Dynamic Behavior Models and Their Modeling Depth in the Design Process of Mechatronic Systems

Abstract

Model-based design is an important part in the development of mechatronic systems. Models are used in various fields of this process, for example, to specify the actuators and sensors, as well as to design the controller, or to describe and analyze the dynamic behavior of the system. These models consist of mechanical, electrical and software components and many more. Hence, developers need a discipline-spanning knowledge in order to model and analyze mechatronic systems, as well as methods for the development of these systems are needed. We divide the design process of mechatronic systems into a discipline-spanning design phase and a concurrent engineering phase in the respective disciplines [1]. Both phases rely on model-based design. The selection of an appropriate modeling depth is essential, and one of the most challenging parts during modeling process [2,3]. An aim of the design method [1] is the preparation of solution knowledge in order to reuse it in future developments. The challenge is to find and select an appropriate dynamic behavior model with a sufficient modeling depth in order to fulfill all modeling objectives. A quantification of modeling depth is required in order to be able to classify models for reuse, and to automate their selection. Therefore, our first step is to classify models by defining four levels of detail. Based on the modeling objectives a level of detail is recommended to the developer. However, models typically consist of many elements from several disciplines and of different levels of detail. In this case, mapping to the four levels of modeling depth is not obvious. That is why our second step defines an index to describe the modeling depth of combined models. This index is based on the state variables of the dynamic behavior models and independent from the composition/modularization of the system model. With these two criteria, it is possible to classify the modeling depth of dynamic behavior models. Our method is shown and validated with the help of an application example.