Controller for controlling battery based on adaptive sampling electrochemical impedance spectroscopy battery measurement

Abstract

The present disclosure provides a controller for controlling a battery based on adaptive sampled electrochemical impedance spectroscopy battery measurements. A system, such as a motorized vehicle, includes a battery, such as a traction battery. A controller controls the battery based on adaptively sampled electrochemical impedance spectroscopy (EIS) measurements of the battery. The controller may perform adaptively sampled EIS measurements. The controller may use the adaptively sampled EIS measurements to identify values of parameters of a model of the battery, such as an equivalent circuit model (ECM). The controller may detect an internal state of the battery using the model using the identified values of the parameters, whereby the controller detects the internal state of the battery based on the adaptively sampled EIS measurements.

Description

Technical Field

[0001] The present disclosure relates to detecting an internal state of a traction battery of an electrified vehicle. Background Art

[0002] An electrified vehicle includes a traction battery for providing power to a motor of the vehicle to propel the vehicle. Electrochemical impedance spectroscopy (EIS) is used to detect an internal state of the traction battery. The detected internal state of the traction battery is used in controlling the operation of the traction battery and / or the vehicle. Summary of the invention

[0003] A system includes a battery and a controller. The battery may be composed of sub-component individual battery cells or virtual battery cells composed of parallel battery cells. The controller is configured to control the battery based on adaptively sampled electrochemical impedance spectroscopy (EIS) measurements of the battery.

[0004] The controller may be configured to perform adaptively sampled EIS measurements. In this case, the controller may perform adaptively sampled EIS measurements until the adaptively sampled EIS measurements indicate that the impedance spectrum of the battery is within a given tolerance. The controller may perform a first set of adaptively sampled EIS measurements to detect an amplitude component of the adaptively sampled EIS measurements, and then perform a second set of adaptively sampled EIS measurements to detect a phase component of the adaptively sampled EIS measurements. The controller may perform the adaptively sampled EIS measurements to iteratively detect the amplitude and phase components of the adaptively sampled EIS measurements.

[0005] The controller may use the adaptively sampled EIS measurements to identify values ​​of parameters of a model of a battery, such as resolved into its subcomponent battery cells. In this case, the controller may use the model to detect an internal state of the battery using the identified values ​​of the parameters, whereby the controller detects the internal state of the battery based on the adaptively sampled EIS measurements.

[0006] The internal state of the battery may be a power capability of the battery, whereby controlling the controller of the battery based on the adaptively sampled EIS measurements involves controlling the controller of the battery based on the power capability of the battery and / or subcomponent battery cells.

[0007] The internal state of the battery may be a state of charge of the battery, whereby controlling the controller of the battery based on the adaptively sampled EIS measurements involves controlling the controller of the battery based on the state of charge of the battery and / or subcomponent battery cells.

[0008] The battery may be a traction battery. In some variations, the battery and the controller are both on the vehicle. In other variations, such as a charging station, the battery is on the vehicle and the controller is not on the vehicle, such as at a charging station.

[0009] A method for a vehicle having a traction battery and a controller is also provided. The method includes performing, by the controller, an adaptively sampled EIS measurement of the traction battery. The method also includes controlling, by the controller, the traction battery and / or another component of the vehicle based on the adaptively sampled EIS measurement of the traction battery.

[0010] The method may further comprise detecting a range-to-empty energy estimate for the vehicle based on the internal state of the traction battery, and controlling the traction battery and / or another vehicle component as a function of the range-to-empty energy estimate.

[0011] A vehicle having a traction battery and a controller is also provided. The controller is configured to charge and / or discharge the traction battery based on adaptively sampled EIS measurements of the traction battery. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 A block diagram of a battery electric vehicle (BEV) having a traction battery is shown;

[0013] Figure 2 A block diagram showing an arrangement for a traction battery controller to monitor a traction battery is shown;

[0014] Figure 3 A schematic diagram showing an equivalent circuit model (ECM) of a traction battery;

[0015] Figure 4 A flow chart depicting operations performed by a traction battery controller for detecting an internal state of a traction battery using adaptive sampling (or adaptive learning) to reduce the time of electrochemical impedance spectroscopy (EIS) measurements of the traction battery over a frequency range of interest is shown;

[0016] Figure 5A showing a Nyquist plot including a graph of a magnitude component of measurement samples of the impedance of the traction battery over a frequency range of interest and a graph of a phase component of measurement samples of the impedance of the traction battery over a frequency range of interest;

[0017] Figure 5B shows a Nyquist plot comprising a graph of real and imaginary components of measured samples of the impedance of the traction battery over a frequency range of interest;

[0018] Fig. 6Agraphs showing first, second and third graphs of percentage of variance of the magnitude component of a surrogate model of the impedance of a traction battery as obtained by a traction battery controller using first, second and third adaptive sampling algorithms for initially learning a graph of the magnitude component relative to high density uniform sampling as a function of the number of measurement samples of the impedance;

[0019] Figure 6B Shows Fig. 6A an enlarged version of a portion of the diagram;

[0020] Figure 6C a diagram showing first, second and third graphs of measurement time taken to obtain a magnitude component of a surrogate model of the impedance of a traction battery using a first adaptive sampling algorithm, a second adaptive sampling algorithm and a third adaptive sampling algorithm as a function of the number of measurement samples of the impedance, the diagram also having a fourth graph of measurement time taken to obtain the magnitude component of the impedance of the traction battery using high density uniform sampling;

[0021] Fig.6D graphs showing first, second and third graphs of percentage of variance of a phase component of a surrogate model of an impedance of a traction battery as obtained by a traction battery controller using a first adaptive sampling algorithm, a second adaptive sampling algorithm and a third adaptive sampling algorithm for subsequent learning of a graph of the phase component relative to high density uniform sampling as a function of the number of additional measurement samples of the impedance;

[0022] Fig. 6E a diagram showing first, second and third graphs of the measurement time taken to obtain a phase component of a surrogate model of the impedance of a traction battery as a function of the number of additional measurement samples of the impedance, the diagram also having a fourth graph of the measurement time taken to obtain the phase component of the impedance of the traction battery using high density uniform sampling;

