US20190170826A1

US20190170826A1 - Battery state-of-health determination upon charging

Info

Publication number: US20190170826A1
Application number: US16/211,995
Authority: US
Inventors: Tina Shoa Hassani Lashidani
Original assignee: Cadex Electronics Inc
Current assignee: Cadex Electronics Inc
Priority date: 2017-12-06
Filing date: 2018-12-06
Publication date: 2019-06-06
Also published as: US20190170827A1; US11079437B2

Abstract

A rechargeable battery is charged and at the same time, steps are taken to assess its state of health. The state of charge is first estimated, then repeatedly measured and predicted during charging until its value converges to a reliable value. A further known state of charge is later determined, during or after the remainder of the charging. Using the two states of charge and the amount of charge supplied to the battery, the state of health is calculated based on the rated capacity of the battery.

Description

TECHNICAL FIELD

This application relates to a method and system for determining the state of health (SoH) of rechargeable batteries. In particular, it relates to the determination of the SoH based on an estimated state of charge (SoC) value that has converged to the actual SoC during charging, a subsequent SoC and a quantity of charge supplied.

BACKGROUND

The main challenges with battery SoH indicator methods are the different varieties of battery types and chemistries, with different rated capacities and nominal voltages. One common method is to use coulomb counting during the charge cycle to obtain full charge capacity. The challenge with this method is when the battery under charge has not been fully discharged prior to charging. Using the voltage method to estimate the initial SoC is not always reliable, particularly for Li ion batteries. In addition, batteries with low SoH do not show the same voltage profile as healthy batteries.

SUMMARY OF INVENTION

The determination of the SoH of rechargeable batteries requires first estimating the SoC of the battery to be charged. During the initial stage of charging, the SoC is repeatedly estimated until it converges to a more reliable value. The value of the SoC at convergence serves as a first data point. At least one further SoC value obtained after convergence, during or after the remainder of the charging, serves as a further data point. The amount of charge supplied to the battery is measured between the two data points. The SoH can then be calculated from the charge supplied, the two data points and the rated capacity of the battery. As an intermediate step, the present capacity of the battery may also be calculated.
Disclosed herein is a method of determining state of health (SoH) of a battery of rated capacity C comprising the steps of: estimating an initial value of a state of charge (SoC) of the battery; starting to charge the battery; repeatedly measuring a voltage across terminals of the battery as the battery is charged; obtaining a converged value (SoC_C) of the SoC from the repeated measurements; determining a later value (SoC₂) of the SoC after further charging the battery; determining a charge (Q₂) supplied to the battery between SoCc and SoC; calculating, using SoC₂, SoC_C, Q₂and C, a value of the SoH of the battery.
Also disclosed is a battery charger for determining state of health (SoH) of a battery of rated capacity C comprising: a battery interface that connects to the battery; a user interface; a processor connected to the battery interface and the user interface; and a computer readable memory storing computer readable instructions. When the instructions are executed by the processor, they cause the battery charger to: estimate an initial value of a state of charge (SoC) of the battery; start to charge the battery; repeatedly measure a voltage across terminals of the battery as the battery is charged; obtain a converged value (SoC_C) of the SoC from the repeated measurements; determine a later value (SoC₂) of the SoC after further charging the battery; determine a charge (Q2) supplied to the battery between SoC_Cand SoC₂; calculate, using SoC₂, SoC_C, Q₂and C, a value of the SoH of the battery; and output the calculated value of the SoH on the user interface.

BRIEF DESCRIPTION OF DRAWINGS

The following drawings illustrate embodiments of the invention, which should not be construed as restricting the scope of the invention in any way.

FIG. 1 is a flowchart showing the main steps of the process for determining the SoH of a battery, according to an embodiment of the present invention.

FIG. 2 is a graph of open circuit voltage against SoC for a rechargeable battery.

FIG. 3 is a process for obtaining the graph of FIG. 2.

FIG. 4 is a more detailed flowchart of a process for determining the SoH of a battery, according to an embodiment of the present invention.

FIG. 5 is a partial flowchart of a process for determining the SoH of a battery, according to an alternate embodiment of the present invention.

FIG. 6 is a partial flowchart of a process for determining the SoH of a battery, according to another alternate embodiment of the present invention.

FIG. 7 is a graph of SoC vs. inputted charge for a battery charged from 0% to 100% SoC.

FIG. 8 is a graph of SoC vs. inputted charge for another battery charged from 0% to 100% SoC.

FIG. 9 is a partial flowchart of a process for determining the SoH of a battery, according to yet another alternate embodiment of the present invention.

FIG. 10 is a graph of SoC vs. inputted charge for the battery of FIG. 8 charged from 50% to 100% SoC.

FIG. 11 is an electrical circuit model of a rechargeable battery.

FIG. 12 is a flowchart representing a Kalman filter used in an embodiment of the present invention.

FIG. 13 is a flowchart for returning the values of components in the electrical circuit model, according to an embodiment of the present invention.

FIG. 14 is a schematic block diagram of a battery charger, according to an embodiment of the present invention.

