Harmonic minimizing neural network SVC controller for compensating unbalanced fluctuating loads

This paper presents the development of an artificial neural network (ANN) controller for a static VAr compensator (SVC) system compensating unbalanced fluctuating loads. The proposed controller can balance and reduce the reactive power drawn from the source under unbalanced load conditions while keeping the harmonics injection to the system due to SVC operation low. First stage of controller development is a fuzzy logic based control algorithm. This algorithm involves determining a number of possible operating states of SVC for a given load condition and calculation of corresponding harmonics injections. Then a fuzzy logic system is used to rank those operating states in terms of the magnitude of the reactive power drawn from the source and the harmonic injection level indicated by total demand distortion (TDD). The operating point is selected based on the ranking score. Finally, the computational speed of the controller is improved by replacing the analytical and fuzzy computations by a set of neural networks trained with data generated with fuzzy logic based controller.