Mechatronic system parameter identification via genetic algorithms

This thesis develops a novel way to identify both the joint friction parameters and a built in torque sensor gain and offset. The identification method is based on a genetic algorithm (GA). A model based friction compensation method and a real coded GA are selected from a variety of methods available. A model of a single degree of freedom mechatronic joint with a link is presented. Numerical simulations are run to determine the optimum configuration of the GA with respect to the population size and maximum number of generations necessary to identify the parameters to within 5% of their actual value. The GA identification technique is then used on an experimental mechatronic joint with a harmonic drive and built-in torque sensor. The friction parameters as well as the sensor gain and offset are identified in the experimental system and the position tracking error is reduced. Based on the experimental results, the method is found to be an effective way of identifying system parameters in a mechatronic joint.