This work describes the potential and the flexibility of online controller optimization strategies, using computational hardware in master-slave configuration, with use of the xPC Target tool. This is an inexpensive alternative to the use of DSP solutions. The control algorithm runs on the slave PC, while the real time optimization runs on the master PC, which modifies parameters in the slave to optimize the dynamic response of the controlled plant. To illustrate methods and procedures parametric optimization strategies without derivative use were adapted and compared in the scope of a magnetic levitation control system. Special attention was given to techniques to speed up performance function calculation and to prevent (temporary) instability of the controlled system.
control; host-target environment; magnetic levitation; numerical methods; parametric optimization; xPC Target
