Design of online supplementary adaptive dynamic programming for current control in power electronic systems

Large-scale penetration of power electronic based systems demands the need for sophisticated control algorithms. Conventional current control techniques may not give the required dynamic performance for modern power electronic systems. This paper designs an online supplementary adaptive dynamic programming (ADP) controller based on neural networks which improves the dynamic performance of a conventional proportional-integral (PI) current controller. A case study with a 1 kW shunt active filter (SAF) was developed and simulated in MATLAB/Simulink environment to verify the investigated controller. Compared to an SAF with a conventional PI controller, the proposed approach obtained a better dynamic response for the current control loop, and thus showed a reduced total harmonic distortion (THD) of the source current from 4.24% to 3.64%.