DESCRIPTION

A. Glossary

The term “capacity”, “rated capacity”, “nominal capacity” or “C” refers to the rated maximum charge that a battery can hold when it is new. Capacity is measured in Coulombs, Ah or mAh.
The term “C-rate” refers to the rate of charge or discharge of a battery. It equals the fraction of the battery's capacity that is charged or discharged in one hour.
The term “CC” means constant current.
The term “CV” means constant voltage.
The term “CCCV” means applying a constant current to charge a rechargeable battery until the voltage reaches some limit; after that point, applying a constant voltage to the battery until the charging current drops below a threshold.
The term “EKF” or “filter” refers to an Extended Kalman Filter.
The term “module” can refer to any component in this invention and to any or all of the features of the invention without limitation. A module may be a software, firmware or hardware module.
The term “nominal voltage” refers to the mid-point between the voltage when fully charged and the voltage when fully discharged based on a discharge rate of 0.2 C per hour, where C is the rated capacity of the cell. The term “nominal voltage” may also be referred to as the voltage rating of the battery, or simply the voltage of the battery.
The term “maximum cut-off voltage” refers to the OCV when the battery is fully charged. A battery should not be charged to a voltage higher than its maximum cut-off voltage.
The term “open circuit voltage” (OCV) refers to the voltage across the terminals of a battery without any current being drawn from the battery. Typically, as the SoC of a battery declines, its OCV falls. Also, as a battery's SoH deteriorates, the maximum OCV to which the battery can be charged declines.
The term “processor” is used to refer to any electronic circuit or group of circuits that perform calculations, and may include, for example, single or multicore processors, multiple processors, an ASIC (Application Specific Integrated Circuit), and dedicated circuits implemented, for example, on a reconfigurable device such as an FPGA (Field Programmable Gate Array). The processor performs the steps in the flowcharts, whether they are explicitly described as being executed by the processor or whether the execution thereby is implicit due to the steps being described as performed by code or a module.
The term “recovery period” or “rest period” refers to a duration of time after which a current pulse has been discharged from a battery. During the recovery period, no current is drawn from the battery, i.e. substantially no current is drawn from the battery.
RMS—Root mean square
RMSE—Root mean square error
The term “state of charge” (SoC) is a percentage that refers to the amount of charge available in a rechargeable battery. Typically, the SoC is measured as a percentage, with 100% being fully charged and 0% being either fully discharged to a minimum cut-off voltage or discharged to the point beyond which damage may occur.
The term “state of health” (SoH) is a percentage that refers to the maximum amount of charge a rechargeable battery can presently hold compared to its rated charge, or its maximum charge when new. As the battery ages, and as it is cycled through discharge and charge cycles, the SoH falls. Eventually, the SoH drops so low that the battery becomes unfit for its purpose.
The term “system” when used herein refers to a system for determining the SoH of rechargeable batteries, the system being a subject of the present invention.

B. Overview

Referring to FIG. 1, the main steps of the process for determining the SoH of a battery are shown. In step 10, an estimate is made of the initial SoC of an uncharged or partially charged battery. The battery has a known rated capacity C and model. The estimate of the SoC is made using a known voltage method, for example, which involves reading the voltage of the battery and looking up the value of the SoC that it corresponds to. This may be done, for example, by referring to a graph such as the one shown in FIG. 2, in which the OCV (open circuit voltage) of the specific battery model is plotted against its SoC. It is accepted that this type of SoC determination is highly prone to error, which is why it is used only as an estimate.
In step 12, the charging of the battery is started. There are many known techniques for charging a battery that can be employed. In step 14, while charging is in process, the SoC of the battery is repeatedly evaluated. After multiple measurements, the value of the SoC converges to a considerably more accurate estimate than the initial estimate. The converged value of the SoC may be obtained using Kalman filtering, for example, in which the SoC is repeatedly predicted and corrected as the charging proceeds. When the RMS (root mean square) error of a series of measurements of the battery terminal voltage, compared to simulated values of the terminal voltage, is within a predefined tolerance, then the terminal voltage, and hence the SoC, can be considered to have converged. The converged value of the SoC is then taken to be the first reliable data point, SoC_c. In some embodiments, initial SoC is one of the state variables of the extended Kalman filter and its corrected value is reported at the end of charging.
In step 16, the battery charging continues. As the battery is charged, the amount of charge entering the battery is measured.
At a later point in time, in step 18, a second SoC data point is obtained (SoC₂), which, for example, may be the SoC when the battery is fully charged, i.e. 100%, or when the battery has been charged more but not completely. The amount of charge (Q_actual) the battery takes between the SoC of the two data points, divided by the expected charge that the battery should take if its capacity is equal to its rated capacity, results in the SoH of the battery.
SoH=Q _actual /Q _expected (equation 1)
The expected charge that the battery should take is calculated from its rated capacity C and the first and second SoC data points:
Q _expected =C(SOC₂−SOC_c) (equation 2)
It can therefore be seen that the SoH can be calculated either after the battery has been completely charged or when it has been partially charged.

