Modelling and design of shunt active power filters using genetic algorithms

One of the main issues when designing a control strategy for a power electronic system is the development of a reliable model of the real system. It ensures that a simulation presents the same behavior as the real system, providing a useful instrument for control design. Consequently, the designer would avoid wasting time manually adjusting parameters of controllers which have achieved good performance in simulation but poor performance in the real system. However, the evaluation of the actual plant parameters is always a hard job for the designer due to the mismatch between rated and actual values and the presence of unmodelled dynamics. This paper presents a technique for the optimization of the parameters of the simulation model of a shunt active power filter system using genetic algorithms (GAs). The optimization process is based on experimental data measured during the normal operation of the real system. After this, the identified model has been used to optimize in simulation the controllers of the shunt active power filter. Experimental results prove that the optimized controllers greatly improve the system performance