Early Detection of Engine Anomalies - A Case Study for AI-Based Integrated Vehicle Health Management
Sasha Apartsin, Hilik Stein, Gil Reiter, Kyle A. Williams, Noam Moscovich · SAE technical papers on CD-ROM/SAE technical paper series · 2022
As vehicle warranty claims, recalls, and maintenance costs continue to grow, new methods are needed to predict, detect, and diagnose vehicle health issues. By integrating artificial intelligence (AI) technology into the vehicle’s embedded electronics, automakers and fleet owners can benefit from highly effective and adaptable vehicle health management capabilities that are not available today. This paper describes how embedded AI-based signal integrity monitoring can be used to detect complex anomalous patterns in engines. It introduces a novel end-to-end signal integrity monitoring solution, which is based on a pipeline of machine learning models that are trained in an unsupervised manner. It also describes how unsupervised deep learning technology can simplify the data collection and labeling process that is needed to train the AI-based vehicle health management models.