Claims:We claim:
1. A method for estimating the State of Health (SOH) of a battery cell, the method comprising:
computing, by a Battery Management System (BMS) (400), a plurality of values of a parameter indicative of depreciation of voltage of a battery cell during a first discharge cycle, with respect to discharge of the battery cell, wherein the depreciation of voltage is observed within a plurality of ranges of discharge; and
estimating, by a Deep Neural Network (DNN) in the BMS (400), a capacity of the battery cell after the first discharge cycle, based on the plurality of values of the parameter, wherein the DNN is trained based on a plurality of values of the parameter pertaining to a plurality of battery cells, obtained during each of a plurality of discharge cycles of the plurality of battery cells.
2. The method, as claimed in claim 1, wherein the method further comprises
estimating a capacity of the battery cell after a second discharge cycle based on a plurality of computed values of the parameter; and
determining a capacity fade of the battery cell by comparing the capacity of the battery cell after the first discharge cycle and the capacity of the battery cell after the second discharge cycle.
3. The method, as claimed in claim 1, wherein each of the plurality of values of the parameter are obtained based on values of the voltage of the battery cell within each of the plurality of ranges of discharge, and a maximum value and a minimum value of each of the plurality of ranges of discharge.
4. The method, as claimed in claim 3, wherein the voltage of the battery cell within each of the plurality of ranges of discharge is determined based on at least one of number of discharge cycles undergone by the battery cell, a depth of discharge of the battery cell, and a capacity of the battery cell.
5. The method, as claimed in claim 1, wherein the DNN comprises of two hidden layers comprising of a plurality of neurons.
6. A Battery Management System (BMS) (400) for estimating the State of Health (SOH) of a battery cell, the BMS (400) configured to:
compute a plurality of values of a parameter indicative of depreciation of voltage of a battery cell during a first discharge cycle, with respect to discharge of the battery cell, wherein the depreciation of voltage is observed within a plurality of ranges of discharge; and
estimating, by a Deep Neural Network (DNN) in the BMS (400), a capacity of the battery cell after the first discharge cycle, based on the plurality of values of the parameter, wherein the DNN is trained based on a plurality of values of the parameter pertaining to a plurality of battery cells, obtained during each of a plurality of discharge cycles of the plurality of battery cells.
7. The BMS (400), as claimed in claim 6, wherein the BMS (400) is further configured to
estimate a capacity of the battery cell after a second discharge cycle based on a plurality of computed values of the parameter; and
determine a capacity fade of the battery cell by comparing the capacity of the battery cell after the first discharge cycle and the capacity of the battery cell after the second discharge cycle.
8. The BMS (400), as claimed in claim 6, wherein each of the plurality of values of the parameter are obtained based on values of the voltage of the battery cell within each of the plurality of ranges of discharge, and a maximum value and a minimum value of each of the plurality of ranges of discharge.
9. The BMS (400), as claimed in claim 8, wherein the voltage of the battery cell within each of the plurality of ranges of discharge is determined based on at least one of number of discharge cycles undergone by the battery cell, a depth of discharge of the battery cell, and a capacity of the battery cell.
10. The BMS (400), as claimed in claim 6, wherein the DNN comprises of two hidden layers comprising of a plurality of neurons.
11. A method for estimating the State of Health (SOH) of a battery cell, the method comprising:
computing, by a Battery Management System (BMS) (400), a plurality of values of a parameter indicative of rise of voltage of a battery cell, during a first charging cycle, with respect to charge stored in the battery cell, wherein the rise of voltage is observed within a plurality of ranges of charge stored in the battery cell; and
estimating, by a Deep Neural Network (DNN) in the BMS (400), a capacity of the battery cell after the first charging cycle, based on the plurality of values of the parameter, wherein the DNN is trained based on a plurality of values of the parameter pertaining to a plurality of battery cells, obtained during each of a plurality of charging cycles of the plurality of battery cells.
12. The method, as claimed in claim 11, wherein the method further comprises
estimating a capacity of the battery cell after a second charging cycle based on a plurality of computed values of the parameter; and
determining a capacity fade of the battery cell by comparing the capacity of the battery cell after the first charging cycle and the capacity of the battery cell after the second charging cycle.
13. The method, as claimed in claim 11, wherein each of the plurality of values of the parameter are obtained based on values of the voltage of the battery cell within each of the plurality of ranges of charge stored in the battery cell, and a maximum value and a minimum value of each of the plurality of ranges of charge stored in the battery cell.
14. The method, as claimed in claim 13, wherein the voltage of the battery cell within each of the plurality of ranges of charge stored in the battery cell is determined based on at least one of number of charging cycles undergone by the battery cell, amount of charge stored in the battery cell, and a capacity of the battery cell.
15. The method, as claimed in claim 11, wherein the DNN comprises of two hidden layers comprising of a plurality of neurons.
16. A Battery Management System (BMS) (400) for estimating the State of Health (SOH) of a battery cell, the BMS (400) configured to:
compute a plurality of values of a parameter indicative of rise of voltage of a battery cell, during a first charging cycle, with respect to charge stored in the battery cell, wherein the rise of voltage is observed within a plurality of ranges of charge stored in the battery cell; and
estimate, by a Deep Neural Network (DNN) in the BMS (400), a capacity of the battery cell after the first charging cycle, based on the plurality of values of the parameter, wherein the DNN is trained based on a plurality of values of the parameter pertaining to a plurality of battery cells, obtained during each of a plurality of charging cycles of the plurality of battery cells.