C. Exemplary Embodiments

FIG. 2, which shows the OCV of a rechargeable battery as a function of its SoC, is obtained in the preprocessing steps prior to determining the SoH of a battery during charging. Line 19 is the OCV of the battery, or a battery of identical model, during discharge. Line 20 is the OCV of the battery or identical model during charging. Line 21 is the average of the two lines 19, 20.
The steps undertaken to obtain line 21 are shown in the flowchart of FIG. 3. In step 22, a new battery of the model in question is fully charged in a CCCV mode. In step 23, the battery is fully discharged in a CC mode with a 1 C rate until its voltage hits its minimum voltage. In step 24, all the cycler current accumulators for counting the charge in the battery (integrating current as a function of time) are reset to zero, and the battery is considered to be at a zero SoC.
In step 25, the battery is fully charged slowly, at a C-rate of C/15, in a CCCV mode, while recording the OCV. Charging is not stopped to record the OCV. An alternative method is to charge the battery with the manufacturer recommended C-rate, for a short period of time (e.g. 5 min), stop the charging and record the OCV after the battery voltage has stabilized (e.g. after 15 min rest), and then repeat the charge-stop steps until the battery is fully charged.
In step 26, the battery is then left to relax for one hour. In step 27, the battery is fully discharged at the same C/15 rate, again while recording the OCV. In step 28, the average of data obtained in steps 25 and 27 is calculated. In step 29, a 10^thorder polynomial is fitted to the data of step 28, using a least squares fit. A 10^thorder polynomial has been chosen because a lower order polynomial would be less accurate and if a higher order were used, then extra computational power would be needed without any significant benefit to having the higher accuracy. Using the polynomial, the expected SoC can be determined analytically from a measurement of the OCV during the process for assessing the SoH of the battery upon charging.
The process of FIG. 3 is then repeated twice, once for a battery with medium SoH and once with a battery of poor SoH, so that there are a total of three graphs. In some embodiments, the SoH can be broadly classed into poor, medium and good before the charging starts, which allows the corresponding one of the three graphs to be chosen for determining SoC from the OCV. This improves the accuracy of the final SoH determination.
FIG. 4 is a more detailed flowchart of a process for determining the SoH of a battery. In this first embodiment, the battery is fully charged during the process, and the SoH is determined after charging.
In step 30, various parameters are input into the extended Kalman filter (EKF) that is to be used. The extended Kalman filter needs certain parameters from the battery for the electrical model and measurement. These parameters are obtained beforehand through testing batteries of each type and model, and stored in the charger unit (preferably in the battery adaptor unit that goes with the charger) for specific battery models. These parameters include the graph of FIG. 2 for each type of battery to be tested.
In step 31, a battery with a known rated capacity C and model is connected to the charger. The user specifies the capacity and model of the battery, unless the charger can automatically detect them. The initial SoC is estimated by, for example, measuring the OCV of the battery and then referring to the graph of FIG. 2 to read off the corresponding SoC. If multiple graphs are used, i.e. one for each of for good, medium and poor batteries, a quick test such as the pulse method is used beforehand to categorize the battery as good, medium or poor, which then indicates which of the three graphs are to be chosen. For example, the pulse method disclosed in US Patent Application Publication 2018/0149708 may be used to categorize the battery as good or poor. This pulse method can be extended to categorize the battery into one of three bins (i.e. good, medium, poor). In step 32, the charger starts charging the battery. In step 34, the present value of the SoC is repeatedly estimated using the extended Kalman filter, while charging is in progress. Charging is not stopped while the measurements are taken. The EKF keeps predicting the SoC and correcting the prediction based on measurement as the charging progresses. The prediction is based on the measurement of the battery terminal voltage. As the battery is being charged, the prediction changes accordingly.
In step 36, the system determines whether the measured SoC has converged over the latest set of measurements to a value within a predetermined threshold from the expected value.
The root mean squared error (RMSE) for the latest predetermined number of samples is repeatedly calculated and compared to a threshold value, which is set to indicate the convergence:
$\begin{matrix} RMSE = \sqrt{\frac{1}{n} \sum_{i = 1}^{n} {({\hat{y}}_{i} - y_{i})}^{2}} & (equation 3) \end{matrix}$
Here, ŷ is the array/list holding the simulated terminal voltages of size n, and y is the array/list holding the measured terminal voltages of size n.
Alternatively, the P matrix which is the covariance of expected error in the EKF states is used to confirm convergence (See FIG. 12 for the Kalman filter process). When all parameter states converge, so do their expected error. The trace of P provides a simple single value, which can be used to confirm convergence, or alternatively a sum of the parameter states could be used for the same purpose.
