Convolutional Neural Networks : Encoding Battery Information
- Tech Stack: Pytorch, Pandas, Matplotlib, Scipy
- Members: Me, Dr. Carlos Rufino, Dr. Hudson Zanin,
Known classification mechanisms include convolutional networks, but since we are dealing with time series, it is recommended to find image representations for each of the features.
A signal transformation technique, yields images that can be convoluted to classify the labels of each state of charge.
Classification of the State-of-Health becomes necessary in a different domain than just time, in order to obtain a feature map that can be trained to identify in which cases the batteries will reach their second life state of use. This makes the convolutional classifier a powerful tool in identifying defective batteries.