Endurance range estimation for battery-powered vehicles

Abstract

The invention provides a method and system for estimating the driving range of an electric vehicle. Estimating travelled range based on each of a Kalman filter based method and a rolling horizon based method and providing travelled range data to a vehicle driver is described. Methods configured to be used when destinations are unknown and when destinations are known are described.

Description

Background Art

[0001] Limited driving range, long charging times, and, at least in some areas, limited charging infrastructure make range anxiety a problem for many electric vehicle (EV) drivers. Due to range anxiety, some drivers reserve up to 20% of battery capacity as a buffer. Rather than responding to the EOR alert (which typically means 20% battery capacity remaining in some EVs) by (re)charging, some drivers try to avoid the EOR alert altogether, even on long journeys. Issues related to range anxiety may limit the acceptance and / or popularity of EVs and may also lead to suboptimal operating decisions. For example, shorter charging cycles may accelerate battery aging.

[0002] Improving EV range estimation is challenging due to the uncertainty of future driving conditions, including driving patterns, traffic conditions, environmental factors such as temperature, wind, and precipitation, and battery state, including nonlinearities and aging. The relationship between remaining usable battery capacity and the circuit voltage measured from the EV battery can be nonlinear. During operation, only loaded battery voltage measurements may be available. In many EV installations, the current drawn from the EV battery and the current delivered to the EV battery during charging are monitored to determine the battery's remaining usable charge. However, measurement errors can occur, and the battery itself is susceptible to self-discharge. Converting remaining usable battery capacity (remaining usable battery charge) to distance / range is not straightforward because factors influencing this conversion include road characteristics (grade, curvature), environmental factors, driver decisions (aggressive or conservative driving style), and secondary current draw due to cabin comfort and infotainment needs. New and alternative methods, controllers, and systems for electric vehicle range estimation are needed. Summary of the Invention

[0003] The present inventors have recognized that a primary unsolved problem is the need to develop new and / or alternative methods, controllers, and systems for range estimation in electric vehicles.

[0004] A first illustrative and non-limiting example employs a method for generating an end of range estimate for an electric vehicle, the method comprising: obtaining a set of data from previous trips of the electric vehicle; using the previous trip data, estimating a set of model parameters for battery capacity reduction relative to distance traveled from a starting battery capacity (charge) to a battery capacity (charge) at the end of the range; using the previous trip data, calculating an error covariance matrix for the set of model parameters; applying the set of model parameters and the error covariance matrix for a new trip to generate an estimated end of range distance; and displaying the estimated end of range distance to a driver of the vehicle, wherein the displayed prediction is updated as the new trip is made.

[0005] Additionally or alternatively, the method may include recording distance, speed, and battery parameters during the new trip for use in performing the estimating and calculating steps for subsequent trips. Additionally or alternatively, the displayed prediction is updated as the new trip progresses: determining a change in actual distance traveled and battery state of charge (SOC) during the new trip; updating the set of model parameters and the error covariance matrix using the change in actual distance traveled and battery SOC during the new trip; reapplying the updated set of model parameters and the updated error covariance matrix to generate an updated estimated end of range distance; and displaying the updated estimated end of range distance to a driver of the vehicle.

[0006] Additionally or alternatively, the error covariance matrix is ​​calculated using a Kalman filter operating on the set of data from previous trips of the electric vehicle, and wherein the step of updating the set of model parameters and the error covariance matrix using the actual distance traveled and the change in battery SOC during the new trip includes updating the model parameters in the discrete-time model by calculating process noise for the corresponding parameters calculated by the Kalman filter and adding it to each model parameter. Additionally or alternatively, the error covariance matrix is ​​calculated using a Kalman filter operating on the set of data from previous trips of the electric vehicle.

[0007] Additionally or alternatively, the set of data for previous trips of the electric vehicle is for previous trips having at least a minimum distance traveled or a shortest duration of travel. Additionally or alternatively, the step of applying the set of model parameters and the error covariance matrix to generate an estimated end-of-range distance for the new trip is performed when the destination is unknown to the control device of the electric vehicle. Additionally or alternatively, the set of model parameters is parameters a and b in the following formula: distance = a*SOC U +b, where SOC U is the available battery charge at the start of a new trip.

[0008] Another illustrative and non-limiting example employs a method of generating an end of range estimate for an electric vehicle, the method comprising: obtaining a set of data from a previous trip of the electric vehicle; using the previous trip data as input to a moving horizon observer to estimate a set of model parameters for battery capacity reduction from a starting battery capacity to an end of range battery capacity relative to a distance traveled; applying the set of model parameters for a new trip to generate an estimated end of range distance; and displaying the estimated end of range distance to a driver of the vehicle, wherein the displayed prediction is updated as the new trip progresses by obtaining new data from the new trip and inputting the new data into the moving horizon observer while removing the oldest data from the moving horizon observer.

