Data-Driven Battery Analysis with Machine Learning: From Laboratory to Field Application

Machine learning has emerged as a pivotal force within the battery industry, spanning research endeavors from the material level to the system level, encompassing both production processes and practical applications. Recent advancements in machine learning and data-driven methodologies have enabled innovative solutions for complex problems, particularly those scenarios where conventional physics-based models have fallen short.<br/><br/>In my talk, I will delve into our latest endeavors in the realm of data-driven battery analysis, encompassing both laboratory experimentation and real-world application in electric vehicles and stationary energy storage systems. Our research is centered on the development of deep learning models designed to augment and denoise CT images for in-depth battery electrode analysis. Additionally, we are actively engaged in harnessing machine learning and statistical learning techniques to scrutinize data from various testing and field sources, thereby improving aging diagnostics in the cloud.<br/><br/>This talk aims to offer a comprehensive overview of the diverse array of machine learning approaches employed in battery-related analyses while also shedding light on the unique challenges faced and the abundant prospects awaiting exploration.