17. The BMS (400), as claimed in claim 16, wherein the BMS (400) is further configured to
estimate a capacity of the battery cell after a second charging cycle based on a plurality of computed values of the parameter; and
determine a capacity fade of the battery cell by comparing the capacity of the battery cell after the first charging cycle and the capacity of the battery cell after the second charging cycle.
18. The BMS (400), as claimed in claim 16, wherein each of the plurality of values of the parameter are obtained based on values of the voltage of the battery cell within each of the plurality of ranges of charge stored in the battery cell, and a maximum value and a minimum value of each of the plurality of ranges of charge stored in the battery cell.
19. The BMS (400), as claimed in claim 18, wherein the voltage of the battery cell within each of the plurality of ranges of charge stored in the battery cell is determined based on at least one of number of charging cycles undergone by the battery cell, amount of charge stored in the battery cell, and a capacity of the battery cell.
20. The BMS (400), as claimed in claim 16, wherein the DNN comprises of two hidden layers comprising of a plurality of neurons.
, Description:TECHNICAL FIELD
Embodiments herein relate to estimation of State of Health (SOH) of battery packs, and more particularly to methods and systems for estimating the SOH a battery cell using a Deep Neural Network (DNN).

BACKGROUND
Several Li-ion based cell chemistry are available commercially which can be used for energy storage purposes. A plurality of cells belonging to any of the Li- ion cell chemistries can be assembled together to create a battery pack. The battery cells serve as the power source of an electronic device or an electric vehicle. The battery cells of the electronic device or the electric vehicle is managed by a Battery Management System (BMS) unit, which is as a part of the power distribution unit of energy management system of the electronic device or the electric vehicle. The Li-ion cells lose the capability of storing energy progressively with usage (increasing number of charging cycles/discharging cycles), due to degradation of the chemical materials in the cells. Therefore, the Total Available Capacity (TAC) of the Li-ion cells reduces along with usage. The progressive loss of the capability to store energy can be referred to as Capacity Fade or Capacity Degradation. The ratio of the TAC of a cell and the maximum available capacity of the cell (the cell is having the maximum capacity when it is fresh), expressed in percentage, is referred to as the State of Health (SOH) of the cell.
Currently, the measurement or prediction of the SOH of a cell requires monitoring of rise/depreciation of voltages of the battery cells throughout the charging cycle or discharging cycle. Such monitoring may not be possible as the user is likely to either charge the cell prior to the State of Charge (SOC) dropping below the critical levels (close to 0% SOC) or cease the charging of the cell prior to the cell attaining full charge (close to 100% SOC). For example, the user of the electric vehicle is likely to charge the battery of the electric vehicle sufficiently prior to the SoC of the battery dropping to/below the critical levels. In another example, a user of an electronic device may cease to charge the battery of the electronic device if the user is in a hurry or if the user feels that the SOC of the battery is sufficient at that time instant. Therefore, in these scenarios, the measurement or the prediction of the SOH of the battery cell may not be accurate.
If the SOH has been incorrectly estimated after a charging/discharging cycle, it is likely that the SOH error will be propagated in the subsequent charging/discharging cycles. Thus, the existing methods of SOH estimation are susceptible to cumulative errors. Further, there can be variations in parameters pertaining to the cells such as capacity, SOC, temperature, resistance, and so on, amongst the cells of a battery pack. The variations in the parameters can introduce variations in SOH values amongst the cells of the battery pack. The existing methods of SOH estimation may not account for the variations in the parameters amongst the cells of the battery pack, thereby contributing to erroneous estimation of the SOH. Thus, precise computation of SOH of the battery pack and the SOH of each cell in the battery pack using the existing methods remains a challenge.

 
OBJECTS
The principal object of the embodiments herein is to disclose methods and systems for estimating and/or predicting the State of Health (SOH) of a battery cell by estimating the capacity fade of the battery cell over a plurality of charging/discharging cycles.
Another object of the embodiments herein is to accurately estimate and/or predict the capacity fade of the battery cell, irrespective of the type of chemistry of the battery cell.
Another object of the embodiments herein is to estimate or predict the capacity fade of the battery cell using a Deep Neural Network (DNN), wherein the estimation or prediction is based on Area-Under-Curve (AUC) values and battery cell voltage.
Another object of the embodiments herein is to observe variations in battery cell voltage with respect to variations in the amount of charge that is stored in, or drained from, the battery cell during progressive charging/discharging cycles of the battery cell, in order to compute the AUC values.
Another object of the embodiments herein is to compute a plurality AUC values for each charging/discharging cycle of the battery cell based on rise/depreciation of the battery cell voltage within a plurality of ranges of charge stored in, or drained from, the battery cell, during each charging/discharging cycle.
Another object of the embodiments herein is to build the DNN model based on AUC values obtained during the charging or discharging cycles of a plurality of battery cells of a battery pack, wherein the battery cell, whose corresponding capacity fade is estimated or predicted, is one of the battery cells of the battery pack.

SUMMARY
Accordingly, the embodiments provide methods and systems for estimating and/or predicting the State of Health (SOH) of a battery cell by estimating the capacity fade of the battery cell. The embodiments include determining the capacity fade of the battery cell, which is one of a plurality of battery cells in the battery pack, based on Area-Under-Curve (AUC) values, which are obtained during charging cycles or discharging cycles of the battery cell. The embodiments include observing the variation of the voltage of a battery cell with respect to variations in the amount of charge that is stored in, or drained from, the battery cell, during different the charging cycles or discharging cycles of the battery cell.
