End-of-Line (EOL) Vehicle Component Tests Anomaly Detection

Motivation & Problem Description

• Reducing warranty rates caused by critical vehicle components, such as transmission and engines

• Current practice is manual and rely on mainly engineers' experience, 

   

Objective

Automate and scale EOL testing procedure by utilizing machine learning (AI/ML) methods.

Solution

• Multichannel time-series anomaly detection by spectral feature learning.
• Dimensionality reduction and unsupervised learning techniques for high-dimensional data.

Impact

• Reduce warranty rates; 
• Improve defect detection; 
• Enhance manufacturing efficiency; 
• Scalable across vehicle components.