University of Twente Student Theses
A Computer Aided Innovation Tool for Generating Solutions for Mechanical Engineering Functions
Linschoten, D. (2019) A Computer Aided Innovation Tool for Generating Solutions for Mechanical Engineering Functions.
|
PDF
19MB |
| Abstract: | Many computer aided innovation (CAI) tools exist but all either rely on a creative input of the designer or make use of some sort of database of working principles. This research proposes a CAI-tool that generates artifacts, without introducing any working principles. The artifacts generated are solutions to a mechanical engineering problem. A framework for the new CAI-tool is developed that disengages the problem description from the actual artifact generation and optimization. For the problem description, the functional basis for engineering design by Hirtz et al. [1] is used to describe a mechanical engineering problem as a mechanical engineering function which converts a certain input flow to an output flow. This function determines the objective function for the computational optimization. Artifacts are designed to consist of basic elements that have behavior and characteristics but no functionality of their own. A computational optimization algorithm is used to create and optimize new versions of the artifacts. Better artifacts are selected based on their fitness determined by the objective function. Because no working principle is introduced to the CAI-tool, it is hypothesized that these working principles will be generated by the tool itself. A proof of concept was made for the proposed CAI-tool based on Evosoro [2]. This software uses compositional pattern-producing networks (CPPN) that produce three-dimensional multimaterial artifacts made up of voxels which are optimized using neuro-evolution of augmenting topologies (NEAT). These voxels are the basic elements of which the artifacts consist, in this proof of concept. A different population selection method, passing multiple objects and new fitness calculations are among the newly introduced or changed features to the software. As a proof of concept, a simple mechanical problem was tested: solutions for the function ‘rotational transmission’ were tried to be generated. Analysis of solutions was done using a modified version of the contact and channel model to identify the difference in working principles. Multiple times, in single runs without a change of parameters, more than one working principle was identified among the generated solutions. This proves that no single working principle was predetermined and proves the ability of the proposed CAI-tool to fulfill a mechanical engineering function without prior knowledge of working principles. Keywords: Computer Aided Innovation, Genetic algorithms, Concept development, Functional analysis |
| Item Type: | Essay (Master) |
| Faculty: | ET: Engineering Technology |
| Subject: | 52 mechanical engineering |
| Programme: | Mechanical Engineering MSc (60439) |
| Link to this item: | https://purl.utwente.nl/essays/93996 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page