[0023] Fig. 7A a graph showing first, second and third graphs of percentage of variance of a magnitude component of a surrogate model of an impedance of a traction battery as obtained by a traction battery controller using first, second and third adaptive sampling algorithms for iteratively learning a graph of the magnitude component relative to high density uniform sampling as a function of the number of measured samples of the impedance;

[0024] Figure 7Ba graph showing first, second and third graphs of percentage of variance of a phase component of a surrogate model of an impedance of a traction battery as obtained by a traction battery controller using first, second and third adaptive sampling algorithms for iteratively learning a phase component graph relative to high density uniform sampling as a function of the number of measured samples of the impedance; and

[0025] Figure 7C A diagram showing first, second and third graphs of measurement times taken to obtain magnitude and phase components of a surrogate model of the impedance of a traction battery using a first, second and third adaptive sampling algorithm as a function of the number of measurement samples of the impedance, and a fourth graph of measurement times taken to obtain magnitude and phase components of the impedance of a traction battery using high density uniform sampling. DETAILED DESCRIPTION

[0026] Detailed embodiments of the present disclosure are disclosed herein; however, it should be understood that the disclosed embodiments are merely examples of the present disclosure that may be embodied in various forms and alternative forms. The drawings are not necessarily drawn to scale; some features may be exaggerated or minimized to show details of particular components. Therefore, the specific structural and functional details disclosed herein should not be interpreted as limiting, but merely as a representative basis for teaching those skilled in the art to adopt the present disclosure in different ways.

[0027] Reference now Figure 1 , a block diagram of an electrified vehicle 12 in the form of a battery electric vehicle (BEV) is shown. The BEV 12 includes a powertrain having one or more traction motors ("motors") 14, a traction battery ("battery" or "battery pack") 24, and a power electronics module 26 (e.g., an inverter). In the BEV configuration, the traction battery 24 provides all propulsion power, and the vehicle does not have an engine. In other variations, the electrified vehicle may be a plug-in hybrid electric vehicle (PHEV) that also has an engine.

[0028] The traction motor 14 is part of the powertrain of the BEV 12 that is used to power the movement of the BEV. In this regard, the traction motor 14 is mechanically connected to a transmission 16 of the BEV 12. The transmission 16 is mechanically connected to a drive shaft 20, which is mechanically connected to wheels 22 of the BEV 12. The traction motor 14 can provide propulsion capability to the BEV 12 and can operate as a generator.

[0029] The traction battery 24 stores electrical energy that can be used by the traction motor 14 to propel the BEV 12. The traction battery 24 typically provides a high voltage (HV) direct current (DC) output. The traction battery 24 is electrically connected to a power electronics module 26. The traction motor 14 is also electrically connected to the power electronics module 26. The power electronics module 26 (such as an inverter) provides the ability to transfer energy bidirectionally between the traction battery 24 and the traction motor 14. For example, the traction battery 24 may provide a DC voltage, while the traction motor 14 may require a three-phase alternating current (AC) current to function. The inverter 26 may convert the DC voltage into a three-phase AC current to operate the traction motor 14. In the regenerative mode, the inverter 26 may convert the three-phase AC current from the traction motor 14 acting as a generator into a DC voltage compatible with the traction battery 24.

[0030] In addition to providing electrical energy for propulsion of the BEV 12 , the traction battery 24 may also provide electrical energy for use by other electrical systems of the BEV, such as HV loads like an electric heater and air conditioning system, and low voltage (LV) loads like an auxiliary battery.

[0031] The traction battery 24 may be recharged by an external power source 36 (e.g., an electrical grid). The external power source 36 may be electrically connected to an electric vehicle supply equipment (EVSE) 38. The EVSE 38 provides circuitry and controls for controlling and managing the transfer of electrical energy between the external power source 36 and the BEV 12. The external power source 36 may provide DC or AC power to the EVSE 38. The EVSE 38 may have a charging connector 40 for plugging into the charging port 34 of the BEV 12.

[0032] A power conversion module 32 of the BEV 12, such as an onboard charger having a DC / DC converter, may condition power supplied from the EVSE 38 to provide appropriate voltage and current levels to the traction battery 24. The power conversion module 32 may interface with the EVSE 38 to coordinate power delivery to the traction battery 24.

[0033] The various components described above may have one or more associated controllers to control and monitor the operation of the components. The controller may be a microprocessor-based device. The controller may communicate via a serial bus (e.g., a controller area network (CAN)) or via discrete conductors.

[0034] For example, there is a system controller 48 ("vehicle controller") to coordinate the operation of the various components. The controller 48 includes electronics, software, or both to perform the necessary control functions for operating the BEV 12. The controller 48 may be a combined vehicle system controller and powertrain control module (VSC / PCM). Although the controller 48 is shown as a single device, the controller 48 may include multiple controllers in the form of multiple hardware devices, or multiple software controllers with one or more hardware devices. In this regard, references to "controller" herein may refer to one or more controllers.

[0035] The controller 48 implements a battery energy control module (BECM) 50. The BECM 50 communicates with the traction battery 24. The BECM 50 is a traction battery controller operable to monitor operating characteristics (e.g., internal states) of the traction battery 24 and to manage the charging and discharging of the traction battery 24. The BECM 50 implements algorithms to measure and / or estimate operating characteristics of the traction battery 24. The BECM 50 controls the operation and performance of the traction battery 24 based on the operating characteristics of the traction battery. The BECM 50 or other controllers may control the operation and performance of other systems and components of the BEV 12 based on the operating characteristics of the traction battery 24.

[0036] The operating characteristics of the traction battery 24 include the charge capacity and the state of charge (SOC) of the traction battery 24. The charge capacity of the traction battery 24 indicates the maximum amount of electrical energy that can be stored in the traction battery. The SOC of the traction battery 24 indicates the current amount of charge stored in the traction battery. The SOC of the traction battery 24 may be expressed as a percentage of the maximum amount of charge that can be stored in the traction battery.

