To address bottlenecks in the battery development process and to reduce time in battery R&D, researchers at Toyota Research Institute developed a self-optimizing software system, BEEP, which stands for Battery Evaluation and Early-prediction. The system has the ability to make early outcome predictions in order to reduce testing time and conduct closed loop optimization to minimize the total number of experiments. This talk focuses on the BEEP battery analytics platform and how automation and machine learning are combined to accelerate battery development by making battery testing faster and less expensive. 

Over the last decade, pharmaceutical co-innovation partnerships have gotten new life-saving products to market faster. Given the importance decreasing time-to-market for battery innovations, we discuss how the playbook for accelerated drug discovery is being adapted to the battery industry. We compare and contrast these two industries from both business and technical perspectives, discuss how co-innovation partnerships are being successfully structured for battery R&D, and showcase several case studies from Aionics partners.

Here, we give an overview of recent methodological advancements of ML techniques for atomic-scale modeling and materials design. We review applications to materials for solid-state batteries, including electrodes, solid electrolytes, coatings, and the complex interfaces involved.

A hurdle to achieving profitability for new battery manufacturing plants is ramping battery production to meet rising market demand and competition. EBI accelerates production yield ramp and quality optimization by automating analysis, pinpointing root causes of defects, and generating insight by combining performance, traceability, and process & equipment data. We also discuss how EBI enables scaling of these analyses to handle the large volumes of data generated from battery production.

Solid-state electrolytes are a promising technology to enable safe lithium metal batteries, but their performance and safety is hindered by void formation at the metal/electrolyte interface, which leads to dendrites. In this work, we develop a coupled electrochemical-morphogenic model of void formation in a symmetric cell. The model can be used to inform the design of interfacial layers that increase the critical current density of the cell.

This talk will review past and ongoing research studies on battery capacity forecasting and early life prediction and discuss the long-term testing and methodology development efforts led by a team of researchers at the University of Connecticut and Iowa State University.

By collecting battery data from the field and building up the battery digital twin in the cloud, the degradation and safety of batteries can be monitored online and the information regarding the degradation modes can be extracted from the data. Physics-based models are gaining more and more success in describing cell behavior and early-stage capacity fade, while the emergence of machine learning models further generates rapid predictions of future health based on indicators learned purely from data. Blending the physics-based model and machine learning is challenging. Here, we will present our work in last years to show the methodology and benefits of integrating physics-based models and machine learning based data-driven methods for battery management.

With AVL Battery Life Cycle Management, a holistic approach of optimizing the battery lifecycle from material mining, battery design, development & testing, through the whole in-use-phase to second-life utilization and recycling is addressed. It will be shown how intelligent modelling a cloud based environment, based on federated learning enables highly accurate prediction of battery degradation for single vehicles and complete fleets. This allows to improve cell and battery design and also in time maintenance scheduling and second-life usage.

Industrial AI, Big Data Analytics, Machine Learning, and Cyber-Physical Systems are changing the way we design product, manufacturing, and service systems. It is clear that as more sensors and smart analytics software are integrated in the EV systems, predictive technologies can further learn and autonomously optimize mobility and performance. This presentation will address the trends of Industrial AI for smart battery realization. First, Industrial AI systematic approach will be introduced. Case studies on advanced predictive analytics technologies for smart battery health management and mobility optimization will be discussed. In addition, issues battery data quality in future predictive mobility will be discussed.

A challenge to bring novel battery materials to market is the incredible complexity of materials design and selection required for a high-performance battery. Materials Informatics (MI) is a flourishing field which utilizes advanced data management and machine learning to enhance materials research. Battery development is a compelling use-case for MI due to the high-dimensionality of materials selection and the often outsized effect that these materials have on downstream cycle life performance. This talk will outline some of these use-cases of MI in battery materials development and demonstrate in particular how electrolyte formulation can be accelerated to yield novel materials combinations with high predicted performance.

This talk will explore how multi-scale models can capture coupled thermal-electrochemical-mechanical battery degradation processes across scales. This includes exploring the interplay between particle cracking, lithium-plating, and solid-electrolyte interphase growth, highlighting the heterogeneous behaviour across scales, from single particles to entire battery packs, path dependency, and positive/negative feedback effects.