FleetTherm
Predictive thermal anomaly detection for wheels, bearings, and brakes with depot integration for scheduled maintenance interventions.
FleetTherm is depot-based thermal monitoring for rolling stock: vehicles pass through thermal imaging arrays at the depot, every bearing, wheel and brake is profiled at rest, and developing hot or cold spots become planned maintenance instead of a hot box on the road.
Who is this for?
Rolling stock maintenance managers at transit agencies and regional operators whose vehicles return to a depot daily, and freight terminals with captive locomotive fleets or private car owners at home terminals. It is built to sit beside your wayside hot box detectors, not replace them.
What problem does it solve?
A wayside hot box detector sees each bearing for a fraction of a second at 40 miles an hour. That catches the failure that is already hot; it cannot see the bearing that is trending up a few degrees a day, and it cannot see a cold wheel at all, which is the brake that never applied.
So the first sign is an emergency bearing change, a service disruption and a crew pulled off planned work. The thermal history that would have shown the trend three weeks earlier was never recorded, because nothing images the vehicle when it is standing still.
How does it work?
Thermal arrays at depot entry or exit image the full truck or bogie as the vehicle passes at up to 5 miles an hour: inboard and outboard bearings per axle, wheel surfaces and disc brakes. Models compare each component against configurable thresholds (absolute bearing temperature aligned with AAR 170 and 200 F guidance, a 95 F differential against the same-axle mate), against the vehicle's own baseline and against its peers.
Each anomaly is classed Monitor (trending up), Elevated (schedule an inspection) or Critical (hold for service), and the alert reaches the maintenance planner with the vehicle ID, the component location, the thermal image and a recommended action. Trend analysis across successive depot visits gives 3 to 14 days of warning before a condition would force an emergency intervention. Cold wheel detection flags brakes that failed to apply, which hot box systems do not see.
Accuracy on bearing and wheel anomalies runs 92 to 96 percent with a tunable false positive rate of 5 to 10 percent, and every alert carries the imagery, because a qualified technician dispositions every alert before the vehicle is released. Overrides are logged with a justification, and high-severity overrides need a supervisor.
Capabilities
- Thermal anomaly detection on wheels, bearings, brakes, and HVAC systems
- Pre-failure alerts with component-level tracking and failure prediction
- Depot integration for scheduled maintenance interventions and pull-aside triage
- Trend analysis dashboard showing thermal degradation patterns over time
- Alert tuning by fleet type, service conditions, and maintenance history
Outcomes
- Service disruptions ↓ 58%
- Emergency repairs ↓ 60%
- Planned maintenance ↑ 73%
What does it connect to?
- CMMS: IBM Maximo, SAP PM, Optram and custom systems through REST APIs, with severity, location, readings, trend and recommended action ready for a work order
- Fleet maintenance scheduling and parts inventory: high-priority alerts trigger a parts availability check and a maintenance slot
- Thermal cameras: 640 by 480 or better, NETD under 50 mK, FLIR and thermal line-scan arrays
- Compliance records for APTA RT-VIM inspection practices and FTA State Safety Oversight documentation
How is it deployed, and how long does it take?
Week one is a site survey and the power and data assessment, week two the array installation at one depot lane, week three a baseline capture across a representative fleet sample, week four accuracy validation and threshold calibration. The pilot then brings live inference, mechanic training and CMMS integration, and the rollout adds lanes and seasonal recalibration for summer and winter baselines.
Thermal imaging detects heat from friction, so a defect must be generating abnormal heat to show. For the earlier, pre-heat warning on internal bearing defects, acoustic monitoring is the complementary layer, and we say so rather than claim both.
Deployment in 60 to 90 days. An 18-month pilot on a 200-car transit fleet cut service disruptions 58 percent and emergency repairs 60 percent while planned maintenance rose 73 percent.
Industry context
Explore the related workflows and other solutions for your industry.