[0037] Another operational characteristic of the traction battery 24 is the power capability of the traction battery. The power capability of the traction battery 24 is a measure of the maximum amount of power that the traction battery can provide or receive in a specified time period. Thus, the power capability of the traction battery 24 corresponds to charge and discharge power limits that define the amount of electrical power that can be supplied by or to the traction battery at a given time. These limits can be provided to other vehicle controls, such as by a vehicle system controller (VSC), so that the information can be used by systems that can draw power from or provide power to the traction battery 24. The vehicle controls will know how much power the traction battery 24 can provide (discharge) or absorb (charge) in order to meet driver demand and optimize energy usage. Thus, knowing the power capability of the traction battery 24 allows management of electrical loads and sources so that the requested power is within the limits that the traction battery can handle.

[0038] Reference now Figure 2 ,, continue to refer to Figure 1, a block diagram of an arrangement for the BECM 50 to monitor the traction battery 24 is shown. The traction battery 24 is composed of a plurality of battery cells 52. The battery cells 52 are physically connected together (e.g., as shown in FIG. Figure 2 In various variants, the traction battery 24 is composed of individual battery cells of subassemblies or of virtual battery cells composed of battery cells connected in parallel.

[0039] The BECM 50 is operable to monitor pack level characteristics of the traction battery 24, such as a battery current 56, a battery voltage 58, and a battery temperature 60. The battery current 56 is the current output from (i.e., discharging) or input to (i.e., charging) the traction battery 24. The battery voltage 58 is the terminal voltage of the traction battery 24.

[0040] The BECM 50 may also be operable to measure and monitor cell level characteristics of the battery cells 52 of the traction battery 24. For example, the terminal voltage, current, and temperature of one or more of the battery cells 52 may be measured. The BECM 50 may measure the cell level characteristics using battery sensors 54. The battery sensors 54 may measure characteristics of one or more battery cells 52. The BECM 50 may utilize Nc battery sensors 54 to measure characteristics of all of the battery cells 52. Each battery sensor 54 may transmit measurements to the BECM 50 for further processing and coordination. The functionality of the battery sensors 54 may be internally incorporated into the BECM 50.

[0041] The traction battery 24 may have one or more temperature sensors (such as thermistors) in communication with the BECM 50 to provide data indicative of the temperature of the battery cells 52 of the traction battery 24 for the BECM to monitor the temperature of the traction battery and / or the battery cells. The BEV 12 may also include a temperature sensor to provide data indicative of the ambient temperature for the BECM 50 to monitor the ambient temperature.

[0042] The BECM 50 controls the operation and performance of the traction battery 24 based on the monitored traction battery level characteristics and battery cell level characteristics. For example, the BECM 50 may use the monitored characteristics to detect (estimate) the internal state (e.g., charge capacity, SOC, and / or power capability) of the traction battery 24, such as for use in controlling the traction battery and / or the BEV 12.

[0043] As known to those of ordinary skill in the art, the BECM 50 can detect the internal state of the traction battery 24 by using an observer, wherein a battery model (i.e., an "equivalent circuit model" (ECM)) is used to construct the observer using measurements of battery current, battery terminal voltage, and battery temperature. Values ​​of parameters of the ECM can be identified by recursive estimation based on such measurements.

[0044] The BECM 50 is configured to infer estimates of internal states of the traction battery 24 using the ECM. In order for the internal state estimates to be accurate, the ECM must accurately model the traction battery 24. In order for the ECM to accurately model the traction battery 24, (i) the ECM must have a set of parameters that are at least substantially similar to the parameters of an ECM that accurately models the traction battery 24 (i.e., from a set of candidate ECMs, a "suitable" ECM that is reasonably representative of the traction battery must initially be selected), and (ii) the estimated values ​​of the parameters of the ECM must be at least substantially similar to the values ​​of the parameters of the ECM that accurately models the traction battery (i.e., the estimated parameter values ​​must be at least substantially similar to the actual parameter values).

[0045] The BECM 50 is configured to implement electrochemical impedance spectroscopy (EIS) to estimate parameter values ​​of the ECM in order to detect the internal state of the traction battery 24 using the ECM. EIS is a non-invasive technique for measuring the internal state of the traction battery 24. EIS is sensitive to internal state factors of the traction battery 24, including state of charge, state of health, temperature, etc.

[0046] The BECM 50 implements EIS to measure the impedance of the traction battery 24 by applying a relatively small amplitude current or voltage sinusoidal stimulus to the traction battery over a range of frequencies and then measuring the voltage or current response of the traction battery. The impedance response can be considered a characteristic of the battery assembly and can also be resolved into parallel groups of individual battery cells. The impedance data (i.e., EIS measurements) of the traction battery 24 is then "fitted" to the ECM to estimate the values ​​of the parameters of the ECM. The goal is to find the ECM parameters that result in the best agreement between the impedance spectrum of the ECM and the measured impedance data of the traction battery 24.

[0047] Reference now Figure 3 , continue to refer to Figure 1 and Figure 2 , a schematic diagram of an ECM 70 for a traction battery 24 is shown. As indicated, the traction battery 24 may be modeled as a circuit, and the ECM 70 is one possible traction battery ECM. The ECM includes one or more circuit elements (i.e., one or more “parameters”). According to the ECM 70, the traction battery 24 is modeled as having a voltage source (“V OC ”) (open circuit voltage (OCV)) 72, inductor (“L”) 74, resistor (“R S ”) 76, one or more RC circuit elements 78, and a constant phase element (CPE) (“A”) 84, each circuit element comprising a resistor (“R X ”)80 and capacitor (“C X ”)82. Voltage (“V t”) 86 is the voltage across the terminals of the traction battery 24 (ie, the terminal voltage).

[0048] The equation for the impedance (Z) of the ECM 70 is as follows:

[0049]

[0050] The parameters of ECM 70 include L, R S , R X , C X and A. These parameters will have values ​​whereby the calculated impedance of the ECM corresponds to the measured impedance of the traction battery 24. Therefore, it is required that the ECM 70 is an ECM with a suitable set of parameters corresponding to the traction battery 24, and that the values ​​of the parameters of the ECM 70 are accurate so that the ECM 70 accurately models the behavior of the traction battery 24. This algorithmic process may be repeated using a circuit model of each component battery cell in the traction battery 24. The combined behavior of the component battery cells comprises the internal state of the traction battery 24.

