Battery Cortex

Battery Cortex: On-road predictive battery analytics for Electric Vehicles, is a project with the aim to provide manufacturers and fleet operators of electrical vehicle with insights into quality, faults and anomalies of the battery packs.


Battery Cortex is an anomaly and fault detection platform for automated analysis on electrical vehicles (EV) battery packs data to identify battery condition and battery performance. This is done through predictive fault detection and early discovery of issues affecting lifetime of battery packs & cells, as well as understanding the conditions affecting capacity and charging/discharging behaviour under different usages and environments. These new insights are invaluable for vehicle manufacturers (OEMs), fleet operators & EV product/service providers.


The battery pack is at the very core of EVs and its performance determines lifespan and cost of ownership. We already now see warranty disputes increase even though the first EVs will be under warranty for years to come. Automated battery condition and performance analysis will be critical for OEMs to mitigate the risks of this immature technology and control warranty costs, while for the operators it will ensure uptime and enable and fair warranty claims when necessary.

The project aims to provide EV OEMs and fleet operators with insights into quality, faults and anomalies of the battery packs, a critical component in EVs. It enables faster R&D iterations (vehicle improvements) and a continuous enhancement of the offer available to end-users, providing better insights to their EV assets. As incentives for automotive electrification decrease it will become more and more difficult for OEMs and Fleet operators to be profitable and competitive. It will require products to mature and battery fault rates rapidly decrease as well as higher vehicle utilisation. The output of this project is an anomaly and fault detection product (software platform) that runs automated analysis on the data generated by EV battery packs.

The core component of this platform, to be developed at Halmstad University, are the machine learning and AI models. They are fed by battery signals and parameters such as the specifications, pack voltage and current, the cells voltage and current, its temperature, the state of charge, charge/discharge power, etc. Based on all this data, they output an assessment of the condition and performance of the batteries. Todays’ algorithms and rules, which already proven useful in the limited scope, will not suffice when both vehicle and battery are analysed together. We believe the next breakthrough AI will disrupt this space, allowing the automation of complex analysis, enabling less warranty costs and higher operational efficiency as less vehicles fail on the road.

Battery manufacturers today make assumptions with regards to usage conditions and the ambient environment in order to design and dimension their products. The manufacturers identify what they believe are the representative use cases which they in return use as product design inputs. However, as for all assumptions, they do not cover all corner cases and batteries used under unexpected circumstances will have an unexpected life time. The BMS in the battery packs has the capacity to stream sensor data and diagnostic trouble codes. However, this data alone is not enough in the automotive setting. The battery is part of a larger system (a vehicle) and components interact and affect each other. Combining the broader vehicle data with battery data will create a better understanding of the expected battery lifetime as well as predicting upcoming problems. This will however require new ways to analyse the data as the type, volume and dimensionality of data increase. And organisations cannot grow the number of analysts / engineers in direct proportion to the volume of the data.

Project period:

  • November 1, 2020, to October 31, 2021


  • EuroStars (EU) project

Involved partners:

  • Stratio Automotive, Portugal
  • Caetanobus, Portugal

Project team at Halmstad University:

The project belongs to Technology Area Aware Intelligent Systems (AIS) at the Department of Intelligent Systems and Digital Design (ISDD) at the School of Information Technology (ITE). All resarch project at ITE are within the research environment Embedded and Intelligent Systems (EIS).