Discrimination of Excessive Exhaust Emissions of Vehicles based on Catboost Algorithm

This paper proposes a classification model based on CatBoost algorithm to discriminate excessive exhaust emission of the vehicles. The carbon monoxide (CO), hydrocarbon (HC) concentrations and other related exhaust data emitted by vehicles are accumulated from the Urban Road Network Vehicle Emissions Monitoring System. The original exhaust data is pre-possessed to yield clean and accurate data. The CatBoost algorithm we applied is a new Boosting algorithm, which has the advantage of higher discrimination accuracy than traditional machine learning algorithms. We collected exhaust data for two separate districts in Beijing and Hefei, China. The experimental results confirm that the discrimination model based on CatBoost algorithm is superior to the traditional machine learning classification algorithms, which can achieve higher discrimination accuracy.