Operational Data Based Anomaly Detection for Locomotive Diagnostics

Locomotives are complex electromechanical systems. Continuously monitoring the health state of locomotives is critical in modern cost-effective maintenance strategy. A typical locomotive is equipped with the capability to monitor their state and generate fault messages and a snapshot of sensed parametric readings in response to anomalous conditions. In our previous studies, we have developed and deployed a case-based reasoning system for locomotive diagnostics where fault codes were used as the inputs to the system. In order to increase the lead-time from detection to failure and allow for more proactive actions, one important effort in locomotive diagnostics is to perform anomaly detection on parametric operational data. In this paper, we present an anomaly detection strategy that is based on a combination of nonparametric statistical testing and machine learning methodology. We demonstrate the effectiveness of the anomaly detection strategy using real-world operational data from locomotives.