Alternatively still, the normalized innovation squared (NIS) is used to confirm convergence. The NIS is calculated as follows:
ϵ_ν(k)=ν(k)^T S(k)⁻¹ν(k) (equation 4)
where S(k) is the innovation covariance and v(k) is the innovation, the difference between measurement and prediction:
ν(k)=z(k)− z (k) (equation 5)
If the error of a filter is consistent with the variances calculated by the filter, the filter is said to be consistent, and hence the state parameter is said to be converged.
Typically, it takes about 5-10 minutes to converge if the time taken to obtain a sample measurement is 200 ms. After each sample is taken, the measured value is fed back into the filter. If, in step 36, it is determined that the SoC has not converged, then the system waits, in step 38, while the battery continues to charge. After a period of waiting, the SoC is once again estimated by the extended Kalman filter, in step 34. If however, in step 36, it is determined that the SoC has converged, then the process moves to step 40 in which the number of coulombs that have been supplied to the battery up until this point is determined. This is determined by multiplying the current supplied by the time taken to supply the current, where the current is supplied at a constant rate.
In step 42, the number of coulombs entering the battery as it continues to charge is counted, until charging finishes in step 44.
In step 46, the system then refines the initial SoC estimate based on the increase in SoC from convergence and the relative amounts of charge supplied before and after convergence:
SoC_i=SoC_c−(SoC_f−SoC_c)Q ₁ /Q ₂ (equation 6)
Here, SoC_iis the refined value of the initial SoC;
SoC_cis the SoC at the moment of convergence of the SoC;
SoC_fis the final SoC (i.e. 100%);
(SoC_f−SoC_c) is the amount of charge supplied after convergence (as a percentage of the present capacity of the battery);
Q₁is the charge supplied prior to convergence; and
Q₂is the charge supplied from convergence to the end of charging.
In step 48, the present or actual capacity C_actualof the battery is calculated from the refined initial SoC value and the total charge supplied:
C _actual =Q _i +Q ₁ +Q ₂ (equation 7)
Q _i =Q ₂×SoC_i/(SoC_f−SoC_c) (equation 8)
where Q_iis the initial charge stored in the battery.
In step 50, the system then determines the SoH of the battery:
SoH=C _actual /C (equation 9)
Instead, equations 1 and 2 can be used to calculate the SoH, where Q_actual=Q₂and SoC₂=SoC_f.
In a second embodiment, the initial SoC does not need to be refined. Instead, the converged SoC and the final SoC are used. The process follows FIG. 4 until step 44, when the battery is fully charged. Now, referring to FIG. 5, the process continues with step 52, in which the actual capacity is determined from the converged SoC, the final SoC and the charge supplied:
C _actual =Q ₂×SOC_f(SOC_f−SOC_C) (equation 10)
The process then reverts to step 50, in which the SoH is determined as per equation 9. Instead, equations 1 can be used to calculate the SoH, where Q_actual=(Q₁+Q₂) and Q_expected=C(SOC₂−SOC_i).
In a third embodiment, the charging of the battery is fully completed and a linear fit of the coulombs supplied versus measured SoC is used for the SoH determination. This is because the SoC and inputted charge have a linear relationship. The process of FIG. 4 is followed until step 44, when the charging of the battery is finished. Referring now to FIG. 6, in step 60, the SoC values are linearly fitted with the coulombs that are counted as the battery is charged.
FIG. 7 shows a linear fit of the estimated SoC and coulombs counted data, with SoC on the y-axis and coulombs counted on the x-axis. The battery used in this graph has a maximum charge of 10,000 coulombs. Using the slope (S) of the linear fit, the SoH is calculated:
SoH=1.0/(S×C) (equation 11)
This is derived from the maximum input charge Q_fbeing:
Q _f=SoH×C (equation 12)
and the slope being:
S=1/(SoH×C) (equation 13)
If we assume that the battery in the graph of FIG. 7 has a rated capacity of 10,000 Amp seconds, then this means that the battery has an SoH of 100%.
FIG. 8 shows a linear fit of the estimated SoC and coulombs counted data, with SoC on the y-axis and coulombs counted on the x-axis, for another battery rated at 10,000 Amp seconds. The battery used in this graph takes a maximum charge of 5,000 coulombs. Using the slope of the linear fit, the SoH is calculated using equation 11, which leads to a value of SoH=50%. A linear fit compensates for uncertainty in individual measurements of the SoC.
In a fourth embodiment, the charging of the battery does not need to be fully completed for the SoH to be determined. The process of FIG. 4 is followed until step 36 and the SoC has converged. Referring now to FIG. 9, the SoC has converged in step 70. In step 72, the battery continues to be charged as coulombs are counted. Some time later, in step 74, a linear fit is performed on the measured values of the SoC against the coulombs counted. For example, ten more data points for the measured SoC may be taken before the fit is performed. The SoH is then determined from the slope of the fit and the rated capacity of the battery, in step 76. In step 78, the charging of the battery continues until charging is complete.
