AI-based end to end drivetrain monitoring and optimization

Digitalized automotive durability testing from virtual sensors to mileage accumulation

AI-based end to end drivetrain monitoring and optimization | COMPREDICT


In a very traditional approach of reliability engineering, the weakest part is designed to cover the maximum distance for the most demanding driver. This results in overdesign, and increased CO2 emissions.

One possible solution consists of digitalized automotive durability testing from end-to-end, with the possibility to connect durability tests on prototypes or pre-series cars with in-use data of series cars and advanced development of the next vehicle generations.

Download the full presentation here

Step 1 – training of virtual sensors

Loading conditions of various components in a car are measured over limited distance and time with laboratory-type hardware sensors. In parallel, virtual sensors using only standard on-board signals are trained to reflect the hardware measurements reliably.

Step 2 – virtual fleet generation and mileage accumulation

A virtual fleet of vehicles is generated in our machine learning hub AI-CORE. The whole fleet is operated as a simulation in the cloud in order to produce representative statistical fatigue data through virtual mileage accumulation.

Step 3 – durability dashboards

As a result, fatigue damage curves as well as predicted reliability information can be generated for each component.

Furthermore, representative multi-dimensional test cycles are generated as input for future development cycles.
The paper shows the results of this method on a virtual fleet of 500 Tesla model 3 cars, in terms of potential weight reduction, cost reduction and CO2 footprint saving in production.

Get in touch with us to discuss about your application !


+49 (0)6151 3844614


Rheinstra├če 40-42

64283 Darmstadt, Germany