Estimating Vehicle Vibration Parameters With Machine Learning

The challenge

A leading vehicle manufacturer was developing an automated vehicle maintenance process. It depended on estimating vehicle vibration parameters accurately, which is difficult because vehicle dynamics are complex and vary widely between vehicle types.

What we did

We advised on the experimental design: which sensors to use, where to place them, and which test conditions would produce representative data across vehicle types and driving scenarios. We then built a Python pipeline that aggregates and processes more than 20 GB of test data and prepares it for machine learning, and trained, tested and refined a series of models. See our advanced software algorithms service.

The result

The models estimated the vibration parameters to the target accuracy for a subset of vehicle types. We also delivered a roadmap for extending the approach across the full range of vehicles.

Work With Us