Modeling and optimization of a light-duty diesel engine at high altitude with a support vector machine and a genetic algorithm

Abstract

The engine performances and emissions of light-duty diesel engines in plateau regions have attracted more attention due to the upcoming China VI emission regulations for light-duty vehicles. In order to obtain a superior performance for a diesel engine running at high altitude, in this research, multi-objective optimization was conducted in an entire operating region for a light-duty diesel engine operating at an altitude of 1960 m. A support vector machine (SVM) was employed to set up a surrogate model between the calibration parameters and the engine performance parameters. The multi-objective optimization of the fuel consumption and the emissions was carried out using a genetic algorithm with the premise of keeping the same power performance of the original engine within durability constraints and with a minimum smoke limit. The results showed that the SVM regression model had excellent predictive performance and generalization abilities, and that the model could accurately predict the various performance parameters of the diesel engine. The diesel engine running in the plateau region could achieve a good comprehensive performance with the proposed multi-objective optimization method. In comparison with the base engine in the plateau region, a simultaneous reduction of 52.92% for the brake specific NOx emission and 0.67% for the brake specific fuel consumption was achieved, with an acceptable increase of smoke emission.