The embodiments include determining the voltage of the battery cell when charge is stored in the battery cell during a charging cycle. The embodiments include determining the voltage of the battery cell when charge is drained from the battery cell during a discharging cycle. If the battery cell is being charged, the embodiments include determining the rise of the voltage of the battery cell with respect to the amount of charge that is stored in the battery cell. If the battery cell is being discharged, the embodiments include determining the depreciation of the voltage of the battery cell with respect to the amount of charge that is drained from the battery cell. The embodiments include storing the variations of the voltage of the battery cell with respect to the amount of charge that is stored in, or drained from, the battery cell, during the charging cycle or discharging cycle of the battery cell.
The embodiments include computing a plurality AUC values during the charging cycle of the battery cell based on rise of the battery cell voltage within a plurality of ranges of charge stored in the battery cell, during the charging cycle, wherein the plurality of AUC values can be computed based on the variations of the voltage of the battery cell with respect to the amount of charge that is stored in the battery cell. Similarly, the embodiments include computing a plurality AUC values during the discharging cycle of the battery cell based on depreciation of the battery cell voltage within a plurality of ranges of charge drained from the battery cell, during the discharging cycle. The embodiments include computing a plurality AUC values during a plurality of charging cycles and a plurality of discharging cycles. In an embodiment, the plurality of AUC values can be computed when the charge stored in, or drained from, the battery cell falls within predefined ranges.
The embodiments include predicting the capacity of a battery cell based on the received AUC values. The neural network processor 403 can predict the capacity fade of the battery cell based on previously predicted values of capacity of the battery cell, wherein the neural network processor 403 had predicted the values of capacity of the battery cell during the previous charging cycles or discharging cycles of the battery cell. Similarly, the neural network processor 403 can estimate and/or predict the capacity fade of other battery cells in the battery pack 406.
These and other aspects of the embodiments herein will be better appreciated and understood when considered in conjunction with the following description and the accompanying drawings. It should be understood, however, that the following descriptions, while indicating embodiments and numerous specific details thereof, are given by way of illustration and not of limitation. Many changes and modifications may be made within the scope of the embodiments herein without departing from the spirit thereof, and the embodiments herein include all such modifications.

BRIEF DESCRIPTION OF FIGURES
Embodiments herein are illustrated in the accompanying drawings, throughout which like reference letters indicate corresponding parts in the various figures. The embodiments herein will be better understood from the following description with reference to the drawings, in which:
FIG. 1 depicts example plots, which demonstrate the phenomenon of capacity fade in battery cells of a battery pack, with increasing number of discharge cycles undergone by the battery cells, according to embodiments as disclosed herein;
FIG. 2 is an example graph depicting variations of voltage of a battery cell with respect to amount of charge drained from the battery cell during different discharge cycles, according to embodiments as disclosed herein;
FIG. 3 depicts example scatter plots demonstrating the variations of Area-Under-Curve (AUC) with respect to the capacities of two battery cells during discharge cycles of the two battery cells, according to embodiments as disclosed herein;
FIG. 4 depicts a Battery Management System (BMS) configured to estimate and/or predict the State of Health (SOH) of a battery cell by estimating the capacity fade of the battery cell using a Deep Neural Network (DNN) model, according to embodiments as disclosed herein;
FIG. 5 is a flowchart depicting a method for estimating and/or predicting the SOH of the battery cell by estimating the capacity fade of the battery cell using the DNN model, according to embodiments as disclosed herein;
FIGS. 6a and 6b are graphs depicting comparisons between actual values of capacity fade (loss) and predicted values of capacity fade for two battery cells operating in different ambient temperatures, according to embodiments as disclosed herein; and
FIGS. 7a and 7b are graphs depicting absolute error involved in predicting the capacity fade, after modification, for the two battery cells operating in the different ambient temperatures, according to embodiments as disclosed herein.

DETAILED DESCRIPTION
The embodiments herein and the various features and advantageous details thereof are explained more fully with reference to the non-limiting embodiments that are illustrated in the accompanying drawings and detailed in the following description. Descriptions of well-known components and processing techniques are omitted so as to not unnecessarily obscure the embodiments herein. The examples used herein are intended merely to facilitate an understanding of ways in which the embodiments herein may be practiced and to further enable those of skill in the art to practice the embodiments herein. Accordingly, the examples should not be construed as limiting the scope of the embodiments herein.
Embodiments herein disclose methods and systems for estimating and/or predicting the State of Health (SOH) of a battery cells by estimating the capacity fade of the battery cells over a plurality of charging or discharging cycles. The embodiments include estimating or predicting the capacity fade of the battery cell using a Deep Neural Network (DNN), wherein the estimation or prediction is based on Area-Under-Curve (AUC) values. The AUC values are computed based on the voltage of the battery cell. The embodiments include accurately estimating and/or predicting the capacity fade of the battery cell, irrespective of the type of chemistry of the battery cell. The embodiments include computing the AUC values based on variation of the voltage of the battery cell with respect to variation in the amount of charge that is stored in, or drained from, the battery cell, during progressive charging cycles or discharging cycles of the battery cell. The embodiments include computing a plurality AUC values for each charging cycle of the battery cell based on rise of the battery cell voltage within a plurality of ranges of charge stored in the battery cell, during each charging cycle. The embodiments include computing a plurality AUC values for each discharging cycle of the battery cell based on depreciation of the battery cell voltage within a plurality of ranges of charge drained from the battery cell, during each discharging cycle. The embodiments include building the DNN model based on AUC values that obtained during the charging or discharging cycles of a plurality of battery cells of a battery pack. The embodiments include estimating and/or predicting the capacity fade of a battery cell, which is one of the battery cells of the battery pack using the trained DNN model.
