Automated thermal diagnostics of rolling stock
At one of the metro depots, the technical inspection process for trains took several hours. An automated solution was required.
Project background
At the metro depot, the technical inspection process for trains took several hours. Manual inspection of key components — bogie bearings, gearboxes, traction motors — required significant labor and created a risk of "human error." Each hour of train downtime increased the load on the train fleet and created a risk of disrupting the schedule.
An automated solution was required for thermal monitoring of critical car components in motion, allowing inspection time to be reduced from several hours to minutes and eliminating risks associated with human error.
Implementation process
Specialists from OSMOVIEW LLC developed and implemented a solution for automated rolling stock diagnostics based on the OSMOCV platform. The platform is deployed on the customer's server infrastructure in a local network.
Video streams from IP cameras installed in key areas of the depot are fed into the platform: a wide-angle camera for car number recognition and thermal cameras for running gear diagnostics.
The solution includes the following in-house neural services developed by OSMOVIEW LLC:
- Car number recognition in motion — a neural network reads the number from the wide-angle camera at a train speed of 10–15 km/h, identifying the car and forming the train composition in real time;
- Thermal diagnostics "on the move" — the system automatically detects 6 types of critical components (bogie units, traction motor bearings, gearbox housings, etc.) using thermal cameras installed under and beside the track. Instant temperature measurement allows overheating to be detected before it leads to a failure.
Additionally implemented:
- Integration with the customer's IT infrastructure — analysis results are automatically transmitted to the depot's corporate system, generating a report for each train;
- Automatic event generation upon detection of critical overheating, with notification to responsible personnel.\
System calibration, training of the customer's personnel on working with the platform, and acceptance tests were conducted.
Achieved results
- Train inspection time reduced — diagnostics take minutes instead of several hours, which allowed accelerating train turnover in the depot and reducing the load on the fleet.
- "Invisible" overheating eliminated — thermal monitoring detects critical temperatures that are not recorded during visual inspection.
- Diagnostics reliability improved — automation eliminates the influence of human error and ensures stable quality of inspection for each train.
- Integration with depot systems ensured — diagnostics results are automatically transmitted to the customer's IT infrastructure, generating reporting for each train.
- Failure prediction capability ensured — thermal monitoring data allows emerging faults to be detected before they transition into a critical phase.
Sources:
MT Integration — Automated technical condition monitoring of rolling stock