Due to the growing demand for fuel-efficient combustion engines and strict emission limits new manipulated variables like variable valve trains have been introduced in the past years. During the optimization the manipulated variables have to be calibrated very precisely in order to meet the legal requirements and the power demands. For optimization purposes it is necessary to be able to simulate the high dimensional and nonlinear engine system. Since it is not possible to optimize them manually, suitable mathematical models are required. Physical-oriented analytical models are at least partially not practicable because of too complex chemical and thermodynamic relations. Thus, it is common practice to excite the system with certain signals and to build mathematical models based on measurements (process identification). Therefore different variation points have to be measured. The required time to obtain one variation point of a steady-state measurement consists of a settling time due to the dynamics and a measuring time. Several values are recorded during the measuring time, which are afterwards averaged to get the final value of the variation point. To reduce the overall measurement time of steady-state approaches, Design of Experiments (DoE) methods were introduced. These methods allow a reduction in the number of variation points. A further reduction of the overall measurement time is achieved by dynamic measurements, as the settling time is omitted. In contrast to steady-state experiments the measured values are recorded continuously. Dynamic measurements require new identification and analysis methods. On the one hand dynamic effects have to be taken into account, on the other hand steady-state combustion engine models are required for the calibration process. During this project a strategy for enhanced measurements was developed. In order to change as few methods of the steady-state measurements as possible, the steady-state workflow was used as template. But nevertheless the modeling algorithms had to be changed in comparison to the conventional approach. It was shown, that not all model analysis methods of the steady-state process were applicable for the new concept. However, valid techniques for model adequacy checking are presented. The Design of Experiments was expanded as well. The new strategy was successfully tested at an industrial testbed. The overall measurement time was reduced to 33 % of the time which steady-state DoE-methods require.
Schnelle Motormessung mit optimierten Testsignalen
2008
26 Seiten, 27 Bilder, 1 Tabelle, 17 Quellen
Conference paper
German
British Library Conference Proceedings | 2006
|Mit optimierten Ladehilfsmitteln
Tema Archive | 2009
|Gewichtseinsparung mit optimierten Längswellen
Tema Archive | 2009
|