Invented by Pankaj Kumar, Imad Hassan Makki, Fakhreddine Landolsi, Mark Eifert, Sheida Malekpour, David W. Linden, Ford Global Technologies LLC
Market For Method to Predict Battery Life
The battery industry is expected to experience rapid growth over the coming years. To ensure optimal usage and maximize battery life, a method must be developed that accurately predicts its remaining duration.
Predicting battery life has been achieved using both model-based and data-driven methods, with the latter emerging as a more viable choice due to its ability to accurately forecast remaining service life with little or no input data.
Market Analysis by Application
Researchers at Stanford University and Massachusetts Institute of Technology have devised a method that can accurately forecast how long batteries will remain full before they need replacing. This new algorithm has the potential to speed up research into battery technology as well as improve manufacturing processes.
A study published March 25 in Nature Energy by scientists combined extensive experimental data with artificial intelligence to uncover the key to accurately predicting battery life before its capacities begin to diminish. The machine learning model was trained using several hundred million data points of batteries charging and discharging patterns.
Results revealed that this machine learning model was accurate 95% of the time. Furthermore, predictions were within 9 percent of actual cell cycle length.
Although the machine learning model proved successful, it had some shortcomings. For instance, it could only accurately estimate lithium-ion battery lifespan with a certain degree of precision and did not fully account for nonlinear battery characteristics.
To address these obstacles, different modeling methodologies were tested. These included a machine learning model, semi-empirical approach and artificial neural network model; the latter proved most successful.
This method accurately predicted capacity fade with minimal computational effort required. It utilized a large dataset of nickel-manganese-cobalt oxide (NMC) battery aging tests.
Overall, this machine learning technique accurately predicted the lifespan of Li-ion batteries with high precision. This could be applied to expedite battery research and development, enhance manufacturing processes, and more effectively sort manufactured batteries.
According to current estimates, the market for this technology is worth approximately $1.5 billion as of 2019. This number is predicted to grow at an average annual growth rate of 6.8% until 2022.
Market Analysis by Technology
Predicting battery life expectancy has always been a daunting challenge in the field of battery technology, since batteries are nonlinear and not fully described by characteristic curves. Stanford researchers have recently combined extensive experimental data with machine learning to accurately forecast how long a lithium-ion battery will last. Their model utilizes several hundred million data points to predict how many cycles a battery will undergo before its capacity limit is reached.
The machine-learning approach was then trained on a data set of several hundred million charge/discharge cycles from lithium-ion batteries. Surprisingly, the algorithm correctly classified these batteries into long or short life expectancy categories 95 percent of the time.
However, this model still has some shortcomings; such as its incapability to accurately predict individual characteristics of a battery due to periodicity and specific amplitudes. Furthermore, certain important battery properties cannot be modeled through machine learning methods, such as voltage declines.
Therefore, there is a high degree of uncertainty regarding the development of battery cost. This is understandable given that prices for raw materials and various parameters such as design and format, production process, plant location, and capacity all play a role. Furthermore, certain underlying assumptions about market growth could potentially impact battery cost development; these include R&D funding sources and public subsidies.
Market Analysis by Region
The market for Method to Predict Battery Life by Region is estimated to reach around $120 million in 2020. This growth is primarily driven by the increasing adoption of electric vehicles across commercial and military space, coupled with an increasing need for renewable energy sources and efficient energy storage systems to keep up with technological advancements. Furthermore, increased competition between EV manufacturers and carmakers looking to reduce costs across their battery portfolios are impacting this market as well.
Market Analysis by Company
The market for Method to Predict Battery Life by Company is projected to expand at a compound annual growth rate (CAGR) of 7.6% from 2018-2026. Several factors are driving this expansion, such as the growing demand for electric vehicles and increased research-and-development activities to develop battery technology. Furthermore, rising demands for lithium-ion batteries in automotive applications and increased production at large lithium-ion battery plants will further fuel expansion of this sector.
This study identified 53 relevant publications with original battery cost forecasts from peer-reviewed literature and classified them according to superordinate forecasting methods (technological learning, literature-based projection, expert elicitation and bottom-up modeling). Parameter extracts for each method and its parameter sets were analyzed separately; results revealed that studies using technological learning, literature-based projection or expert elicitation often focus on time dimensions of battery cost while those applying bottom-up modeling tend to include more technical aspects of battery technology.
