Using model-based methods for the development of electronic control systems is state of the art. The classical DoE method has established itself. The classical DoE method comprises the use of ASCMO (Advanced Simulation for Calibration, Modelling and Optimization). ASCMO is a tool available from the ETAS Group that serves to model the input/output behaviour of unknown systems on the basis of measured data that are obtained by means of a statistical design of experiments. ASCMO allows with its advanced and parameter-free modelling algorithms application of databased simulation methods for the development and calibration of engine electronic control units and is widely used at Bosch. Because the possible application is restricted to steady-state or quasi-steady state behaviour, though, only a limited part of the development process can actually be performed with a DoE. This article explains how the steady-state DoE approach can be used to describe and optimize dynamic behaviour, despite the known restrictions. In the process, characteristic numbers are determined on the basis of the dynamic behaviour of the system, which will be referred to here as criteria. These criteria can subsequently be modelled and optimized with the DoE method. In this context, it is important that the chosen criteria represent the properties that will be later optimized in the system as well as possible. In a manner of speaking, the dynamic behaviour of the system is thereby transformed to steady-state behaviour by means of these criteria. The multi-criteria optimization integrated in ASCMO facilitates the comprehension of conflicts of objectives between several values and provides optimum solutions. Last but not least, the decision is left to the calibration engineer to select the best compromise. The article has shown that the steady-state DoE method can also be used to model and optimize dynamic processes. The procedure has been explained by using the simple example of a PT2-element and has been introduced on the basis of the practical application of the Bosch platform driveability function of the EDC. Specific characteristics of the method have been explained, and the tool chain that has been developed for the method has been covered. The results achieved show a very good reproducibility of the results. This enables the calibration engineer to describe the distinct trade-off of the system and, depending on the specifications of the target function, always determine optimal parameters. Transfer of the method to other tasks and the recognized advantages have been explained. A factor that should not be neglected is the fact that the model of behaviour is based on real measured data that can be generated with "on-board means" of the Bosch engine management system. Elaborate modelling and parameterization can be omitted. The measurements can be taken on a given experimental vehicle, and the best compromise can subsequently be directly set in the vehicle.
Extension of the classical DoE method to dynamic conditions by means of characteristic numbers
2011
16 Seiten, 11 Bilder, 9 Quellen
Conference paper
English
Engineering Index Backfile | 1926
|Dynamic modeling and optimization: the natural extension to classical DoE
Tema Archive | 2007
|Extension of classical tip loss formulas
Tema Archive | 1989
|