[0051] As explained, tracking the traction battery internal state can be used to control charging rate, control traction battery power delivery, monitor traction battery health (including capacity estimation), etc. However, external measurements of the traction battery internal state are obscured by the long relaxation times and state variables defined by the integration algorithm. EIS is a technique that promises to yield information about the traction battery internal state relatively more quickly by probing complex electrical responses (via small amplitude current or voltage sinusoidal stimulation) over a range of frequencies.

[0052] As noted, the BECM 50 implements EIS by applying a sinusoidal electrical stimulus to the traction battery 24 and measuring the electrical output of the traction battery in response to the stimulus. For example, the stimulus applied to the traction battery is a voltage, in which case the output of the traction battery is a current, thereby obtaining an impedance measurement (i.e., Z=V / I, where Z is the impedance, V is the voltage applied to the traction battery, and I is the current output from the traction battery). Alternatively, the stimulus applied to the traction battery is a current, in which case the output of the traction battery is a voltage, thereby obtaining an impedance measurement (i.e., Z=V / I, where Z is the impedance, V is the voltage output from the traction battery, and I is the current applied to the traction battery). Thus, the input stimulus is a variable of one type (i.e., one of voltage or current), and the output is a variable of another type (i.e., the other of voltage or current). The BECM 50 is capable of performing pack-resolved EIS using full pack current and voltage sensors, and is capable of performing cell-resolved EIS due to individual cell current or voltage sensors. Alternatively, an application specific integrated circuit may be used to perform the EIS measurement.

[0053] Each measured impedance is a "complex impedance" because the impedance comprises both magnitude and phase. Alternatively, according to another reference frame, each measured impedance comprises a real component and an imaginary component.

[0054] Each measured impedance is a "sample". The measurement process is repeated using sinusoidal electrical stimulation at different frequencies to obtain additional impedance measurements (i.e., additional samples) at each frequency. That is, the frequency of the sinusoidal electrical stimulation is varied to produce a measured spectrum of impedance as a function of frequency. The range of frequencies is useful because the varying frequencies can stimulate various internal processes of the traction battery 24 (i.e., charge transfer processes, diffusion processes, etc.) that can be calibrated or modeled to produce useful information about the internal state of the traction battery (e.g., state of charge, state of health, internal temperature, etc.).

[0055] Typically in a laboratory environment, the frequency range can span several dozen times in logarithmic space and range from the highest frequency (e.g., 1.0 MHz) to the lowest frequency (e.g., 1.0 mHz). Ten to fifteen measurements are typically collected per decade. It takes a relatively long time to collect low-frequency data because the sinusoidal cycle is relatively long in time. In addition, for low-frequency data, several cycles may be required to average out measurement noise or compensate for voltage drift due to changes in system polarization. As an example, it takes approximately one hour to make fifteen measurements per decade (i.e., high-density uniform sampling) over the frequency range of 0.01 Hz to 1000 Hz using a commercial EIS measurement device.

[0056] One hour is an excessive amount of time to perform EIS measurements on a vehicle or at a public charging station. Performing a full EIS scan for one hour on a BEV 12 is prohibitive because the traction battery 24 changes state while the BEV is driven. Performing a full EIS scan for one hour at a charging station is prohibitive because a user may not be able to charge the traction battery 24 within that length of time. Therefore, in order to be able to utilize EIS on a BEV 12 or at a charging station, the EIS measurement time for a full scan must be reduced.

[0057] According to the present disclosure, the BECM 50 is operable to use adaptive sampling (or adaptive learning) to reduce the EIS measurement time to detect the internal state of the traction battery 24. That is, the BECM 50 uses adaptive sampling to reduce the EIS measurement time for a given frequency range. The EIS results using adaptive sampling are substantially equivalent to the EIS results using high-density uniform sampling. Therefore, the internal state detected by the BECM 50 using adaptive EIS sampling is substantially the same as the internal state that the BECM would detect using uniform EIS sampling. The reduction in EIS measurement time is a result of adaptive EIS sampling, which in many cases involves a 50% to 75% reduction in the number of samples compared to high-density uniform EIS sampling.

[0058] By using adaptive sampling for EIS measurements, the BECM 50 implements a method for learning the impedance of the traction battery 24 over the full frequency range of interest in a reduced amount of time compared to high density uniform sampling. In a common manner, the BECM 50 fits the impedance of the traction battery 24 learned using adaptive EIS sampling to the ECM of the traction battery so that the BECM estimates the values ​​of the parameters of the ECM. The BECM 50 then uses the ECM to detect the internal state of the traction battery 24. In this way, the BECM 50 uses the ECM of the traction battery based on the EIS measurement information of the adaptive sampling to detect the internal state of the traction battery 24. Adaptive sampling and EIS measurements can resolve information down to the virtual cell level.

[0059] When implemented by the BECM 50, the method of reducing EIS data acquisition time makes EIS useful on-board a BEV. The method of reducing EIS data acquisition time also makes EIS useful when implemented off-board, such as at a BEV charging station. In this document, for convenience, the BECM 50 is primarily referred to as using adaptive sampling to reduce EIS measurement time.

[0060] As an overview, the BECM 50 uses an adaptive sampling (or active learning) approach to reduce the number of impedance samples required to define a metamodel or surrogate model that is equivalent to a uniform high-density sampling of the impedance. To achieve this, the BECM 50 begins with EIS measurements at a sparse uniform grid containing the minimum and maximum frequencies of interest. As an alternative to a uniform grid, a set of sparse frequency points from a previous measurement can be used for the initial plot. This can include points based on the immediately preceding EIS measurement or based on calibrations that depend on operating parameters such as temperature and internal battery state. The BECM 50 fits surrogate models of the magnitude and phase components of the impedance to the measured impedance points. For example, the BECM 50 uses splines or other functions to fit the magnitude and phase to the points. This can also be done using the real and imaginary components of the impedance, as they are mathematically interchangeable with the magnitude and phase components of the impedance.

