Computing the dynamic AC of an electrical network via a fuzzy adaptive recurrent neural network

• Published in: ITM web of conferences Vol. 72; p. 03003
• Main Author: Mourtas, Spyridon D
• Format: Conference Proceeding Journal Article
• Language: English
• Published: Les Ulis EDP Sciences 01.01.2025
• Subjects: Alternating current; Convergence; Electrical networks; Parameters; Recurrent neural networks
• ISSN: 2431-7578; 2271-2097
• DOI: 10.1051/itmconf/20257203003

The convergence and durability of zeroing neural networks (ZNN), a special family of recurrent neural networks, have been the subject of much recent research. Numerous time-varying problems in science and engineering have been successfully solved by ZNN dynamics. An improvement of the ZNN design for calculating the dynamic alternating current (AC) of an electrical network, which is a specific time-varying linear matrix equation problem, is proposed in this paper by utilizing a suitable defined neutrosophic-logic system (NS). In particular, the gain parameter in the ZNN architecture can be dynamically adjusted over time to accelerate the convergence of the ZNN model using an appropriate value that is acquired as the outcome of an adequately built NS. The results of the application demonstrate that the NS-based ZNN model defines the varying-gain parameter more effectively than the corresponding standard ZNN model.