A Synthesis of Planar Multilink Mechanisms Based on Genetic Algorithms.

Abstract

This paper presents a method of synthesizing planar multilink mechanisms which can generate desired planar paths. In previous studies, various optimization methods were applied to synthesize those path-generation planar mechanisms. The techniques used were often very cumbersome and computationally costly to apply. Furthermore, most of the former methods devised ad hoc techniques for each individual design or application. These problems are mainly due to the highly nonlinear nature of synthesizing mechanisms and the existence of many local solutions. Genetic algorithms have recently been noted as a good method of solving nonlinear problems. This method can search for plural approximate answers simultaneously and widely. Furthermore, the method does not require differentiating objective functions. Namely, this method can be applied to various problems. From the above point of view, a new method for synthesizing planar multilink mechanisms is presented based on genetic algorithms. The present method does not require initial mechanisms and can search for plural appropriate mechanisms simultaneously. The programs can be easily made because the method does not involve complicated calculations. Furthermore, it can derive appropriate mechanisms and be applied to various synthesizing problems. This paper particularly denotes a closed curve adapting synthesis such that the objective values are given as a closed loop curve. As an example, configurations of 4-bar and 6-bar planar mechanisms are decided in a practical application.