THE EFFECTS OF MODIFIED CHARGE ON LITHIUM â ION BATTERY FOR HEALTH AND LIFE PREDICTION PURPOSES OF BATTERY USING MACHINE LEARNING
Lithium batteries are well known that have been used in recent years and today there is a transition period where fossil energy has been taken over gradually by renewable energy. In this condition, batteries play an important role in energy storage. Lithium-Ion is one of the batteries that is ver...
Saved in:
Main Author: | |
---|---|
Format: | Theses |
Language: | Indonesia |
Online Access: | https://digilib.itb.ac.id/gdl/view/36591 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Institution: | Institut Teknologi Bandung |
Language: | Indonesia |
Summary: | Lithium batteries are well known that have been used in recent years and today
there is a transition period where fossil energy has been taken over gradually by
renewable energy. In this condition, batteries play an important role in energy
storage. Lithium-Ion is one of the batteries that is very suitable for storing electrical
energy. But lithium batteries are quite expensive so a lot of research aims to
examine the performance degradation of the lithium-ion battery. Capacity
degradation as a result of the modified charging and its influence on internal
resistance was the object of this research. The research used commercial batteries.
The obtained result showed that the modified charging method was proven to be
faster in increasing battery resistance than the prior CC-CV charging method.
In this research machine learning such as Support Vector Machine (SVM), linear
regression, and Seasonal ARIMA (SARIMA) had been used to predict battery's life.
Linear regression in predicting for 20 cycles ahead, gave MSE 624.01, RMSE
24.98, MAPE 0.77%, and coefficient of determination (R2)of the predictive model
was 34%. Meanwhile, Support Vector Machine for 20 cycles gave MSE 612.37,
RMSE 24.75, MAPE 0.74% and coefficient of determination (R2) of the predictive
model was 35%. By applying the SARIMA prediction method for 20 cycles gave
MSE 202.37, RMSE 14.23, MAPE 0.43%, and the coefficient of determination (R2)
of the predictive model was 76%.
Modified charge method could increase battery life when observed from its
capacity. The method named as multi-current step III could make the lithium-ion
battery has a trend of capacity degradation slower than CC-CV charge and multicurrent
step II. The battery life could be extended when used modified charge
method such as multi-current step III. In this research multi-current step III could
prolong battery life up to 29 cycles with total cycles were 435 cycles. It was higher
than the battery that uses the CC-CV charging method which had total cycles of
406 cycles and battery with multi-current step II charging method which had total
cycles of 325 cycles. The longest life was 435 cycles because this research applied
a fast charging method with 100% deep of discharge. |
---|