[0009] Additionally or alternatively, the set of data of previous trips of the electric vehicle is for previous trips having at least a minimum traveled distance or a shortest travel time. Additionally or alternatively, the set of model parameters is the parameters a and b in the following formula: distance = a*SOC U +b, where SOC U is the available battery charge at the start of the new trip. Additionally or alternatively, the method may further comprise recording the distance, speed and battery parameters during the new trip for use in performing the estimation and calculation steps for subsequent trips.

[0010] Another illustrative and non-limiting example employs a method for generating an updated estimate of the end of range of an electric vehicle, the method comprising: receiving a destination for a current trip of the vehicle; in a digital twin simulation: generating a speed profile for the vehicle to the destination using a model of the vehicle and road data for a path between the vehicle's current location and the destination; generating a battery charge consumption using a battery model of the vehicle to determine an expected charge consumption to reach the destination or end of range location; and setting an end of range (EOR) reference value and a battery state of charge (SOC) EOR value; communicating the EOR reference value and the battery SOC EOR value from the digital twin to the vehicle; as the vehicle travels along the path: collecting actual distance traveled and battery state of charge (SOC) measurements; fusing the EOR reference value, the battery SOC EOR value, the battery SOC measurements, and the actual distance traveled to update a model that relates battery SOC to distance traveled; estimating the EOR of the vehicle for the current trip; and displaying the estimated EOR of the vehicle for the current trip to a driver of the vehicle.

[0011] Additionally or alternatively, the fusion step is performed in a Kalman filter having an R matrix, a Q matrix and an error covariance matrix by constructing a first model using the EOR reference value and the battery SOC EOR value: 1,k =a*SOC EOR +b+v 1,k, where y 1,k is the EOR reference value, SOC EOR is the battery SOC at EOR, v 1,k is the white noise defined by the variance of the Kalman filter R matrix, and a and b are model parameters; the first model is paired with the second model, the second model is of the form: y 2,k =a*SOC k +b+v 2,k , where y 2,k is the actual distance traveled at sample k, SOC k is the battery SOC at sample k, v 2,k is white noise defined by the variance of the Kalman filter R matrix.

[0012] Additionally or alternatively, the model parameters a and b are treated as constants and updated sample by sample using the following formula: k =p k-1 +w p,k-1 , b k =b k-1 +w b,k-1 , where w p,k-1 is the process noise of parameter a, w b,k-1 is the process noise of parameter b, each of which is given by the Kalman filter Q matrix. Additionally or alternatively, the method may further include: updating the estimated EOR of the vehicle for the current trip using the updated model parameters, and displaying the updated estimated EOR to the driver of the vehicle at least once. Additionally or alternatively, the collecting, fusing, estimating, and displaying steps are repeated at sampling times as the vehicle travels along the route.

[0013] Additionally or alternatively, the digital twin uses traffic data in addition to road data to generate the speed profile, and performs the collection, fusion, estimation, and display steps without acquiring or using traffic or road data. Additionally or alternatively, the set of data for previous trips of the electric vehicle is for previous trips with at least a minimum travel distance or a shortest travel time. Additionally or alternatively, the digital twin may be computed at a fleet monitor or data processing center remote from the electric vehicle.

[0014] Additional illustrative examples include a controller configured to perform the above method, and Figure 1 An electric vehicle having such a controller is shown.

[0015] This summary is intended to introduce the subject matter of this patent application. This summary is not intended to provide an exclusive or exhaustive explanation. The detailed description is intended to provide further information about this patent application. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In the accompanying drawings, which are not necessarily drawn to scale, like reference numerals may describe similar components in different views. Like reference numerals with different letter suffixes may represent different instances of similar components. The accompanying drawings generally illustrate various embodiments discussed in this document by way of example and not limitation.

[0017] Figure 1 An electric vehicle is shown in simplified block diagram form;

[0018] Figure 2-Figure 4 A first illustrative example of a trip with an undefined destination is shown in block diagram and graphical form;

[0019] Figure 5-Figure 6 and Figure 7A-7B A second illustrative example of a trip with a defined destination is shown in block diagram and graphical form; and

[0020] Figure 8 A method using a moving time-domain observer is shown. DETAILED DESCRIPTION

[0021] Figure 1 The electric vehicle is shown in simplified block diagram form. Those skilled in the art will recognize that the following discussion may not necessarily describe every feature that would be present in the vehicle 10 to avoid overelaborating features that are not necessary to understand the following examples.

