The upholstered parts of a train wear out over time and reach a significant level of wear that prevents them from delivering their maximum performance rate.
The customer wanted to automatically predict the wear and tear of the upholstered part to alert the operator when a replacement was required. There was no historical data. In order to validate the solution, it was necessary to simulate the whole system to prove the feasibility of building a robust predictive model and to define data collection parameters.
Amiral Technologies used its knowledge and expertise in industrial, mechanical and control theory to build a digital twin that simulated several possible real-world equipment behaviours including all sources of deviation, noise and uncertainty. The digital twin generated realistic operational data which was used to train the DiagFit predictive model and simulate its performance to predict wear conditions.
We have proven that the system will generate data that can be used to create a robust model to predict wear on upholstered parts and trigger maintenance alerts.