Abstract

One of the ways to improve the fuel economy of heavy duty trucks is to operate the combustion engine in its most efficient operating points. To do that, a mathematical model of the engine is required, which shows the relations between engine speed, torque and fuel consumption in transient states. In this paper, easy accessible exploitation data collected via CAN bus of the heavy duty truck were used to obtain a model of a diesel engine. Various polynomial regression, K-Nearest Neighbor and Artificial Neural Network models were evaluated, and based on RMSE the most relevant sets of parameters for the given algorithm were selected. Finally, the models were compared by using RMSE and Absolute Relative Error scores for 5 test samples. These represent the whole engine’s operating area. Apart from goodness of fit, the models were analyzed in terms of sensitivity to the size of the training samples. ANN and KNN proved to be accurate algorithms for modeling fuel consumption by using exploitation data. The ANN model was ranked best, as it required less observations to be trained in order to achieve an absolute relative error which was lower than 5%. A conventional method, i.e. polynomial regression, performed significantly worse than either the ANN or the KNN models. The approach presented in this study shows the potential for using easy accessible exploitation data to modeling fuel consumption of heavy duty trucks. This leads to the reduction of fuel consumption having a clear positive impact on the environment.

Citations

Authors (5)