A Comparative Modeling Study of Gas Turbine Using Adaptive Neural Network, Nonlinear Autoregressive Exogenous, and Fuzzy Logic Approaches for Modeling and Control

This paper focuses on the identification and modeling of gas turbine dynamics, specifically those used in power generation plants. The approach utilizes experimental data and employs fuzzy reasoning systems. The resulting model serves the purpose of approximating nonlinear gas turbine systems and ensuring reliable system control. By incorporating uncertainties associated with human reasoning, such as fuzzy systems based on Takagi-Sugeno reasoning, it is possible to achieve highly reliable control systems. The primary objective of this work is to develop an effective monitoring system by employing nonlinear identification techniques, namely fuzzy systems and neuro-fuzzy systems, based on real-time on-site experimental data. Additionally, the proposed identification approaches are evaluated through a comparative study, where the results obtained using the nonlinear autoregressive exogenous neural network (NARAX-NN Modeling) technique are compared with those obtained using the ANFIS approach. The obtained results further facilitate the comprehension and analysis of the nonlinearities present in these complex systems, ultimately aiding in the prediction of their dynamic behavior.