Referring now to the drawings, and more particularly to FIGS. 1 through 7, where similar reference characters denote corresponding features consistently throughout the figures, there are shown preferred embodiments.
FIG. 1 depicts example plots, which demonstrate the phenomenon of capacity fade in battery cells of a battery pack, with increasing number of discharge cycles undergone by the battery cells, according to embodiments as disclosed herein. As depicted in plot 101, the capacity fade of a cell 445 has been observed in an environment where the ambient temperature is 250 C. The initial capacity of the cell 445 was greater than 72 Ampere-Hour. The capacity of the cell 445 fades below 68 Ampere-Hour after 1000 cycles. As depicted in plot 102, the capacity fade of a cell 1446 has been observed in an environment where the ambient temperature is 250 C. The initial capacity of the cell 1446 was greater than 72.5 Ampere-Hour. The capacity of the cell 1446 is close to 65 Ampere-Hour after 1300 cycles.
As depicted in plot 103, the capacity fade of a cell 1493 has been observed in an environment where the ambient temperature is 450 C. The initial capacity of the cell 1493 was close to than 75 Ampere-Hour. The capacity of the cell 1493 is close to 62.5 Ampere-Hour after 475 cycles. As depicted in plot 104, the capacity fade of a cell 1583 has been observed in an environment where the ambient temperature is 450 C. The initial capacity of the cell 1583 was close to than 75 Ampere-Hour. The capacity of the cell 1583 is close to 62.5 Ampere-Hour after 450 cycles.
The cells 445, 1446, 1493, and 1583, are the cells of a battery pack. It is observed in the plots that the capacity of the battery cells degrades with usage. The battery cells undergo an increasing number of charging/discharging cycles, with continued usage of the battery cells. The degradation of the capacity of the battery cells with usage is influenced by various factors such as temperature, C-rate (discharge rate), depth of discharge, and so on. Amongst the aforementioned factors, the temperature is the greatest influence on the capacity fade. With an increase in the ambient temperature, in which the battery cell is operating, the rate of degradation of the capacity of the battery increases. This can be observed in the plots 103 and 104, wherein the capacity of the cell 103 and the capacity of the cell 104, operating in an environment where the temperature is 450 C, decreases sharply (from close to 75 Ampere-Hour to 62.5 Ampere-Hour) within 500 discharge cycles.
It is to be noted that the plots 101-104 were used to train a DNN model. The trained DNN model could be used for prediction of capacity fade of other cells of the battery pack.
FIG. 2 is an example graph depicting variations of voltage of a battery cell with respect to amount of charge drained from the battery cell during different discharge cycles, according to embodiments as disclosed herein. As depicted in FIG. 2, the voltage of the battery cell depreciates as charge is drained from the battery cell. The rate of voltage depreciation increases with increase in the number of charging/discharging cycles undergone by the battery cell. In this example, the battery cell is the cell 1583, and the depreciation of the voltage of the cell 1583 is observed in an environment in which the ambient temperature is 450 C. The graph depicts a comparison of two plots, viz., voltage depreciation of a battery cell with respect to amount of charge drained from the cell 1583 at 450 C during the first discharge cycle (cycle 0), and voltage depreciation of the cell 1583 with respect to amount of charge drained from the cell 1583 at 450 C during the 443rd discharge cycle.
If the voltage of the cell 1583 is close to 3.4 Volts, the State of Charge (SOC) of the cell 1583 is close to 100% (cell 1583 is fully charged). If the voltage of the cell 1583 is below 2.6 Volts, the State of Charge (SOC) of the cell 1583 is close to 0% (cell 1583 is drained of all charge). There is a capacity fade, i.e., the capacity of the cell 1583 progressively decreases as the number of discharge cycles undergone by the cell 1583 increases. As depicted in the plots, after the first discharge cycle, the capacity of the battery is close to 75 Ampere-Hour, whereas, after the 443rd discharge cycle, the capacity of the battery is close to 64 Ampere-Hour. The capacity of the cell 1583 can be obtained based on the amount of charge that needs to be drained from the cell 1583, such that the voltage of the cell drops from (close to) 3.4 Volts to (close to) 2.6 Volts. The amount of charge drained from the battery after the first discharge cycle is close to 75 Ampere-Hour. The amount of charge drained from the battery after the 443rd discharge cycle is close to 64 Ampere-Hour. Thus, the capacity fade of the cell 1583 after 443 discharging cycles is close to 11 Ampere-Hour.
The embodiments include predicting/estimating the SOH of the cell 1583 by estimating the capacity fade of the cell 1583. The capacity fade of the cell 1583 can be measured by observing the depreciation of the voltage of the cell 1583 during the discharge cycle. The embodiments include selecting certain ranges of amount of charge drained from the cell 1583. In this example, the embodiments include selecting the ranges 7-10 Ampere-Hour, 22-25 Ampere-Hour, 42-45 Ampere-Hour and 57-60 Ampere-Hour. It can be noted that each of the four ranges is 3 Ampere-Hour.
