This talk describes efforts to organize and synthesize results during our extensive testing of electrolyte materials designed to improve the safety and performance of next-gen batteries. I will discuss how an intelligent systems approach allows us to develop better materials more efficiently. This approach allows for improved cell safety coming from the electrolyte level, so they themselves are indeed ‘intelligent battery materials’.

At Forge Nano, the use of ALD to become an industry leader in materials optimization in the battery space is predicated on our ability to optimize coating solutions through an iterative process between coating the materials and electrochemical testing. In this work we have employed Voltaiq analysis software as a tool to efficiently explore ALD coatings as a means of stabilizing Ni-rich NMC surfaces, enabling increased capacity retention and high voltage utilization.

Sepion Technologies is developing advanced membranes to overcome barriers in the path to wide-spread electrochemical energy use and storage. Known fade mechanisms in Li-ion batteries associated with transitional metal crossover currently limit cycle life, thereby, reducing the impact of these technologies in electric vehicles. Our proprietary membrane technology works in concert with current separator technology to effectively block transition metal crossover resulting in increased energy density and cycle life.

The cell formation and grading section of the cell manufacturing line is the largest, most expensive, and most costly-to-operate portion of the manufacturing line. Today, some manufacturing lines are relying on 20 year old technology because it is safe and reliable. But advances in factory automation, cell forming electronics, measurements, data collection, and control systems offer the promise of improved productivity and efficiency, better flexibility and agility, and increased profitability. With the increased availability of process data and big data machine learning, it is further possible to feedback real-time insight and adapt the process on the fly to give the highest quality cell output.

Batteries are not a one-fit-all solution. CUSTOMCELLS develops tailor-made and optimized battery configurations that can meet very specific requirements such as high energy density and C-rates as well as installation space or temperature requirements. In order to offer the highest quality and thus a long cycle life and safety, we strive for high transparency and corresponding traceability during the development and production process.

This work illustrates turning standard cells into intelligent cells, through the integration of in situ sensors and wireless communication systems during manufacturing, thus enabling significant advancements in performance mapping and cell monitoring technology. Furthermore, the technology demonstrated can be transferred to many cell chemistry and form factors.

Applied AI technology is catapulting the automotive battery pack industry forward to meet customer demands in range extension, fast charging capabilities, and worldwide electrification. At Electra, we work to improve current and future battery pack systems by leveraging widely available automotive data to inform machine learning control models. Our flagship product, EVE-Ai 360 Adaptive Controls, consists of 6 modules that each optimize battery pack performance in consumer demanded areas using our ML algorithms, and the EVE-Ai Analytics platform supports the controls by managing fleet data. We carry our expertise in energy storage into the realm of battery pack design with EnPower Design Suite and MATLAB Application. Join us to learn more about the future of intelligent battery pack systems!

An electrified vehicle is potentially more susceptible to data breach and cyber attacks than one with a traditional propulsion system due to additional attack surfaces. These attack surfaces include smartphone apps that monitor and control charging, V2V and V2X interfaces, payment apps and possible vulnerabilities in charging infrastructure. The increased severity of resulting incidents could cause significant damage to the battery system and the vehicle itself, the occupants and bystanders, as well as the grid. Monitoring the health of the battery and systems pose an opportunity to mitigate dangerous events from happening and/or alerting the driver to seek maintenance, repair or replacement. How can we keep valuable data secure and critical systems safe from hackers, but available for diagnosis, maintenance and advanced warnings of trouble?

The increased amount of battery testing data and growth of machine learning based tools has made it easier to apply such tools to model battery cells and to perform battery state estimation. The first part of the presentation will show methods into modelling battery cells using machine learning approached based on feed forward neural network (FNN) and recurrent neural networks such as Gated Recurrent Unit (GRU) and Long-term Short-term Memory (LTSM). Practical details such as design, collection and processing of the data for effective training/testing of neural networks shall be discussed. The second part of the presentation shall focus on neural network based approaches for battery state-of-charge (SOC) estimation. Pragmatic approaches designed to train for robustness based on augmented data generation, drive cycle profile design, and repeated random seeding shall be discussed.

We demonstrated the use of the powerful machine learning tool to develop the “safety envelope” of lithium-ion batteries for electric vehicles that provides the range of mechanical loading conditions ensuring safe operation. The daunting challenge of obtaining a large databank of battery tests was overcome by utilizing a high-accuracy finite element model of a pouch cell to generate over 2500 numerical simulations. The safety envelope will serve as important guidelines to the design of EV and batteries.

The field of battery development and manufacturing is full of opportunities for the application of machine learning. Machine learning techniques have accelerated materials discovery at the fundamental atomic scale, and have also impacted the commercial and manufacturing scale, accelerating failure predictions. In this talk, we will be covering some case studies of impactful machine learning applications in the battery field, spanning these time, length, and cost dimensions.

Discovering promising new materials is central to our ability to design better batteries, but research progress can be limited by an incomplete understanding of structure/property relationships, slow testing cycles, and overwhelmingly large numbers of candidate materials. New machine learning (ML) approaches offer a route to accelerated materials discovery by training predictive models on existing experimental data and then using these models to screen databases of candidate materials. Our research at Stanford University suggests that careful ML modeling can provide a significant acceleration in the rate of new materials discovery, even when trained on small amounts of data. In this talk, we present our research in using ML to accelerate electrode and electrolyte discovery, discuss best practices for the application of ML to materials design, and highlight the Aionics materials design software platform.

Battery lifetime testing is a major bottleneck in battery development due to both the number and the duration of required experiments. In this talk, I present work from my time at Stanford on both early prediction, which reduces the time per experiment by predicting the final cycle life using data from early cycles, and Bayesian optimization, which reduces the number of experiments by balancing exploration and exploitation to efficiently learn the parameter space.

The key to optimally manage batteries is to provide insights early on and with increased accuracy over time. TWAICE deploys a hybrid approach of empirical-analytical models which are combined with machine learning approaches to enable accurate simulation and prediction of battery conditions from day one. Furthermore it enables the accurate tracking of the battery health along the entire lifetime and prediction of the remaining useful lifetime depending on the usage at any time as well as identifying failure causes.