Research On Model-based Virtual Calibration Of Diesel Engines With Common Rail System

Diesel engines, with high pressure common rail injection system, is capable of flexible injection timing, injection pressure and injection rate control. But one of development difficulties is the calibration of so many MAPs.In order to improve the calibrating efficiency, a Model-based virtual calibration method is presented in this thesis. The method consists of the following steps: experiment of design, virtual experiments, response modeling and global optimization. The main achievements can be detailed as follows.First, a diesel engine simulation model using GT-power software in one dimensional thermodynamic simulation was built depending on the basic dimensions of the engine and some capability experiments on test bed. Test bed experiments show that the model can meet the requirements of virtual experiments.Then, the orthogonal experiments design, class design, optimal design and space-filling design were compared and analyzed. Results show that the space-filling method is more preponderant.Two stage response model was built, and for the global modeling, an artificial network model was built to simulate, and for the local modeling, a polynomial model was built to analyze data .Genetic algorithm (GA),based on the two stage model,was used for global optimization and MAPs generating. Evaluating results on the MAPs via GT-Power simulating show advantages of the Model-based virtual calibration method.