The voltage of the cell 1583 can be observed once 7 Ampere-Hour amount of charge has been drained from the cell 1583. The observation can be ceased after 10 Ampere-Hour amount of charge has been drained from the cell 1583. The voltage of the cell 1583 can be observed once 22 Ampere-Hour amount of charge has been drained from the cell 1583. The observation can be ceased after 25 Ampere-Hour amount of charge has been drained from the cell 1583. The voltage of the cell 1583 can be observed once 42 Ampere-Hour amount of charge has been drained from the cell 1583. The observation can be ceased after 45 Ampere-Hour amount of charge has been drained from the cell 1583. The voltage of the cell 1583 can be observed once 57 Ampere-Hour amount of charge has been drained from the cell 1583. The observation can be ceased after 60 Ampere-Hour amount of charge has been drained from the cell 1583.
The embodiments include computing a plurality of AUC values during each discharge cycle pertaining to a plurality of the ranges of amount of charge drained from the cell 1583. In this example, the embodiments include four AUC values during each discharge cycle corresponding to the four ranges of amount of charge drained from the cell 1583, i.e., 7-10 Ampere-Hour, 22-25 Ampere-Hour, 42-45 Ampere-Hour and 57-60 Ampere-Hour. An AUC value in a range can be computed based on the voltages of the cell 1583, as charge continues to drain from the cell 1583 within the range, during the discharge cycle; the maximum value of the charge drained from the cell 1583 in the range; and the minimum value of the charge drained from the cell 1583 in the range. For the range 22-25 Ampere-Hour, the AUC value can be obtained based on the voltages of the cell 1583, as charge from the cell 1583 is drained from 22 Ampere-Hour to 25 Ampere-Hour during the discharge cycle; the maximum value of the amount of charge drained from the cell 1583 in the range 22-25 Ampere-Hour, i.e., 25 Ampere-Hour; and the minimum value of the charge drained from the cell 1583 in the range 22-25 Ampere-Hour, i.e., 22 Ampere-Hour.
The value of AUC for a particular range of amount of charge drained from the battery cell in a particular charging/discharging is given as:
〖AUC〗_(cycle n)^(range (AHmax-AHmin))=(∫_AHmax^AHmin▒〖V_t d(AH)〗)/(AHmax-AHmin ) ------ (equation-1)
Here, Vt represents the voltage, which depreciates as the charge in the battery cell continues to drain from AHmin to AHmax. The value of AUC corresponding to the range 22-25 Ampere-Hour, wherein the charge drained from the battery is in the range 22-25 Ampere-Hour, during the 443rd cycle is given as:
〖AUC〗_443^((22-25)AH)=(∫_25^22▒〖V_t d(AH)〗)/(25-22)=1/3 ∫_25^22▒〖V_t d(AH)〗
The embodiments include estimating the capacity of the cell 1583 operating in the environment with temperature of 250 C, based on four AUC values obtained each discharge cycle. The capacities of the cell 1583, estimated using four AUC values in the first discharge cycle and four AUC values in the 443rd discharge cycle can be compared for estimating/predicting the capacity fade of the cell 1583, and thereby the SOH of the cell 1583.
FIG. 3 depicts example scatter plots demonstrating the variations of AUC with respect to the capacities of two battery cells during discharge cycles of the two battery cells, according to embodiments as disclosed herein. A first scatter plot demonstrates the variation of AUC values with respect to the capacity of the cell 445, which is operating in the environment with temperature of 250 C. A second scatter plot demonstrates the variation of AUC values with respect to the capacity of the cell 1583, which is operating in the environment with temperature of 450 C. The capacities of the cells 445 and 1583 are represented in Ampere-Hour. The capacity of the cell 445 is depicted in the left ordinate and the capacity of the cell 1583 is depicted in the right ordinate.
The points in the scatter plots represent the values of the AUC and the corresponding capacity of the cell, operating in the environment having an ambient temperature of either 250 C of 450 C, after a particular discharge cycle. The scatter plots depict those AUC values and the corresponding capacities of the cells 445 and 1583, which have been obtained in the range of 7-10 Ampere-Hour.
The AUC values of the cell 445 are obtained for about 800 discharge cycles. Therefore, the first scatter plot includes about 800 points, i.e., AUC values. The observation of the depreciation of voltage of the cell 445 is started when the amount of charged drained from the cell 445 is 7 Ampere-Hour, and the observation of the depreciation of voltage of the cell 445 is ceased when the amount of charged drained from the cell 445 is 10 Ampere-Hour. This observation can be repeated for 800 cycles for obtaining the 800 AUC values in the first scatter plot.
The AUC values of the cell 1583 are obtained for about 1400 discharge cycles. Therefore, the second scatter plot includes about 1400 points, i.e., AUC values. The observation of the depreciation of voltage of the cell 1583 is started when the amount of charged drained from the cell 1583 is 7 Ampere-Hour, and the observation of the depreciation of voltage of the cell 1583 is ceased when the amount of charge drained from the cell 1583 is 10 Ampere-Hour. This observation can be repeated for 1400 cycles for obtaining the 1400 AUC values in the second scatter plot.
