Simultaneous Type and Dimensional Synthesis of Planar 1DOF Mechanisms Using Evolutionary Search and Convertible Agents (DETC2009-86722)

摘要

This paper introduces a novel approach to automated mechanism synthesis called "convertible agents." The evolutionary computing technique has been developed specifically for the unique design challenges encountered when synthesizing a mechanism for both type and dimensionality. Several case studies are presented, which demonstrate the approach 's effectiveness over earlier solution strategies. In these studies, six different planar single-degree-of-freedom mechanism types are considered: a four-bar mechanism, Stephenson's six-bar-mechanisms (types I, II, and III), and Watt's six-bar-mechanisms (types I and II). The synthesis technique selects the best suited mechanism type from this set and optimizes its dimensions to meet the design objective at hand. The method is readily scalable to account for any number of different mechanism types and complexities.