Optimization of Magnetic Levitation System

In magnetically levitated system in certain applications like magnetically levitated train or bearingless motors the distance between the magnet and the levitated object has to be constant. The nonlinear system in such cases can be linearized and transfer function of the system is obtained. But a proper controller is required to manipulate the current through the magnet coil. The simplest controller to achieve this is the PID controller. But it was observed that as the levitated object deviated from the desired position, the gain of the PID controller set for one position did not hold good for some other position. Hence particle swarm optimization technique was implemented to determine the gains. Though optimal performance can be obtained for only one value of desired position y1d of the object, but acceptable performance can be there for a small range of values of y1d. PID with PSO is found to be easy to implement, has stable convergence and good computational efficiency.