[0022] Vehicle 10 features an electric motor 12 (or multiple electric motors 12) powered by a battery 14 to provide propulsion for vehicle 10. Battery 14 can be recharged via a connection 16 to an off-board power source, as is known in the art, and can have any suitable chemistry and / or design. Battery 14 can be associated with various secondary features, such as heating and / or cooling devices to maintain a suitable temperature therein. Regenerative braking 18 can be provided to at least partially recharge battery 14 under appropriate braking conditions.

[0023] Controller 20 is coupled to each of these blocks (boxes) and may further be linked to control blocks for communications 22, navigation 24, infotainment 26, and cabin 28. Controller 20 is configured to send and receive information, as well as provide and / or control power used by, for example, the air conditioning unit used to cool cabin 28 or other environmental controls for cabin 28. Communications 22 may include any of satellite, cellular, Bluetooth, broadband, WiFi, and / or various other wireless communication circuitry, antennas, receivers, transceivers, transmitters, and the like, as needed. Communications 22 may allow controller 20 to send and receive data associated with one or more internet, private, and / or cloud-based data reception and / or processing centers, such as fleet monitors. In some examples described below, communications 22 may be used to upload and / or download various types of data. Navigation system 24 may store, retrieve, receive, and / or display various types of data as needed, including, but not limited to, weather / environmental data, road data including curvature, speed limit signs, and grade, and traffic data. Navigation system 24 may also be used to provide route instructions to the vehicle's driver and / or provide routes for use by the autonomous driving controller. The navigation system 24 may include a global positioning system (GPS) device for determining and tracking the location of the vehicle 10 .

[0024] Figure 2-Figure 4 A first illustrative example of a trip without a defined destination is shown in block diagram and graphical form. In this example, the vehicle has not yet received a defined destination from the driver, and therefore cannot formulate a comprehensive road plan. At this point, vehicle 100 communicates data about one or more previous trips to cloud 120, which includes servers / processors hosting a digital twin modeling vehicle 100 and the components it will use to reach the destination along the route.

[0025] As used herein, a "previous trip" may be a specific instance of driving a vehicle, or may be a previous instance in which the vehicle battery was depleted to a selected degree. For example, if a driver stops several times during a day out and then returns home to (re)charge the vehicle battery, the "previous trip" may include all activities from the departure point to the return trip. If the battery is not charged immediately, the "previous trip" may include all activities between the last battery charging session (process) and the battery charging session (process) before the last battery charging session (process). Other formulas and / or input restrictions may be used. For example, as desired, a "previous trip" may be a vehicle trip that exceeds a minimum distance (e.g., 10 to 100 kilometers or other distance) or duration (30 minutes to 3 hours, or other duration).

[0026] The previous trip data may optionally include multiple battery utilization measurements, including one or more state-of-charge (SOC) reports, such as battery SOC reports based on battery charge and discharge conditions and / or battery voltage readings. The previous trip data may include distance traveled during the previous trip. Typically, both battery SOC data and distance traveled data are provided. Additional data may also be communicated (transmitted), such as speed, road grade, curvature, ambient conditions, and secondary current draw (e.g., due to cabin heating / cooling).

[0027] The cloud-based digital twin 120 is used to generate a model of battery SOC and driving distance and extrapolate from the provided data to determine how far the vehicle can travel if driven to the battery's range limit by selecting a battery SOC for the end of range (EOR). The battery SOC at the EOR point may be referred to herein as the battery SOC EOR , which can be a battery SOC determined by the manufacturer or the driver. Typically, this range can be 15% to 30% of the total battery SOC, and in some illustrative examples, 20%. If desired, the driver can be given the option to select a larger or smaller battery SOC EOR percentage.

[0028] In one example, the digital twin 120 relates the total available range to the battery SOC using a formula of the type shown in Equation 1:

[0029] Distance = a*SOC U +b(1)

[0030] Where, as further explained below, a and b are the coefficients of the equation, "SOC U ” is the available SOC at the start of the trip, for example, as shown in Equation 2:

[0031] SOC U =SOC ST -SOC EOR (2)

[0032] Among them, SOC STis the state of charge at the start of the trip, which may be the maximum SOC if the battery is fully charged. Equation 1 is illustrative only, and other more complex and / or more accurate models may be used as needed. For example, the available distance may include additional terms that use the square of the actual or average vehicle speed multiplied by a third coefficient. In some examples, the coefficients a and b in Equation 1 can be determined using a best fit or least squares analysis based on previous trip data and / or initial settings for a and b. For example, for a new vehicle or new battery, the manufacturer may provide values ​​for a and b based on the manufacturer's testing. As the vehicle and / or battery ages, a and b may be updated over time. In some examples, a and b in Equation 1 are determined in a best fit manner based solely on the most recently available trip data.

