This paper employs artificial intelligence and machine learning techniques to predict the dominant oscillation modes in AC microgrids.The dominant modes are highly dependent on the droop gains and only slightly affect...This paper employs artificial intelligence and machine learning techniques to predict the dominant oscillation modes in AC microgrids.The dominant modes are highly dependent on the droop gains and only slightly affected by the loading conditions.This paper utilizes the least absolute shrinkage and select operator(LASSO)algorithm to extract the key features contributing directly to dominant modes.The adaptive neuro-fuzzy inference system(ANFIS)is employed as a nonlinear regression technique to train a model that relates the system’s key features to the dominant modes of the AC microgrid.The data obtained from a 6-bus AC microgrid test system is used to train the LASSO-based ANFIS model.The results show that the proposed method can substantially reduce the data volume of the training set due to LASSO sparse feature.The precision of the proposed algorithm is de-termined by comparing its output to the modes deter-mined by the derived small-signal model of the system.展开更多
基金supported by the Deanship of Scientific Research(DSR)at King Abdulaziz University(KAU),Jeddah,Saudi Arabia(No.GPIP:676-135-2024).
文摘This paper employs artificial intelligence and machine learning techniques to predict the dominant oscillation modes in AC microgrids.The dominant modes are highly dependent on the droop gains and only slightly affected by the loading conditions.This paper utilizes the least absolute shrinkage and select operator(LASSO)algorithm to extract the key features contributing directly to dominant modes.The adaptive neuro-fuzzy inference system(ANFIS)is employed as a nonlinear regression technique to train a model that relates the system’s key features to the dominant modes of the AC microgrid.The data obtained from a 6-bus AC microgrid test system is used to train the LASSO-based ANFIS model.The results show that the proposed method can substantially reduce the data volume of the training set due to LASSO sparse feature.The precision of the proposed algorithm is de-termined by comparing its output to the modes deter-mined by the derived small-signal model of the system.