FIG. 10 shows a graph of the linear fit performed in step 74 (FIG. 9). The graph is for the same battery as for FIG. 8, but it starts at a state of charge of 50%. The slope of this graph remains the same as in the second graph (FIG. 8), and so the SoH can be calculated using equation 11. The SoH can be calculated with any initial SoC.
FIG. 11 is an electrical circuit model of a rechargeable battery (Li-ion cell), which is used in the Extended Kalman Filter. Higher order R-RC models can also be used.
FIG. 12 shows the steps that relate to the operation of an exemplary extended Kalman filter in step 34 of FIG. 4.
In step 100, the filter obtains the initial OCV, which is used in step 31 to estimate the initial SoC. In step 102, an initial SoH range is obtained, based on the SoH classification into which the battery has been classed. For example, a good SoH is in the range>90%; a medium SoH is in the range 70-90%; and a poor range is <70%.
In step 104, the values for the P, Q, R, and X matrices are selected. In step 106, the equivalent circuit model is selected. Use of the OCV/SoC model for poor batteries along with the RC values of a good battery is adequate. However, in other embodiments, RC parameters can be obtained beforehand for poor and medium batteries. In step 108, the initial states of the filter are set, which include the initial SoH classification and the initial estimate of the SoC.
In step 110, the filter waits N ms, which is the time between each iteration of the filter. In step 112, the voltage across the battery terminals is read, and in step 114, the current through the battery is read. In step 116, the filter makes an a priori estimate of X, where F is the prediction matrix, B is the control matrix and u is the control vector (the current). In step 118, the state transition matrix H is updated. In step 120, the measurement matrix is updated.
In step 122, the simulated voltage is calculated from the electrical circuit model. Specifically, the charge supplied to the battery along with the estimated SoC at the previous iteration are used to predict the SoC, which is then used to determine the OCV from the OCV vs. SoC graph for the battery; the OCV is then used in the electric circuit model with the measured current through the battery to calculate an expected (i.e. simulated) battery terminal voltage. The Kalman filter can be used for both charging and discharging, but it is only used for charging in the present embodiment. The objective is to update SoC predication on every iteration based on the error between the measured and the simulated terminal voltage, until the error falls below a certain threshold and the SoC is said to be converged.
In step 124, the error in the simulated voltage is calculated. In step 126, the Kalman gain matrix K is updated. In step 128 the a posteriori estimate of X is made. In step 130, the state variance matrix is updated. The process then loops back to step 110.
The values of R, R₁, and C₁are determined according to the steps in FIG. 13. In step 150, voltage and current data is collected from a new battery using a C/2 charge and discharge rate. The voltage is measured in mV and the current in mA. Typically, 200-400 data points are obtained. Other charge and discharge rates may be used in other embodiments and according to the battery manufacturer's recommendation.
First the battery is fully discharged at a constant current (CC) mode of C/2 until the voltage hits the minimum voltage. The cell is then left to rest for one hour. After this, the cell is fully charged in a CCCV (constant current followed by constant voltage) mode. The charging switches from CC to CV when the battery reaches the maximum terminal voltage that is set (e.g. 4.2V).
In step 152, the battery is then modeled using, for example, a 1st order R-RC Equivalent Circuit Model (ECM) and Particle Swarm Optimization (PSO). In step 154, a swarm of size 49 is initialized with random initial values within the solution range:
0<R<5Ω
0<R1<1Ω
0<C1<100,000F
For 5,000 iterations (steps 156, 186) the process does the following loops (A), (B) and (C) below:
(A) For each particle in the swarm (steps 160, 168), the process does (1) and (2): (1) For each voltage/current data point (steps 162, 174) obtained through discharging and charging, the RMSE between measured and simulated voltage of the battery is updated in step 170. In step 172, if at any point the voltage across R is greater than 2V then the process reverts to initialization of the swarm, in step 154. (2) After the data point number has reached a maximum in step 162, then if, in step 164, the RMSE is the lowest of all the particles so far this iteration, store the particle's position and RMSE in step 166.
(B) Compare the global lowest RMSE with the best RMSE from the present iteration in step 180, and if the present iteration has a lower RMSE, then store the particles position and RMSE in step 182.
(C) In step 184, for each particle in the swarm, the velocity and position of each particle are updated to values based on a set of criteria, such as best individual position, best swarm position, cognitive and social parameters and inertia weight. The position of a particle corresponds to the best R, R1 and C1 values. The velocity of a particle is how fast the particle moves towards the global best position.
When the iterations have been completed, as determined in step 156, the process returns the position where the global lowest RMSE was calculated in step 190, which provides the optimized values for R, R1 and C1.