[0033] Go to Figure 4 , illustrates the operation of using a digital twin 120. Using communication data from a previous trip 200, a recursive least squares or Kalman filter algorithm is run to learn the error covariance matrix P of equation (1). The result is both an estimation model 202 and an associated error covariance matrix 204, such as in equation 1.

[0034] Next, in the vehicle 210 and / or in online operation, the model and error covariance matrix are input at 212, for example, by downloading from a digital twin / cloud / remote server using the vehicle's communication capabilities. At 214, for the new trip, the predicted EOR and estimated remaining range (RDR) are calculated. The RDR is calculated by subtracting the actual distance traveled from the EOR. The RDR is then displayed to the driver at 216. The trip data is recorded at 218 and used as previous trip data for subsequent iterations of the method, as indicated by the arrow from box 218 to box 200.

[0035] The method can be internally recursive, as shown by the return arrow from 216 to 214. In its simplest form, the RDR is updated as the distance traveled accumulates without adjusting the original EOR estimate. In another example, as new vehicle state data accumulates, the downloaded model can be iteratively updated using the vehicle's current state measurements. This may result in adjustments to the error covariance matrix and / or coefficients in the downloaded model. Thus, the EOR estimate can be tailored as the journey progresses. The model parameters can be stored in a database and classified, and their initial values ​​can be scheduled or calculated on-board based on location, weather conditions and / or power demand. For example, the above equation 1 can be updated to adapt variables a and / or b to actual data reflecting the current vehicle state, environmental conditions and route data. In addition, equation 1 can be modified during the journey by monitoring the amount of power drawn from the drive battery and tracking the distance traveled using that power.

[0036] The initial state of the Kalman filter can be described by the following equation:

[0037] x=[a;b] (3)

[0038] Using Equation 1 above; other models / equations with appropriate modifications may be used instead of Equation 1. The estimated covariance matrix P will be based on the digital twin’s analysis of previous trip data. By extrapolating the Kalman filter model to the SOC EOR , get the estimated EOR and remaining range (RDR). Then the lower limit prediction interval is displayed to the driver through the vehicle's human-machine interface.

[0039] Go to Figure 2 , Figure 4 The illustrative examples of can be augmented by various factors. For example, the cloud / digital twin can access environmental condition data 122, such as, but not limited to, data related to wind, temperature, and / or precipitation. In windy conditions, it can be assumed that the vehicle is less efficient than otherwise because more force may be required to maintain speed. In warm or extremely cold conditions, the passenger compartment may require greater heating or cooling, and the energy required to maintain the battery within its optimal performance range may vary. In precipitation conditions, the maximum vehicle speed may be reduced, meaning that over time, more energy will be required to manage secondary functions in the vehicle because the journey may take longer than otherwise. GPS data 106 can also be accessed in the cloud 120 to provide an overall understanding of the likely driving type (e.g., city or highway). These factors can be used to classify and modify (schedule) the coefficients in, for example, Equation 1, and / or adjust the available power as in Equation 2, as needed.

[0040] The cloud / digital twin 120 provides data to initialize the onboard range model 102 in the vehicle. Estimation and filtering operations are performed at 104, such as applying a Kalman filter to update the model and provide an output estimate. GPS data 106, or an onboard odometer or any other suitable distance traveled technology, provides the distance traveled as the vehicle moves. Then, at 108, the estimation and filtering block 104 provides an onboard extrapolation of the EOR and RDR. The RDR can then be displayed to the driver of the vehicle.

[0041] Figure 3A graphical representation of this approach is shown. Here, the cloud digital twin 150 uses, for example, traffic, GPS, and / or weather data to provide an initial model to onboard analytics 160 via batch processing. The batch growth or backoff window at 162 reflects ongoing data from the start of the trip, with multiple observed data points 164. The range model extrapolation is shown at 166, tracking to the modeled EOR at 168, while the actual data proceeds to the actual EOR at 170. In the graphical portion, the lines representing distance traveled and battery SOC should be understood to move from right to left across the page and up the graph, such that as distance traveled increases, battery SOC decreases. As described above, data is captured throughout the process for subsequent trips.

