Applying a neural network-based approach for estimating instantaneous emissions of the heavy-duty buses: A case study

The lack of a chassis and a transient engine dynamometer for heavy-duty vehicles in Vietnam is a significant impediment to develop the country-specific emission factors. This study presents a novel approach for modeling the emission rate of heavy-duty bus engines based on the artificial neural network (ANN) to overcome the above limitation while ensuring emission prediction's accuracy and minimizing test costs. The ANN-based models with high reliability (R-2 > 0.98 and MAPE <20%) were built based on the experimental data collected from the engine emission measurement on the steady-state engine dynamometer. The developed models were used to estimate the bus's emission rate according to the real-world driving characteristic that was taken into the typical transient engine cycle. The estimated average emission factors in terms of distance-based ones for buses in Hanoi according the ANN-based models were close to those measured.