AI for Energy Storage

We will discuss how new computational approaches enabled by high-performance computing and machine learning algorithms are accelerating the traditional materials design and commercialization process for battery materials.

In this work, we describe the evolution of AI/ML tools in Li-Ion life prediction and show how they naturally culminate in the need for a unified foundation model (FM) that supports diverse applications across the ecosystem. We present our first steps on the path towards a true FM and offer a definitive demonstration that our proto-FM, implemented as a Hybrid Transformer architecture and trained on dramatically different datasets, outperforms the present paradigm of the single dataset model.

Artificial intelligence is emerging as a key enabler of battery digital twins and next-generation battery management systems. This presentation highlights recent advances in AI-assisted battery diagnostics and prognostics, including non-invasive core temperature estimation, physics-guided degradation prediction, and AI-based fault detection. By combining electrochemical sensing, thermal analysis, and machine learning, these approaches provide improved visibility into battery health, safety, and remaining life, enabling safer, more reliable, and longer-lasting energy storage systems for transportation and grid applications.?

Lithium-ion battery thermal runaway poses significant safety risks in electric vehicles, energy storage systems, and consumer electronics. This presentation introduces an acoustic-based early warning framework for detecting thermal runaway initiation using deep learning. The approach identifies precursor signatures that emerge prior to catastrophic failure, enabling earlier detection and intervention compared with conventional monitoring methods. Results demonstrate robust detection performance across multiple battery conditions, highlighting the potential of acoustic sensing as a non-invasive and scalable solution for future battery safety monitoring and hazard mitigation.

While deep learning achieves state-of-the-art battery state-of-health estimation, the emerging EU Digital Battery Passport introduces a critical constraint: model auditability. Black-box predictions, however accurate, cannot satisfy the transparency and traceability requirements that regulators, OEMs, and recyclers will demand. This presentation proposes a counterfactual explainability framework that maps model outputs onto interpretable, condition-level degradation drivers—identifying precisely which operating scenarios accelerate or preserve battery health. Structured for DBP compliance, these explanations deliver actionable decision support across warranty assessment, second-life qualification, and end-of-life routing, closing the gap between predictive performance and regulatory-grade transparency.

Understanding lithium-ion battery aging under real-life driving conditions is essential for EV development. Using WLTC-based cycling, this work reveals that 100% and 45% DoDs accelerate cathode degradation and interfacial degradation, respectively, while 75% DoD balances both and achieves the longest cycle life (>2200 EFCs). The results provide insights for optimizing BMS cycling protocols and extending battery lifetime.

Manufacturing anomalies in battery cells can compromise safety and longevity of energy storage systems. Ultrasound-based quality inspection enables rapid, cost-effective, and reliable in-line inspection of battery cells. AI-based feature extraction from ultrasonic time-series data, combined with clustering algorithms, allows for automated anomaly identification, localization, and classification, reducing production waste and enhancing battery safety.

Rapid diagnostic tests use short (10s-100s) alternating- or direct-current signals to probe a battery, from which battery capacity, state-of-charge, or safety may be estimated. Many methods have been proposed, but a holistic evaluation of these techniques has not yet been presented. Here, we will present a study on the effectiveness of using rapid diagnostics for state-of-charge and temperature agnostic battery state estimation, using both previously published and newly collected data.

This presentation discusses leveraging machine learning and robotic experimentation to accelerate innovation in battery materials. It explores how these advanced techniques streamline material discovery, optimize properties, and expedite the development of next-generation battery technologies.

Battery packaging design has to consider space-claim limitation for large structures like energy storage or machines other than automobile industry. AI-enabled processes would greatly save design time and also take into account the structure integration assessment and meet thermal and electrical requirements.

The transition from laboratory innovation to high-volume manufacturing of solid-state batteries demands intelligent systems that operate across the complexity of the production environment. This presentation describes a multimodal AI approach integrating vision-based inspection, electrical characterization, and process telemetry into a unified view of quality and process health. Underpinning it are machine learning systems built for production reliability: robust data pipelines that absorb manufacturing variability, continuous monitoring, and feedback mechanisms that let models improve as data accumulates. We discuss how this architecture improves inspection, process optimization, and electrical performance, supporting a faster, more reliable path to high-volume solid-state manufacturing.

Lithium-ion (LIB) battery degradation is often characterized at three distinct levels: mechanisms, modes, and metrics. ML can provide a unique multi-level perspective on characterizing LIB degradation and it can improve accuracy especially when combined with perturbation techniques. This pulse-injection aided machine learning (PIAML) technique can be used for battery diagnostic and prognostics with high fidelity. LIB management systems can leverage this estimation framework to extend lifetime and reduce costs.