For a global offroad vehicle leader, defects in new engine products need to be anticipated during the design stage to trigger proactive design improvements and accelerate the Design cycle.
Alumnus created an AI-based solution to identify engine failure signatures using data read every second from over 100 sensors placed around the product. This is transferred to an Azure cloud and processed using ML and Rule Engines to anticipate engine breakdowns. When engineers get a real-time alert for a possible defect, engineers can query datasets around that instance to ratify the issue and diagnose underlying causes.
The solution’s capabilities include Correlation matrices, PCA & LDA for dimensionality reduction Random Forest / XGBoost and RNN-based Time series analyses.
Azure Cloud, Python