuel type. Other viable substances include water, methanol, mixtures of alcohol and water, mixtures of water and methanol, mixtures of alcohols, etc.

[0043] In yet another example, the two fuels can be alcohol mixtures with different alcohol compositions, where the first fuel type can be a gasoline-alcohol mixture with a lower alcohol concentration, such as E10 (which is approximately 10% ethanol), while the second fuel type can be a gasoline-alcohol mixture with a higher alcohol concentration, such as E85 (which is approximately 85% ethanol). Additionally, the first fuel and the second fuel may also differ in other fuel qualities, such as differences in temperature, viscosity, octane number, etc. Furthermore, the fuel characteristics of one or both fuel tanks may change frequently, for example, due to daily variations in refueling the tank.

[0044] The controller 12 is shown in Figure 1A as a microcomputer, which includes a microprocessor unit 106, input / output ports 108, an electronic storage medium for executable programs and calibration values (shown in this particular example as a non-transitory read-only memory chip 110 for storing executable instructions), a random access memory 112, a keep-alive memory 114, and a data bus. In addition to the signals previously discussed, the controller 12 can also receive various signals from sensors coupled to the engine 10, including measurements of the intake mass air flow (MAF) from the mass air flow sensor 122; the atmospheric pressure from the BP sensor 137; the engine coolant temperature (ECT) from the temperature sensor 116 coupled to the coolant jacket 118; the surface ignition pickup signal (PIP) from the Hall effect sensor 120 (or other type) coupled to the crankshaft 140; the throttle position (TP) from the throttle position sensor; and the absolute manifold pressure signal (MAP) from the sensor 124. The engine speed signal RPM can be generated by the controller 12 from the signal PIP. The manifold pressure signal MAP from the manifold pressure sensor can be used to provide an indication of the vacuum or pressure in the intake manifold. The controller 12 receives signals from Figure 1A various sensors and employsFigure 1A Various actuators adjust engine operation based on received signals and instructions stored in the controller's memory. For example, based on a pulse-width signal commanded by the controller to a driver coupled to a direct injector, fuel pulses can be delivered from the direct injector into the corresponding cylinder. Refer to Figure 3 and Figure 4 illustrates an example program that can be executed by the controller.

[0045] As described above, Figure 1A only one cylinder of a multi-cylinder engine is shown. Thus, each cylinder can similarly include its own set of intake / exhaust valves, fuel injectors, spark plugs, etc. It will be understood that engine 10 can include any suitable number of cylinders, including 2, 3, 4, 5, 6, 8, 10, 12, or more cylinders. Additionally, each of these cylinders can include some or all of the various components described and illustrated with reference to cylinder 14 via Figure 1A described and illustrated.

[0046] In some examples, vehicle 5 can be a hybrid vehicle having multiple torque sources available for one or more wheels 55. In other examples, vehicle 5 is a conventional vehicle having only an engine, or an electric vehicle having only an electric motor. In the example shown, vehicle 5 includes engine 10 and electric motor 52. Electric motor 52 can be a motor or a motor / generator. When one or more clutches 56 are engaged, the crankshaft 140 of engine 10 and electric motor 52 are connected to wheels 55 via transmission 54. In the example shown, a first clutch 56 is disposed between crankshaft 140 and electric motor 52, and a second clutch 56 is disposed between electric motor 52 and transmission 54. Controller 12 can send signals to the actuators of each clutch 56 to engage or disengage the clutch, so as to connect or disconnect crankshaft 140 from electric motor 52 and the components connected thereto, and / or to connect or disconnect electric motor 52 from transmission 54 and the components connected thereto. Transmission 54 can be a gearbox, a planetary gear system, or other type of transmission. The powertrain can be configured in various ways, including parallel, series, or series-parallel hybrid vehicles.

[0047] Electric motor 52 receives electrical power from traction battery 58 to provide torque to wheels 55. Electric motor 52 can also operate as a generator to provide electrical power to charge battery 58, such as during a braking operation.

[0048] In some examples, a battery management system (BMS) can be present on a vehicle, where the BMS is electrically and thermally coupled to a battery and communicates with a vehicle controller. A battery monitoring sensor (BMS) monitors and calculates the actual battery conditions (state of charge (SOC), state of health (SOH), and state of function (SOF)). It consists of hardware and software. The hardware includes a single-chip solution for measuring battery voltage, battery current, and temperature. These inputs are used to calculate the actual battery state. The BMS learns the battery state of the connected battery over time and maintains the actual battery state in RAM, and periodically saves the learned and adjusted battery parameters to non-volatile memory (NVM).

[0049] Battery 58 can also be used for various engine operation-related functions. For example, battery 58 can be coupled to a starter motor (not shown) that is used to start the engine during engine starting. Battery 58 can be implemented with Li-ion technology. The methods for predicting end-of-life and the associated control and communication strategies described herein can be adapted to this type of propulsion battery and to chemistries other than those used in lead-acid batteries. A battery monitoring sensor (BMS) 184 can be coupled to battery 58 for estimating one or more conditions associated with the degradation state of the battery. Based on inputs from BMS 184, controller 12 can calculate the inferred end-of-life of the battery, as Figures 3 to 4 detailed herein.

[0050] In an alternative embodiment, such as in the case where the vehicle is a conventional gasoline-fueled engine, the engine system can include an electrical architecture, such as Figure 1B the example architecture. Figure 1B The electrical system shown shows a conventional 12V-based alternator 186 coupled to a single lead-acid battery 188. The battery can be used to power one or more conventional 12V electrical loads 182 (such as an electric power steering system (EPAS), electric power brakes, roll control, etc.). Battery 188 can be coupled to a battery monitoring sensor (BMS) 184. As detailed herein, the degradation state of the battery can be inferred via the BMS. Additionally, the end-of-life of the battery can be calculated based on data received from the BMS. It will be understood that in some examples, Figure 1A the vehicle system of Figure 1B can also include the system of

[0051] Batteries may need to be periodically serviced and diagnosed. Additionally, based on their service or degradation history, the remaining life of a battery may vary. Due to an incorrect estimation of the end-of-life (EOL) of a battery, an unexpected battery failure may occur. For example, if a battery health monitor is inaccurate, then the EOL of the battery may be predicted later than when it actually occurs. This may result in customer dissatisfaction due to loss of mobility (such as when a vehicle or engine fails to start) and reduced electrical functions that affect driving performance (such as when the assistance of an electric power assisted steering (EPAS) is lost). If the service life of a battery can be accurately predicted, then the battery can be replaced before a failure occurs. This may be particularly advantageous in the case of a vehicle system in which a central controller coordinates various vehicle systems to minimize power demand. For example, in the case of an autonomous vehicle system, in the situation where there is limited interaction between a vehicle driver and a vehicle controller, autonomous functionality may be selectively activated only when a redundant power source supplied by a battery is available. If a degraded battery is identified, or if it is predicted that the end-of-life of the battery will occur within a short pre-defined time period, then the autonomous functionality may be disabled or reduced. For example, a low battery voltage condition may cause an electric power assisted steering system to reduce the steering assistance it provides, or deactivate non-safety critical electrical functions. Other battery-related problems may include excessive battery gassing, which may result in an unpleasant odor during vehicle operation, and in the case of a plug-in vehicle charged in a garage, an unpleasant odor in the garage.

[0052] In addition to the problems associated with an incorrect prediction of the end-of-life of a battery where the EOL is predicted later than when it actually occurs, there may be other problems associated with an incorrect prediction of the end-of-life of a battery where the EOL is predicted earlier than when it actually occurs. For example, an on-board battery health monitor may annoy customers in the situation where they are alerted that a battery is defective, when the defective battery becomes healthy when the vehicle is serviced.

[0053] The inventors have recognized herein that current battery health monitoring methods may only evaluate battery characteristics such as voltage response or internal resistance. However, based on the usage history of a battery, temperature limitations, and problems associated with the chemistry of the battery, the battery may fail for a variety of reasons. For example, a lead-acid battery may degrade due to corrosion, sulfation, active mass loss, etc. By using an EOL prediction method that evaluates multiple battery characteristics associated with battery degradation, as referenced Figures 3 to 4described in the program details, which improves the reliability of EOL prediction. In addition, by relying on multiple parallel paths for fault prediction, the range of detectable fault mechanisms is extended. In addition, the interaction of different characteristics can also be considered (such as the effect of the corrosion rate on the sulfation rate of the battery, and vice versa, the effect of the sulfation rate on the corrosion rate of the battery). The performance of the battery in a vehicle is a function of multiple characteristics. If the end-of-life is defined only for each characteristic without considering other characteristics, the end time of the end-of-life may be overestimated because a small deterioration in multiple characteristics may lead to a significant reduction in the overall battery performance. For example, doubling the internal resistance of the battery may cause the vehicle to fail to start due to low voltage at the starter. Halving the battery capacity may also cause the vehicle to fail to start due to the depletion of the ignition-switch-off load. However, increasing the internal resistance by only 50% and reducing the capacity by 25% may also prevent the vehicle from starting. Therefore, it is beneficial to define the end-of-life with respect to multiple battery characteristics, as Figure 6 shown.

[0054] For example, the controller can identify one or more battery characteristics that can be measured and affect battery health. The controller compares the data on those characteristics collected during vehicle operation with corresponding thresholds. As described in reference Figures 3 to 4 , the vehicle controller can define the thresholds based on battery history and statistical data collected from various sources (such as fleet data, dealer vehicle data, warranty laboratory data, etc.). The controller can be configured to use an algorithm that estimates the rate of convergence of the measured data to the defined thresholds, and use the estimated rate, along with the previous history of the battery's degradation behavior, the sensed data of the battery-related parameters, and the vehicle driving statistics based on mapping (such as real-time vehicle driving statistics, or those compiled within the current vehicle driving cycle), to make a statistical prediction of the remaining life of the battery. Then, the controller can provide the vehicle driver with meaningful information about the battery life, such as in the form of an estimate of the distance or time to repair. Therefore, the expected remaining life of the battery can be estimated via a method that runs online in the vehicle, and its accuracy can be enhanced by performing periodic parameter adaptation using battery data obtained from the vehicle fleet on the road and the batteries replaced during repair.

[0055] In this way, Figures 1A to 1B the components of implement a vehicle system, which includes: an alternator driven by an engine; a battery, sensors for measuring battery voltage and current; and a controller. The controller can be configured with computer-readable instructions stored on a non-transitory memory for performing the following operations: predicting the degradation state of the battery based on changes in capacity and internal resistance.

