Nonlinear Dynamic Modeling of Diesel Engines

摘要

Nowadays, the method Design of Experiment (DoE) has evolved as a standard in the development of combustion engines. Although the measurement time is significantly reduced, compared to a grid-based approach, it is still necessary to wait for stabilization during the measurement. For future applications and Rapid Measurement it is therefore of essential importance to apply dynamic models for transient engine processes. However, the process of setting up the nonlinear dynamic model requires elaborateness in dynamic test plan design, signal synchronization, and model term selection. We exemplarily show the results of NO{sub}x-emission modelling and exhaust temperature modeling. For that purpose, two possible model types, i.e. Parametric Volterra Series and recurrent RBF (radial basis functions), are introduced and the design of suited excitation signals is discussed. Subsequently, the realization at the test bench system is shown and the results of the modeling approaches are compared.