D. Exemplary System

Referring to FIG. 14, an exemplary system of the battery charger 200 that assess SoH is shown. The charger 200 include a processor 202 operably connected to a computer-readable memory 204 in which is stored a sequence of computer readable instructions in the form of an application 206. The memory also stores the Kalman filter 208. A user interface 220 allows users to input the details concerning a battery, such as the battery's capacity and/or model number. The user interface 220 also includes an output display, which indicates the SoH as determined by the charger 200.
There is an interface 230 to the battery, which may be a battery adapter. The battery adapter in some embodiments is removable from the charger. The battery adapter in some embodiments stores data specific to the batteries to which it is adapted, such as the Kalman filter, the rated capacity of the battery and the OCV-SoC curve(s).
The charger 200 also includes a current source 232 for charging the battery and a voltmeter 234 for measuring the voltage across the battery terminals during the charging process. The charger has a load module 236 for discharging the battery, for example if a discharge current pulse is required to be drawn from the battery when classifying its SoH.
The battery charger 200 optionally includes a network interface 240 through which the charger is connected via a network 242 (e.g. internet and/or cellular network) to a server 244. The server in some embodiments stores data specific to the batteries to be charged and for which the SoH is to be determined, such as the Kalman filter, the rated capacity of the battery and the OCV-SoC curve(s).

