This thesis introduces new methods to calculate online design of experiments and to identify online the dynamic behaviour of combustion engines. These methods are developed to support the electronic control unit calibration in terms of modelling and optimizing the dynamics of the combustion engine.
The model-based online design of experiments allows gathering maximum information with minimum measurements, to consider constraints on the inputs and outputs and to optimize the design of experiment in respect to the indented use of the models. The online-identification enables to identify continuous time, input-to-state stable state-space models using neural networks, which are disretized by Runge-Rutta methods. It is possible to show, that the proposed online design of experiment and the online identification are superior to state of the art methods.
«
This thesis introduces new methods to calculate online design of experiments and to identify online the dynamic behaviour of combustion engines. These methods are developed to support the electronic control unit calibration in terms of modelling and optimizing the dynamics of the combustion engine.
The model-based online design of experiments allows gathering maximum information with minimum measurements, to consider constraints on the inputs and outputs and to optimize the design of experiment...
»