[0061] The BECM 50 then uses an adaptive sampling algorithm to select the next measurement point based on two competing metrics: 1) grid exploration (i.e., measuring in areas of low point density); and 2) model development (i.e., adding points where the model indicates that more points may be needed due to high curvature, rapid functional changes, etc.). These competing metrics define the properties of the adaptive sampling algorithm. For reference, many adaptive sampling (or active learning) methods have been published in the academic literature.

[0062] The BECM 50 may apply the adaptive sampling algorithm to both graphs (i.e., the magnitude graph and the phase graph) in series (i.e., first magnitude, then phase), iteratively (i.e., magnitude, phase, magnitude, phase, ...), or by some other method (e.g., using the Kramers-Kronig relationship to infer one graph from the other). The BECM 50 repeats the process of adding points until some stopping criterion is reached. The stopping criterion may be a set number of samples, a difference between the two norms of the surrogate function between consecutive iterations, etc. The stopping criterion may depend on the full battery pack behavior or on the behavior of a single battery cell or group of battery cells.

[0063] For more details, refer now to Figure 4 , a flow chart 100 depicting operations performed by the BECM 50 for detecting the internal state of the traction battery 24 using adaptive sampling to reduce EIS measurement time is shown. The operations begin with the BECM 50 measuring a set of initial impedance samples of the traction battery 24 over a frequency range of interest in accordance with process block 102. As noted, each impedance sample measured by the BECM 50 requires the BECM to apply a sinusoidal electrical stimulus to the traction battery 24 at a given frequency and measure the electrical output of the traction battery to the stimulus. Thus, the set of initial impedance samples is measured by the BECM 50 applying a sinusoidal electrical stimulus to the traction battery 24 at a set of corresponding different frequencies and the BECM measuring the electrical output of the traction battery to each stimulus. The different frequencies are, for example, at corresponding decades (or smaller intervals) over the frequency range of interest. As an example, the number of impedance samples in the set of initial samples measured in process block 102 may be approximately 5% to 15% of the number of samples measured in high-density uniform EIS sampling.

[0064] The BECM 50 generates a surrogate model of the impedance spectrum of the traction battery 24 based on the set of initial impedance samples according to process block 104. As described above, each measured impedance sample includes a magnitude component and a phase component (or alternatively, a real component and an imaginary component). Therefore, the surrogate model of the impedance spectrum includes a graph of the magnitude component of the set of initial samples and a graph of the phase component of the set of initial samples. Since each graph is based on, for example, 5% to 15% of the number of samples measured in uniform high-density EIS sampling, the graphs of the surrogate model will be at most a rough approximation of the magnitude and phase components of the impedance spectrum of the traction battery 24.

[0065] The BECM 50 may also consider generalized information about the expected behavior of the impedance spectrum of the traction battery 24 when generating the surrogate model. For example, such expected properties of the impedance spectrum of the traction battery 24 include that there are expected to be no discontinuities in the graph of the surrogate model, that the graph of the surrogate model is expected to be a well-behaved smooth graph, that the graph of the surrogate model is expected to increase / decrease within certain frequency intervals, etc. Thus, with the BECM 50 further considering the expected behavior of the impedance spectrum of the traction battery 24, the graph of the surrogate model will better approximate the magnitude and phase components of the impedance spectrum of the traction battery 24.

[0066] The BECM 50 then performs adaptive sampling of the impedance of the traction battery 24 according to process block 106. Adaptive sampling involves the BECM 50 measuring samples of the impedance of the traction battery 24 at selected frequencies within a frequency range of interest. Thus, the BECM 50 applies a sinusoidal electrical stimulus to the traction battery 24 at the selected frequency and measures the electrical output of the traction battery in response to the stimulus, thereby obtaining measured impedance samples.

[0067] "Adaptive" sampling involves the selection of a selected frequency at which impedance samples are measured. Specifically, as described above, the BECM 50 selects the selected frequency (i.e., selects the next impedance measurement point) based on grid exploration (taking measurements in an area of ​​the frequency range where the density of points is low, i.e., the number of impedance samples in this area of ​​the frequency range is lower than the number of impedance samples in other areas of the frequency range) and model development (taking measurements in an area of ​​the frequency range where an alternative model indicates that more points may be needed, i.e., the approximation of the magnitude and phase components of the impedance spectrum of the traction battery 24 is relatively poorer in this area of ​​the frequency range than in other areas of the frequency range).

[0068] Upon completion of process block 106, the total number of measured impedance samples includes (i) the number of measured impedance samples in the initial set according to process block 102, and (ii) the adaptively sampled measured impedance samples according to process block 106. The BECM 50 then updates the surrogate model by further taking into account the adaptively sampled measured impedance samples according to process block 108. Thus, the graph of the magnitude component of the surrogate model also includes the magnitude component of the adaptively sampled measured impedance samples, and the graph of the phase component of the surrogate model also includes the phase component of the adaptively sampled measured impedance samples. Thus, the graph of the surrogate model will now be a better approximation of the magnitude and phase components of the impedance spectrum of the traction battery 24 because each graph includes one or more additional measurements.

[0069] The BECM 50 then determines whether the surrogate model of impedance is sufficiently equivalent to the impedance spectrum of the traction battery 24 (ie, whether the surrogate model sampling is sufficiently equivalent to the high density uniform EIS sampling), as indicated by decision block 110 .

[0070] If the surrogate model of impedance is not sufficiently equivalent to the impedance spectrum of the traction battery 24, another cycle of the adaptive sampling operation of process block 106, the surrogate model updating operation of process block 108, and the comparison decision operation of process block 110 is performed. Specifically, the BECM 50 performs another cycle of adaptive sampling of the impedance of the traction battery 24 according to process block 106. The adaptive sampling in this cycle involves the BECM 50 measuring another sample of the impedance of the traction battery 24 at another selected frequency within the frequency range of interest. The BECM 50 selects the selected frequency based on grid exploration and model development. Now, the total number of measured impedance samples includes (i) the number of measured impedance samples in the initial group according to process block 102, (ii-a) the measured impedance samples of the initial adaptive sampling according to the initial iteration of process block 106, and (ii-b) the measured impedance samples of the additional adaptive sampling according to the initial iteration of process block 106. Then, according to process block 108, the BECM 50 further considers the additional adaptively sampled measured impedance samples to update the surrogate model. Thus, the graph of the magnitude component of the surrogate model also includes the magnitude component of the additional adaptively sampled measured impedance samples, and the graph of the phase component of the surrogate model also includes the phase component of the additional adaptively sampled measured impedance samples. Thus, the graph of the surrogate model will now be an even better approximation of the magnitude and phase components of the impedance spectrum of the traction battery 24 because each graph includes one or more additional measurements. Then, according to decision block 110, the BECM 50 decides whether the surrogate model of impedance is now sufficiently equivalent to the impedance spectrum of the traction battery 24.

