BACKGROUND
[0001] Electric vehicles (EVs) rely on rechargeable batteries, the health
of which is integral to the performance and longevity of the EVs. The stateof-health (SoH) of a battery is a metric that reflects current state of the
5 battery relative to an ideal state. Accurate assessment of the SoH is of
paramount interest in the context of EVs, as the SoH directly influences an
EV’s operational range, safety, and longevity.

We claim:
1. A system for determining an actual state-of-health (SoH) of a battery of
an electric vehicle, the system comprising:
a processor;
5 a determination engine coupled to the processor, to:
for an electric vehicle in operation, obtain a data set
corresponding to a set of health indicators of the battery, wherein a
health indicator is a parameter indicative of the SoH of the battery of
the electric vehicle;
10 provide the obtained data set for interpretation by a first
machine learning model and a second machine learning model, to
respectively provide first predicted SoH value and second predicted
SoH value of the battery; and
compute a mean value of the first predicted SoH value and
15 the second predicted SoH value of the battery to determine the actual
SoH of the battery.
2. The system as claimed in claim 1, wherein the set of health indicators of
the battery corresponds to at least one of a temperature, operational
20 voltage, operational current of the battery of the electric vehicle in operation,
or a combination thereof.
3. The system as claimed in claim 1, wherein the set of health indicators of
the battery are obtained from at least a plurality of sensors associated with
25 the battery of the electric vehicle.
4. The system as claimed in claim 1, wherein the system further comprises
a training engine, coupled to the processor, to train the first machine
learning model and the second machine learning model, wherein the
30 training engine is to:
34
obtain simulation data representing SoH of the battery of the vehicle;
train the first machine learning model and the second machine
learning model based on the obtained simulation data;
obtain a first SoH indicator and a second SoH indicator value from
5 the first machine learning model and the second machine learning model
respectively, based on the data set corresponding to the set of health
indicators of the battery of the electric vehicle;
re-train the first machine learning model based on the simulation data
and the second SoH indicator to obtain a first matured machine learning
10 model; and
concurrently re-train the second machine learning model based on
the simulation data and the first SoH indicator to obtain a second matured
machine learning model.
15 5. The system as claimed in claim 4, wherein the training engine is to further:
determine a performance score of the first machine learning model
and the second machine learning model; and
when one of the performance score of the first machine learning
model and the second machine learning model is less than a predefined
20 threshold score, continue to re-train the first machine learning model and
the second machine learning model.
6. The system as claimed in claim 4, wherein the simulation data comprises
at least one of number of charging events of the battery, standard deviation
25 in voltage, skewness in voltage, C-rate of the battery, Battery pack effective
temperature, charging time, or a combination thereof.
7. The system as claimed in claim 1, wherein each of the first machine
learning model and the second machine learning model is a regression
30 model.
35
8. A method comprising:
obtaining simulation data for a battery of an electric vehicle, wherein
the simulation data represents state of health (SoH) of the battery;
5 training a first machine learning model and a second machine
learning model based on the obtained simulation data;
obtaining a first SoH indicator and a second SoH indicator of the
battery, from the first machine learning model and the second machine
learning model, respectively, based on a data set corresponding to a set of
10 health indicators of the battery of the electric vehicle in operation;
subsequently re-training the first machine learning model based on
the simulation data and the second SoH indicator to obtain a first matured
machine learning model; and
concurrently re-training the second machine learning model based
15 on the simulation data and the first SoH indicator to obtain a second matured
machine learning model.
9. The method as claimed in claim 8, wherein the set of health indicators of
the battery of the electric vehicle corresponds to at least one of a
20 temperature, operational voltage, operational current of the battery of the
electric vehicle in operation, or a combination thereof.
10. The method as claimed in claim 8, wherein the set of health indicators
of the battery are obtained from at least a plurality of sensors associated
25 with the battery of the electric vehicle.
11. The method as claimed in claim 8, wherein the simulation data
comprises at least one of number of charging events of the battery, standard
deviation in voltage, skewness in voltage, C-rate of the battery, Battery pack
30 effective temperature, charging time, or a combination thereof.
36
12. The method as claimed in claim 8, further comprising:
determining a performance score of the first machine learning model
and the second machine learning model; and
5 when one of the performance score of the first machine learning
model and the second machine learning model is less than a predefined
threshold score, continue to re-train the first machine learning model and
the second machine learning model.
10 13. The method as claimed in claim 8, wherein each of the first machine
learning model and the second machine learning model is a regression
model.
14. An electric vehicle comprising:
15 a battery for powering the electric vehicle;
a determination engine coupled to a processor, wherein the
determination engine is to:
for the electric vehicle in operation, obtain a data set
corresponding to a set of health indicators of the battery, wherein a
20 health indicator is a parameter indicative of the SoH of the battery of
the electric vehicle;
provide the obtained data set for interpretation by a first
machine learning model and a second machine learning model, to
respectively obtain a first predicted SoH value and a second
25 predicted SoH value of the battery; and
compute a mean value of the first predicted SoH value and
the second predicted SoH value of the battery to determine an actual
SoH of the battery.
37
15. The electric vehicle as claimed in claim 14, further comprises a training
engine coupled to the processor, to train the first machine learning model
and the second machine learning model, wherein the training engine is to:
obtain simulation data representing the SoH of the battery of the
5 vehicle;
train the first machine learning model and the second machine
learning model based on the obtained simulation data;
obtain a first SoH indicator and a second SoH indicator from the first
machine learning model and the second machine learning model,
10 respectively, based on the data set corresponding to the set of health
indicators of the battery of the electric vehicle;
re-train the first machine learning model based on the simulation data
and the second SoH indicator value to obtain a first matured machine
learning model; and
15 concurrently re-train the second machine learning model based on
the simulation data and the first SoH indicator value to obtain a second
matured machine learning model.