E. Variations

The present embodiments include the best presently contemplated mode of carrying out the subject matter disclosed and claimed herein, however, other embodiments are possible.
If the battery being charged does not have its parameters stored, then a self-learning mode can be added as an option to the charger. In this, a battery is fully charged and discharged by C/15 or slower rate and the charger automatically fits a polynomial to the curve. Subsequently, a C/2 charge or any standard charge rate used by the charger is performed, and the resulting data is fitted to the electrical circuit model. These parameters are saved to the adaptor for that particular battery model and can be used in the future as well.
In some embodiments, the electrical circuit model used may be of an order different to the example given, e.g. 2^ndor 3^rdorder.
In some embodiments, for simplicity instead of an electrical circuit model, a curve can be fitted to a portion of the charge profile (e.g. in constant current mode or constant voltage mode), and the Kalman filter only operates during a corresponding portion of the charging.
It is possible to input parameters into the Kalman filter for just new batteries, and extrapolate the parameters of poor and medium SoH batteries from the new ones. This will be done through mapping to the profiles of other pre-tested batteries. For instance, a battery with 50% SoH would take half the time to charge compared to a new battery. So, the charge profile will be shrunk in the x-axis, but in the y-axis it is not obvious how it should be scaled. A study on Li-ion batteries with different SoHs gives a predication model for the charge profile of batteries at different SoHs. This also applies to the circuit model parameters for poor versus good batteries. A database of the study's results shows a trend in the model parameters with SoH.
Once the SoC of the battery is established using the estimation method during the charge process, it can be deployed to tune or control an ultra-fast charging procedure for the battery. Ultra fast charging protocols usually involve charging fast up until a certain SoC and then either continue slowly or stop. The present invention helps to find the actual SoC of the battery so that the charger knows when it reaches the limit of the fast charging.
The SoH determination during ultra-fast charging could be used to evaluate if the battery is suitable for continued ultra-fast charging, or further ultra-fast charging procedures. If the battery's SoH is below a certain threshold, then it is not safe to charge it fast anymore as it speeds up the aging process. The present invention shows the SoH of the battery every time it is charged and can therefore evaluate whether the SoH has fallen below the safe threshold for fast charging.
In some embodiments, the parameter relating to the initial SoC is computed through measurement of the battery open circuit potential.
In some embodiments, the parameter relating to SoC of the battery is computed based on the battery's terminal voltage and battery electrical circuit model, wherein a nonlinear estimation method such as extended Kalman filtering or Smooth variable structure filtering is employed to correct the SoC prediction during the charging process.
In some embodiments, the electrical circuit model used in the estimation algorithm, such as Extended Kalman Filter, is obtained through preprocessing of the data relating to the battery's SoC/OCV in conjunction with the electrical circuit model.
In some embodiments, the data relating to the battery's SoC/OCV relationship and/or the electrical circuit model parameters of both new and aged batteries (at various SoHs) are used during the Kalman filtering and the filter decides the appropriate model through an interacting multiple model (IMM) algorithm.
In some embodiments, a multiple order polynomial is fitted to the average of SoC/OCV of the charge and discharge graphs.
In some embodiments, the hysteresis effects of the SoC/OCV of charge and discharge graph are considered in the OCV model of the electrical circuit model.
In some embodiments, a fitting algorithm such as the particle-swarm method is used to extract parameters from the electrical circuit model.
In some embodiments, the overall charging time is used to calculate the battery remaining capacity (SoH), wherein the time during which the battery had reached its initial SoC is also calculated using the estimation algorithm and considered in the SoH calculation.
In some embodiments, the SoH of the battery is first classified into classes through a test prior to charging.
In some embodiments, a signal is applied to the battery in the form of pulse or continuous wave prior to charging.
In some embodiments, a signal is applied to the battery in the form of pulse or continuous wave after charging.
In some embodiments, the battery response from the excitation signal is evaluated through machine learning algorithms in order to classify the SoH.
In some embodiments, the parameters obtained from exciting the battery prior to charging (or exciting the battery after charging) are normalized based on the battery rated parameters such as rated capacity.
In some embodiments, the parameter relating to classified SoH is employed to adjust the parameters of the estimation method for accurate estimation of SoH within full or partial charging.
In some embodiments, the parameter relating to classified SoH is employed to adjust the SoC/OCV profile and its polynomial fit, and/or battery electrical circuit model for accurate estimation of SoH.
In some embodiments, the parameter relating to the battery charging period is obtained only during the constant current mode, and the estimation algorithm only applies during constant current period
In some embodiments, a simple polynomial fit is used instead of an electrical circuit model to fit to the battery charging profile in a constant current mode.
In some embodiments, the parameter relating to the battery charging period is obtained during the complete charge cycle, and the estimation algorithm applies during the full charging process.
In some embodiments, the parameter relating to the battery temperature is used to compensate for the heat loss during charging.
In some embodiments, the parameter relating to the battery SoH is obtained only during the constant voltage mode, and the estimation algorithm only applies during constant voltage period
In some embodiments, a simple polynomial fit is used instead of an electrical circuit model to fit to the battery charging profile in a constant voltage mode.
Also disclosed is a diagnostic charger unit for charging batteries and reporting the SoC and SoH of the battery during and at the end of charging process, wherein the parameters relating to the battery diagnostic method including electrical circuit models, SoC/OCV profile and nonlinear estimation are obtained through testing batteries, preprocessed, and then utilized during the charging process.
In some embodiments, the parameters relating to the battery and the filter are stored in the battery adaptor connected to the charger unit.
In some embodiments, the parameters relating to the battery and the filter are stored in a cloud-based server and obtained through network connectivity.
Embodiments include a method and an apparatus for a charger equipped with a battery health monitoring system. The battery health monitoring method comprises analyzing the battery electrical information during charging to evaluate the battery state of health. The apparatus comprises a battery charger and is configured to charge the battery and obtain the battery electrical information. In the first embodiment the battery information is obtained prior and during charging. In some embodiments the apparatus comprises additional electronics, wherein the battery discharge response and/or small signal frequency response can be extracted from the battery and is employed to evaluate the condition (i.e. poor, medium or good SoH) of the battery.
Thermal information relating to the battery may be included for cases where the battery heats up during charging. Depending on the temperature, this will trigger an adjustment to the parameters used. The battery charger is configured to obtain the battery's thermal information.
One embodiment is a method for evaluating the condition of a battery during charging, the method comprising: measuring the battery voltage, and temperature prior to charging; recording the battery current, voltage and time during at least part of the charging period; applying an estimation algorithm during the charging process, computing a measure of a condition of the battery based on acquired parameters wherein the first parameter relates to the battery initial SoC; a second parameter relates to the voltage across the RC branch of the battery's electrical circuit model during charging; the third parameter relates to the battery SoC during charging; and the fourth parameter comprises either the duration of charging period, or directly relates to SoH.
In general, unless otherwise indicated, singular elements may be in the plural and vice versa with no loss of generality.
Throughout the description, specific details have been set forth in order to provide a more thorough understanding of the invention. However, the invention may be practiced without these particulars. In other instances, well known elements have not been shown or described in detail and repetitions of steps and features have been omitted to avoid unnecessarily obscuring the invention. Accordingly, the specification and drawings are to be regarded in an illustrative, rather than a restrictive, sense.
The detailed description has been presented partly in terms of methods or processes, symbolic representations of operations, functionalities and features of the invention. These method descriptions and representations are the means used by those skilled in the art to most effectively convey the substance of their work to others skilled in the art. A software implemented method or process is here, and generally, understood to be a self-consistent sequence of steps leading to a desired result. These steps require physical manipulations of physical quantities. Often, but not necessarily, these quantities take the form of electrical or magnetic signals or values capable of being stored, transferred, combined, compared, and otherwise manipulated. It will be further appreciated that the line between hardware and software is not always sharp, it being understood by those skilled in the art that the software implemented processes described herein may be embodied in hardware, firmware, software, or any combination thereof. Such processes may be controlled by coded instructions such as microcode and/or by stored programming instructions in one or more tangible or non-transient media readable by a computer or processor. The code modules may be stored in any computer storage system or device, such as hard disk drives, optical drives, solid-state memories, etc. The modules may alternatively be embodied partly or wholly in specialized computer hardware, such as ASIC or FPGA circuitry.
It will be clear to one having skill in the art that further variations to the specific details disclosed herein can be made, resulting in other embodiments that are within the scope of the invention disclosed. Steps in the flowcharts may be performed in a different order, other steps may be added, or one or more steps may be removed without altering the main function of the system. Steps may be made to occur in parallel. Flowcharts from different figures may be combined in different ways. Modules may be divided into constituent modules or combined into larger modules. All parameters, factors, thresholds and configurations described herein are examples only and actual values of such depend on the specific embodiment. Accordingly, the scope of the invention is to be construed in accordance with the substance defined by the following claims.

