Abstract
The power system industry is one of the pioneers in implementing machine learning (ML) methods which have acquired a lot of attention across various facets of contemporary life. Low-frequency oscillation (LFO) is purportedly a non-threatened but slowly poisoning concern with electrical power networks that, if not adequately addressed in a reasonable timeframe, could appear as the cause of a total network failure. The function of a well-known member of the ML family, the adaptive neuro-fuzzy inference system (ANFIS), is proposed in this paper for optimizing LFO damping in real-time in the power system networks. It adopts two power system networks, one in which the synchronous machine is equipped solely with a power system stabilizer (PSS). In the second one, PSS relates to a second-generation flexible alternating current transmission system (FACTS) device named the unified power flow controller (UPFC). The well-known genetic algorithm (GA) supports the proposed ML algorithm in generating the dataset and training them well. The designed approach is examined using various statistical performance measures and well-recognized stability performance measures, such as the minimal damping ratio, eigenvalue, and time-domain simulation. The article also includes comparing and discussing the results from the reference works to make inferences on the viability of the proposed GA-tuned ANFIS technique for improving real-time power system stability.