[0056] Figure 2An exemplary map 200 is shown that can be used by a vehicle controller to monitor the end-of-life of a vehicle battery, such as a lead-acid battery. In Figure 2 The method applied in Figure 3 can represent the minimum solution implementation of the method of

[0057] which depends only on 2 of the multiple measurable characteristics of a lead-acid battery, but it will be understood that the method can be extended to include additional characteristics. Figure 2 In the example of

[0058] Figure 2 , the battery internal resistance (R) and the battery capacity (Q) are monitored. The battery resistance is defined as the ratio of the voltage change to the current change that causes significant voltage and current transients measured during a starting crank or other event. On the other hand, the battery capacity is defined as the amount of charge it can deliver at the rated discharge current and temperature such that the battery terminal voltage is higher than a defined threshold. The starting ability of the battery may be reduced due to corrosion, which typically manifests as an increase in internal resistance or a decrease in the cold-cranking current. Similarly, if the battery becomes sulfated or experiences active mass loss (both of which can manifest as a loss of capacity), then the battery may lose its charging ability. Therefore, by monitoring the changes in the internal resistance and the total capacity, the health of the battery can be monitored. As the battery ages, it is expected that the resistance of the battery increases while the capacity is expected to decrease.The mapping diagram 200 shows an algorithm 202 that receives an input BSBattRiTNom regarding the internal resistance 204 of the battery, which is normalized at 100% battery SOC and 25°C. The algorithm 202 can be stored in the memory of the vehicle controller, such as stored in the battery management module. The algorithm 202 also receives an input BSBattQCapAhTNom regarding the capacity 206 of the battery, which is normalized at 100% battery SOC and 25°C. The normalized capacity represents the amount of charge that can be removed from the battery at 25°C when discharging from 100% battery SOC at the nominal current of I20. I20 is defined as the current magnitude obtained by dividing the nominal capacity by 20. In one example, data can be received from a battery management system (BMS) on the vehicle. The BMS is coupled to the battery and communicatively coupled to the vehicle controller. The BMS measures current, voltage and can also measure the battery terminal temperature. It can use a built-in model such as an equivalent circuit model to infer the battery SOC, capacity and capacity loss due to sulfation and loss of active mass. Data can be collected during specific situations to estimate specific battery characteristics. For example, when discharging occurs in a situation where the current changes over time (the change is greater than a threshold), only the internal resistance can be estimated, and when the battery is fully charged and allowed to rest without a minimum amount of charging and discharging, only the capacity loss due to sulfation can be calculated. A plausibility strategy can be employed, which only transmits an estimate of the battery characteristics after the measurement has occurred multiple times, or can transmit the average value of the estimates within a defined moving time range. Some battery monitoring sensors can use an extended Kalman filter to remove outliers from the estimate of the battery characteristics and provide a stable value.

[0059] The normalized internal resistance 204 and the normalized capacity 206 are plotted on the mapping diagram 210 to determine the current position of the battery on the mapping diagram 210. Additionally, resistance and capacity thresholds are determined to identify the region on the mapping diagram 210 where the battery is in a usable healthy state (e.g., where the battery health monitor quality passes). For example, the controller can define a region 212 on the mapping diagram 210, and the region 212 is bounded by a resistance threshold 214 (R0) and a capacity threshold 216 (Q0). As Figure 3 detailed, the thresholds can be determined based on various factors, including sensed data, statistics, battery repair history, vehicle driving history, data retrieved from other vehicles in the fleet, other vehicles at the dealership, etc. In the example shown, based on the data, it is determined that the battery health state is within the region 212, at the position 218, at a distance 220 from the resistance threshold 214, and at a distance 222 from the capacity threshold 216.

[0060] In addition, based on the sensed and statistical data, and via Figure 3For the algorithms discussed herein, the controller can estimate the rate of convergence of each parameter towards its corresponding threshold, such as the rate at which the internal resistance changes from position 218 towards the resistance threshold 214, and the rate at which the internal capacity changes from position 218 towards the capacity threshold 216. Based on the estimated convergence position and rate, the algorithm can determine the state of health of the battery 224, including whether the battery monitor has passed or failed (e.g., whether the battery will cause the vehicle to start or not start) and the remaining battery life (e.g., the predicted end-of-life estimate). The estimated state of health of the battery is then displayed to the vehicle driver.

[0061] It will be appreciated that although Figure 2 illustrates the battery capacity monitored as a function of sulfation and active mass loss, in additional examples, additional signals can be used to monitor sulfation-induced capacity loss that is different from the capacity loss due to active mass loss. This is because the effects of sulfation can be reversed. Additionally, although R0 and Q0 are shown herein as temperature-independent calibratable thresholds normalized to a predefined temperature and state of charge, in other examples, the thresholds for battery characteristics can be temperature-independent. Turning now to Figure 3 , an example method 300 for predicting the end-of-life (EOL) of a vehicle battery is illustrated. The instructions for performing method 300 and the remaining methods included herein can be executed by a controller based on instructions stored on the controller's memory and in conjunction with signals received from battery monitoring sensors, as Figures 1A to 1BAs shown. According to the method described below, the controller can use the actuators of the vehicle system and the engine system to diagnose the battery health state. This method enables predicting the deterioration state of the vehicle battery based on the rate of change of the value of a metric from its initial value when installed in the vehicle system during the duration of vehicle operation. The prediction is also based on the distance traveled by the vehicle during the duration, and the metric is obtained from sensed vehicle operating parameters. The method also enables converting the predicted deterioration state into an estimate of the remaining time or duration before the vehicle battery needs repair (or replacement) for display to the vehicle driver. The method includes an advanced learning-based approach that can be applied to all types of automotive batteries, including lead-acid and lithium-ion batteries. Equivalent circuit parameters are periodically identified at fixed intervals, and the assumptions are explicitly based on battery temperature and state of charge (SOC). In the method shown, two general performance metrics are applied to battery diagnosis / pre-diagnosis. They are calculated using statistics associated with replaced batteries and batteries that caused vehicle failures. The metrics are used to adjust the threshold for defining end-of-life (EOL) for each battery characteristic. After the threshold has been adjusted, the performance metrics are used to determine whether the adaptation is sufficient. Since battery statistics only converge to stable values after a representative number of batteries have been estimated over a long enough time, a monitoring period is associated with each metric. Once the metric has been estimated for at least the relevant monitoring period, it can be used to quantify system performance and calculate further adaptation to the EOL threshold. Other metrics for measuring system performance and facilitating parameter adaptation can also be applied. As new requirements for battery diagnosis / pre-diagnosis systems emerge, those described herein can be used as prototypes for new metrics.

[0062] At 302, the method includes selecting one or more battery characteristics (from a plurality of possible battery characteristics) for battery EOL prediction. Characteristics may include battery properties that can be used to measure battery degradation. Characteristics may be selected based on the nature or configuration of the battery, the computing power available on the vehicle for EOL prediction, and the time constraints for EOL prediction. As an example, when the battery is a lead-acid battery, the selected characteristics may include the minimum internal resistance and minimum capacity of the battery. If additional computing power or time is available for estimation, additional lead-acid battery characteristics may be added, such as the battery sulfation level that is different from the effective mass loss of the battery. As another example, when the battery is a lithium-ion battery, the selected characteristics may include the battery impedance or Li-ion charge concentration estimated by the embedded model. Additional characteristics may also be added to the prediction algorithm to include additional failure mechanisms. For example, as novel battery technologies emerge, and / or as previously unknown degradation mechanisms for existing batteries are discovered, additional characteristics may be selected. In other words, the prediction algorithm may not be limited to using only two features such as internal resistance and capacity. Additional features may be added as more information about battery end of life is collected and as existing technologies change. For example, additional functionality may be added to include new battery monitoring techniques, new battery sensors, and new information correlating battery characteristics to battery end of life.

[0063] By increasing the number of features, the accuracy of EOL prediction can be improved when computing power is not a constraint. Additional features that can extend the prediction may include, for example, battery gassing rate (e.g., in grams of H2 per hour). When overcharged, lead-acid batteries release oxygen and hydrogen. It is possible that the battery has enough capacity to start a plug-in vehicle, but a lot of gassing still occurs in the garage when it is plugged in. As another example, the battery water loss rate (in grams / minute water loss) can be combined. Lead-acid batteries release oxygen and hydrogen, and when overcharged, experience water loss. When the water level of the battery becomes too low, it may fail due to reduced capacity and increased internal resistance. As another example, the occurrence of an internal short circuit can be evaluated (such as via a set flag). An internal short circuit is characterized as a severe drop in the open circuit voltage of the battery after the power of the vehicle is turned off. This symptom may be accompanied by a high internal resistance corresponding to the battery voltage and the average discharge current. As another example, the capacity loss caused by sulfation (e.g., in terms of capacity loss of Ah caused by sulfation) can be characterized. Sulfation is a process that causes a reduction in battery capacity and a drop in voltage level relative to the state of charge. It is reversible to some extent by charging at high voltage for a sustained period of time.

[0064] As yet another example, the active mass loss can be characterized (e.g., in terms of the loss of Ah capacity due to the loss of active material). The active mass loss results in a reduction in the battery capacity. As another example, the battery repair time can be characterized (e.g., in terms of the battery time of a vehicle in motion expressed in hours). The statistical distribution of the expected life of the battery in a particular vehicle can be used to predict the distance to end-of-life. The prediction can be improved by using the statistics of particular vehicles operating in a particular area to define the end-of-life as a function of vehicle usage. For example, the same battery may deteriorate faster in a warm / hot environment due to corrosion and water loss compared to cold climate conditions. As yet another example, the charge throughput weighted with respect to the depth of discharge can be characterized (e.g., in terms of the loss of battery capacity in Ah). The charge throughput can be weighted with respect to the battery depth of discharge and integrated over time to estimate the aging defined as the loss of battery capacity.

[0065] The selection of characteristics can also include the method of selecting the characteristics to be measured, the sampling frequency, identifying one or more algorithms required to calculate the signal or metric from the data collected, and the computational power and time required. For example, if it is found that many batteries fail due to water shortage, then a new advanced algorithm for measuring water loss or even a water loss sensor can be implemented. On the other hand, if many batteries fail due to grid corrosion, then an equivalent circuit model can be implemented to track the corrosion. The degree of corrosion in the positive electrode is related to the change in the values of the resistance and capacitance of the RC pair in the standard Randle equivalent circuit model. The RC pair is associated with a time constant that falls within a defined range. The sampling period for voltage and current measurements can be selected to adequately sample those values for the algorithm used to estimate the values of the RC pair. These values will then be used to track the corrosion.

[0066] It will be appreciated that the battery can be evaluated periodically, such as based on the vehicle travel time or distance elapsed since the last battery evaluation. In other examples, the selection of the characteristics used to evaluate the battery can be based on an active request received from the driver. This may be in addition to or independent of the periodic evaluation. For example, the driver can request a pre-diagnosis of the system battery before starting a planned driving route.

[0067] At 304, the method includes measuring selected characteristics online during vehicle operation. For example, the selected characteristics can be measured via one or more battery sensors at a determined sampling frequency. For example, a Hall-based or shunt-based current measurement and a voltage sensor can be used to measure the internal resistance of a battery at a sampling frequency greater than 1 kHz. As another example, the loss of battery capacity due to sulfation can be measured by a single voltage measurement after the battery is fully charged and has been allowed to rest for several hours without charging or discharging. The change in the open-circuit voltage of a fully charged battery as it ages is a measure of the capacity loss due to sulfation. As another example, an internal short circuit (and the severity of the short) can be identified based on the degree of voltage relaxation. At 306, the method includes normalizing the collected data to obtain temperature and state of charge. For example, the data can be normalized to 25 °C and 100% SOC. In this way, the data sensed on the vehicle can be used to determine the health state of the battery. For example, where the determined metric is one or more of battery resistance and battery capacity, the sensed vehicle operating parameters can include one or more of battery current and battery voltage. In some examples, after sensing one or more parameters associated with battery degradation, the controller can compare the data sensed on the current iteration of the program with the data sensed on a previous iteration of the program to update the rate of battery degradation (from a base rate) in real time. For example, the controller can predict the degradation state of the vehicle battery based on a determined metric derived from the sensed vehicle operating parameters (including the past history of the determined metric).

[0068] At 308, the method includes defining thresholds for each selected battery characteristic. These are the thresholds for the metrics that define the end of life (EOL) of the battery. The EOL thresholds can be determined from one or more sources, including battery experts, vehicle electrical requirements that cascade feature requirements to battery requirements, a set of fixed aged battery data that describes the characteristics, and iteratively online monitoring of the characteristic data from end-of-life batteries replaced at a dealership. As an example, in the case of an initially fully charged battery with a capacity C 新 , it may be required to be able to start the vehicle after 30 days of parking. After 30 days, the ignition switch off load consumes a known amount of battery charge (C KOL ). When the capacity decreases such that C < C 新 - C KOL , end of life is determined to have occurred. For example, when capacity loss is the only deteriorating battery characteristic, end of life can be defined when the battery capacity has decreased 30% from its new value. The ability of a cold battery to start the vehicle can be hampered by a high internal resistance. If the internal resistance increases 75% from its new value, end of life can be identified, or if it reaches a value such as 9 milliohms, end of life can be identified.