As the number of discharge cycles undergone by a battery cell increases, the rate of depreciation of voltage of the battery cell increases. For example, the rate of depreciation of voltage of the battery cell during the 443rd discharge cycle will be higher compared to the depreciation of voltage of the battery cell during the first discharge cycle. The voltage of a battery cell after 40 AH discharge during the 443rd will be considerably lower than the voltage of the battery cell during the first discharge cycle.
Further, within a particular discharge cycle, the rate of depreciation of voltage of the battery cell is more for ranges indicating higher values of amount of charge (in Ampere-Hour) drained from the battery cell, compared to ranges indicating lower values of amount of charge (in Ampere-Hour) drained from the battery cell. For example, during the 443rd discharge cycle, the rate of depreciation of voltage of the battery cell in the range 22-25 Ampere-Hour will be higher compared to the rate of depreciation of voltage of the battery cell in the range 7-10 Ampere-Hour. This can be observed by computing a first difference between the voltages of the battery cell when the amount of charge drained from the battery cell is 7 Ampere-Hour and 10 Ampere-Hour respectively. Similarly, a second difference can be computed between the voltages of the battery cell when the amount of charge drained from the battery cell is 22 Ampere-Hour and 25 Ampere-Hour respectively. By comparing the first and second differences, it can be noted that the second difference is higher compared to the first difference.
Thus, the capacity of the battery cell decreases with the increase in the number of discharge cycles undergone by a battery cell. This is because; a faster rate of depreciation of voltage of the battery cell (with increasing in the number of discharge cycles undergone by a battery cell) decreases the amount of charge that is drained from the battery cell prior to the complete discharge of the battery cell. The fluctuation of voltage of the battery cell increases with the increase in the ambient temperature in which the cell is operating. Therefore, the fluctuation of voltage of the cell 1583, which is operating in an environment with an ambient temperature of 450 C, is higher compared to that of the cell 445 operating in an environment with an ambient temperature of 250 C.
The AUC value corresponding to a range of amount of charge drained from a battery cell is directly proportional to the rate of depreciation of voltage within the range of amount of charge drained from a battery cell (equation-1). In an example, the AUC value that corresponds to the range 7-10 Ampere-Hour is lower than the AUC value that corresponds to the range 22-25 Ampere-Hour. The rate of depreciation of the voltages of the cells 445 and 1583 and actual values of the voltages of the cells 445 and 1583, within the range 7-10 Ampere-Hour, is greater than the rate of depreciation of the voltages of the cells 445 and 1583 and actual values of the voltages of the cells 445 and 1583, within the range 22-25 Ampere-Hour (FIG. 2).
As depicted in FIG. 3, the values of the AUC and the values of the capacity decrease with increase in the number of charging/discharging cycles. This is because the voltage depreciates with the increase in the number of charging/discharging cycles. This is because the voltage of the cells 445 and 1583 after undergoing 500 discharging cycles is lower than the voltage of the cells 445 and 1583 after undergoing 250 discharging cycles (equation-1).
In the first scatter plot, i.e., cell 445 operating at 250 C, the values of AUC and the values of capacity vary within a narrow range. The values of AUC varies in the range 9.55-9.59 i.e., 0.04. The values of capacity vary in the range 67-71, i.e., 5 Ampere-Hour. In the second scatter plot, i.e., cell 1583 operating at 450 C, the values of AUC and the values of capacity vary within a wider range, compared to the first scatter plot. The values of AUC vary in the range 9.52-9.67, i.e., (0.15). The values of capacity vary in the range 64-76, i.e., 12 Ampere-Hour. This is because there is a greater fluctuation of voltage in the cell 1583 operating at 450 C, compared to the cell 445 and operating at 250 C.
FIG. 4 depicts a Battery Management System (BMS) 400 configured to estimate and/or predict the SOH of a battery cell by estimating the capacity fade of the battery cell using a DNN model, according to embodiments as disclosed herein. As depicted in FIG. 4, the BMS 400 includes a Power Management Integrated Circuit (PMIC) 401. The PMIC 401 includes a computing unit 402, a neural network processor 403, a memory 404 and a display 405. The BMS 400 can be a part of an electronic device or an electric vehicle, which includes a battery pack 406. The battery pack 406 comprises of a plurality of battery cells. The PMIC 401 can utilize the neural network processor 403 to estimate and/or predict the SOH of battery cells of a battery pack by estimating the capacity fade of the battery cells over a plurality of charging or discharging cycles.
The neural network processor 403 can accurately estimate and/or predict the capacity fade of the battery cells of the battery pack 406, irrespective of the type of chemistry of the battery cells (such as Lithium-Iron phosphate, Lithium-Nickel-Manganese-Cobalt oxide, Lithium-Nickel-Cobalt-Aluminum oxide, and so on).
In an embodiment, the neural network processor 403 can be a trained DNN model. The neural network processor 403 can be configured to estimate or predict the capacity fade of the battery cells based on AUC values and voltage of the battery cells. Consider that the battery pack comprises of six battery cells. The neural network processor 403 can be trained using battery cells of the battery pack 406 or other battery packs. In an example, consider that four cells of the battery pack 406 are used for training the neural network processor 403. The neural network processor 403 can be trained using AUC values obtained during a plurality of charging or discharging cycles of the four cells of the battery pack 406. The neural network processor 403 can predict the capacity fade of the four cells after a certain number of charging or discharging cycles.
