A Methodology Based on Robust Design and Optimization Statistics for Fitting of Parameters in Mechanical Systems

摘要

The development and analysis of mechanical systems exposes the designer to a series of unknown parameters from several sources such as material properties, environmental and operational conditions. Therefore, the qualification and quantification of these inherent sources of design uncertainties become very important in several aspects in the context of design development and so, a system is reliable and robust if it allows a certain range of uncertainties before the first failure occurs. With this in mind, we propose here the development of a methodology that can be identified the sources of uncertainties and parameters that largely influence the whole design. An initial study focuses on a simple oscillatory system that consists of a mass, a spring and a damper. The first step was the choose the element or mechanical system and in choosing the experiment design for identifying the critical parameters (factorial designs or fractional designs). This led to the development of polynomial models (linear and quadratic) that fit the experimental results from factorial/fractional designs. Once the critical parameters are obtained there is a search for optimum regions maximum, minimum or singular point. The steps used in the search interval occur along maximum or minimum lines that describe a region of interest or experimentation. A sensitivity analysis also takes place using canonical analysis and for parameter fittings can be used optimization constrained methods. Thus are obtained confidence limits for parameters through the reliability concepts with respect to the critical parameters in the robust design concepts. In the future extended to other applications such foundation structures and rotor-bearings systems.