Claims

1. A method of determining state of health (SoH) of a battery of rated capacity C comprising the steps of:

estimating an initial value of a state of charge (SoC) of the battery;

starting to charge the battery;

repeatedly measuring a voltage across terminals of the battery as the battery is charged;

obtaining a converged value (SoC_C) of the SoC from the repeated measurements;

determining a later value (SoC₂) of the SoC after further charging the battery;

determining a charge (Q₂) supplied to the battery between SoC_Cand SoC₂;

calculating, using SoC₂, SoC_C, Q₂and C, a value of the SoH of the battery.

2. The method of claim 1 further comprising:

determining a charge (Q_i) supplied to the battery between starting to charge the battery and SoC_C; and

calculating a refined value (SoC_i) of the initial SoC using Q₁, Q₂, SoC_Cand SoC₂;

wherein the SoH is calculated as (Q₁+Q₂)/[C(SoC₂−SoC_i)].

3. The method of claim 1 wherein the SoH is calculated as Q₂/[C(SoC₂−SoC_C)].

4. The method of claim 1, wherein SoC₂is 100%.

5. The method of claim 1, wherein the battery is fully discharged when the initial value of the SoC is estimated.

6. The method of claim 1, wherein estimating the initial value of the SoC comprises:

measuring an open circuit voltage (OCV) of the battery to result in value V; and

using V to determine the initial value by accessing a relationship between the OCV and SoC, the relationship being represented by a graph, an analytic function or a table.

7. The method of claim 6, wherein the relationship is defined by a 10th order polynomial.

8. The method of claim 7, wherein the 10th order polynomial represents an average of OCV and SoC data measured during charging and discharging the battery or another battery that is comparable to the battery.

9. The method of claim 7 comprising:

charging the battery fully at a rate of C/15 or lower;

discharging the battery fully at a rate of C/15 or lower;

deriving the 10^thorder polynomial from data recorded said charging and discharging;

subsequently charging the battery while recording further data; and

fitting the further data to an electrical circuit model of the battery that is used to predict the voltage across the terminals.

10. The method of claim 1, comprising:

repeatedly predicting values of the voltage across the terminals using a Kalman filter;

repeatedly comparing the predicted values with the measured voltages; and

determining that a difference between the predicted values and the measured voltages is below a threshold;

wherein the SoC_Cis obtained when said difference is below the threshold.

11. The method of claim 10, wherein the voltage across the terminals is predicted using an electric circuit model of the battery, a current supplied to the battery and a 10th order polynomial relationship between the OCV and SoC.

12. The method of claim 11, comprising automatically selecting the electric circuit model from a choice of multiple electric circuit models using an interacting multiple model algorithm.

13. The method of claim 11, wherein the electric circuit model is a first, second or third order model.

14. The method of claim 11, wherein a particle-swarm method is used to extract parameters from the electric circuit model for use in prediction of the voltage across the terminals.

15. The method of claim 10, wherein the difference is a root mean square difference of multiple sequential comparisons of the predicted values and the measured voltages.

16. The method of claim 10, wherein:

the battery is charged with constant current; and

the voltage across the terminals is predicted using a polynomial.

17. The method of claim 1, wherein the voltage across the terminals is measured multiple times per second.

18. The method of claim 1 comprising, prior to charging the battery, categorizing the SoH of the battery as good, medium or poor;

wherein estimating the initial value of the SoC comprises:

using V to determine the initial value by accessing a relationship between the OCV and SoC, the relationship being defined by a corresponding one of three different 10th order polynomials, each polynomial representing a good, medium or poor SoH.

19. The method of claim 1, wherein the battery is charged with a constant current.

20. The method of claim 1, wherein the SoC_Cis obtained when the battery is being charged in constant voltage mode, and the further charging occurs in constant voltage mode.

21. The method of claim 1, comprising:

obtaining multiple converged values of the SoC of the battery;

creating a graph of the multiple converged values of the SoC against charge supplied to the battery; and

calculating a slope (S) of the graph;

wherein the SoH is calculated as 1.0/SC.

22. A battery charger for determining state of health (SoH) of a battery of rated capacity C comprising:

a battery interface that connects to the battery;

a user interface;

a processor connected to the battery interface and the user interface;

a computer readable memory storing computer readable instructions, which, when executed by the processor, cause the battery charger to:

estimate an initial value of a state of charge (SoC) of the battery;

start to charge the battery;

repeatedly measure a voltage across terminals of the battery as the battery is charged;

obtain a converged value (SoC_C) of the SoC from the repeated measurements;

determine a later value (SoC₂) of the SoC after further charging the battery;

determine a charge (Q₂) supplied to the battery between SoC_Cand SoC₂;

calculate, using SoC₂, SoC_C, Q₂and C, a value of the SoH of the battery; and

output the calculated value of the SoH on the user interface.

23. The battery charger of claim 22, configured to estimate the initial value of the SoC by:

using V to determine the initial value by accessing a 10th order polynomial relationship between the OCV and SoC;

and further configured to:

repeatedly predict values of the voltage across the terminals using a Kalman filter;

repeatedly compare the predicted values with the measured voltages; and

determine that a difference between the predicted values and the measured voltages is below a threshold;

wherein the SoC_Cis obtained when said difference is below the threshold.

24. The battery charger of claim 23, wherein the battery interface is a battery adapter and the Kalman filter is stored in the battery adapter.

25. The battery charger of claim 23, further comprising an interface to an internet-based server, wherein the Kalman filter is stored on the server.

26. The battery charger of claim 22, wherein the battery charger is an ultra-fast charger and configured to control a rate of charging using an estimate of the SoC obtained during charging.