THE USE OF EVOLUTION ALGORITHMS FOR IDENTIFICATION OF TRANSMISSION BACKLASH

Abstract

Mechanical propulsion system including, inter alia, an electric motor, shaft and transmission is characterized by certain parameters whose values cannot be calculated using mathematical methods (transmission backlash) or may vary during operation time. Such parameters have to be known to ensure proper operation of the system. In this article a method for identification of these parameters is described. The method is based on an evolution algorithm. This algorithm utilizes natural selection methods and natural transformation processes within a population, similar to those in animal world. The least-fit individuals are rejected by applying the fitness function. The population will improve, and in infinite time it will be perfect. That means that the perfect solution can be attained in a very long time. This time can be shortened by defining the termination criteria, e.g. the identification can be stopped if one of the population individuals has a very low error. Such identification requires a transition function (model), as well as reference measurement made on a real system.