FleetLynq: AI-driven Decision Support Solution for Transit Fleet

Transit agencies connect millions of Americans daily to their critical needs. However, services are unreliable because the fleet at over one-third of agencies is old, has a 10-15% spare ratio, and experiences recurring major mechanical failures. Skilled technician shortages and unreliable tools worsen this situation. Transit agencies have deployed vehicle health monitoring (VHM) solutions over the past 2 decades for real-time monitoring and prognosis of component failures. However, these solutions suffer from false positives and have not met agency expectations. Recent deployments of predictive maintenance solutions are promising but models require significant human feedback for training, are black boxes, and are component-specific. Further, both these approaches heavily depend on OEM data, which is either noisy, inconsistent, or insufficient. FleeyLynq is an AI-first solution that will deliver 1) an Edge AI unit to filter and process OEM and IoT data to train Cloud models; 2) Cloud AI models to train on long-term sensor data and maintenance/inspection records; 3) Self-learning knowledge base as a distilled LLM to augment edge/cloud processing and enable applications such as predictive maintenance and remaining useful life (RUL) estimation. FleetLynq will enable efficient, safe, and reliable transit services and deliver an approach scalable to other industries.