Design Parameter Selection in the Presence of Noise

Kevin N. Otto and Erik K. Antonsson

Research in Engineering Design
Volume 6, Number 4 (1994), Pages 234-246.


The it method of imprecision is a design method whereby a multi-objective design problem is resolved by maximizing the overall degree of designer preference: values are iteratively selected based on combining the degree of preference placed on them. Consider, however, design problems that exhibit multiple uncertainty forms (noise). In addition to degrees of preference ( imprecision) there are probabilistic uncertainties caused by, for example, measuring and fabrication limitations. There are also parameters that can take on any value it possible within a specified range, such as a manufacturing or tuning adjustment. Finally, there may be parameters which must necessarily satisfy all values within the range over which they vary, such as a horsepower requirement over a motor's different speeds. This paper defines a ``best'' set of design parameters for design problems with such multiple uncertainty forms and requirements.