The Ford Global Technologies LLC invention works as follows
Methods and systems can be used to reliably predict the expected life expectancy of a vehicle’s battery. The rate at which the battery’s state of degradation is predicted is determined from the vehicle operating parameter. This metric is then compared to a threshold that has been established based upon past experience. The predicted state is then converted to an estimate of the time or distance that the component will need to be serviced and presented to the vehicle operator. The predicted state of degradability may affect vehicle control and communication strategies.
Background for Method to predict battery life
Vehicle engines contain an energy storage device such as a lead acid battery for powering a starter engine as well as supporting electrical load transients. This type of battery is often referred to in literature as an SLI?battery. In hybrid electric vehicles, a propulsion battery can be used to power the electric motor and driveline. Over time, batteries can become less efficient and need to be replaced or serviced. Multiple factors can affect the rate at which a battery is degrading, including the amount of usage, age, temperature, and the nature of the battery.
Many approaches have been used to predict the health of a vehicle battery. Uchida, U.S. Pat., is an example of such an approach. No. 8,676,4825. The fuel economy of hybrid vehicles is used to predict the health of the battery. Kozlowski and colleagues offer another example. US 20030184307. The frequency of system battery charging and discharge and their effects on parameters like impedance and electrolyte status, are used to predict the battery’s health. This is how the battery’s remaining useful cycles are figured out.
However, the inventors have found many problems with these approaches. One example is that there could be many mechanisms that impact a battery’s health. Some of these mechanisms may be interdependent while others can be independent. These examples do not take into account different battery characteristics, which can have an impact on the battery’s health. Conditions that cause battery corrosion may affect a battery?s internal resistance more than those that increase its capacity. However, conditions that cause a battery to lose more of its active mass or sulfation may have a greater effect on a battery?s overall capacity. Some of these conditions can be reversed. The level of battery sulfur may change, for example. Some characteristics can be affected by temperature, while others may not be affected at all. A functional battery might need to be serviced sooner than anticipated if the battery’s end date is not predicted. A degraded battery might not be sent to the right place at the correct time. Driver satisfaction could be affected by the battery not starting, resulting in a loss of mobility and degraded electrical functionality that affects driving performance (such loss of electric assist to power steering, or electrically-boosted brakes).
The issues could be worsened in autonomous vehicles where the battery (e.g. a 12V SLI) supports essential operating system and safety critical systems in certain modes of operation. If the battery is nearing its end-of-life, or defective, then the vehicle’s autonomous functions may be disabled. Customers may be dissatisfied and annoyed if autonomous functions are suddenly disabled, especially if the vehicle is not able to be driven manually.
In one instance, some of these issues can be addressed using a method for a car battery that is coupled to a motor vehicle. This involves: Predicting a state of degrading a vehicle from a plurality battery metrics, derived based upon past driving history data, including past performance of each of the plurality battery metrics; and Converting the predicted state to a time or duration estimate based a rate convergence towards a threshold to define the end of life to display to the vehicle operator. This will allow for a better prediction of the remaining useful life and may also enable the vehicle operator to receive the information in a timely fashion.
This method allows for accurate prediction of the battery’s remaining life without having to rely on complex algorithms. The state of health of the battery can be more accurately calculated by using data from the vehicle along with fleet driving statistics and vehicle data. Accounting for temperature effects may help to determine the internal resistance or capacitance of the batteries. Technically, defining thresholds for each characteristic of the battery based on statistical, empirical, and machine-learning methods and then estimating the battery’s end of life based upon a trajectory of each characteristic towards the threshold to determine the expected threshold, is that different mechanisms of battery decay can be taken into account. Battery degradation due to corrosion effects might be different than battery degradation due sulfation effects. This may allow for a more reliable calculation of overall battery health. A vehicle operator might be more informed about the condition of a component by converting its state of health to an estimate of the remaining time or duration before servicing is necessary. This may allow for prompt component servicing, which can improve vehicle performance. The timely notification can also be useful if the battery characteristic is affected due to operating driving habits. This may allow the operator to modify their driving to prolong the battery life. The remaining life of a component of a vehicle can be predicted using a recursive estimation based on statistical features. This allows for a lower computation intensity and a higher accuracy. This allows for a greater margin that will ensure the component’s continued health and safety. To help customers plan ahead and prevent component failure, the prognostics feature can provide an indication of the battery’s remaining life to aid in planning for maintenance. An easy-to-use package may also offer online estimation.