[0042] Figure 5-Figure 6 and Figure 7A-7B A second illustrative example of a trip with a defined destination is shown in block diagram and graphical form. Figure 5 Initially, here, as described above, the digital twin 300, which may be cloud-based, now receives a known destination, which can then be used to determine traffic, weather, and / or route / road data. The digital twin 300 is used to generate initial range model parameters, for example using a Kalman filter, and then communicates (transmits) the model parameters and associated error data, such as an error covariance matrix if a Kalman filter is used, to the vehicle.

[0043] In one example, a digital twin simulation can acquire / determine vehicle characteristics, such as, but not limited to, vehicle mass, vehicle drag, transmission type or other details, motor performance (e.g., power / current ratio), battery capacity and / or type, and secondary power usage in the vehicle, which may include battery temperature control, and cabin controls. A route map can be determined based on a known destination using GPS and / or map data. A power (electricity) profile can then be generated, for example, by determining road characteristics (curvature, grade, current / legal speed limits, traffic control data such as for roundabouts, stop lights, and stop signs, and existing or predicted traffic and / or weather information), using the vehicle route and road characteristics to create a speed profile for the vehicle, with the power (electricity) profile being directly derived from the speed profile. The speed profile can be augmented using driver data or previous vehicle usage data uploaded from the vehicle, for example, by referencing previous similar (or identical) trips. The speed profile can also be augmented using comfort information, for example, by determining a permissible speed range where a bend in the road occurs.

[0044] The battery state of health and SOC data are then used to simulate the battery SOC throughout the route, and the end of range is determined based on the battery SOC at the end of range. The digital twin simulation is provided to the vehicle for information fusion 310 with an onboard end of range model, which in turn can be based on previous trip data or an EOR model stored in the vehicle.

[0045] Go to Figure 6 In one example, a Kalman filter is used in the digital twin. The Kalman filter measurement model for the actual mileage and the EOR cloud reference value can be given by the following equation:

[0046] y 1,k =a*SOC EOR +b+v 1,k (4)

[0047] y 2,k =a*SOC k +b+v 2,k (5)

[0048] Among them, y 1,k is the EOR cloud reference measurement provided by the digital twin, and y 2,k SOC is the actual distance traveled at the current time k provided by the vehicle GPS or odometer. k is the SOC measurement value at the current time k, and SOC EOR is the SOC value at EOR. Discrete time white noise is shown as v 1,k and v 2,k , defined by the variance in the Kalman filter R matrix. Treating the model parameters a and b as constants (except for process noise), in the discrete-time model it is given by:

[0049] a k =a k-1 +w a,k-1 (6)

[0050] b k =b k-1 +w b,k-1 (7)

[0051] Among them, w a,k-1 is the process noise of parameter a, w b,k-1 is the process noise of parameter b, each of which is given by the Kalman filter Q matrix. Figure 7A-7B The information fusion at block (box) 310 is illustrated. In the batch growth or backoff window 312, as the vehicle progresses along the driving route, multiple data points are constructed, where the range model shown at 314 is extrapolated to the estimated range endpoint y* based on the model fusion. 1,kThe actual or true range endpoint is shown at 316. As the vehicle progresses along the route, the graph again moves from right to left along the battery SOC (which is decreasing) and rises as the distance traveled. Data fed back to the information fusion 310 from window 312 may include, for example, but is not limited to, GPS and / or odometer data for vehicle speed, location, and distance traveled. The information fusion block 310 may also receive battery SOC data to continue updating the information fusion process.

[0052] Figure 6 The digital twin process is illustrated. Vehicle characteristics are collected at 350. These vehicle characteristics may include vehicle weight / mass, battery capacity and / or type, aerodynamics and other data that affects vehicle motion and efficiency (such as tires, brakes, differentials, transmission / transmission), and battery state of health (SOH). Then, at 352, the path to the destination and its details are collected, including the starting and ending points, road curvature and slope. Additionally, in 352, road quality (pavement, road type, gravel), traffic conditions, speed limits, current (real-time) traffic / driving speeds, such as traffic lights, roundabouts, stop signs, merging and lane splitting, and weather that affect vehicle trajectory can also be integrated.

