Modal Activity-Based Vehicle Energy/Emissions Estimation Using Sparse Mobile Sensor Data

Microscopic energy/emissions models have emerged as a promising tool for evaluating the environmental impact of the traffic, but require high-resolution speed/acceleration rate profile as the input. The massive but low-frequency global positioning system (GPS) data from commercial real-time traffic and navigation applications could be a potential data source for those models. In this paper, the authors developed a modal activity based method to estimate the vehicular emissions and energy consumption using sparse mobile sensor data. The proposed model estimate the vehicle dynamic state stochastically based on GPS data and traffic knowledge. The Speed and acceleration rate profile estimation models are developed for both acceleration/deceleration process and cruising process. The Motor Vehicle Emission Simulator (MOVES) model is then applied to calculate the emissions and energy consumption. The proposed model is calibrated with Next Generation Simulation (NGSIM) Lankershim dataset, and validated with the data from another time period. Numerical results show that the performance is good for emission estimation. Especially for aggregated measures over a short time period, the Mean Absolute Percentage Error (MAPE) is about 4%-8%.