Using Machine Learning-Based Methods to Quantify Particulate Matter Emissions from Vehicles Post-Electrification

Abstract

It is essential to develop a comprehensive methodology to quantify the exhaust emission of road motor vehicle electrification. This study established a machine learning-based (ML-based) model that integrates social, economic, meteorological and motor-related. It quantified the inhalable particles (PM10) emission of Beijing from on-road vehicle in different electrification scenarios, including its heavy metal and microplastics. The results revealed that the proportion of exhaust PM10 emissions can be reduced from 49.33% in 2020 to 15.16% in 2035. Passenger cars at low-speeds and heavy-duty trucks at high-speeds contribute significantly to brake wear PM10 (TBPM10) and road and tire wear PM10 (TRWPM10) emissions. From 2020 to 2035, heavy metals and microplastics in non-tailpipe PM10 are increasing at rates exceeding 4% and 9% per year, respectively. Sensitivity analysis indicated that the application of regenerative braking technology holds significant potential for future vehicle emission reductions.

Keywords: Tire Wear, Traffic Flow, Electrification Policy, Emission Factors, Particulate Matter