[0069] Figure 5 An example is shown of using a statistical representation of battery characteristics to define corresponding thresholds during EOL prediction. In map 500, a statistical distribution of the normalized internal resistance of batteries replaced by a dealer is shown. The controller can select a threshold (dashed arrow) based on a warranty and customer satisfaction model to correspond to a defined percentage of the population. For example, the resistance corresponding to 10% of the batteries replaced at the dealer can be used to identify end-of-life. In this case, many batteries that actually still have some useful life can be replaced, but the number of customer vehicles that cannot start due to defective batteries will be small. On the other hand, the resistance corresponding to 90% of the batteries replaced at the dealer will be higher than the 10% resistance. Selecting this value will result in allowing some batteries to continue to serve in the vehicle even if they may fail, and thus, the number of vehicles stranded due to defective batteries may increase. However, the warranty cost may be reduced because batteries that can still maintain service will not be replaced unnecessarily. The adjusted EOL thresholds are shown at maps 510 and 520 and will be described later with reference Figure 7 to the EOL threshold update procedure.

[0070] Return Figure 3 , machine learning techniques for pattern recognition can also be implemented to define the threshold. For example, pattern learning within a machine learning framework can be applied, which is also known as supervised learning. Given the probabilities associated with each classification with an associated confidence level, supervised learning can define the "N best classification labels" for a given set of features. The classification labels for supervised learning for battery end-of-life prediction can include "before (outside) end-of-life" and "at (within) end-of-life". More than one feature can be used for classification identification with supervised learning. An example output of a supervised learning algorithm can include "the probability that the battery is at end-of-life is 78%, the probability that the battery is not at end-of-life is 22%, and the confidence level of this prediction is 90%".

[0071] There may be different classes of supervised learning algorithms for threshold calculation. For example, there may be linear, quadratic, or parametric algorithms that rely on maximum entropy calculations. As another example, there may be non-parametric algorithms, such as those using neural networks or support vector machines. Thus, any of these algorithms can be used to calculate the threshold that defines end-of-life. For example, a linear parametric algorithm of the linear discriminant analysis type can be used.

[0072] If it is found that the thresholds and calibration parameters do not provide the predicted target accuracy requirements, then the thresholds and calibration parameters can also be updated. In one example, this can occur if changes in battery technology or vehicle electrical system technology change the characteristics of the battery or battery failure modes. The advanced architecture can use data from a large central database (e.g., the cloud) as the source for the updated thresholds and calibration parameters.

[0073] To adaptively optimize the parameters of the diagnostic / prediction system, an appropriate infrastructure must be available to facilitate measuring the characteristics of a battery replaced or malfunctioning in a vehicle. The measurements are analyzed, and the algorithm can use performance metrics to recalculate the thresholds defining end-of-life.

[0074] It will be understood that the algorithm can run in the cloud or on a handheld device without a continuous data stream.

[0075] Various implementations of the architecture are possible. A first example shows such an architecture at Figure 12 Example 1200. Example 1200 shows the general form of the architecture, where battery characteristics are collected and processed by a central entity external to the vehicle. In the example shown, the thresholds can be calculated at a central location (e.g., a database, such as in the cloud) and transmitted to the vehicle. The thresholds can be calculated at the database based on inputs received from database sources such as vehicle BMS, dealer battery tests, warranty analysis laboratories, etc. The thresholds are then transmitted to an on-vehicle algorithm, which also receives inputs from various on-vehicle battery monitoring sensors. The algorithm further estimates the battery life based on the characteristic gradients stored in the memory of the controller.

[0076] Another example is shown at Figure 12 Example 1210, where battery characteristics can be collected and processed at a central location. Then, parameters (e.g., the mean, standard deviation, etc. of a given characteristic) describing the battery data and characteristic statistics can be transmitted to the vehicle, and the calculation of the parameters and thresholds can occur locally on the vehicle. Simple characteristics such as pattern recognition, supervised learning, or characteristic statistical parameterization (such as percentiles) are used to determine the parameters of the characteristics, such as the mean and standard deviation of each selected battery characteristic. The parameters can be calculated at the database based on inputs received from database sources such as vehicle BMS, dealer battery tests, warranty analysis laboratories, etc. The parameters are then transmitted to the vehicle controller, where the on-vehicle algorithm is used to locally calculate the thresholds.

[0077] Transmitting the thresholds or parameters to the vehicle can be performed via an on-vehicle modem, through a diagnostic port, wireless communication, or by reflashing the on-vehicle controller (or ECU) in which the prediction algorithm runs.

[0078] As discussed above, the database can receive input from various database sources. For example, data can be received from a vehicle battery monitoring system (BMS) including battery monitoring sensors. Wherein, before the battery is replaced by the dealer, the maintenance personnel can trigger a data collection program to save the battery characteristic data. This represents the imprint of the battery at the end of its life. Then, the data is transmitted to the cloud via an in-vehicle modem or by downloading it from the diagnostic port.

[0079] As another example, the data can be sourced from a dealer battery tester. Wherein, when the battery tester diagnoses a defective battery, a program can be triggered that saves the information used for diagnosis and may cause the tester to collect other information. Then the information is transmitted to the cloud via a modem or by other means.

[0080] As yet another example, the data can be sourced from a warranty analysis laboratory. Wherein, a subset of the batteries replaced at the dealer can be analyzed and disassembled by the laboratory for quality control. The information collected during the analysis can be transmitted to the cloud via the Internet or by other means.

[0081] At 310, the method includes estimating the convergence rate of the normalized battery data towards a corresponding threshold. This allows the controller to infer how fast the battery is approaching the end of its life. At 312, the battery EOL can be estimated based on the convergence rate. As an example, in the case where the normalized internal resistance of the battery is the battery characteristic being monitored, for a given battery, the EOL threshold can be set to 6 milliohms. The battery degradation as a function of the convergence rate of the internal resistance can be described by the gradient:

[0082]

[0083] Then the distance from the EOL can be described by the equation threshold – RiNorm = x.

[0084] Then the remaining life can be estimated as:

[0085] Reference Figure 2 For the example of, the current condition or state of the battery can be defined by location 218. The resistance threshold 214 (R0) and the capacity threshold 216 (Q0) can be calibratable thresholds for the internal resistance and the battery capacity, as identified from the aged battery data and / or the advice of battery experts. The distance of the current capacity at location 218 from the threshold is herein referred to as x1 (and is shown as 222 in Figure 2 ), and is defined as: x1 = Q0 - Q. Then the rate at which the capacity approaches the threshold (i.e., the convergence rate) is determined as:

[0086] Similarly, the distance of the current internal resistance at location 218 from the threshold is herein referred to as x2 (and inFigure 2 is shown as 222) is defined as: x2 = R0 - R. Then the rate at which the circuit approaches the threshold (i.e., the convergence rate) is determined as: The derivative that describes the rate of convergence towards the threshold can also be referred to in this context as the gradient (of the corresponding feature).

[0087] Since a battery may fail due to low capacity or high resistance, the end - of - life (EOL) of the battery is defined as the earliest failure time due to either mechanism. In other words,

[0088] The prediction algorithm can be similarly extended to multiple features, as shown in reference Figure 6 Although the end - of - life (EOF) prediction algorithm for the lead - acid battery shown in Figure 2 measures battery characteristics or features and compares them to thresholds of end - of - life values that define features (specifically, the features "normalized battery internal resistance" and "normalized capacity"), Figure 6 the mapping diagram 600 of

[0089] As shown at mapping diagram 600, the end - of - life thresholds predicted relative to multiple features may not necessarily be linear boundaries in the "feature space" (such as the case in Figure 2 , see the linear feature space 212). Instead, the shape of the feature space 612 can be determined by the pattern recognition / supervised learning algorithm used for its definition.

[0090] The feature state represents the values of the features {feature1, feature2,...} or {f1(t), f2(t), f3(t),..., f N (t)}, which are used to determine the expected life of the battery. They may be internal resistance and capacity or other characteristics. Each axis of the feature space corresponds to the value of a specific feature. The feature measurement at a certain point in time corresponds to a point in the feature space. The point f(t) = {f1(t), f2(t), f3(t),..., f N (t)} and f1(t)...f N (t) represent the values of the battery's features at time t.

[0091] Mapping diagram 600 also includes the gradient of the battery's features. The gradient describes the change of the battery's feature state (in the feature space) with respect to time in the direction orthogonal (shortest path) to the end - of - life threshold in the feature space

[0092] The prediction of the time to end - of - life is defined as the remaining time before the battery reaches its end - of - life. It is obtained by dividing the orthogonal length Divided by the convergence rate along the orthogonal direction is calculated. The length of the orthogonal direction can be expressed as the distance from the end-of-life threshold, as shown. Figure 6 shown. The direction of can be parallel to the direction of a specific feature, as Figure 2 shown, or as Figure 6 shown, it can proceed in another direction. In this case, it can be assumed that the battery degradation is captured by multiple features.

[0093] When implementing a battery diagnosis / pre-diagnosis system, the threshold can be initially estimated. To monitor its performance, the characteristics of the batteries replaced during maintenance and vehicle failure statistics can be used to define a metric. If the metric indicates that the system is not performing properly, then the threshold can be adaptively adjusted.

[0094] Return Figure 3 , in some examples, the convergence rate can be further updated based on the battery degradation or repair history. For example, the convergence rate can be further adjusted according to the time or duration elapsed since the battery was first installed or operated in the vehicle. As another example, the time or duration elapsed since the battery was last serviced, repaired, or reset can be considered. Additionally, the repair history can include details about: the degradation rate of the battery before the most recent repair event, the basic degradation rate of the battery, the average degradation rate of the battery over the entire vehicle life, and any diagnostic codes associated with the battery enabled during the vehicle life.

[0095] In a further example, the convergence speed can be further updated based on vehicle driving statistics. Vehicle driving statistics can include, for example, the distance covered over the life of the vehicle (e.g., based on odometer readings), the number and frequency of maintenance events that have occurred over the entire life of the vehicle (e.g., how many refueling services have occurred, how frequently they were performed, the odometer readings at which they were performed, the average fuel economy of the vehicle, the average speed of the vehicle, the average gear usage of the vehicle, the average number of miles driven per day, the average tire pressure of the vehicle, etc.). Vehicle driving statistics can also include, for example, driver-specific driving patterns and habits. For example, this can include the driver's preference for fuel economy versus performance, the frequency and degree of pedal stomping and depression (e.g., whether the driver "likes to drive fast"), the aggressiveness with which the driver tends to drive, the average speed at which the driver drives, etc. Vehicle driving statistics can also include details about the weather conditions in which the vehicle is typically driven, such as whether the vehicle is typically operated in rain or snow, dry or wet conditions, etc. Vehicle driving characteristics can reflect the driver's driving tendencies and the average conditions that the vehicle battery has experienced, which can affect the convergence speed of one or more of the selected characteristics at different rates.

[0096] In the case of monitoring a propulsion battery in a hybrid or electric vehicle and predicting its time to end-of-life, future (e.g., predicted) driving patterns can be taken into account, such as terrain, ambient altitude and temperature, predicted pedal events, and the incline / grade expected along a selected navigation route. For example, if the driver frequently stomps on the accelerator pedal and the brake pedal (or is expected to do so given the selected driving route), then the battery may experience above-average charge throughput. If the battery is used for propulsion in an electric vehicle, then the battery will only be depleted under these conditions. In other vehicle configurations, traffic predictions can be used to estimate charge throughput in all types of vehicles, and charge throughput can be directly related to battery degradation and aging. As another example, if the driver drives aggressively, then the battery may heat up faster and wear due to the higher average battery temperature. Therefore, the convergence speed can be increased, indicating that due to the vehicle driving statistics, the battery can be expected to degrade faster.