[0053] Block 354 shows the generated expected power profile for the travel route. As an intermediate point in the analysis, block 354 may include estimating a vehicle speed profile along the travel route using traffic data, legal speed limits, traffic controls, and vehicle comfort determination using road type / curvature. Some examples of generating vehicle speed and / or power (electricity) profiles along a route or path can be found in U.S. Patent Application No. 17 / 969,398, entitled "DRIVER AND LEADVEHICLE PREDICTION MODEL," U.S. Patent Application No. 17 / 969,359, entitled "HIERARCHICAL OPTIMAL CONTROLLER FOR PREDICTIVE POERSPLIT," and / or U.S. Patent Application No. 17 / 969,181, entitled "ENERGY EFFICIENT PREDICTIVE POER SPLIT FOR HYBRID POWERTRAINS." Additional factors that may be considered in block 354 may include environmental factors such as ambient temperature that is expected to affect the power required to drive battery heating / cooling to maintain the battery in a desired operating state, as well as passenger compartment requirements. If the vehicle is used for freight transportation, additional factors related to the cargo (load mass and any required power consumption associated with the load, such as heating / cooling factors) may be included in determining the power profile requirements, where the cargo may affect the speed profile or the power required to maintain speed, and / or require additional power to maintain the cargo, such as at a desired temperature.

[0054] The digital twin then simulates the battery SOC to the destination at 356. If necessary, stop(s) can be planned and considered in block 356. The simulation results are then sent to the vehicle for data fusion.

[0055] Figure 7A-7B An illustrative digital twin fusion process is shown in block diagram form. Using a set destination, a speed profile to the destination can be calculated using the digital twin vehicle model at 400, taking into account road / route details such as road curvature, road grade, applicable speed limits and other traffic controls, existing traffic conditions and weather, and any known driver attributes such as the likely speed at which the driver will want to travel. An expected power consumption along the route is then generated, as shown at 402. A range end reference value y is then set. 1,k , where the determination is based on simulating the vehicle's travel toward the destination until an EOR value of SOC is reached (e.g., 20% to 30% of the maximum battery SOC). The flow then moves to Figure 7B .

[0056] At block 410 , while driving the vehicle, the system collects the actual distance traveled value y 2,k and battery SOC, SOC k GPS or odometer data can be used to determine the actual distance traveled, and the battery SOC can be monitored by tracking the current flowing out of / to the battery over time, or using other measurements. The model is then applied at 412, such as by using a Kalman filter or a moving time domain observer to fuse the EOR measurements (values) y 1,k And the actual measured current range y 2,k . Data fusion is then used to update the model parameters a and b, for example by using equations 6 and 7 above. The error covariance matrix of the Kalman filter, typically denoted as the P matrix, can be initiated (at least for the first sample) or updated at 416. At 418, the input data weights are generated using the Kalman filter R matrix. By integrating the input data and updating the model parameters a and b, an EOR estimate is generated at 420 and displayed to the driver / user at 422. The method then iterates back to block 410 and increments k to the next sample. For example, sampling can be performed at intervals of 0.1 to 10 seconds, or longer or shorter intervals, as desired. The EOR estimate can be displayed directly from the analysis, or a smoothing / weighting function can be further applied so that the user does not see a continuously changing EOR estimate, although this smoothing function can also be provided by the input data weights generated at 418.

[0057] As mentioned above, a moving time-domain observer can be used instead of the Kalman filter. Figure 8 Examples are provided. The moving horizon observer uses batches of receding (rolling, receding) windows of data (rather than using the current sample as in the Kalman filter). The rolling window exploits model prediction ideas, similar to model predictive controllers. The moving horizon observer avoids the use of the state vector error covariance matrix (P matrix) or any other historical information beyond the data window and the prior state estimate, resulting in a conceptually simple problem formulation and parameter tuning. The model parameters a and b (and any additional parameters if more complex evaluation is used) can be computed from a limited memory moving window of current and historical measurement data using a least squares minimization procedure.

[0058] exist Figure 8A graph of stored data is shown in the example of . Each data point can be data from a previous trip (DP) or data from a new trip (DN), and has a first subscript corresponding to the time sample (1, 2, 3, 4) and a second subscript corresponding to the data type (also 1, 2, 3, 4). The data type can be any of the data discussed previously, including battery charge usage or state of charge, distance traveled (accumulated or in time samples), and appropriate acceleration, drag, non-propulsion current information, etc., as appropriate for the specific application. An initial data horizon is shown at 520, which includes data from the previous trip (part or all of this data may be used). The data at 520 is used to generate a first set of model parameters, which can take the form of parameters a and b discussed previously. As new data from a new trip is captured, the oldest data from the previous trip is removed from view, as indicated by brackets 522 and 524. Thus, the data is continuously updated, with the oldest data discarded as new data is input. Model parameters a and b can be repeatedly recalculated and updated based on the new trip.

