Download files
Abstract
In today's highly competitive landscape, new product development strategies are imperatives for companies to create and sustain competitive advantages. The objective of this research is to develop an integrated approach to automate, or aid, the design problem solving process. An Integrated Design System (IDS) is proposed focusing on the parametric and detail design. In this system, generation and evaluation of new design problems occur quickly and easily by changing the inputs for the design model. The IDS provides an integrated platform to incorporate available application programs such as CAD and FEM tools into a single system. Four major frameworks, namely information handling, problem decomposition, design space characterization, and problem solving, are proposed and embedded in it to implement the product development process.
The information handling includes five aspects. A naming protocol is devised to organize the historical design cases. A search algorithm is proposed to retrieve a design case. A system-generated report is used to distribute the design information. A constraint definition frame is presented to define the relationships between design parameters. Two schemas, information matrix and constraint tree, are developed to represent information in the IDS. A diagonal-centered decomposition scheme is developed utilizing a Genetic Algorithm to decompose a complex design problem. In addition to the conventional genetic operators, two novel genetic operators, unequal position crossover and insertion mutation, are proposed. To characterize the design space, two methods, Incremental Response Method (IRM) and Artificial Neural Network (ANN), are presented. The IRM is derived from response surface method, while the back-propagated ANN is coded to be self-evaluated. The presented problem solving algorithm constitutes the solving mechanism of the IDS. Based on the assessment of the design objectives, all design parameters are given a priority index to facilitate the parameter selection. An independent recursive method is introduced in this algorithm to handle the design constraints. The case studies are performed on two design problems: a hard disk drive actuator arm and a shaft. The results show that the system can automatically align parameter values towards the objective values in a reasonable manner, and thus verify the feasibility of the embedded frameworks.