Impementation of Self-Learning ANN-ANFIS Based Dynamic Voltage Restorer for Voltage Power Quality Improvement

This paper proposes a novel metaheuristic design of an adaptive control technique (ACT) to enhance the compensation capability of the Dynamic Voltage Restorer (DVR). The proposed Artificial Neural Network (ANN) technique is an implementation of optimized architecture which is used for the fundamental extraction and Adaptive Neuro-Fuzzy Interface System (ANFIS) is utilized for voltage regulation under dynamics. The drawback of traditional system SRF-PI has manual intervention for tuning and mathematical modeling. The conventional SFR-PI based DVR is not a failsafe mechanism for supply voltage variation conditions. The proposed controllers, based on the main principle of ANN-ANFIS, overcome the problems of SRF-PI by working in collaboration with each other which significantly increases the tracking ability and robustness of the system uncertainties. In the proposed design of DVR, ANN interconnected weights are optimized by Corona Virus Herd Immunity algorithm for forecasting the fundamental weight quantity and ANFIS is optimized by Hybrid learning algorithm and gaussian membership functions are applied to obtain the best fitted ANFIS models. The model performance is evaluated by employing all the statistical indices such as Mean Square Error (MSE), Root Mean Square Error (RMSE), Mean Error (ME), Standard Deviation (SD) and Regression (R). ANN controller is used for fundamental extraction of load voltage components whilst ANFIS controllers are used to regulate DC and AC voltage link. The controllers are self-learning by adapting weights and bias of neurons to handle the complexity and uncertainty in non-linear system. Compared to the conventional control strategies, ANN-ANFIS based DVR performs better by improving the output voltage profiles and producing less THD of 2.13%. The simulation and experimentation results are discussed to measure the efficacy of proposed control scheme.