Shunt active power filter with enhanced dynamic performance using novel control strategy

To realise the accurate and real-time compensation of the harmonics of a power system, this study proposes a novel control strategy for shunt active power filter (SAPF), adopting the composite strategy of neural network proportional-integral (PI) control and dual-repetitive controller (DRC). In DRC, one repetitive controller is used to ensure current tracking accuracy and the other one is used to enhance dynamic response. The neural network PI control is adopted to improve response speed by turning the PI parameters adaptively and setting optimisation parameters online. Through the optimisation of this approach and the rational configuration of the hardware and software systems, the SAPF could achieve rapid dynamic response with high tracking accuracy and excellent compensation performance. A 100 kVA industrial prototype active power filter has been developed by using this control strategy. System compensation could effectively reduce the total harmonic distortion values from 26 to 27.7%, 27.6 to 4.25% and 4.57 to 4.35% for each phase of the current. The full response time of the system is < 10 ms.