Digital twin, machine learning and fleet management

Innovations are also happening in software implementations within the BMS. Acquired pack- and cell-measurement accuracies are the basis for more advanced state-of-X algorithms than a Kalman filter or Coulomb counting.

The ability to monitor individual driving behavior, traffic situations, and geographical and road conditions enables more precise vehicle range predictions and battery state-of-health data and state-of-charge estimations. If centralizing data in the cloud, machine learning algorithms can monitor a whole fleet of vehicles and enable predictive service. For example, if a failure pattern had been observed and stored before, the algorithms can detect early indications and calculate the likelihood of future failures of other vehicles to ask for garage service proactively. This function, known as creating a digital twin, enables further commercial models such as temporary vehicle range upgrades in a software-defined vehicle.

TI works with Electra, which makes artificial intelligence-powered battery-pack solutions, to make the BMS smarter and more connected by bringing EV batteries online. Electra’s EVE-Ai 360 fleet analytics software is a battery analytics tool that harnesses vehicle-specific and fleetwide battery-pack data to generate battery state-of-health trends and predictive models. It uses data from the battery, the vehicle and the environment, along with machine learning, to identify potential battery issues and failures before they occur, optimizing fleet efficiency and performance.

TI’s AM263P4-Q1 Arm-based, AutoSAR-enabled MCU includes a library to use an adaptive cell modeling system and enables machine-learning services to improve fleet and vehicle state-of-X measurements, helping enable smarter charging and optimizing battery health as well as range.