[0059] Each of these non-limiting examples may stand alone or may be combined with one or more of the other examples in various permutations or combinations.

[0060] The above detailed description includes references to the accompanying drawings, which form part of the detailed description. The accompanying drawings show specific embodiments in an illustrative manner. These embodiments are also referred to herein as “examples”. Such examples may include elements other than those shown or described. However, the inventors also contemplate examples in which only those elements shown or described are provided. In addition, the inventors also contemplate examples using any combination or arrangement of the elements shown or described (or one or more aspects thereof), or with respect to a specific example (or one or more aspects thereof), or with respect to other examples shown or described herein (or one or more aspects thereof). If there is a discrepancy between the usage of this document and any document incorporated by reference, the usage in this document shall prevail. As is common in patent documents, the terms “a” or “an” are used to include one or more than one, and are not related to any other instances or usages of “at least one” or “one or more”. In addition, in the claims, the terms “first”, “second”, and “third” are used merely as labels and are not intended to impose numerical requirements on their objects.

[0061] The method examples described herein may be at least partially implemented by a machine or computer. Some examples may include a computer-readable medium or machine-readable medium encoded with operable instructions for configuring an electronic device to perform the methods described in the above examples. The implementation of such methods may include code, such as microcode, assembly language code, high-level language code, etc. Such code may include computer-readable instructions for performing various methods. The code may form part of a computer program product. In addition, in an example, the code may be tangibly stored on one or more volatile, non-transient or non-volatile tangible computer-readable media, such as during execution or at other times. Examples of these tangible computer-readable media may include, but are not limited to, a hard disk, a removable disk or optical disk, a magnetic tape, a memory card or memory stick, a random access memory (RAM), a read-only memory (ROM), etc.

[0062] The above description is intended to be illustrative and not restrictive. For example, the above examples (or one or more aspects thereof) may be used in combination with each other. Other embodiments may be used, such as one of ordinary skill in the art would do upon reading the above description. The Abstract is provided to comply with 37 C.F.R. § 1.72(b) to enable the reader to quickly ascertain the nature of the technical disclosure. It should be understood that the Abstract is not intended to interpret or limit the scope or meaning of the claims.

[0063] In addition, in the above detailed description, various features may be grouped together to simplify the disclosure. This should not be interpreted as intending to indicate that unclaimed, disclosed features are essential to any claim. On the contrary, innovative subject matter may lie in fewer than all features of a particular disclosed embodiment. Therefore, the following claims are incorporated into the detailed description as examples or embodiments, with each claim standing on its own as a separate embodiment, and it is contemplated that these embodiments may be combined with each other in various combinations or permutations. The scope of protection should be determined by reference to the appended claims and the full scope of equivalents to which such claims are entitled.

Claims

1. A method for generating an end-of-range estimate for an electric vehicle, comprising: Obtaining a set of data on previous trips of the electric vehicle; using data from the previous trip, estimating a set of model parameters for a reduction in battery capacity relative to distance traveled from a starting battery capacity to an end-of-range battery capacity; Calculating an error covariance matrix for the set of model parameters using the data from the previous trip; For a new trip, applying the set of model parameters and the error covariance matrix to generate an estimated end-of-range distance; and Displays the estimated end-of-range distance to the driver of the vehicle, The forecast displayed therein is updated as the new trip is made.

2. The method according to claim 1, characterized in that Also included is recording distance, speed, and battery parameters during the new trip for use in performing estimation and calculation steps for subsequent trips.

3. The method according to claim 1 or 2, characterized in that The displayed forecast is updated as the new trip is made in the following ways: determining an actual distance traveled and a change in a battery state of charge (SOC) of the vehicle during the new trip; updating the set of model parameters and the error covariance matrix using the actual distance traveled and the change in the battery SOC during the new trip; Reapplying the updated set of model parameters and the updated error covariance matrix to generate an updated estimated end-of-range distance; as well as The updated estimated end-of-range distance is displayed to a driver of the vehicle.

4. The method according to claim 3, characterized in that The error covariance matrix is ​​calculated using a Kalman filter operating on the set of data for the previous trip of the electric vehicle, and wherein the step of updating the set of model parameters and the error covariance matrix using the actual distance traveled and the change in the battery SOC during the new trip includes updating model parameters in a discrete-time model by calculating process noise for the corresponding parameters calculated by the Kalman filter and adding it to each model parameter.

5. The method according to any one of the preceding claims, characterized in that The error covariance matrix is ​​calculated using a Kalman filter operating on the set of data for the previous trips of the electric vehicle.

