Analysis of vehicle-related non-exhaust PM10 and emerging pollutants in Beijing with machine-learning.

With the growth of electric vehicles, non-exhaust PM10 is becoming the dominant contributor to vehicle-related particulate pollution and a source of emerging pollutants. This study evaluated four machine-learning algorithms and selected Random Forest (RF) to estimate road traffic flow. By coupling RF with the MOVES model, we developed an RF-MOVES model to quantify the emissions of non-exhaust PM10, heavy metals, and microplastics, and assessed the characteristics of the emissions under three electrification scenarios. Research shows that temporal variations in tire-road wear PM10 (TRWPM10) and tire brake PM10 (TBPM10) are attributed to travel behavior and road conditions, while spatial heterogeneity reflects road-network structure and vehicle-type distribution. Vehicle electrification increased the proportions of TRWPM10 and TBPM10 to total vehicle-related PM10 due to reduced exhaust PM10. The fractions of heavy metals and microplastics in non-exhaust PM10 increased by over 4% and 9%, respectively, indicating a growing potential for environmental contamination. Furthermore, increasing regenerative braking reduces non-exhaust PM10 and heavy metal emissions, while its effects on microplastic mitigation remain limited. This study provides a model for calculating high-resolution non-exhaust PM10. Our results highlight the potential environmental contamination risks of vehicle-related non-exhaust PM10 and offer insights for managing non-exhaust PM10, heavy metals and microplastics in future electrification scenarios.