Research on the Emissions and Fuel Efficiency Sensitive Factors of Gasoline Engine Using a Support Vector Machine Coupling Sobol Method

Abstract

With the increasing prominence of energy and environmental issues, research in the field of internal combustion engines is becoming more and more refined. Engines are affected by the coupling of many factors, and it is necessary to decouple and quantify the impact of inputs on the objectives. In this paper, the effect of engine configuration parameters, operating conditions, and fuel parameters on fuel efficiency and emissions was investigated based on a support vector machine coupling Sobol method. The results of Sobol sensitivity analysis show that the most sensitive parameters for both brake specific fuel consumption and carbon monoxide emission are excess air ratio, total hydrocarbon emission and nitrogen oxides emission, engine load, and intake pressure; the first-order indices are 0.72, 0.27, 0.17, and 0.20, respectively. The most sensitive combustion parameters are maximum pressure in the cylinder, indicated mean effective pressure, maximum temperature in the cylinder, and high-temperature range, and the first-order indices are 0.40, 0.25, 0.39, and 0.57, respectively. It can be summarized through all the Sobol indices that, on the one hand, some input parameters, such as excess air ratio, affect fuel efficiency and emissions through the combustion process and, on the other hand, such as oxygen mass fraction directly affects carbon monoxide emission and total hydrocarbon emission by affecting the oxygen concentration in the cylinder. Sensitivity analysis based on the Sobol method coupling support vector machine was proved to be feasible and will provide valuable guidance for the optimization of internal combustion engines.