A Method for Pre-Calibration of DI Diesel Engine Emissions and Performance Using Neural Network and Multi-Objective Genetic Algorithm

Diesel engine emission standards are being more stringent as it gains more publicity in industry and transportation. Hence, designers have to suggest new controlling strategies which result in small amounts of emissions and a reasonable fuel economy. To achieve such a target, multi-objective optimization methodology is a good approach inasmuch as several types of objective are minimized or maximized simultaneously. In this paper, this technique is implemented on a closed cycle two-zone combustion model of a DI (direct injection) diesel engine. The main outputs of this model are the quantity of NOx, soot (which are the two main emissions in diesel engines) and engine performance. The optimization goal is to minimize NOx and soot while maximizing engine performance. Fuel injection parameters are selected as design variables. A neural network model of the engine is developed as an alternative for the complicated and time-consuming combustion model in a wide range of engine operation. Finally design variables are optimized using an evolutionary genetic algorithm, called NSGA-II.