Real time recurrent learning based algorithm for control of DSTATCOM

The real time recurrent learning (RTRL) based control algorithm is developed for the control of distribution static compensator (DSTATCOM). This algorithm minimizes instantaneous squared error between actual and estimated quantity at each instant during the process. It is a forward gradient learning method, because gradient information at some instant is forward propagated. The performance of DSTATCOM using RTRL based control algorithm is observed to mitigate power quality problems such as injection of harmonics into grid, reactive power, load unbalancing and load fluctuation in both power factor correction and voltage regulation modes. A hardware prototype of DSTATCOM is developed in the laboratory to verify performance of proposed control algorithm.