6. The method according to any one of the preceding claims, characterized in that The set of data of the previous trips of the electric vehicle is for previous trips having at least a minimum travelled distance or a shortest travel duration.

7. The method according to any one of the preceding claims, characterized in that The step of applying the set of model parameters and the error covariance matrix to generate the estimated end-of-range distance for the new trip is performed with the destination unknown to a control device of the electric vehicle.

8. The method according to any one of the preceding claims, characterized in that The set of model parameters is the parameters a and b in the following formula: Distance = a*SOC U +b Among them, SOC U is the available battery charge at the start of the new trip in question.

9. A method for generating an end-of-range estimate for an electric vehicle, comprising: Obtaining a set of data on previous trips of the electric vehicle; estimating a set of model parameters for battery capacity reduction relative to distance traveled from a starting battery capacity to an end-of-range battery capacity using the previous trip data as input to a moving time-domain observer; For a new trip, applying the set of model parameters to generate an estimated end-of-range distance; and Displays the estimated end-of-range distance to the driver of the vehicle, The displayed predictions are updated as the new trip is made by acquiring new data from the new trip and inputting the new data into the moving horizon observer while removing the oldest data from the moving horizon observer.

10. The method according to claim 9, characterized in that The set of data of the previous trips of the electric vehicle is for previous trips having at least a minimum travelled distance or a shortest travel duration.

11. The method according to claim 9 or 10, characterized in that The set of model parameters is the parameters a and b in the following formula: Distance = a*SOC U +b Among them, SOC U is the available battery charge at the start of the new trip in question.

12. The method according to claim 9, 10 or 11, characterized in that Also included is a step of recording distance, speed, and battery parameters during the new trip for use in performing estimation and calculations for subsequent trips.

13. A method for generating an updated estimate of the end-of-range range of an electric vehicle, comprising: Receive the destination of the vehicle's current trip; In digital twin simulation: generating a speed profile for the vehicle to reach the destination using a vehicle model and road data for a path between the vehicle's current position and the destination; generating a battery power consumption using a battery model of the vehicle to determine an expected power consumption to reach the destination or end-of-range location; as well as Set the end of range (EOR) reference value and battery state of charge (SOC) EOR value; communicating an EOR reference value and a battery SOC EOR value from the digital twin to the vehicle; When the vehicle travels along the path: Collect actual distance traveled and battery state-of-charge measurements; fusing an EOR reference value, a battery SOC EOR value, a battery SOC measurement value, and an actual traveled distance to update a model relating battery SOC to traveled distance; estimating an EOR for the vehicle during the current trip; as well as An estimated EOR for the vehicle during the current trip is displayed to a vehicle operator.

14. The method according to claim 13, characterized in that The fusion step is performed in a Kalman filter with an R matrix, a Q matrix and an error covariance matrix in the following way: The first model is constructed using the EOR reference value and the battery SOC EOR value: y 1,k =a*SOC EOR +b+v 1,k Among them, y 1,k is the EOR reference value, SOC EOR is the battery SOC at EOR, v 1,k is the white noise defined by the variance of the Kalman filter R matrix, and a and b are model parameters; Pair the first model with a second model of the form: y 2,k =a*SOC k +b+v 2,k where y 2,k is the actual distance traveled at sample k, SOC k is the battery SOC at sample k, v 2,k is the white noise defined by the variance of the Kalman filter R matrix.

15. The method according to claim 14, characterized in that The model parameters a and b are considered constant and updated sample by sample using the following formula: p k =p k-1 +w p,k-1 b k =b k-1 +w b,k-1 Among them, w p,k-1 is the process noise of the parameter a, w b,k-1 is the process noise of the parameter b, each given by the Kalman filter Q matrix.

16. The method according to claim 15, characterized in that Also includes: An estimated EOR for the vehicle during the current trip is updated using the updated model parameters, and the updated estimated EOR is displayed at least once to a driver of the vehicle.

17. The method according to claim 15 or 16, characterized in that The collecting, fusing, estimating, and displaying steps are repeated at sampling times as the vehicle travels along the path.

18. The method according to any one of claims 13 to 17, characterized in that In addition to using road data, the digital twin also uses traffic data to generate the speed profile, and performs the collecting, fusing, estimating, and displaying steps without acquiring or using traffic data or road data.

19. The method according to any one of claims 13 to 18, characterized in that The set of data of the previous trips of the electric vehicle is for previous trips having at least a minimum travelled distance or a shortest travel duration.

20. The method according to any one of claims 13 to 19, characterized in that The digital twin is calculated at a fleet monitor or data processing center remote from the electric vehicle.

Images