Research Topic

Multidimensional Optimization: Automatic Optimizer for Combustion Engines

Car engine on a test bench

The optimization of inner combustion engines (ICEs) is a great challenge for etablished automotive manufacturers. In order to fulfill various exhaust emission standards, the level of application of car engines is steadiliy increasing. Additionally, due to increasing power densities (Downsizing) the protection of engine components is getting important. Modern engines are equipped with various new actuators, i.e. CAM Shifts, EGRs, Wastegate, Pilot injections, aso. Therefore the degree of the mathematical optimization problem is typically between 6 till 14.
Classical measurement cycles are inappropriate for this high complexity. Additionally, reproducibility has to be guaranteed, therefore objective criteria have to be defined in front of the automatic measurement process.
Due to increasing data volume, optimization algorithms have to be highly efficient.

Recent Results

  • Data Acquisition

    Data space is covered by multidimensional measurement paths, which grant a homogeneous measurement point density. The system is defined by phyiscal conditions, i.e. maximal gradients, which have to be contained. Additionally, automatic detection and correction meachnisms supervise the measurement process, in oder to eliminiate regulator oscillations.

  • Data Analysis

    Data Acquisition have to be generilized to a data room description. This is done by Support Vector Regression, Hyperplan rendering and multidimensiona tesselation algorithms.

    Furthermore, classification criteria due to engine charactersitics and physical subgroups have to be defined and validated.

  • Data Optimization

    The data is optimized on the foundation of measurement data points or data room description. Random Walker algorithms derive optimization samples, which gain insight to possible global optima. Finally, optimization samples have to be completed and combined to steady solution functions, which fullfil all boundary conditions of the phyical system.

Key Research Area

Optimization – Combustion Engines – Data room descriptions

Contact

Timo Burggraf
Dipl.-Phys.

Address:

Dolivostra├če 15

D-64293 Darmstadt

Germany

Phone:

+49 6151 16 - 24401 or 24402

Fax:

+49 6151 16 - 24404

Office:

S4|10-206

Email:

burggraf (at) gsc.tu...

 Print |  Impressum |  Sitemap |  Search |  Contact |  Privacy Policy
zum Seitenanfangzum Seitenanfang