[0097] In one example implementation, the end-of-life prediction can be implemented in the vehicle, where a display on the dashboard indicates the remaining battery life. Where the end-of-life threshold and other calibration parameters can be refreshed in the vehicle on the assembly line and can only be updated during an occasional service visit (if any). The feature data can come from battery monitoring sensors and any battery monitoring algorithms or controllers running in the vehicle's control module or another on-board ECU. The feature gradient describing the convergence speed of a given feature towards the corresponding threshold can be calculated online and stored in the body control module or another ECU.

[0098] In another embodiment, the prediction algorithm can run externally, where battery parameters are transmitted from the vehicle to the cloud or to an external device where the algorithm runs. The external device can include, for example, a handheld device such as a smart phone or a tablet. Additionally, the algorithm can be configured as an application running on the device. If the algorithm runs in the cloud, then the battery parameters can be transmitted continuously or can be scheduled for transmission during regular intervals. The transmission can also be triggered by an operating mode such as ignition switch on, ignition switch off, or initialization of a service procedure at a dealership. If the algorithm runs on an external device, then data transmission will occur when a wired or wireless link is established with the vehicle. This can occur at any time but will typically occur during vehicle servicing.

[0099] In some examples, instead of calculating the EOL of the battery, the state of health (SOH) of the battery can be output, where the SOH of the battery can be expressed as a percentage of remaining life varying from 100% of a new battery to 0% of a scrapped battery. As the battery ages, its internal resistance increases, its internal capacity decreases, and correspondingly, its SOH decreases.

[0100] At 318, the method includes converting the estimated EOL or SOH of the battery into an EOL estimate that can be easily understood by the vehicle driver, such as the remaining time of the remaining vehicle operating distance before complete battery degradation (when the battery is scrapped). At 320, the remaining time or distance can be displayed to the vehicle driver. For example, the estimated time / distance remaining before battery degradation can be displayed to the vehicle driver on a display screen of the vehicle's central console.

[0101] In one example, the controller can use an algorithm to convert the state of health into an estimate of the remaining time / distance before component degradation occurs. For example, the controller can convert the predicted degradation state into an estimate of the remaining time or duration to display to the vehicle driver based on past driving history data and predicted future driving (including the past history of determined metrics). Additionally, the predicted degradation state can be converted into the remaining number of fuel tank refueling events to display to the vehicle driver based on past driving history data and predicted future driving.

[0102] For example, it can be displayed that "the battery will need to be replaced within 120 miles". This can provide the driver with a more easily understandable estimate of when the battery needs to be serviced. Additionally, the displayed estimate can prompt the vehicle driver to adjust their driving mode. For example, it can prompt the driver to drive less aggressively.

[0103] In some examples, using pattern recognition to determine end-of-life also enables the controller to determine if a battery belongs to a specific group of failed batteries given its characteristics and the rate of convergence of those characteristics (or the trend of their rates of convergence). A dealer or fleet controller can use this data to schedule the repair of other batteries for other vehicles in the same fleet. Similarly, an individual vehicle owner can also use the information provided to plan trips and appropriate repair or maintenance arrangements. For example, the EOL prediction displayed to the vehicle driver can include "The probability that the battery belongs to this group of failed batteries is 73%. The assigned confidence level is 95%."

[0104] A variety of communication strategies can be used to alert drivers and maintenance personnel of impending battery failures, with the various strategies taking into account marketing, customer satisfaction, warranty costs, and other factors. Generally, when a corresponding threshold is exceeded, the strategy can communicate directly with maintenance personnel and drivers on separate occasions. The communication can be triggered during the countdown period before end-of-life is reached.

[0105] In other examples, such as when the vehicle is an autonomous vehicle, the controller can also limit or reduce the autonomous functions of the vehicle based on the predicted EOL. For example, if the predicted EOL of the battery is within a pre-specified amount of time, then the autonomous operation limit can take effect.

[0106] In some examples, when the expected end-of-life of the battery is short, various alert, communication, and control strategies for sending messages to the driver / maintenance personnel and reducing functionality can be performed in a step-by-step manner. As an example, at the corresponding EOL threshold, the vehicle driver may receive a warning about an impending battery failure and can expect to lose assistance for the steering system or have the climate control or entertainment system periodically turned off. Reference Figure 9 shows an example of providing notifications step-by-step and restricting the functionality of an autonomous vehicle step-by-step. At 322, after learning the performance characteristics, the EOL threshold can be updated, as Figures 7 to 8 detailed in.

[0107] In Figure 4 is shown an example implementation of the Figure 3 method. Method 400 represents a basic or minimal implementation that only evaluates two characteristics.

[0108] At 402, start the engine. At 404, read the battery resistance and capacity, such as based on the outputs of battery voltage and current sensors. At 406, confirm the entry conditions for the diagnostic program. In one example, when no flag associated with battery resistance or capacity measurement is set, confirm the entry conditions. For example, it can be confirmed based on the flag and other diagnostic codes that the data provided by the voltage and current sensors is accurate and reliable. If the entry conditions are not met, then delay the execution of the program until the conditions are satisfied.

[0109] At 408, retrieve the thresholds corresponding to the measured battery characteristics, such as retrieving from the memory of the controller or via establishing communication with an external device or cloud where the thresholds are calculated. At 410, calculate the difference from the corresponding threshold. For example, the distance of the resistance from the threshold is determined as x1 = R0 - R, and the distance of the capacity from the threshold is determined as x2 = Q0 - Q.

[0110] At 412, determine the convergence rate of each of the battery and the resistance towards the corresponding threshold. At 414, select the minimum of the two rates and use it to predict the EOL. Then at 416, display the determined EOL to the user. The user can be a vehicle driver, a fleet owner, a dealer, etc. At 418, compare the EOL with the threshold life. The threshold life is a non-zero calibratable threshold. In one example, the threshold life is set to 75% of the reserved capacity or is defined based on the vehicle requirements that the battery must meet. If the determined EOL is less than the threshold life, then at 420, flag the battery for warranty. For example, a malfunction indicator lamp (MIL) can be lit to request the vehicle driver to take the battery to a service station or a dealer for repair.

[0111] It will be appreciated that the above method can be used to predict the remaining life of multiple battery types in various implementation scenarios. For example, the method can be implemented for lead-acid starting, lighting, and ignition (SLI) batteries in conventional, hybrid, and electric vehicles, for propulsion batteries based on Li-ion or other technologies in hybrid and electric vehicles, and for any type of battery used in marine, aircraft, or stationary power applications. Other scenarios can include running a battery health prediction algorithm on a vehicle, displaying battery health and end-of-life predictions on a vehicle dashboard, sending health assessments and end-of-life predictions for reading on an external device (e.g., a handheld device in the absence of continuous data streams or others), running the prediction algorithm externally, and transmitting parameters to an external device having a display or cloud where the algorithm runs.

[0112] In one example, the vehicle controller can define the service metric for a general battery as:

[0113]

[0114] Where F(f) represents the number of failed batteries in vehicles that meet the following conditions: (1) the failure mechanism of the battery is associated with feature f; (2) the value of feature f identifies a healthy battery at end-of-life. N(f, G) represents the total number of in-field batteries monitored by the diagnostic / pre-diagnostic system with respect to feature f and falling within group G. The group definition can define battery size and can also include vehicle type, electrical content (options), and other constraints. The performance metric is designed to measure and improve the system performance of the vehicle or platform defined by G. Once the EOL threshold is adjusted, new failures that are to occur in a large number of in-field batteries can be monitored over a long enough time span regardless of the updated threshold or at the updated threshold. The monitoring period Tf is associated with a given feature and can be analytically selected to ensure the maximum possible change in the metric over several monitoring periods or by other means.

[0115] The metric BISFMf can be used to define the percentage of battery failures occurring due to the mechanism identifiable by feature f. High diagnostic / pre-diagnostic system performance can occur at low value performance metrics and can be defined by the threshold BISFThf for the minimum acceptable performance. If the metric exceeds this threshold, it can be assumed that the EOF threshold is not sensitive enough and too many batteries in the vehicle are failing without warning.

[0116] Therefore, the false replacement metric (FRM) is determined, which measures the performance of the battery diagnostic / pre-diagnostic algorithm to avoid excessive battery replacement due to false alarms or over-sensitivity caused by the calibration of the associated performance threshold. In one example, the general false replacement metric (FRF) can be defined as:

[0117]

[0118] Where I(f) represents the number of batteries that meet the following conditions: (1) the battery is identified as failed due to the value associated with feature f and the associated EOF threshold; (2) a bench test or laboratory examination of the health state of the battery indicates whether it is only discharging and whether replacement is necessary. N(f, G) retains the same definition as above. It represents the total number of in-field batteries monitored by the diagnostic / pre-diagnostic system with respect to feature f and falling within group G. The monitoring period Tf can be associated with a given feature and can be analytically selected to ensure the maximum possible change in the metric over several monitoring periods or by other means. Therefore, the calculated value of FRMf is only used when the feature is monitored within the period Tf.

[0119] For the performance metrics defined above, unacceptable performance may occur when the metric exceeds its associated threshold. Therefore, the EOL threshold can be adjusted to improve performance. If there is no system knowledge mapping the change in the EOL threshold to a change in performance, then a small incremental change ΔEOLThf in the EOL threshold can be applied. If a relationship exists between the change in the EOL threshold EOLThf and the general performance PM f for a given feature f, then it can be expressed as a transfer function of the following form: ΔPM f = h(ΔEOLTh f ). In the absence of an available transfer function, the application of an initial incremental change in the EOL threshold can be used as a basis for creating a transfer function for future use. The process of estimating the necessary change in the EOL threshold to obtain a desired change ΔPM f* in the performance metric can be defined by an inverse function of the following form: Using the transfer function to adaptively update the EOL threshold is referred to as intelligent adaptation, and is elaborated with reference to the example of Figure 5 and the procedure of Figure 7 .

[0120] Turning now to Figure 7 , method 700 illustrates an example end-of-life adaptation process for a particular battery feature.

[0121] At 702, the method includes starting an adaptive diagnostic / pre-diagnostic system operation, including starting a timer. At 704, it can be determined whether the elapsed time is greater than a defined monitoring period. For example, it can be confirmed that a large number of vehicles with the battery system have been monitored for the defined monitoring period. In one example, the monitoring period is a duration of more than one year (and has a value of 12 months) to cover each season. If the defined period has not elapsed, then the method returns to the monitoring system and compares the output of the timer with the defined monitoring period. In other words, the method causes the system performance metric to be updated only after the vehicle population with the system has been monitored for at least the specified monitoring period.

[0122] After confirming the elapsed defined period, at 706, the method includes comparing a performance metric of a system characteristic (e.g., a battery in service failure metric BISFf or a false replacement metric FRMf) to a corresponding threshold. The threshold applied here can be selected according to the characteristic whose performance is being evaluated. Thus, the battery resistance threshold for a battery performance metric can be different from the battery capacity threshold for a battery performance metric. For a performance metric, when the metric exceeds its corresponding threshold, unacceptable performance may occur. Thus, if satisfactory performance is identified, such as when the performance metric does not exceed the corresponding threshold, the method resumes monitoring. Otherwise, if satisfactory performance is not identified, such as when the performance metric does exceed the corresponding threshold, the method continues to update the EOL threshold. Specifically, at 708, it can be determined whether a transfer function that maps an incremental change in the EOL threshold to the performance metric is available. If it is available, then at 714, the method includes a smart adaptation of estimating the EOL threshold using the available transfer function mapping. Otherwise, at 710, a calibrated initial value is used. The sign of the adaptation of the EOL threshold can be according to the specific characteristic and metric being monitored, and a change in the EOL threshold for one characteristic may cause a change in the threshold for a second characteristic. Refer to Figure 5 illustrates an example case.

