Künstliche Intelligenz zur Struktur- und Maßsynthese ebener Führungs- und Übertragungsgetriebe

  • Artificial intelligence for type- and dimensional synthesis of planar guidance and transfer mechanisms

Müller, Mario; Hüsing, Mathias (Thesis advisor); Corves, Burkhard (Thesis advisor); Modler, Karl-Heinz (Thesis advisor)

Aachen (2020)
Dissertation / PhD Thesis

Dissertation, Rheinisch-Westfälische Technische Hochschule Aachen, 2020


Mechanisms are mechanical solutions for the transformation of motion or for the transmission of forces and torques. The increasing requirements on mechanisms in practice can hardly be met by a developer without in-depth knowledge in mechanism theory. Therefore, the research field of artificial intelligence is used to support the development process. It provides solutions to help both experts and nonprofessionals to solve complex problems. Such IT systems are known in engineering and science as expert systems. Within this work, an artificial intelligence was developed in the form of an expert system for mechanism synthesis. Starting from a motion task, the artificial intelligence can autonomously go through the mechanism design process. The motion task is thereby independent from potential synthesis methods, and can be formulated for both transfer and guidance mechanisms. Just as in classical mechanism synthesis, the artificial intelligence addresses the two topics of structural and dimensional synthesis. In the subject area of the structural synthesis, potentially suitable mechanism structures are determined for the following step of the dimensional synthesis. Therefore, in the context of this thesis, a database has been developed which contains mechanism structures for the fulfillment of transfer and guidance mechanisms. To generate such mechanism structures, a graph-based algorithm has been developed. The structures identified by this algorithm can consist of revolute joints, prismatic joints as well as cam joints. Due to the specific properties of the mechanism structures, the number of potentially suitable structures can be drastically limited within the structural synthesis by means of the set of rules of artificial intelligence. These mechanism structures can then be dimensioned within the dimensional synthesis by using optimization-based algorithms. For this, they are divided into recurring modules. Because of this procedure, solutions for complex mechanism structures can be determined with comparatively few dimensioning algorithms. The validation of the developed artificial intelligence shows that it is not only able to find known solutions of a problem, but also to extend the known scope of solution by further mechanisms. Thus, the artificial intelligence does not only provide a support for non-professionals but also for experts.


  • Chair and Institute of Mechanism Theory, Machine Dynamics and Robotics [411910]