An Experimental Comparison of a Model Based Controller and a\ud Fuzzy Logic Controller for Magnetic Bearing System Stabilization

In this paper, the structure and performance of a model based controller have been compared experimentally with that of a fuzzy logic controller (FLC) for stabilization of a MBC500 magnetic bearing system. The model based controller was designed on the basis of a reduced order model of the MBC500 magnetic bearing system. In this approach, notch filters are necessary to cancel the resonant modes of the active magnetic bearing system. The FLC uses error and rate of change of error in the position of the rotor as inputs and produces an output voltage to control the current of the amplifiers driving the magnetic bearing system. In this approach, no analytical model of the MBC500 magnetic bearing system was required. This can greatly simplify the controller design process. Furthermore, the FLC can stabilize the magnetic bearing system without the use of any notch filters. In the experiment, both the model based controller and the FLC have been coded in C and implemented in real time via a Digital Signal Processor (DSP) card. Experimental results have demonstrated that, despite its simplicity, the FLC has higher disturbance rejection performance and better robust stabilty than the model based controller. This leads to the following open question which we believe is yet to be answered: Why does the FLC possess the common characteristics of high authority linear controllers without at the same time having the low stability robustness usually exhibited by such linear controllers? We believe an answer to the above question may be the first step to better controller design methods for stabilizing active magnetic bearing systems. Conference held: 9-11 December 2009, Christchurch, New Zealan