Physically consistent parameter optimization for the generation of pose independent simulation models using the example of a 6-axis articulated robot

摘要

In the context of shortening product life cycles, simulation of machine tools is more and more in the conflict between economic efficiency and technical requirement. Due to this both complex simulation tools and simulation methods with minimal effort are used. In the model alignment-an essential part of simulation-the highly parameterized models are adapted to the measurement data using adequate fitting algorithms from optimization theory. Thereby the pose dependence of the adapted simulation models is the result and thus multiple model alignments in the other relevant poses of the researched machine are necessary. This is associated with a high measuring expenditure, a high personal expenditure and an increasing development time of the machine. In this paper a method for the generation of pose independent simulation models by the example of a 6-axis articulated robot is presented. Basically, the model alignment is done in a way that the fitting model parameters are disturbed systematically with respect to physical sense and the resulting models have to be proved in terms of their physical plausibility. First of all, simulation models with different optimal parameter configurations related to the measured data are generated. After this, the resulting parameters are analyzed concerning their behavior in all poses and physically consistent parameter configurations of every pose are identified. Finally, the discovered configurations are used to create an exclusive parameter configuration respectively an exclusive simulation model. The pose independence of this model is postulated and the developed method is verified by the example of a 6-axis articulated robot. The implementation of the method is based on a new way of modeling mechatronic multi body dynamics using the bond graph theory. Therefore it also contains the description of the machine structure by symbolic equations of motion. This enables a transfer of the method to other machine structures without consideration of special software tools.