A Dual-Stream Physics-Augmented Unsupervised Architecture for Runtime Embedded Vehicle Health Monitoring
This work solves the problem of accurate predictive maintenance for commercial and heavy-duty fleets by distinguishing between transient anomalies and sustained mechanical loads, though it appears incremental as it builds on existing unsupervised and physics-based approaches.
The paper tackles the problem of vehicle health monitoring by addressing the blind spot where traditional unsupervised learning fails to detect high-load steady states that cause mechanical fatigue, proposing a dual-stream architecture that combines unsupervised anomaly detection with physics-based load estimation. The result is a low-overhead system validated on a RISC-V embedded platform, enabling edge-based monitoring without cloud dependency.
Runtime quantification of vehicle operational intensity is essential for predictive maintenance and condition monitoring in commercial and heavy-duty fleets. Traditional metrics like mileage fail to capture mechanical burden, while unsupervised deep learning models detect statistical anomalies, typically transient surface shocks, but often conflate statistical stability with mechanical rest. We identify this as a critical blind spot: high-load steady states, such as hill climbing with heavy payloads, appear statistically normal yet impose significant drivetrain fatigue. To resolve this, we propose a Dual-Stream Architecture that fuses unsupervised learning for surface anomaly detection with macroscopic physics proxies for cumulative load estimation. This approach leverages low-frequency sensor data to generate a multi-dimensional health vector, distinguishing between dynamic hazards and sustained mechanical effort. Validated on a RISC-V embedded platform, the architecture demonstrates low computational overhead, enabling comprehensive, edge-based health monitoring on resource-constrained ECUs without the latency or bandwidth costs of cloud-based monitoring.