Online Battery Data Analytics Pipeline using Bigdata Tools for Electric Vehicles

This paper proposes an online big data analytics pipeline for machine learning (ML)-based battery health monitoring in electric vehicles (EVs). The proposed battery data analytics pipeline systematically integrates open-source big data tools including: 1) Apache Kafka (confluent) and Zookeeper for online battery data ingestion from multiple data sources, 2) KSQL database for streaming and sending multiple rows of data to the ML feature engineering stage, 3) Jupyter notebook training/running ML algorithms, and 4) Docker container to run all the pipeline software images. The proposed nearly real-time ML pipeline platform will enable a new level of battery intelligence system used for battery product development and lifetime battery health monitoring using the open-source bigdata tools. The feasibility of the proposed pipeline for an EV battery cell is validated by experiments.