[0123] At Figure 5 , the map 500 shows the measured internal resistance values of failed batteries (e.g., at a vehicle fleet or dealership) and the probability distribution function of the EOL threshold. Here, depending on the metric considered, the initial selection of the EOL threshold for the internal resistance of the battery may be too high or too low. A first example of the initial EOL threshold 502 is shown at the map 500 relative to the probability distribution of the internal resistance values observed in a large sample of failed batteries.

[0124] Water loss and corrosion are failure mechanisms associated with high battery internal resistance. Assume a diagnostic / pre-diagnostic system is in place to monitor the normalized battery internal resistance (RiTNom), and it follows the Figure 7 process described therein. Also assume the monitoring has been carried out for at least a specified monitoring period TRiTNom. If the BISF metric corresponding to water loss indicates too high a percentage of in-field vehicles with starting failures due to water loss, then the EOL failure threshold must be made smaller so that a larger percentage of batteries are diagnosed as battery degradation. This is shown at the map 510 as the adapted EOL threshold 504. By updating the EOL threshold, more failed batteries will come to the attention of the owner and the repair personnel, and thus can be replaced before the vehicle breaks down.

[0125] In one example, a transfer function that estimates a change in a performance metric relative to a change in the EOL threshold can be used to effect a change in the EOL threshold. If this is not available, then the magnitude of the change can be estimated. For the example herein, if an accurate transfer function is not used to adapt the EOL threshold, then the adaptation may be too much or too little. This must be determined after the adaptation from the initial EOL threshold 502 to the updated EOL threshold 504 has been observed in the in-field vehicle for at least another minimum monitoring period TRiTNom. If the adaptation is excessive, then too many vehicles may be diagnosed with a battery fault. This will result in excessive warranty costs and possible owner dissatisfaction due to unnecessary work being performed on the vehicle. A misdiagnosis of a healthy battery will be flagged by the FRMf metric exceeding its performance threshold. If this occurs, then the EOL threshold can be adapted again by slightly increasing the threshold, as shown in the map 520. Specifically, the updated threshold 504 can be further adjusted to an updated threshold 506, where the updated threshold 506 is closer to the initial threshold 502 than the updated threshold 504. The magnitude of the second adaptation (i.e., the adaptation from threshold 504 to threshold 506) can use information from the first adaptation (i.e., the adaptation from threshold 502 to threshold 504) to better ensure that it is not excessive and that the percentage of undiagnosed problems as defined by the BISF metric does not become excessive either. This is Figure 7 part of the intelligent adaptation algorithm described in. The change in the performance metric relative to the EOL threshold can be recorded and used for future adaptation. In one example, in the context of control theory, the adaptation can include a model predictive control algorithm with adaptive model parameters.

[0126] One example scenario where the EOL threshold of the replaceable battery can be updated includes a situation where battery water loss occurs. Battery water loss is a significant degradation mechanism for a set of batteries including a specific vehicle platform and battery size. If the number of undetected problems reflected by the BISF metric is too high, then the fault mechanism (water loss) can be directly addressed by reducing the charged voltage set point of the battery at high temperatures. This will reduce the number of faults caused by this mechanism, and the value of the performance metric will decrease proportionally. However, reducing the charged voltage may cause sulfation, which will effectively reduce the capacity of the battery over time. Battery failure may be caused by capacity loss rather than increased internal resistance. This type of failure will occur when the vehicle is parked and the battery is depleted very quickly due to the ignition switch turning off the load. This fault mechanism can be monitored by the battery characteristic "normalized battery capacity" (QCapAhTNom). Due to sulfation to a certain extent, reducing the charged voltage will increase vehicle failures in a large number of vehicles. If the number of undetected vehicle failures becomes excessive due to variations, then the BISF metric corresponding to the battery capacity will exceed its performance threshold. Therefore, the adaptation of the threshold of the first characteristic can affect the adaptation of the threshold of the second different characteristic. Following Figure 7 the process shown in Figure 10 , the BISF metric for battery capacity can be updated only after being monitored for at least a specified monitoring period for a large number of vehicles equipped with a diagnostic / pre - diagnostic system after a change in set - point control has occurred. After this monitoring period, the BISF metric for battery capacity is compared with its corresponding threshold. If an excess is detected, then the EOL threshold of the characteristic "normalized capacity" can be reduced from the initial EOL threshold 1002 to the updated EOL threshold 1004 in a manner similar to the change in the EOL threshold of the internal resistance described in Figure 5 the mapping diagram 1000. As in the case of water loss, several adaptations may be required to achieve acceptable performance with respect to all performance metrics, and new techniques can be applied to mitigate the fault mechanism associated with the loss of battery capacity.

[0127] The performance of the diagnostic / pre - diagnostic system may also be affected by variations in the set of monitored vehicles. The variations can be regarded as noise factors and classified into several categories. These variations may cause the shape of the distribution function of battery parameters to evolve, and adaptive EOL thresholds may be required to maintain an acceptable performance level. Technical variations that can affect performance include new battery charging strategies or changes in the packaging of the battery that affect the operating temperature. Variations in this category may also include improvements in battery design.

[0128] Changes in the operation of the battery in a fleet of vehicles can also affect performance. These changes include upgrades to the electrical content of the vehicle (e.g., more entertainment devices operating in the ignition-off condition), or when a fleet of vehicles is sold in a new market with extreme ambient temperatures (e.g., in a new environment that is colder or hotter). This can occur if a vehicle designed to operate in a temperate market is sold in a land with a very hot or cold average temperature. If battery monitoring technology changes, then system performance can also be affected. New battery monitoring technology for basic functions such as internal resistance or capacity can cause the distribution of values of a given characteristic at EOL to also change. Each vehicle and electrical system is expected to undergo these changes as it evolves. For this reason, EOL threshold adaptation should be used to continuously monitor and improve system performance. Performance thresholds should also be periodically reviewed in order to adjust them relative to the state of the art in terms of diagnostic capabilities and warranty goals. They can be defined to make the system as accurate as possible without resulting in a large number of warranty costs.

[0129] As discussed above, a degraded or failed battery can be analyzed using an aging model or other methods to determine the EOL threshold. The threshold can then be represented as a set of linear boundaries, such as Figure 2 those shown in. Alternatively, the threshold can be represented by a curve or a manifold, as Figure 8 shown. In some examples, the controller can also define one or more intermediate thresholds, as discussed with reference to Figure 8 the mapping diagrams 800, 810, and 820 of.

[0130] In mapping diagram 800, h represents the value of the EOL curve for a general characteristic. One or more intermediate thresholds can be selected, which define the predicted time before reaching the end of life. Intermediate thresholds can be introduced between the current state f(t) of the battery and the end-of-life threshold. The distance from the intermediate end-of-life threshold is determined according to the convergence rate IntThres and the selected intermediate time threshold T

[0131]

[0132] as defined by the following relationship: Figure 8 When the state of the battery reaches a pre-defined time before the predicted end of life or a pre-defined distance from failure, communication and control activities can be triggered. In the case of a time-defined threshold, as Figure 8 shown in mapping diagram 810, if the battery state fails within a given number of hours, days, or weeks, then a trigger may occur. For example, when the battery characteristic is at a first intermediate threshold T IT1When it is possible to trigger a first communication or control activity. When the battery characteristics are at a second intermediate threshold T over a second duration (h) starting from the predicted EOL IT2 it is possible to trigger a second communication or control activity, the second duration being less than the first duration and thus closer to h.

[0133] In the case of a threshold defined by the distance from the occurrence of a fault, if certain battery characteristics deteriorate, then the occurrence of a trigger can be defined. As an example, if the internal resistance of the battery is the monitored characteristic, then if the internal resistance is within a calibrated distance from the internal resistance corresponding to the end of life, the trigger can be activated, as Figure 8 shown in the mapping diagram 820. Where the threshold defines a first intermediate internal resistance threshold, defines a second intermediate internal resistance threshold, and h represents the threshold at the predicted end of life. When the battery characteristics are at the first intermediate threshold, a first communication or control activity can be triggered, while when the battery characteristics are at the second intermediate threshold, a second communication or control activity can be triggered, the second threshold being closer to h than the first threshold.

[0134] The distance from the occurrence of a fault can be defined by multiple characteristics. If the battery capacity is the second monitored characteristic, then the trigger can occur based on both the internal resistance and the capacity. If the battery resistance or the battery capacity loss exceeds a separate threshold, or if both the battery resistance and the battery capacity loss exceed a separate threshold, or if the values of the battery resistance and capacity in the "characteristic space" are within a predefined distance from the end-of-life threshold, then a trigger may occur, as Figure 6 shown. In this way, the end-of-life threshold in the characteristic space can be expanded by intermediate thresholds representing a specific time before reaching the end of life or the distance from the end of life. Thus, the distance from the end of life can be interpreted as the distance from the component failure.

[0135] When an intermediate end-of-life threshold representing the distance or time from the occurrence of a fault is exceeded, a communication strategy can be triggered. The communication can be directed to the driver and to an entity such as the vehicle manufacturer. Additionally, if a radio data link (e.g., a modem) is implemented to communicate with the outside world, then the communication can be directed to a local car dealership or a fleet owner. In addition to the communication, one or more actions can be recommended in view of the upcoming EOL of the battery.

[0136] Since the 12V accessory battery in an autonomous vehicle supports safety-critical systems in certain operating modes, if a defective battery or a battery approaching its end-of-life is identified, then the autonomous function can be disabled. Suddenly disabling the autonomous function can lead to customer dissatisfaction and inconvenience, especially in cases where the vehicle cannot be driven manually. Therefore, the driver and maintenance personnel must be warned of an impending battery failure before it occurs, so that the battery can be replaced. For this reason, predicting the time before the battery reaches its end-of-life is of particular importance in autonomous vehicles.

[0137] In addition to customer satisfaction issues, it has been shown that if the end-of-life of the battery is impending but has not been reached, then the autonomous function can be disabled or scaled back. In such a case, if the EOL of the battery is predicted within a predefined time, then the restriction will take effect.

[0138] Therefore, communication strategies can be used to alert the driver and maintenance personnel of an impending battery failure while taking into account marketing, customer satisfaction, warranty costs, and other aspects. Generally, when a corresponding threshold is exceeded, the strategy can communicate directly with the maintenance personnel and the driver on separate occasions. The communication can represent a countdown before reaching the end-of-life. Similarly, the autonomous function can be restricted in a step-by-step manner as the end-of-life of the battery is approached. Each time a threshold is exceeded, the restriction can be stepped up until the vehicle is no longer allowed to drive itself, and manual driving can also be scaled back.

[0139] These types of communication and control strategies that send messages to the driver and maintenance personnel and scale back functionality in a step-by-step manner as a predicted end-of-life of the battery can be applied to conventional vehicles. The driver may receive a warning about an impending battery and expect to lose the assistance of the steering system or have the climate control or entertainment system shut down periodically. Both of these types of events are symptoms of a failing 12V accessory battery.

[0140] When an intermediate end-of-life threshold representing the distance or time to failure is exceeded, communication strategies, such as Figure 13 the example communication strategy of method 1300, may be triggered. If a radio data link is implemented to communicate with the outside world, then the communication can be directed to the driver and entities such as the vehicle manufacturer or the automobile. Additionally, the communication can take the form of a marker written to the non-volatile memory in the vehicle, which indicates when the threshold has been exceeded. The maintenance personnel can read these markers to facilitate the testing and replacement of the battery.

