Remote monitoring on-board diagnostics data-driven models for NOx emission evaluation and high-emitting vehicle screening

Heavy-duty diesel vehicles (HDDVs) are among the largest sources of nitrogen oxide (NOx) emissions, underscoring the need for effective real-time monitoring. Portable emissions measurement systems (PEMS) provide reliable measurements but are costly and unsuitable for large-scale deployment. Remote monitoring On-Board Diagnostics (OBD) systems represent a practical alternative, offering continuous, low-cost data streams. This study develops a systematic framework to process remote OBD data and proposes five emission assessment models: OBD-based Moving Average Window (OBD-MAW), OBD-based Three-Bin MAW (OBD-3BMAW), OBD-based Vehicle Specific Power (OBD-VSP), OBD-based Fuel Consumption Factor (OBD-Fuel), and an OBD-based 50 ppm threshold model (OBD-50 ppm). One month of second-by-second OBD data from 50 China VI diesel vehicles was analyzed. Seven vehicles were consistently identified as vehicles with markedly elevated NOx emissions relative to typical fleet levels by all models, with inspection records confirming aftertreatment malfunctions. In contrast, several vehicles flagged by only one model showed condition-specific above-threshold cases under the adopted criterion, revealing model sensitivity to operating conditions. Among the five approaches, OBD-Fuel and OBD-50 ppm demonstrated the greatest potential for regulatory application due to their simplicity, minimal data dependency, and real-time compatibility. The proposed methodology supports scalable, data-driven monitoring and provides actionable tools for policymakers to enhance precision in emission supervision.