Software / AI

BMS Diagnostics and Prognostics BMS diagnostics now leverage multiple paths towards improved fidelity, including neural networks AI) deployed onto edge devices and cloud analytics integration. Purely on device methods: Claims of real-time capable algorithms to detect li-plating, torn cell tabs, and predict risk of thermal runaway in BMS have been made, although no scaled deployment of these algorithms in large fleets have been claimed. Linked cloud/edge device methods: It is clear that automotive OEMs are deploying OTA updates to diagnostic algorithms in response to field events, leveraging field data. Fully closed loop processes, between embedded algorithms, to cloud based batch processing capabilities and full fleet context have been proposed, in order to build closed-loop performance on the basis of fleet prognostics.

Linked Cloud And Embedded Solutions

On Device AI Deployment Edge AI Accelerator Leverages powerful Infineon Aurix TC4 Processor, to achieve compute levels necessary for onboard ML features

Edge and Cloud Compute to increase accuracy

OTA Update Strategies

Software Updates Over Replacements Safety recalls are increasingly handled by diagnostic software updates rather than full battery replacements. Example: The Porsche Taycan recall

Linked ecosystem approach to diagnostics, prognostics, SOH, and RUL Adaptive BMS, based on prognostics foundation Leverages online and offline compute capacity for batch analytics and increased accuracy, reduced false positives and negatives Deploys through OTA updates to calibration

Source: Accelerating eMobility with Automotive Grade AI for Batteries - EE Times, Elysia: Better Software Can Unlock Hidden Potential In EV Batteries, https://www.linkedin.com/pulse/batteries-lets-get-know-each-other-better-elysiabi/, AABC US2025 talks: Uwe Weidemann Munich Electrification), Tom Maull Elysia Battery Intelligence)

Extracted from the PDF; charts and images may read imperfectly.