[0141] Communication with the driver himself can take the form of warning lights or text on the dashboard, emails sent to the driver's account or SMS or notifications to an application on the driver's mobile device. If the end of life is exceeded, then when the threshold is closer to the predicted end, the communication strategy can trigger new or different means of communication with the driver or the outside world. For example, if the estimated time to failure is within three months, then an indicator on the dashboard can be activated. If the battery is not replaced and the predicted time to failure is within less than two months, then when the vehicle is started, a text message can appear on the dashboard. Exceeding subsequent thresholds can trigger sending an email to the driver and alerting the dealer.

[0142] In Figure 13 TTF represents the estimated time to failure (end of life time), and T IT1 , ……, T IT4 represent intermediate thresholds expressed in terms of time. The communication strategy can be activated whenever the vehicle is in motion or periodically when the vehicle is parked. When the first intermediate threshold is exceeded at 1302, at 1302, an indicator on the dashboard can be activated. When the second intermediate threshold is exceeded at 1306, at 1308, when the vehicle is started, a text message can appear on the dashboard or a backup multifunction display. When the third intermediate threshold is exceeded at 1310, at 1312, the controller can send an email to alert the driver. When the fourth intermediate threshold is exceeded at 1314, the controller can send an alert to the dealer. It will be understood that the communication actions written Figure 13 in the box can include additional complex tasks not described in Figure 13 . For example, the actions of "sending an email to the owner" or "sending an email to the dealer" can include another logic for enabling the email to be set only once a day.

[0143] Alternatively, the controller can use a general communication strategy, such as Figure 14 shown in method 1400. Where the threshold is expressed relative to the distance to failure. The thresholds defined in terms of time can be redefined relative to the distance to failure using the following equation:

[0144]

[0145] Doing so allows defining a communication strategy defined by both the time to failure and the distance to failure. For example, when the time to failure is less than the threshold or the internal resistance rises to a defined level, the strategy can trigger communication.

[0146] The communication can be activated whenever the vehicle is in motion or periodically when the vehicle is parked. Figure 14Communication strategy. When the first intermediate threshold is exceeded at 1402, a first communication Com1 is transmitted at 1402. When the second intermediate threshold is exceeded at 1406, a second communication Com2 is transmitted at 1408. When the third intermediate threshold is exceeded at 1410, a third communication Com3 is transmitted at 1412. When the fourth intermediate threshold is exceeded at 1414, a fourth communication Com4 is transmitted at 1416.

[0147] Figure 14 The communication actions in it are generally designated as Com1, ……, Com4. As Figure 13 In the strategy shown in it, the individual communication actions may be complex and have internal controls to modify or delay their transmission according to external factors. For example, they may communicate only once a day or when the vehicle ignition switch is turned off.

[0148] It will be understood that communication can take various forms, including those described above and other forms.

[0149] In a manner similar to the communication strategy, different control strategies can be activated when an intermediate threshold representing the distance or time to a fault occurrence is exceeded. As an example, a set of control strategies can be implemented to limit the autonomous operation of the vehicle or to limit or deactivate individual systems or functions in a vehicle with an active suspension according to the exceeded intermediate threshold. As an example, if the vehicle is operating autonomously, then steering, braking, and powertrain control functions can be achieved by electric actuators. These can be servo systems for maneuvering the vehicle (just like a by-wire system) and electric pumps and valves for actuating the brakes. In addition to actuation, control is also done by a microprocessor, which also relies on electricity for operation. Electric actuators are typically characterized by high current transients when they are turned on or when they are controlled to apply a high force. The main power source of a modern vehicle can be an alternator or generator when an internal combustion engine is used for propulsion or a high voltage battery and a DC-DC converter when the vehicle has an electric propulsion system. In either case, the alternator, generator, or DC-DC converter generally cannot supply the transients drawn by the active chassis system. Therefore, a 12V SLI battery is implemented in parallel with the (low voltage) distribution network to provide power when the main power source is saturated.

[0150] When these batteries degrade and are no longer able to provide sufficient power when needed, actuation may be limited or may not occur at all. This means that steering and braking may occur slowly and / or the desired level of actuation may not be achieved. Although actuation occurs in degraded batteries, the voltage on the power distribution network can be reduced to significant levels, which deactivates or resets the microprocessor controlling the vehicle and does not allow other actuators to operate correctly. The reduction in actuation and critical voltage levels on the power distribution network may cause the vehicle to deviate from its desired path and pose a safety hazard. Therefore, the 12V SLI battery can be classified as a safety-critical component in autonomous vehicles. If the time or distance to the predicted end-of-life of the battery becomes less than a defined threshold, control strategies can be implemented to limit autonomous functionality.

[0151] In one example, six levels of vehicle automation can be defined (e.g., numbered 0 to 5). For example, the Society of Automotive Engineers (SAE) has defined six levels of vehicle automation, numbered 0 to 5. As the level increases, the vehicle becomes more automated. At level 3 (the midpoint), the vehicle can be autonomously controlled without human intervention, but the driver is required to be constantly ready to take over control of the vehicle in case of problems or when exceeding the limits of the control system. At level 4, full autonomy is possible in certain driving modes. In these modes, the driver does not have to be ready to perform control operations at all times. Level 5 is the highest level, at which no human intervention is required at any stage of vehicle operation. If the time or distance to a failing battery is below a calibrated threshold, autonomous functionality can be restricted along the vehicle automation horizontal line defined above. In other words, if the time or distance to failure falls between calibrated amounts, a vehicle capable of level 4 or 5 operation can be restricted to level 3. According to such a policy, only functionality with driver supervision will be allowed. This will be communicated to the driver and may require his confirmation before driving begins. If control strategies are defined along these lines, the operation of the autonomous vehicle can be restricted to level 3 if the time or distance to the failing battery is below a calibrated intermediate threshold. If the time or distance to failure is below a second intermediate threshold closer to the EOL prediction, functionality can be restricted to a lower level of functionality.

[0152] Reference Figure 15 Method 1500 of [reference] shows an example control strategy. Wherein the controller can use a threshold for activating the control mode, which is expressed relative to the predicted time to failure. As in the communication strategy ( Figures 13 to 14) Similarly, the threshold can be defined relative to the distance from the fault occurrence, and the same applies to the hybrid strategy using two types of thresholds. In the case of a control strategy that restricts the functionality of an autonomous vehicle, it can be expected that the restriction of the functionality may be applied only at the start of driving when the driver has the opportunity to see and confirm the restriction of the functionality. Thus, in one example, the strategy shown in Figure 15 can be activated when the vehicle ignition switch is turned on. The control strategy that restricts the autonomous functionality of the vehicle Figure 15 can be coordinated with the Figure 13 or Figure 14 communication strategy to explain the reason for the restriction to the driver and convey this information and additional relevant information to the maintenance personnel of the dealership.

[0153] Whenever the vehicle ignition switch is turned on, the control strategy shown in Figure 15 can be activated. When the first intermediate threshold is not exceeded at 1502, at 1512, the controller can allow the unrestricted AV functionality of the autonomous vehicle. If the first intermediate threshold is exceeded at 1504 but the second intermediate threshold is not exceeded, then the method moves to 1506, where the controller activates a message in the multifunction display (e.g., on the dashboard) that indicates the application of a Level 2 AV restriction. If, after the message is activated, no driver confirmation is received at 1508, then the method returns to 1506 to continue displaying the message. Otherwise, after receiving the driver confirmation, at 1510, the AV functionality is restricted to Level 2 (out of Levels 0 to 5 defined by SAE).

[0154] If both the first intermediate threshold and the second intermediate threshold are exceeded at 1502 and 1504, then the method moves to 1514, where the controller activates a message in the multifunction display (e.g., on the dashboard) that indicates the application of a Level 3 AV restriction. If, after the message is activated, no driver confirmation is received at 1516, then the method returns to 1514 to continue displaying the message. Otherwise, after receiving the driver confirmation, at 1516, the AV functionality is restricted to Level 3 (out of Levels 0 to 5 defined by SAE).

[0155] As described above, active chassis components that apply high mechanical forces in a chassis system may draw large power transients when they are activated or when their output forces are controlled to rapidly step up. Examples of active chassis components include electric power assisted steering (EPAS), electric brakes, and electrically actuated roll control (eARC). Functional limitations for some non-essential active chassis system functions can be implemented on a conventional or autonomous vehicle to maintain the functionality of those that are necessary for safety. If these components are powered by a 12V electrical distribution network, they may produce large voltage drops when activated unless the network is supported by a fully charged SLI battery. If the battery is at the end of its life, high internal resistance, low capacity, internal short circuits, and other aging-related fault mechanisms may limit its ability to support these loads. Non-essential active chassis systems can be deactivated to support the functionality of systems that are considered more critical for safety. In a vehicle with EPAS, electric brakes, and electric roll control, if the time or distance to the end of the battery's life is less than a threshold, roll control can be deactivated.

[0156] An example of such a control strategy is shown at method 1600 of Figure 16 . As in the case of communication and control strategies for autonomous vehicles, thresholds can also be defined relative to the distance to a failure and a hybrid strategy using both types of thresholds can be used. Figure 16 The strategy shown in

[0157] is a specific implementation of a general control strategy that can limit or deactivate any non-essential high-power loads when the battery reaches the end of its life. Figure 16 Whenever the vehicle ignition switch is turned on, the control strategy of Figure 11 can be activated. When a first intermediate threshold is not exceeded at 1602, all settings are maintained. If the first intermediate threshold is exceeded, then at 1604, the controller activates a message in a multi-function display (e.g., on the instrument panel) that indicates that the electrically actuated roll control eARC is about to be deactivated. At 1606, after the activation message, eARC is deactivated. In this way, when an intermediate threshold is exceeded, non-essential vehicle systems can be deactivated to support the functionality of systems that are considered more essential (e.g., safety-critical). Another type of control action that can occur at the end of the battery's life is to increase the charge voltage to help maintain a high state of charge in the remaining limited capacity. Such a control strategy is shown in method 1100 of IT2If not, then the battery charge voltage is defined by a function Z1 having a lower voltage. Otherwise, if a threshold duration has elapsed, then the battery charge voltage is defined by a function Z2 having a higher voltage. Herein, Z1 and Z2 represent temperature-dependent charge-voltage curves. In this example, the voltage defined by Z2 will be higher than the voltage defined by Z1 to facilitate charging of the battery.

[0158] Although Figures 13 to 16 different communication and control action strategies are shown, in other examples, communication and control action strategies can be combined, such as in Figure 9 the example strategies of. Just as with individual control and communication strategies, Figure 9 method 900 uses intermediate thresholds. Where TTF represents the estimated time to failure (end-of-life time), and T IT1 to T IT4 represent intermediate thresholds expressed in time. Whenever the vehicle is in motion or periodically when the vehicle is parked, the communication strategy can be activated. Although the intermediate thresholds T IT1 to T IT4 are shown as time thresholds at method 900, it will be understood that in alternative examples, without departing from the scope of the present invention, intermediate distance thresholds can be similarly applied. In the example shown, the method is used for battery monitoring of an autonomous vehicle (AV) configured with autonomous driving capabilities.

[0159] At 902, the method includes initiating battery monitoring. At 904, the method includes whether a first intermediate threshold (e.g., a first time or distance threshold) has been reached. If the first threshold has not been reached, then at 920, vehicle operation can continue without any restrictive actions. For example, unrestricted functionality of the autonomous vehicle can be allowed.

