Integration of Genetic Algorithm and Hedge Algebras in controlling mechanical machining robots

Abstract

Robot applications in mechanical processing have become popular. The critical issue when applying robots in mechanical processing is ensuring accuracy. Usually, robot control is based on dynamic models. This method has difficulty accurately determining the system's dynamic model because the robot has a complex structure. Besides, dynamic factors such as cutting force, friction force and machining conditions constantly change. Robot control based on Hedge Algebras gives excellent and reliable results. The critical factors that determine the quality and reliability of the Hedge Algebra controller are the Control Law, the method of Denormalization, and the determination of the Physical Value Domain. The construction of the Control Law and Denormalization is based on expert knowledge. Determining the physical value domain is problematic because it requires many experiments. This article introduces the method of applying genetic algorithms to find the appropriate physical value domain for the controller based on Hedge Algebras. The article presents a robot controller based on Hedge Algebras to do this. Numerical experiments with a mechanical machining robot verify the results.