Optimal Modelling of Mechatronic System Components
Optimal Modelling of Mechatronic System Components in Mobile Applications
The advancing development in the field of driver assistance, active driving dynamics control and especially the trend towards highly automated driving has led to the use of very complex mechatronic systems in motor vehicles in recent years. A model-based approach is often chosen to control the systems. These models sometimes have a very high level of detail and complexity. The consequence of this development is an increasing computational effort as well as the loss of intuitive understanding of prevailing interactions in the system. The desire in the automotive industry for high-performance real-time control models and, at the same time, low-cost target hardware presents development engineers with new challenges. In order to optimally design a system, a comprehensive knowledge of all model components and their interactions is necessary.
The aim of the project is to provide this knowledge in order to ensure optimal parameterisation and modelling of physical interrelationships in automotive applications. An analytical identification of these interrelationships is not possible in the vast majority of cases, so statistical methods are used.
In the sense of a methodology that is as robust and generally applicable as possible, the process of finding an optimum was divided into several steps.
The aim of the project is to provide this knowledge in order to ensure optimal parameterisation and modelling of physical interrelationships in automotive applications. An analytical identification of these interrelationships is not possible in the vast majority of cases, so statistical methods are used.
In the sense of a methodology that is as robust and generally applicable as possible, the process of finding an optimum was divided into several steps.
- Sensitivity analysis The sensitivity analysis has the task of identifying from a set of different parameters those that have the greatest influence on a previously defined criterion.
- Metamodelling The metamodel is a reduced image of the originally investigated model, which reflects the relationship between the parameters to be investigated and the target variables.
- Multo-objective optimisation In multi-objective optimisation, a global optimum can be found between several antagonistic objective values.
On the basis of these three building blocks, a routine could be developed that enables the optimal design of a wide variety of model types.
Project supervisor at the Chair of Mechatronics:
Belongs to profile focus areas of the Faculty of Engineering at the UDE:
Link to the Faculty (research profiles): www.uni-due.de/iw/en/research/psp.php