[0160] If a first intermediate threshold has been reached, then at 905, the method includes performing a first communication action (generally denoted as Com1), such as by activating a dashboard indicator. For example, the first communication with the driver can take the form of a warning light or text on the dashboard. Optionally, at 906, it can be determined whether an acknowledgement has been received, such as can be acknowledged when the vehicle driver actuates a button or interacts with a touch display on the vehicle's dashboard. After acknowledgement, at 907, a first action can be taken. Otherwise, the method can move directly to 908. The first action can include, for example, limiting the autonomous functionality of the vehicle to a first amount to a first level, below the full autonomous functionality level. Limiting the autonomous functionality to the first level can include, for example, limiting the autonomous functionality to "conditional automation" corresponding to level 3 in the vehicle automation levels (SAE standard J3016 as of September 22, 2016) published by the Society of Automotive Engineers. Alternatively, one or more non-essential functions of the vehicle (such as eARC for roll control) can be disabled.

[0161] The method then moves to 908 to determine whether a second intermediate threshold (e.g., a second time or distance threshold) has been reached, the second threshold being closer to the EOL threshold (h) of the component than the first threshold. In other examples, the method can move directly from 905 to 908, and at 905, the limitation of limiting the autonomous functionality to the first level can be included in the first communication with the driver.

[0162] If a second intermediate threshold has been reached, then at 909, the method includes performing a second communication action (generally denoted as Com2), such as by displaying a text message in a vehicle's multifunctional display. For example, the second communication with the driver can take the form of text displayed on the dashboard, an SMS sent to the driver's account, or a notification sent to an application running on the driver's mobile device. Optionally, at 910, it can be determined whether an acknowledgement has been received. After the acknowledgement, at 911, a second action can be performed. Otherwise, the method can move directly to 912. The second action can include, for example, restricting the autonomous functionality of the vehicle to a second level by a second amount greater than the first amount, the second level being lower than the first level. Restricting the autonomous functionality to the second level can include, for example, restricting the autonomous functionality to "partial automation" corresponding to level 2 in the vehicle automation levels published by the Society of Automotive Engineers. Alternatively, one or more non-essential additional functions of the vehicle can be disabled. The method then moves to 912 to determine whether a third intermediate threshold (e.g., a third time or distance threshold) has been reached, the third threshold being closer to the EOL threshold (h) of the component than the first threshold or the second threshold. In other examples, the method can move directly from 909 to 912, and at 909, the restriction of the autonomous functionality to the second level can be included in the second communication with the driver.

[0163] If a third intermediate threshold has been reached, then at 913, the method includes performing a third communication action (generally denoted as Com3), such as by sending an email to the driver's account or an application running on the driver's mobile device. Optionally, at 914, it can be determined whether an acknowledgement has been received. After the acknowledgement, at 915, a third action can be performed. Otherwise, the method can move directly to 916. The third action can include, for example, restricting the autonomous functionality of the vehicle to a third level by a third amount greater than the second amount, the third level being lower than the second level. Restricting the autonomous functionality to the third level can include, for example, disabling all autonomous functions and, when possible, driving the vehicle manually only. Alternatively, all non-essential functions of the vehicle can be disabled.

[0164] The method then moves to 916 to determine whether a fourth intermediate threshold (e.g., a fourth time or distance threshold) has been reached, the fourth threshold being closer to the EOL threshold (h) of the component than any of the first threshold, the second threshold, or the third threshold. In other examples, the method can move directly from 913 to 916, and at 913, the restriction of the autonomous functionality to the third level can be included in the second communication with the driver.

[0165] If a fourth intermediate threshold has been reached, then at 917, the method includes performing a fourth communication action (generally denoted as Com4), such as sending an email to a dealer. Additionally, at 918, a fourth final action may be taken. The fourth action may include, for example, completely disabling the vehicle. Additionally, at 918, the method may include continuing to operate the vehicle with restricted autonomous functionality. Although the method shows four intermediate thresholds, in other examples, fewer or greater numbers of intermediate thresholds may be included.

[0166] In this manner, the controller can predict the time to end-of-life of a vehicle battery (such as a lead-acid 12V SLI battery commonly used in vehicles to start and run accessories when the motor is not running), and support high power load transients. Any of the methods and strategies discussed herein can be applied to autonomous vehicles, as well as conventional gasoline or hybrid electric vehicles. The general form of the control and communication strategies described herein can be applied to any vehicle that uses an SLI battery. Generally, when the predicted time to end-of-life is less than a calibrated threshold, these strategies trigger communication and control actions. The nature of the communication and control actions may change as the predicted time to end-of-life decreases. Gradually increasing warnings sent via dashboard messages and emails as end-of-life approaches can be applied to any vehicle. The same can be said for control strategies. For example, as end-of-life approaches, the charge voltage can be gradually increased, and the battery charging strategy can be modified to extend the ability of the battery to provide basic functionality. The methods herein for predicting the end-of-life of an SLI battery can also be applied to traction batteries or other batteries. This may require selecting new battery characteristics to characterize the end-of-life condition. Any method for predicting the time to end-of-life of a component can be used in combination with the general control and communication strategies described herein.

[0167] It will also be appreciated that although the methods disclosed herein predict the expected life of a vehicle battery, the methods can be similarly used to predict the expected life of multiple components in a road vehicle, such as tires, filters, and lubricants.

[0168] In this way, a prognosis-based method for assessing the remaining useful life of a vehicle battery is provided. The prognosis method can be used to complement any existing prognosis features by estimating the remaining time and / or remaining distance before battery degradation. By defining thresholds for each battery characteristic that affects battery life based on changes in the internal properties of the battery, such as resistance and capacity, diagnosis of the current state of the battery and prognosis of its future health (including an estimate of the time to end of battery life) can be achieved. By relying on the rate of convergence of the sensed battery characteristics towards the estimated corresponding thresholds, the different effects of different battery degradation mechanisms, such as corrosion and sulfation, can be taken into account. In addition, the trajectory of each characteristic can be used to better estimate the end of battery life. By more accurately estimating the remaining life of a component, it can be provided to the vehicle driver as a more understandable metric, which reduces warranty issues and improves customer satisfaction.

[0169] In one example, a method for a vehicle includes: predicting a degradation state of a vehicle battery based on a convergence rate of a plurality of battery metrics derived from sensed vehicle operating parameters toward corresponding thresholds, the thresholds being determined based on past driving history data including a past history of each of the plurality of battery metrics; and converting the predicted degradation state into a remaining time or duration before end of battery life for display to a vehicle driver. In the foregoing example, additionally or optionally, the plurality of battery metrics includes battery resistance and battery capacity, and the sensed vehicle operating parameters include one or more of battery current, battery voltage, and battery terminal temperature. In any or all of the foregoing examples, additionally or optionally, the prediction includes predicting a higher degradation state with an increase in the convergence rate of any one of the plurality of battery metrics. In any or all of the foregoing examples, additionally or optionally, the method further includes updating the corresponding threshold based on vehicle performance after the conversion. In any or all of the foregoing examples, additionally or optionally, the method further includes: comparing the predicted degradation state with an end-of-life threshold and one or more intermediate thresholds between a current degradation state and the end-of-life threshold; and restricting one or more vehicle functions based on the comparison. In any or all of the foregoing examples, additionally or optionally, the vehicle is an autonomous vehicle, and wherein the restriction includes: operating the vehicle without restricting any autonomous vehicle functionality when the predicted degradation state is below each of the one or more intermediate thresholds and the end-of-life threshold; operating the vehicle with a first degree of restriction of autonomous vehicle functionality when the predicted degradation state is above a first of the one or more intermediate thresholds; operating the vehicle with a second degree of restriction of autonomous vehicle functionality higher than the first degree when the predicted degradation state is above a second of the one or more intermediate thresholds greater than the first of the one or more intermediate thresholds; and operating the vehicle with not all autonomous vehicle functionality restricted when the predicted degradation state is above each of the one or more intermediate thresholds and the end-of-life threshold. In any or all of the foregoing examples, additionally or optionally, the sensed vehicle operating parameters are sensed during transient and steady-state vehicle operating conditions, and the parameters sensed during transient conditions are weighted differently than the parameters sensed during steady-state conditions. In any or all of the foregoing examples, additionally or optionally, the method further includes estimating a value of a determined metric based on a most recent estimate of the determined metric retrieved from the past history of the determined metric and a distance traveled by the vehicle since the most recent estimate of the determined metric.In any or all of the foregoing examples, additionally or optionally, the method further includes estimating a value of the determined metric based on an initial estimate of the determined metric retrieved from the past history of the determined metric when the component is installed in the vehicle. In any or all of the foregoing examples, additionally or optionally, the method further includes converting the predicted degradation state into a remaining number of engine start events based on the past driving history data and the predicted future driving to be displayed to the vehicle driver. In any or all of the foregoing examples, additionally or optionally, the threshold is determined outside the vehicle, while the convergence rate is determined on the vehicle. In any or all of the foregoing examples, additionally or optionally, the vehicle is one of a plurality of vehicles in a fleet, and the method further includes: estimating the plurality of battery metrics of each vehicle in the fleet for at least a threshold duration; and predicting the degradation state of the vehicle battery in response to the estimation. In any or all of the foregoing examples, additionally or optionally, the method further includes: updating each of the end-of-life threshold and the one or more intermediate thresholds of the vehicle in response to the performance of each vehicle in the fleet after the prediction.

[0170] Another example method for predicting the battery health of a vehicle includes: using one or more on-vehicle battery monitoring sensors to monitor at least one battery health parameter in real time; determining a threshold for the monitored battery health parameter based on information collected from a vehicle communication network and vehicle operating conditions; defining a battery end-of-life prediction algorithm based on the convergence rate of the monitored battery health parameter towards the determined threshold; and estimating the end-of-life of the battery based on the prediction algorithm. In any or all of the foregoing examples, additionally or optionally, the method further includes: restricting one or more autonomous functions of the vehicle based on the estimation, the degree of restriction being based on the estimated end-of-life relative to the end-of-life threshold. In any or all of the foregoing examples, additionally or optionally, the method further includes: restricting one or more non-essential autonomous functions of the vehicle, such as the vehicle's electro-mechanical anti-roll control system, based on the estimation, the degree of restriction being based on the estimated end-of-life relative to the end-of-life threshold. In any or all of the foregoing examples, additionally or optionally, the vehicle is one of a plurality of vehicles in a vehicle fleet, and wherein determining the threshold includes determining the threshold based on battery end-of-life information collected from each of the plurality of vehicles in the fleet and received via the vehicle communication network.

[0171] An example vehicle system includes: a battery; one or more sensors coupled to the battery; a motor driven by power drawn from the battery; an engine; a network that communicatively couples the vehicle system to one or more additional vehicles of a fleet; a display; and a controller having computer-readable instructions for: predicting a remaining duration before end-of-life of the battery based on a rate of convergence of a plurality of sensed battery parameters toward corresponding parameter thresholds, the thresholds being determined based on the battery history of the vehicle system and each of the one or more additional vehicles of the fleet; comparing the predicted duration to an end-of-life threshold; and restricting one or more functions of the vehicle based on the comparison. In any or all of the foregoing examples, additionally or optionally, the controller includes other instructions for: when the predicted duration is above the end-of-life threshold, displaying the predicted duration to a vehicle driver on the display and restricting autonomous functionality of the vehicle. In any or all of the foregoing examples, additionally or optionally, restricting one or more functions of the vehicle based on the comparison includes restricting one or more of power steering assist, climate control, and entertainment system operation. In any or all of the foregoing examples, additionally or optionally, the controller includes other instructions for: updating the end-of-life threshold based on vehicle performance of the vehicle system after the prediction and each of the one or more additional vehicles of the fleet, the end-of-life threshold being reduced in response to a decline in vehicle performance.