It is important to understand that the summary below is intended to simplify a selection concepts that will be further described in detail. This summary is not intended to identify key features or essential characteristics of the claimed subject matter. The claims that follow the detailed description define the scope of this claim. The claimed subject matter does not include implementations that address any of the disadvantages mentioned above, or any portion thereof.
The following description refers to systems and methods that predict the remaining life of a car battery, such as the vehicle system shown in FIG. 1A, including the FIG. 1B. 1B. An onboard controller can be programmed to perform a control program, such as those shown in FIGS. 3.4) to calculate the remaining life of a vehicle’s battery using statistical and measured data. As shown in FIG. 2. The controller can also estimate the speed at which the battery characteristic is expected to convergence with the threshold. FIG. 6. To predict the end-of-life of the battery. These thresholds can be calibrated using data from both the vehicle and other data as described in FIGS. 5, 7, 10 and 12. Multiple intermediate thresholds may be set before the EOL. As each intermediate threshold is reached, various notifications can be sent. Control actions can be taken as well. 8-9, 11 and 13-16. This will allow for regular battery servicing and may help to reduce warranty issues.
FIG. “FIG. One possibility is that Engine 10 can be used in conjunction with a propulsion system such as vehicle system 5. One example is vehicle system 5. It may be an electric hybrid vehicle system.
Engine 10 can be controlled at most partially by a control device including controller 12, and input from a vehicle operator 130 via a input device 132. This example shows that input device 132 contains an accelerator pedal and a pedal positioning sensor 134 to generate a proportional pedal position signal (PP). Cylinder (herein also ?combustion chamber?) 14 may have combustion chamber walls 136 and piston 138. Piston 138 can be coupled to crankshaft140 so that the reciprocating motion is converted into rotational motion by the crankshaft. A transmission system may allow crankshaft 140 to be connected to at least one passenger vehicle’s drive wheels. A starter motor (not illustrated) can be connected to crankshaft 140 by a flywheel. This will allow engine 10 to start.
Cylinder 14 can receive intake through a series intake air passages 142-144 and 146. Air from intake air passage 142 may be filtered through an air filter 135 before it enters air passages 144 and 146. Intake air passage146 can communicate with other engines 10 cylinders in addition to cylinder 14. One or more intake passages can include a boosting device, such as a turbocharger and/or supercharger. FIG. FIG. 1A illustrates engine 10, which is equipped with a turbocharger, including a compressor (174) arranged between intake passages 142-144 and 144 and an exhaust turbine (176 arranged along the exhaust passage 148). The exhaust turbine 176 may provide some power to compressor 174 via a shaft 180. This is where the turbocharger is installed. In other cases, like when engine 10 has a supercharger installed, exhaust turbine 176 can be optionally omitted. Compressor 174 may instead be powered mechanically by the motor or engine. An intake passage may have a throttle 162 and a throttle plate. This throttle can be used to control the intake air flow rate and pressure. As shown in FIG. 16, throttle 162 can be located downstream of compressor 174. 1A or upstream of compressor 174.
Exhaust passage148 can also receive exhaust gases from other engines 10 cylinders in addition to cylinder 14. Exhaust gas sensor 128 can be shown to the exhaust passage 148 downstream of emission control device (178). One of the many sensors that can provide an indication of the exhaust gas/fuel ratio, Sensor 128, may be chosen from among a variety of suitable sensors, such as a linear oxygen sensor (UEGO) (universal or broad-range exhaust gas oxygen), or a two-state oxygen sensor (EGO) (as shown), or a HEGO, (heated exhaust gas oxygen sensor), for example. Emission control device number 178 could be a TWC, NOx trap, or other emission control devices.
Each cylinder in engine 10 can include one or more intake or exhaust valves. One example is cylinder 14, which shows at least one intake valve 150 and at most one exhaust valve 156, located in the upper region of cylinder 14. Some examples show that each cylinder of engine 10, including the 14th, may have at least two intake poppet vales and at most two exhaust poppet vales located in an upper part of the cylinder.
