Probabilistic Analysis and Design Optimization of Modelica Models
Tomas Johansson, Petter Krus · 2005
Abstract In this paper, the system simulation model is dis-cussed from an engineering design perspective. Spe-cial emphasis will be given Modelica models, and it is exemplified how computational design methods operate on the simulation mode l in order to evaluate different concepts. Model based design optimization and probabilistic analysis are discussed as examples of such computational methods. An XML-based information system for representa-tion and management of design data for use together with the Modelica model is further proposed in order to simplify the use of computational design methods. Finally, an example is presented, where probabilistic analysis is carried out on a Modelica model of an aircraft actuation system using the proposed and im-plemented tools and methods. 1 Introduction In the area of engineering design, a substantial part of the process consists of manual design work in-volving the inspiration and creativity of the designer. However, a large part of the design process can be formalized, and by applying formal design methods, these can be implemented in computer software as computational design methods. By employing com-putational methods in early stages of the design process, it is possible to acquire valuable informa-tion. Such methods could for example include model based design optimization or probabilistic analysis. These computational methods will be described in more detail throughout the paper, but common for the methods is that they operate on simulation mod-els in an automatic, iterative way. This implies new requirements on the simulation tools as well as on the representation and management of data related to the computational methods.