[0172] Note that the example control and estimation routines included herein can be used with a variety of engine and / or vehicle system configurations. The control methods and procedures disclosed herein can be stored as executable instructions in non-transitory memory and can be executed by a control system including a controller in conjunction with various sensors, actuators, and other engine hardware. The specific routines described herein can represent one or more of any number of processing strategies (such as event-driven, interrupt-driven, multi-tasking, multi-threaded, etc.). Accordingly, the various acts, operations, and / or functions illustrated can be executed in the illustrated sequence, executed in parallel, or in some cases omitted. Likewise, the order of processing is not necessarily required to implement the features and advantages of the example embodiments described herein, but is provided for ease of illustration and description. One or more of the illustrated acts, operations, and / or functions can be repeated in accordance with the particular strategy being used. Further, the acts, operations, and / or functions can be graphically represented as code to be programmed into the non-transitory memory of a computer-readable storage medium for an engine control system, where the described acts are executed by executing instructions in a system including various engine hardware components in conjunction with an electronic controller.

[0173] It will be understood that the configurations and procedures disclosed herein are exemplary in nature and these specific embodiments should not be regarded as limiting since many variations are possible. For example, the above techniques can be applied to batteries for starting, supporting transient loads, or propulsion with various chemicals. The subject matter of the present disclosure includes all novel and non-obvious combinations and sub-combinations of the various systems and configurations as well as other features, functions, and / or properties disclosed herein.

[0174] The following claims particularly point out certain combinations and sub-combinations regarded as novel and non-obvious. These claims may refer to "one" element or "first" element or the equivalent thereof. These claims should be understood to include the combination of one or more such elements, neither requiring nor precluding the combination of two or more such elements. Other combinations and sub-combinations of the disclosed features, functions, elements, and / or characteristics may be claimed by modifying these claims or presenting new claims in this application or a related application. These claims, whether broader, narrower, the same, or different in scope from the original claims, are also regarded as included within the subject matter of the present disclosure.

[0175] According to the present invention, a method for a vehicle includes: predicting a degradation state of a vehicle battery based on a convergence rate of a plurality of battery metrics derived from sensed vehicle operating parameters towards corresponding thresholds, the thresholds being determined based on past driving history data including a past history of each of the plurality of battery metrics; and converting the predicted degradation state into a remaining time or duration before end-of-battery life for display to a vehicle driver.

[0176] According to one embodiment, the plurality of battery metrics includes battery resistance and battery capacity, and the sensed vehicle operating parameters include one or more of battery current, battery voltage, and battery terminal temperature.

[0177] According to one embodiment, the prediction includes predicting a higher degradation state as the convergence rate of any one of the plurality of battery metrics increases.

[0178] According to one embodiment, the present invention is further characterized in that the corresponding threshold is updated based on vehicle performance after the conversion.

[0179] According to one embodiment, the present invention is further characterized in that the predicted degradation state is compared with an end-of-life threshold and one or more intermediate thresholds between the current degradation state and the end-of-life threshold; and one or more vehicle functions are restricted based on the comparison.

[0180] According to one embodiment, the vehicle is an autonomous vehicle, and wherein the limitations include: operating the vehicle without restricting any autonomous vehicle functionality when the predicted degradation state is below each of the one or more intermediate thresholds and the end-of-life threshold; operating the vehicle with a first degree of restriction on autonomous vehicle functionality when the predicted degradation state is above a first intermediate threshold of the one or more intermediate thresholds; operating the vehicle with a second degree of restriction on autonomous vehicle functionality that is higher than the first degree when the predicted degradation state is above a second intermediate threshold of the one or more intermediate thresholds that is greater than the first intermediate threshold of the one or more intermediate thresholds; and operating the vehicle without autonomous vehicle functionality when the predicted degradation state is above each of the one or more intermediate thresholds and the end-of-life threshold.

[0181] According to one embodiment, the invention is further characterized in that the value of the determined metric is estimated based on the most recent estimate of the determined metric retrieved from the past history of the determined metric and the distance traveled by the vehicle since the most recent estimate of the determined metric.

[0182] According to one embodiment, the invention is further characterized in that the value of the determined metric is estimated based on an initial estimate of the determined metric retrieved from the past history of the determined metric when the component is installed in the vehicle.

[0183] According to one embodiment, the invention is further characterized in that the predicted degradation state is converted into the remaining number of engine start events based on the past driving history data and the predicted future driving to be displayed to the vehicle driver.

[0184] According to one embodiment, the threshold is determined outside the vehicle, while the convergence rate is determined on the vehicle.

[0185] According to one embodiment, the vehicle is one of a plurality of vehicles in a fleet, and the method further includes: estimating the plurality of battery metrics of each vehicle in the fleet for at least a threshold duration; and predicting the degradation state of the vehicle battery in response to the estimation.

[0186] According to one embodiment, the invention is further characterized in that each of the end-of-life threshold and the one or more intermediate thresholds of the vehicle is updated in response to the performance of each vehicle in the fleet after the prediction.

[0187] According to the present invention, a method for predicting the battery health status of a vehicle includes: using one or more on-vehicle battery monitoring sensors to monitor at least one battery health parameter in real time; determining a threshold for the monitored battery health parameter based on information collected from a vehicle communication network and vehicle operating conditions; defining a battery end-of-life prediction algorithm based on the convergence rate of the monitored battery health parameter towards the determined threshold; and estimating the end-of-life of the battery based on the prediction algorithm.

[0188] According to one embodiment, the present invention is further characterized in that, based on the estimation, one or more autonomous functions or safety-related functions of the vehicle that require power are restricted, and the degree of restriction is based on the estimated end-of-life relative to the end-of-life threshold.

[0189] According to one embodiment, the present invention is further characterized in that, based on the estimation, one or more non-essential power-requiring systems of the vehicle are restricted, and the degree of restriction is based on the estimated end-of-life relative to the end-of-life threshold, and the one or more non-essential power-requiring systems of the vehicle include an electro-mechanical roll control system.

[0190] According to one embodiment, the vehicle is one of a plurality of vehicles in a vehicle fleet, and wherein determining the threshold includes determining the threshold based on battery end-of-life information collected from each of the plurality of vehicles in the fleet and received via the vehicle communication network.

[0191] According to the present invention, a vehicle system is provided, the vehicle system having: a battery; one or more sensors coupled to the battery; a motor driven by power drawn from the battery; an engine; a network communicatively coupling the vehicle system to one or more additional vehicles in a fleet; a display; and a controller having computer-readable instructions for performing the following operations: predicting the remaining duration before the end-of-life of the battery based on the convergence rate of a plurality of sensed battery parameters towards corresponding parameter thresholds, the thresholds being determined based on the battery history of the vehicle system and each of the one or more additional vehicles in the fleet; comparing the predicted duration with an end-of-life threshold; and restricting one or more functions of the vehicle based on the comparison.

[0192] According to one embodiment, the controller includes other instructions for performing the following operations: when the predicted duration is higher than the end-of-life threshold, displaying the predicted duration on the display to the vehicle driver and restricting the autonomous functionality of the vehicle.

[0193] According to one embodiment, restricting one or more functions of the vehicle based on the comparison includes restricting one or more of power steering assist, climate control, and entertainment system operation.

[0194] According to one embodiment, the invention is further characterized by a battery life prediction system communicatively coupled to the controller and the one or more sensors, wherein the controller includes additional instructions for: updating the end-of-life threshold based on monitoring the performance of the battery life prediction system in each of the one or more additional vehicles of the fleet after the prediction, the end-of-life threshold being lowered in response to a decline in the performance of the prediction system.

Claims

1. A method for a vehicle, comprising: Predicting a degradation state of a vehicle battery based on a convergence rate of a plurality of battery metrics derived from sensed vehicle operating parameters towards corresponding thresholds, the thresholds being determined based on past driving history data and end-of-life battery information collected from each of a plurality of vehicles in a fleet and received via a network, the past driving history data including a past history of each of the plurality of battery metrics, the network communicatively coupling the vehicle to one or more additional vehicles in the fleet; Converting the predicted degradation state into a remaining time or duration before end-of-life of the battery for display to a vehicle driver; And Updating the end-of-life threshold based on vehicle performance after the conversion, the updating including adding a change in the end-of-life threshold determined by a transfer function based on a change in the vehicle performance to an initial threshold, wherein the transfer function maps the change in the performance to the change in the end-of-life threshold.

2. The method of claim 1, wherein the plurality of battery metrics includes battery resistance and battery capacity, and the sensed vehicle operating parameters include one or more of battery current, battery voltage, and battery terminal temperature.

3. The method of claim 2, wherein the predicting includes predicting a higher degradation state as the convergence rate of any one of the plurality of battery metrics increases.

4. The method of claim 1, the method further comprising: Comparing the predicted degradation state with the end-of-life threshold and one or more intermediate thresholds between the current degradation state and the end-of-life threshold; And Limiting one or more vehicle functions based on the comparison.

5. The method of claim 4, wherein the vehicle is an autonomous vehicle, and wherein the limiting includes: Operating the vehicle without restricting any autonomous vehicle functionality when the predicted degradation state is below each of the one or more intermediate thresholds and the end-of-life threshold; Operating the vehicle with a first degree of restriction of autonomous vehicle functionality when the predicted degradation state is above a first intermediate threshold of the one or more intermediate thresholds; Operating the vehicle with a second degree of restriction of autonomous vehicle functionality higher than the first degree when the predicted degradation state is above a second intermediate threshold of the one or more intermediate thresholds greater than the first intermediate threshold of the one or more intermediate thresholds; And Operating the vehicle without autonomous vehicle functionality when the predicted degradation state is above each of the one or more intermediate thresholds and the end-of-life threshold.

6. The method according to claim 1, the method further comprising: Estimating a value of a determined metric based on a most recent estimate of the determined metric retrieved from the past history of the determined metric and a distance traveled by the vehicle since the most recent estimate of the determined metric.

7. The method according to claim 1, the method further comprising: Estimate the value of the determined metric based on an initial estimate of the determined metric retrieved from the past history of the determined metric when the component is installed in the vehicle.

8. The method according to claim 1, wherein the method further comprises: Convert the predicted degradation state into the remaining number of engine start events based on the past driving history data and the predicted future driving to display to the vehicle driver.

9. The method according to claim 1, wherein the threshold is determined outside the vehicle, and the convergence rate is determined on the vehicle.

10. The method according to claim 5, the method further comprising: Estimate the plurality of battery metrics of each vehicle in the fleet for at least a threshold duration; and Predict the degradation state of the vehicle battery in response to the estimation.

11. The method according to claim 10, the method further comprising: Update each of the end-of-life threshold and the one or more intermediate thresholds of the vehicle in response to the performance of each vehicle in the fleet after the prediction.

12. A vehicle system, the vehicle system comprising: A battery; One or more sensors coupled to the battery; A motor driven by the power drawn from the battery; An engine; A network communicatively coupling the vehicle system to one or more additional vehicles in a fleet; A display; and A controller having computer-readable instructions for: Predicting the remaining duration before the end of life of the battery based on the convergence rate of a plurality of sensed battery parameters towards corresponding parameter thresholds, the thresholds being determined based on the battery history of the vehicle system and each of the one or more additional vehicles in the fleet; Comparing the predicted duration with an end-of-life threshold; Limiting one or more functions of the vehicle based on the comparison; and Updating the end-of-life threshold based on monitoring the performance of the battery life prediction system of each of the one or more additional vehicles in the fleet after the prediction, the update including adding a change in the end-of-life threshold determined by a transfer function based on the change in the performance to an initial end-of-life threshold, wherein the transfer function maps the change in the performance to the change in the end-of-life threshold.

13. The system according to claim 12, wherein the controller includes other instructions for: When the predicted duration is higher than the end-of-life threshold, display the predicted duration to the vehicle driver on the display and limit the autonomous functionality of the vehicle.

14. The system according to claim 13, wherein limiting one or more functions of the vehicle based on the comparison includes limiting one or more of power steering assist, climate control, and entertainment system operation, wherein the battery life prediction system is communicatively coupled to the controller and the one or more sensors, and wherein the end-of-life threshold is decreased in response to a decrease in the performance of the prediction system.

Images