[0071] Another cycle of the adaptive sampling operation of process block 106, the surrogate model updating operation of process block 108, and the comparison decision operation of process block 110 is performed until the BECM 50 determines, according to decision block 110, that the surrogate model of impedance is substantially equivalent to the impedance spectrum of the traction battery 24. The BECM 50 may determine that the surrogate model of impedance is substantially equivalent to the impedance spectrum of the traction battery 24 after a set number of impedance samples have been measured, the difference between the immediately previous version of the surrogate model and the current version of the surrogate model is sufficiently small, etc.

[0072] Once the surrogate model of impedance is sufficiently equivalent to the impedance spectrum of the traction battery 24, the BECM 50 terminates any further adaptive sampling. The BECM 50 then uses the surrogate model to update the parameters of the ECM of the traction battery 24 according to process block 112. According to process block 114, the BECM 50 uses the ECM with the updated parameters to detect the internal state of the traction battery 24.

[0073] According to process block 116, the BECM 50 controls the traction battery 24 and / or other components of the BEV 12 based on the internal state of the traction battery. For example, the internal state of the traction battery 24 may be the power capability of the traction battery 24, the state of charge (SOC) of the traction battery, and / or a distance to energy (DTE) estimate of the BEV 12, and the BECM 50 controls the traction battery and / or components of the BEV 12 based on the internal state of the traction battery.

[0074] To illustrate the benefits of adaptive sampling operation used by the BECM 50 to reduce EIS measurement time, reference is now made to Figure 5A and Figure 5B . Figure 5A A Nyquist plot 120 is shown, which includes a graph 122 of a magnitude component of measured samples of the impedance of the traction battery 24 over a frequency range of interest and a graph 124 of a phase component of the measured samples of the impedance of the traction battery over a frequency range of interest. As shown, the measured impedance data of the traction battery 24 is plotted in the form of two curves (i.e., graphs 122 and 124). (For completeness, Figure 5B A Nyquist plot 130 is shown, which includes a graph 132 of real and imaginary components of measured samples of the impedance of the traction battery over a frequency range of interest. As indicated, Nyquist plots 120 and 130 depict the same data.

[0075] The magnitude graph 122 and phase graph 124 are experimental data from a traction battery pouch cell taken at 0% SOC and 15°C. This cell state was chosen arbitrarily and similar results will be achieved at different SOCs and temperatures. In this example, experimental data for a 76 point sweep (i.e., 76 impedance samples) is shown, with a high density uniform sampling of fifteen points per decade of the frequency range of interest. The frequency range of interest is f = 0.01–1000 Hz.

[0076] The adaptive sampling operation is used to obtain a surrogate model of impedance that is sufficiently accurate for the impedance spectrum of the traction battery. The magnitude graph 122 and the phase graph 124 represent the impedance spectrum of the traction battery 24. Therefore, the magnitude graph 122 and the phase graph 124 will be learned by using the adaptive sampling operation. That is, the graph of the magnitude component of the surrogate model will be sufficiently similar to the magnitude graph 122, and the graph of the frequency component of the surrogate model will be sufficiently similar to the phase graph 124. For this example, the goal of the adaptive sampling operation is to learn the magnitude and phase graphs in much less than a 76-point swept impedance measurement.

[0077] Fig. 6A , Figure 6B , Figure 6C , Fig.6D and Fig. 6E The results of sequentially learning the amplitude curve and the phase curve sequentially (continuously) using any of three different adaptive sampling algorithms. Learning the amplitude curve and the phase curve sequentially means initially learning the amplitude curve using adaptive sampling, and then subsequently learning the phase curve using adaptive sampling. (Similarly, sequentially learning the phase curve and the amplitude curve means initially learning the phase curve using adaptive sampling, and then subsequently learning the amplitude curve using adaptive sampling.)

[0078] In more detail, Fig. 6A A graph 140 is shown with a first graph 142, a second graph 144, and a third graph 146 of the variance percentage of the magnitude component of a surrogate model of the impedance of the traction battery 24 as obtained by the BECM 50 using a first adaptive sampling algorithm, a second adaptive sampling algorithm, and a third adaptive sampling algorithm for initially learning the magnitude component graph relative to high density uniform sampling as a function of the number of measured samples of the impedance. The first adaptive sampling algorithm is a Voronoi diagram / cross validation sampling algorithm; the second adaptive sampling algorithm is a hybrid adaptive sampling algorithm; and the third sampling algorithm is a Lipschitz criterion sampling algorithm. These sampling algorithms are known to those of ordinary skill in the art. These sampling algorithms are provided as examples, and other sampling algorithms may be used.

[0079] Figure 6B An enlarged version of a portion of Figure 140 is shown. Figure 6B As shown, each of the three adaptive sampling algorithms achieves close parity with the amplitude graph 122 of the 76-point experimental sweep (i.e., R 2 >0.9999).

[0080] Figure 6C A graph 150 is shown with a first graph 152, a second graph 154, and a third graph 156 of the measurement time taken to obtain the magnitude component of the surrogate model of the impedance of the traction battery 24 using the first adaptive sampling algorithm, the second adaptive sampling algorithm, and the third adaptive sampling algorithm as a function of the number of measurement samples of the impedance. Graph 150 also includes a fourth graph 158 of the measurement time taken to obtain the magnitude component of the impedance of the traction battery 24 using high density uniform sampling (i.e., fifteen samples per decade). As shown in graph 150, each of the three adaptive sampling algorithms takes approximately 2 to 4 times less time than the time taken to obtain the 76 point swept impedance measurement.