The weights of the neural network processor 403 can be adjusted based on errors between the predicted values of the capacity fade and the actual values of capacity fade. The actual values of the capacity fade can be obtained experimentally using the AUC values computed during the charging cycles or discharging cycles of the four battery cells and the capacity values of the four battery cells. The AUC values and the capacity values relevant to the four battery cells can be obtained from plots pertaining to each of the four battery cells, which are indicative of voltage of a particular battery cell and the amount of charged stored or drained from the battery cell during the charging cycles or the discharging cycles (FIGS. 2 and 3).
Once the errors between the predicted values of capacity fade and the actual values of capacity fade are adjusted, the DNN model is built. The accuracy of the DNN model (neural network processor 403) is ascertained by validating the predicted values of capacity fade pertaining to the remaining two battery cells of the battery pack 406. The validation of the neural network processor 403 is performed by computing the AUC values during the charging cycles or discharging cycles of the two battery cells. The AUC values obtained during a charging cycle or discharging cycle can be provided as inputs to the trained neural network processor 403. The neural network processor 403 can, thereafter, predict the capacity fade of the two battery cells.
Consider that the neural network processor 403 has been trained and deployed in the electronic device or the electric vehicle. The neural network processor 403 can determine the capacity fade of the battery cells of the battery pack 406 based on the AUC values, obtained from the computing unit 402 during a charging/discharging cycle. The computing unit 402 can compute the voltage of the battery cell as the battery cell is charged or discharged. The computing unit 402 can compute the voltage of the battery cell when the battery cell is charged or discharged to certain predefined levels. In an embodiment, the computing unit can store the variation of the voltage of a battery cell with respect to the amount of charge that is stored in, or drained from, the battery cell, during a charging cycle or discharging cycle of the battery cell, for constructing plots indicating the variations. The BMS 400 can determine the voltage of the battery cell as charge is stored in the battery cell during a charging cycle. Similarly, the BMS 400 can determine the voltage of the battery cell as charge is drained from the battery cell during a discharging cycle.
The computing unit 402 can obtain the values of the voltage of the battery cell during the charging cycle or discharging cycle. The computing unit 402 can determine the variation (rise/depreciation) of the voltage of the battery cell with respect to the amount of charge that is stored or drained from the battery cell. This enables obtaining plots indicating the variation of the voltage of the battery cell with respect to the amount of charge that is stored in, or drained from, the battery cell, during the charging cycle or discharging cycle of the battery cell. Once the variations of the battery voltage with respect to the variations of amount of charge stored in, or drained from, the battery cell have been obtained, the computing unit 402 can compute a plurality AUC values during the charging cycle of the battery cell based on rise of the battery cell voltage within a plurality of ranges of charge stored in the battery cell, during the charging cycle. Similarly, the computing unit 402 can compute a plurality AUC values during the discharging cycle of the battery cell based on depreciation of the battery cell voltage within a plurality of ranges of charge drained from the battery cell, during the discharging cycle.
The computing unit 402 can send the plurality AUC values to the neural network processor 403. The neural network processor 403 can predict the capacity of the battery cell based on the received AUC values. The neural network processor 403 can predict the capacity fade of the battery cell based on previously predicted values of capacity of the battery cell, wherein the neural network processor 403 had predicted the values of capacity of the battery cell during the previous charging cycles or discharging cycles of the battery cell. Similarly, the neural network processor 403 can estimate and/or predict the capacity fade of other battery cells in the battery pack 406.
FIG. 4 shows exemplary units of the BMS 400, but it is to be understood that other embodiments are not limited thereon. In other embodiments, the BMS 400 may include less or more number of units. Further, the labels or names of the units of the BMS 400 are used only for illustrative purpose and does not limit the scope of the invention. One or more units can be combined together to perform same or substantially similar function in the BMS 400.
FIG. 5 is a flowchart 500 depicting a method for estimating and/or predicting SOH of a battery cell by estimating the capacity fade of the battery cell using a DNN model, according to embodiments as disclosed herein. At step 501, the method includes recording a variation of the voltage of the battery cell with respect to variation in amount of charge that is stored in, or drained from, the battery cell, during charging cycles or discharging cycles of the battery cell. The battery cell can be one of a plurality of battery cells in the battery pack 406. The embodiments include constructing plots that depict a variation of the voltage of a battery cell with respect to the variation of amount of charge that is stored in, or drained from, the battery cell, during the different charging cycles or discharging cycles of the battery cell.
The embodiments include determining the voltage of the battery cell when charge is stored in the battery cell during a charging cycle. The embodiments include determining the voltage of the battery cell when charge is drained from the battery cell during a discharging cycle. When the battery cell is charged, the voltage of the battery cell rises. The embodiments include determining the rate of rise of the voltage of the battery cell with respect to the amount of charge that is stored in the battery cell. When the battery cell is discharged, the embodiments include determining the depreciation of the voltage of the battery cell with respect to the amount of charge drained from the battery cell. The embodiments include storing the variation of the voltage of the battery cell with respect to the amount of charge that is stored in, or drained from, the battery cell, during the charging cycles or discharging cycles of the battery cell.
