A Comparison of the Fitness Functions to Identify the Motor-Table System: Simulations and Experiments

Abstract

The different state-error fitness functions (FFs) are proposed and compared numerically and experimentally to identify a motor-table system by using self-learning particle swarm optimization (SLPSO). Firstly, the completed mathematical model containing both of mechanical and electrical equations is successfully formulated. Secondly, the FFs containing different state-errors are compared by using PSO and SLPSO to identify the unknown parameters. It is found that the identify performance of the SLPSO algorithm by using FF with full-state error of displacement, velocity and current is the best than the other methods. Thus, the FF with full-state errors is adopted in experiments for a real mechatronic motor-table system. Then, the unknown parameters are successfully identified by the SLPSO algorithm. The contributions of this paper are: (1) the more states of the system are measured and used in the FF, the more parameters of system are accurately identified by the proposed identification approach, (2) the FF with full-state errors is performed in a real mechatronic motor-table system, and the unknown parameters are successfully identified by the SLPSO algorithm in experimental results.