Investigating a new model-based calibration procedure for optimizing the emissions and performance of a turbocharged diesel engine

Today, diesel engines are no longer mentioned for generating huge amount of soot and high level of noise. These achievements are owing to the employment of numerous mechatronic systems implemented in the engine. Altogether, with the increase of the number of controllable parameters, the complexity of control and calibration tasks has been increased. In the conventional calibration processes, numerous tests are required for calibration of engine controllers, making it a time consuming and expensive procedure. However, in this paper a model-based calibration procedure based on evolutionary algorithms is investigated to fulfill the feed-forward controller look-up tables. The look-up tables obtain the fuel injection and air induction system parameters based on engine speed and relative load and guarantee the optimal operation of engine. The developed procedure guarantees the maximum attainable torque in full load. The proposed method decreased the time, cost and complexity of whole calibration procedure to high extent. Artificial neural network is employed for modeling the combustion process while steady-state mass and energy balance equations are used for inlet and exhaust models. The models have been validated using experimental data. The optimization is done in two phases: full load curve shaping and part load optimization. The aim of former is attaining maximum possible torque with the minimum emissions and fuel consumption in every engine speed while the aim of latter is delivering the required torque with the lowest possible emissions and fuel consumption. The results of tests show that the proposed model-based calibration method can effectively reduce the fuel consumption and emissions in whole engine operation regime and decrease the time and cost of calibration.