Controller 12 may control intake valve 150 via actuator 152. Controller 12 may also control exhaust valve 156 via actuator 154. Controller 12 can change the signals to actuators 15 and 154 in certain conditions to adjust the opening and closing times of respective exhaust and intake valves. Not shown, the position of exhaust valve 156 and intake valve 150 can be determined using respective valve position sensors. The actuators can be either electric or cam-actuated, or a combination of both. You can control the intake and exhaust timing simultaneously, or you may use any combination of variable intake cam timing, variable exit cam timing, dual independent variable timing or fixed timing. Each cam actuation system can include one or several cams. It may use one or more of the following: cam profile switching (CPS), variable timing (VCT), variable valve Timing (VVT), and/or variable valve Lift (VVL), which may be controlled by controller 12. This allows for valve operation to be varied. An alternative to cylinder 14 could include an intake valve controlled by electric valve actuation, and an exhaust valve controlled using cam actuation (CPS/VCT). Other options include a common actuator or an actuation system that controls the intake and exhaust valves, or a variable timing actuator or system that controls the valves.
Cylinder 14 may have a compression ratio. This is the ratio between volumes at the bottom center and top center of piston 138. The compression ratio ranges from 9:1 to 10. The compression ratio can be increased in certain cases where different fuels were used. This could happen when fuels have a higher latent enthalpy for vaporization or higher octane. Direct injection can also increase the compression ratio due to its impact on engine knock.
In some cases, each cylinder may contain a spark plug 192 to start combustion. Under select operating modes, ignition system 190 may provide ignition spark to combustion chamber 14, via spark plug 192, in response to the spark advance signal SA from controller 12. In some cases, however, spark plug 192 may not be required. For example, engine 10 may start combustion by either auto-ignition or injection of fuel.
In some cases, each cylinder may have one or more fuel injections to provide fuel. Two fuel injectors, 166 and 170 are shown in cylinder 14. The fuel injectors 166 or 170 can be set up to receive fuel from the fuel system 8. Fuel system 8 could include fuel rails, fuel pumps, and fuel tanks. The fuel injector 166 can be seen attached to cylinder 14. It injects fuel directly into cylinder 14 in accordance with the pulse width signal FPW-1, which is received via electronic driver 162. Direct injection is what fuel injector 166 provides. of fuel into combustion cylinder 14. FIG. FIG. Because alcohol-based fuels are less volatile, this position can improve combustion and mixing. To improve mixing, the injector can be placed overhead or near the intake valve. Fuel injector 166 may receive fuel from fuel tank 8 via a high-pressure fuel pump and a fuel rail. A pressure transducer may be installed in the fuel tank to provide a signal to controller 12.
Fuel injector 170 can be seen in intake passage 146, rather than in the cylinder 14. This configuration is called port injection of fuel (hereafter,?PFI?)). In the intake port downstream of cylinder 14. The fuel injector 170 can inject fuel from the fuel system 8 in proportion to the pulse width signal FPW-2 that controller 12 receives via electronic driver 171. You can use a single driver (168 or 171) for both fuel injection systems or multiple drivers (e.g. driver 168 for fuel injector 166, driver 171 for the fuel injector 170), as shown.
In another example, each fuel injector 166 and 170 could be configured as direct fuel injections to inject fuel directly into the cylinder 14. Another example is that each of the fuel injectors 170 and 166 may be used as port fuel injections to inject fuel upstream from intake valve 150. Another example is cylinder 14. This cylinder may only contain one fuel injector, which can be configured to receive fuel from different fuel systems in varying relative amounts. It is further configured to inject the fuel mixture directly into the cylinder or upstream of intake valves as either a direct fuel injector, or a port fuel injection. It should be noted that the specific fuel system configurations discussed herein are not intended to limit the functionality of the fuel systems described.
Both injectors may deliver fuel to the cylinder in a single cycle. Each injector could deliver a fraction of the total fuel injection to the cylinder, for example. The distribution and/or relative amounts of fuel delivered by each injector can vary depending on operating conditions such as engine load and knock as well as exhaust temperature. Port injected fuel can be delivered at any time, including during an open or closed intake valve event. Directly injected fuel can also be delivered during an intake stroke. Even for one combustion event, the timing of fuel injection may vary depending on whether it is coming from the port or direct injector. Multiple injections of fuel may be made per cycle for a single combustion event. Multiple injections can be made during compression stroke, intake stroke or any other combination.