[0081] Fig.6DGraph 160 shows first, second, and third graphs 162, 164, and 166 of the variance percentage of the phase component of a surrogate model of the impedance of the traction battery 24 as obtained by the BECM 50 using the first, second, and third adaptive sampling algorithms for subsequent learning of the phase component graph relative to high density uniform sampling as a function of the number of additional measurement samples of the impedance. Fig. 6E Graph 170 is shown with first 172, second 174, and third 176 graphs of the measurement time taken to obtain the phase component of the surrogate model of the impedance of the traction battery 24 as a function of the number of additional measurement samples of the impedance. Graph 170 also includes a fourth graph 158 of the measurement time taken to obtain the phase component of the impedance of the traction battery 24 using high density uniform sampling.

[0082] In operation, first solve the amplitude curve ( Fig. 6A , Figure 6B and Figure 6C After the subject), the sampled points produce a relatively high R in the phase curve 2 Five more impedance samples are measured to enhance the phase curve consistency. Thus, there is a net reduction of 2x to 3x in the sampling time used to learn the magnitude and phase curves compared to the time taken to obtain a 76-point experimental sweep.

[0083] As described, Fig. 6A , Figure 6B , Figure 6C , Fig.6D and Fig. 6E Relates to the amplitude and phase plots as a function of frequency, predicted either individually or sequentially (amplitude first, phase second) relative to the case where fifteen impedance samples are taken per decade. 2 Serves as a comparison between the adaptive sampling prediction and the high-density uniform sample case. Fig. 6E Graph 170 indicates the cumulative sampling time when each measurement is made.

[0084] Fig. 7A , Figure 7B and Figure 7C Results are related to iteratively learning amplitude and phase curves using any of three different adaptive sampling algorithms. Iteratively learning amplitude and phase curves means updating the amplitude curve after an adaptive sampling measurement, then updating the phase curve after the next adaptive sampling measurement, then further updating the amplitude curve, then further updating the phase curve, etc., until the amplitude and phase curves are learned.

[0085] In more detail, Fig. 7AGraph 180 shows first graph 182, second graph 184, and third graph 186 of the variance percentage of the magnitude component of the surrogate model of the impedance of the traction battery 24 as obtained by the BECM 50 using the first adaptive sampling algorithm, the second adaptive sampling algorithm, and the third adaptive sampling algorithm for iteratively learning the magnitude component graph relative to high density uniform sampling as a function of the number of measured samples of the impedance. Accordingly, Figure 7B Graph 190 shows first, second, and third graphs 192, 194, and 196 of variance percentages of phase components of a surrogate model of the impedance of the traction battery 24 as obtained by the BECM 50 using first, second, and third adaptive sampling algorithms for iteratively learning phase component graphs relative to high density uniform sampling as a function of the number of measured samples of the impedance. Figure 7C Graph 200 is shown with first graph 202, second graph 204, and third graph 206 of the measurement time taken to obtain the magnitude and phase components of a surrogate model of the impedance of the traction battery 24 using the first adaptive sampling algorithm, the second adaptive sampling algorithm, and the third adaptive sampling algorithm as a function of the number of measurement samples of the impedance. Graph 200 also includes a fourth graph 208 of the measurement time taken to obtain the magnitude and phase components of the impedance of the traction battery 24 using high density uniform sampling.

[0086] As described, Fig. 7A , Figure 7B and Figure 7C The frequency of the amplitude and phase is sampled continuously in an iterative manner. Fig. 7A , Figure 7B and Figure 7C The result of iteratively learning the amplitude curve and phase curve is similar to that of Fig. 6A , Figure 6B , Figure 6C , Fig.6D and Fig. 6E Results of sequentially learning the amplitude curve and phase curve. For iterative learning, for all three adaptive sampling algorithms, after about fifteen samples, the amplitude and phase curves of the impedance are learned to R relative to the 76-point high-density uniform sampling curve. 2 >0.9999. Figure 7C The reduction in measurement time shown in is about 2 to 3 times faster than the time taken by 76-point high-density uniform sampling. The time can be further reduced by optimizing the algorithm and exploiting the relationship between the amplitude component and the phase component. The time may also be significantly reduced in the case of a reduced frequency range of interest, especially in the case of reduced low-frequency conditions.

[0087] With adaptive sampling, the time taken to acquire spectrally resolved EIS data can be significantly reduced, enabling onboard or charging station applications. This is beneficial because EIS does not require extensive additional hardware in the vehicle, and it can enable spatially distributed measurements in the vehicle by utilizing voltage as the sensing variable, since voltage is resolved to individual battery cell levels.

[0088] As described herein, the conventional method of acquiring EIS data is to make measurements across a high density uniform grid spanning from low frequencies to high frequencies of interest. This can be time prohibitive. Therefore, as the state of the traction battery 24 changes, and thus the impedance of the traction battery changes, the BECM 50 operating in accordance with the present disclosure attempts to minimize the number of measurements required to fully resolve the complex impedance curve. In this way, a robust algorithmic approach to achieving the desired results is provided, and the number of sampling points varies as the complexity of the curve increases or decreases.

[0089] Although exemplary embodiments are described above, these embodiments are not intended to describe all possible forms of the present disclosure. In fact, the words used in this specification are descriptive rather than restrictive, and it should be understood that various changes can be made without departing from the spirit and scope of the present disclosure. In addition, the features of various implemented embodiments can be combined to form other embodiments of the present disclosure.

[0090] According to the present invention, there is provided a system having: a battery; and a controller configured to control the battery based on adaptively sampled electrochemical impedance spectroscopy measurements of the battery.

[0091] According to an embodiment: the controller is further configured to perform electrochemical impedance spectroscopy measurement of the adaptive sampling.

[0092] According to an embodiment: the controller is further configured to perform the adaptively sampled electrochemical impedance spectroscopy measurements until the adaptively sampled electrochemical impedance spectroscopy measurements indicate that the impedance spectrum of the battery is within a given tolerance.

[0093] According to an embodiment: the controller is configured to perform a first set of the adaptively sampled electrochemical impedance spectroscopy measurements to detect an amplitude component of the adaptively sampled electrochemical impedance spectroscopy measurements, and then perform a second set of the adaptively sampled electrochemical impedance spectroscopy measurements to detect a phase component of the adaptively sampled electrochemical impedance spectroscopy measurements.

