Development of Digital Twin Platform for Electric Vehicle Battery System

Abstract

The battery system in electric vehicles needs proper monitoring and control to ensure reliable, efficient, and safe operation. Recent advancement in cyber-physical technology has brought the emerging digital twin concept. This concept opens a new possibility of real-time condition monitoring and fault diagnosis of the battery system. Although it sounds promising, the concept implementation still faces many challenges. One of the challenges is the availability of a platform to develop digital twins, which involves data pipelines and modeling tools. The data pipeline will include the acquisition, storing, and extract-transform-load (ETL) with high velocity, volume, value, variety, and veracity data, known as big data. The modeling tools must provide applications to build the high-fidelity model, one of the required elements of the digital twin. Based on those urgencies, this paper proposes a platform that facilitates a digital twinning of the battery system in an electric vehicle. The platform is built on the open-source framework CDAP, equipped with a data pipeline and modeling tools. It has run several performance tests with different computation resource configurations and workloads. Doubling the processing power can reduce 12% of computation time while increasing memory size by four times only reduces 10% of computation time. The result shows that the processing power affects the performance digital twin platform more than the memory size.