The three most important phases of design are (1) conceptual phase; (2) configuration phase; and (3) parameterization phase. The second and the third phases are well researched. However, little work has been done in the conceptual design phase. In this paper the author deals with a different way of modeling the conceptual design phase. In this research the paradigm of function-to-structure transformation is used. One of the most difficult ideas of design is that of modeling the function-to-structure transformation process. The current research shows that the function-to-structure transformation is accomplished through a process of association. Whenever a designer is faced with finding a solution to a new functional requirement, the designer tries to associate this new function with known functions from his/her memory through a process of association. After having identified the closest function, an associated structure can be retrieved and mutated to form the design solution for the new problem. In essence, the designer associates the new functions with known functions and will try to retrieve the closest and most general design solution from his/her memory through a process of association. The author models the human associative memory through artificial neural networks (ANN) with back-propagation. A simple yet illustrative example of cups and containers is selected to model the function-to-structure transformation process at the conceptual design phase. In this paper the implementation aspects of the ANN are clearly explained. The robustness of the ANN through different schemes is also explored. A performance analysis via simulation by varying the nodes of the hidden layer is also carried out.
All Science Journal Classification (ASJC) codes
- Industrial and Manufacturing Engineering
- Artificial Intelligence