This analysis is based on the challenge of SIEMENS at HackZurich 2021. With the callenge, data from one of their train-communication-systems was provided. The task was to detect, predict and categorize failures. See the challenge documentation for further information about the challenge.
Our team's approach was to transform the provided data from train-based to track-based and then build a predictive maintenance model for the degradation-based emergency-stops.
The original hackathon code can be found here and here, this repo contains an analysis done afterwards.