At step 502, the method includes determining the capacity fade of the battery cell based on AUC values computed from the constructed plots. The embodiments include computing the AUC values during the charging cycles or discharging cycles of the battery cell. The embodiments include computing a plurality AUC values during the charging cycle of the battery cell based on rise of the battery cell voltage within a plurality of ranges of charge stored in the battery cell, during the charging cycle. The embodiments include computing a plurality AUC values during the discharging cycle of the battery cell based on depreciation of the battery cell voltage within a plurality of ranges of charge drained from the battery cell, during the discharging cycle. The embodiments include computing a plurality AUC values during a plurality of charging cycles and a plurality of discharging cycles.
At step 503, the method includes predicting the capacity of the battery cell based on the AUC values computed during the charging cycles or discharging cycles of the battery cell. The embodiments include predicting the capacity fade of the battery cell based on previously predicted values of capacity of the battery cell. The embodiments include predicting the values of capacity of the battery cell during the previous charging cycles or discharging cycles of the battery cell. The embodiments include estimating and/or predicting the capacity fade of other battery cells in the battery pack 406.
The various actions in the flowchart 500 may be performed in the order presented, in a different order, or simultaneously. Further, in some embodiments, some actions listed in FIG. 5 may be omitted.
FIGS. 6a and 6b are graphs depicting comparisons between actual values of capacity fade (loss) and predicted values of capacity fade for two battery cells operating in different ambient temperatures, according to embodiments as disclosed herein. As depicted in FIG. 6a, the plot depicting the predicted values of capacity fade closely follows the plot depicting the actual values of capacity fade. The plots depict that the capacity fade of the cell increases with increase in the number of charging cycles or discharging cycles undergone by the battery cell. The battery cell is the cell 445 and is operating in an environment with an ambient temperature of 250 C. The capacity fade of the cell 445 is predicted to be close to 8% after the battery cell undergoes 1000 charging cycles or discharging cycles. As depicted in FIG. 6b, the plot depicting the predicted values of capacity fade of cell 1571 closely follows the plot depicting the actual values of capacity fade of cell 1571. The plots depict that the capacity fade of the cell 171 increases with increase in the number of charging cycles or discharging cycles undergone by the cell 1571. The cell 1571 is operating in an environment with an ambient temperature of 450 C. The capacity fade of the cell 1571 is predicted to be close to 10% after the battery cell undergoes 1000 charging cycles or discharging cycles.
The predicted capacity loss of the cell 445 (8%) is lower than the predicted capacity loss of the cell 1571 (10%) since the cell 445 is operating a lower temperature (by 200 C) than that of the cell 1571.
FIGS. 7a and 7b are graphs depicting absolute error involved in predicting the capacity fade, after modification, for the two battery cells operating in the different ambient temperatures, according to embodiments as disclosed herein. As depicted in FIGS. 7a and 7b, the absolute error involved in predicting the capacity fade is demonstrated in boxplots. The boxplots depict the distribution of absolute error in the predicted capacity fade in percentage, for the cell 445 and the cell 1571. The modification in the absolute error is performed by checking whether predicted class label for a specific set of feature values is closer to 1 or 0, wherein 1 corresponds to cell 1571 and 0 corresponding to cell 445. If the predicted class label is closer to 0, then the predicted capacity fade value can be modified by subtracting 1.5 (in percentage) from the absolute error value (in percentage). The modified capacity fade value is the predicted value of capacity fade. The absolute error in the predicted value of capacity fade for cell is less than 1.5 %. The predicted capacity fade for cell 445, after modification, is close to 1.5 %, which ensures that the DNN model can predict the values of capacity fade of the battery cells with a maximum absolute error of 1.5-2 %.
The embodiments disclosed herein can be implemented through at least one software program running on at least one hardware device and performing network management functions to control the network elements. The network elements shown in FIG. 4 include blocks which can be at least one of a hardware device, or a combination of hardware device and software module.
The embodiments disclosed herein describe methods and systems for estimating and/or predicting the State of Health (SOH) of a battery cell by estimating the capacity fade of the battery cell over a plurality of charging/discharging cycles. Therefore, it is understood that the scope of the protection is extended to such a program and in addition to a computer readable means having a message therein, such computer readable storage means contain program code means for implementation of one or more steps of the method, when the program runs on a server or mobile device or any suitable programmable device. The method is implemented in a preferred embodiment through or together with a software program written in example Very high-speed integrated circuit Hardware Description Language (VHDL), or any other programming language, or implemented by one or more VHDL or several software modules being executed on at least one hardware device. The hardware device can be any kind of portable device that can be programmed. The device may also include means, which could be, for example, a hardware means, for example, an Application-specific Integrated Circuit (ASIC), or a combination of hardware and software means, for example, an ASIC and a Field Programmable Gate Array (FPGA), or at least one microprocessor and at least one memory with software modules located therein. The method embodiments described herein could be implemented partly in hardware and partly in software. Alternatively, the invention may be implemented on different hardware devices, e.g. using a plurality of Central Processing Units (CPUs).
The foregoing description of the specific embodiments will so fully reveal the general nature of the embodiments herein that others can, by applying current knowledge, readily modify and/or adapt for various applications such specific embodiments without departing from the generic concept, and, therefore, such adaptations and modifications should and are intended to be comprehended within the meaning and range of equivalents of the disclosed embodiments. It is to be understood that the phraseology or terminology employed herein is for the purpose of description and not of limitation. Therefore, while the embodiments herein have been described in terms of preferred embodiments, those skilled in the art will recognize that the embodiments herein can be practiced with modification within the scope of the embodiments as described herein.