Detecting defects on a vehicle front-wheel axle unit while driving

Customer: Confidential
Country: France
Application: Mobility / Land transport

The Issue

A defect in the area of the front-wheel axle unit influences the behaviour of the vehicle and its reliability. The objective was for Metravib to propose and evaluate indicators in order to detect the defects of the front-wheel axle unit during driving on a production vehicle.

Five types of defects have been studied. Monitoring the degradation of these sensitive elements with sensors will allow the customer to reduce costs in the end (manufacturing, maintenance) and ensure good reliability of the produced vehicles.

Several configurations have been analysed by Metravib teams following a methodology in several steps:

  • Comparative analysis of the signals to propose a set of indicators
  • Systematic analysis to confirm the previous results and prepare the data for the classification work
  • Research of the best indicators & the training of classification models

The Challenge

The first challenge was to use and process at best the existing data recorded from sensors already onboard the vehicle.

The second challenge was to start from a blank sheet of paper in order to find the right indicators and detect defects without knowing if the direction taken was the right one.

The Solution

Analyses were used to propose indicators correlated with most defects. These indicators were integrated into a Machine Learning model to detect anomalies in the front-wheel axle unit of vehicles. Machine learning algorithms were used to learn a model and detect from the indicators each anomaly in the front-wheel axle unit of vehicles.