[0094] According to an embodiment: the controller is configured to perform the adaptively sampled electrochemical impedance spectroscopy measurement to iteratively detect amplitude and phase components of the adaptively sampled electrochemical impedance spectroscopy measurement.

[0095] According to an embodiment: the controller is further configured to use the adaptively sampled electrochemical impedance spectroscopy measurements to identify a value of a parameter of a model of the battery.

[0096] According to an embodiment: the controller is further configured to detect an internal state of the battery using the model with the identified value of the parameter, whereby the controller detects the internal state of the battery based on the adaptively sampled electrochemical impedance spectroscopy measurement.

[0097] According to an embodiment: the internal state of the battery is the power capability of the battery, whereby controlling the controller of the battery based on the adaptively sampled electrochemical impedance spectroscopy measurements involves controlling the controller of the battery based on the power capability of the battery.

[0098] According to an embodiment: the internal state of the battery is a state of charge of the battery, whereby controlling the controller of the battery based on the adaptively sampled electrochemical impedance spectroscopy measurements involves controlling the controller of the battery based on the state of charge of the battery.

[0099] According to an embodiment: the battery is a traction battery.

[0100] According to an embodiment: the battery and the controller are on board the vehicle.

[0101] According to an embodiment: the battery is on board the vehicle and the controller is not on board the vehicle.

[0102] According to the present invention, a method for a vehicle having a traction battery and a controller includes: performing, by the controller, adaptively sampled electrochemical impedance spectroscopy measurements of the traction battery; and controlling, by the controller, the traction battery and / or another component of the vehicle based on the adaptively sampled electrochemical impedance spectroscopy measurements of the traction battery.

[0103] In one aspect of the invention, the method includes: identifying a value of a parameter of a model of the traction battery using the adaptively sampled electrochemical impedance spectroscopy measurements; and detecting an internal state of the traction battery based on the model.

[0104] In one aspect of the invention, the method comprises: detecting a range-to-empty energy estimate for the vehicle based on the internal state of the traction battery; and controlling the traction battery and / or another vehicle component as a function of the range-to-empty energy estimate for the vehicle.

[0105] According to the present invention, a vehicle is provided having: a traction battery; and a controller configured to charge and / or discharge the traction battery based on adaptively sampled electrochemical impedance spectroscopy measurements of the traction battery.

[0106] According to an embodiment: the controller is further configured to perform electrochemical impedance spectroscopy measurement of the adaptive sampling.

[0107] According to an embodiment: the controller is further configured to use the adaptively sampled electrochemical impedance spectroscopy measurements to identify a value of a parameter of a model of the traction battery.

[0108] According to an embodiment: the controller is further configured to detect an internal state of the traction battery using the model with the identified value of the parameter, whereby the controller detects the internal state of the traction battery based on the adaptively sampled electrochemical impedance spectroscopy measurements.

[0109] According to an embodiment: the vehicle is a battery electric vehicle.

Claims

1. A system comprising: Battery; and A controller is configured to control the battery based on adaptively sampled electrochemical impedance spectroscopy measurements of the battery.

2. The system of claim 1, wherein: The controller is further configured to perform the adaptively sampled electrochemical impedance spectroscopy measurement.

3. The system of claim 2, wherein: The controller is further configured to perform the adaptively sampled electrochemical impedance spectroscopy measurements until the adaptively sampled electrochemical impedance spectroscopy measurements indicate that an impedance spectrum of the battery is within a given tolerance.

4. The system of claim 2, wherein: The controller is configured to perform a first set of the adaptively sampled electrochemical impedance spectroscopy measurements to detect an amplitude component of the adaptively sampled electrochemical impedance spectroscopy measurements and then perform a second set of the adaptively sampled electrochemical impedance spectroscopy measurements to detect a phase component of the adaptively sampled electrochemical impedance spectroscopy measurements.

5. The system of claim 2, wherein: The controller is configured to perform the adaptively sampled electrochemical impedance spectroscopy measurement to iteratively detect amplitude and phase components of the adaptively sampled electrochemical impedance spectroscopy measurement.

6. The system of claim 1, wherein: The controller is also configured to identify a value of a parameter of a model of the battery using the adaptively sampled electrochemical impedance spectroscopy measurements.

7. The system of claim 6, wherein: The controller is further configured to detect an internal state of the battery using the model with the identified value of the parameter, whereby the controller detects the internal state of the battery based on the adaptively sampled electrochemical impedance spectroscopy measurements.

8. The system of claim 7, wherein: The internal state of the battery is a power capability of the battery, whereby controlling the controller of the battery based on the adaptively sampled electrochemical impedance spectroscopy measurements involves controlling the controller of the battery based on the power capability of the battery.

9. The system of claim 7, wherein: The internal state of the battery is a state of charge of the battery, whereby controlling the controller of the battery based on the adaptively sampled electrochemical impedance spectroscopy measurements involves controlling the controller of the battery based on the state of charge of the battery.

10. The system of claim 1, wherein: The battery and the controller are on board the vehicle.

11. The system of claim 1, wherein: The battery is on board the vehicle and the controller is not on board the vehicle.

12. A method for a vehicle having a traction battery and a controller, the method comprising: performing, by the controller, adaptively sampled electrochemical impedance spectroscopy measurements of the traction battery; as well as The traction battery and / or another component of the vehicle is controlled with the controller based on the adaptively sampled electrochemical impedance spectroscopy measurements of the traction battery.

13. The method of claim 12, further comprising: identifying values ​​of parameters of a model of the traction battery using the adaptively sampled electrochemical impedance spectroscopy measurements; as well as An internal state of the traction battery is detected based on the model.

14. The method of claim 13, further comprising: detecting a range-to-empty estimate of the vehicle based on the internal state of the traction battery; as well as The traction battery and / or another vehicle component is controlled as a function of the range-to-earth estimate of the vehicle.

15. A vehicle comprising: Traction batteries; and A controller is configured to charge and / or discharge the traction battery based on adaptively sampled electrochemical impedance spectroscopy measurements of the traction battery.

Images