The identification of distinctive features in the language underlying mechanical engineering design communication.

摘要

During the design process, the ideas of the inventor must be precisely communicated to other individuals to enable the conceptual design and creation of an artifact. Miscommunication often spells disaster. This project partially quantifies communication in mechanical engineering design.;Some of the distinctive features of the language underlying mechanical engineering design communication were identified. Five experts, four quasi-experts, and seven novices were selected based primarily on their level of theoretical knowledge and practical experience in mechanical design. Verbal protocols were collected when they were describing thirteen two-dimensional drawings of mechanical assemblies representative of a broad spectrum of complexity. The collected data were reduced to uncover the features of the drawings attended to, whether the subjects correctly identified the overall function of the assembly, and their reasoning strategy. The geometric properties of the identified design features also were tallied.;Descriptive statistics, univariate analyses of variance, multivariate discriminant analysis, and non-parametric techniques were used to describe which design features were used by the three groups, the inclusion of which contributed to the distinction between the groups, and which geometric attributes were used by the subjects in the three groups when the design features were identified. The thrust of these analyses was to determine what geometric attributes and design features are used in the transformation from perception to recognition. Those geometric properties playing a major role in this transformation were identified as some of the distinctive features of the language underlying mechanical engineering design communication. Cluster analysis was used to describe how the transformation could occur.;A cognitive model of the transformation was developed and a computational simulation of the model was implemented using the neural network paradigm. The developed model, using adaptive resonance theory, was exercised and compared with the behavioral findings.;Implications for the development of computer-based representations of design communication for both the practicing mechanical engineering designer and the student of mechanical design are discussed. The study also supports the need for multidisciplinary researchers, such as biomedical engineers and knowledge engineers, to know and to employ behavioral science research methods to problems which involve human beings.