Optimizing Gas Turbine Generator data in Uncertain Environments: A Fuzzy Approach

Abstract

This paper describes the use of a new kind of fuzzy logic system namely, a Takagi-Sugeno-Kang (TSK)-based interval type-2 intuitionistic fuzzy logic inference approach with artificial neural network learning for the prediction of electricity generation of a gas turbine engine. Fuzzy logic systems have been used widely to solve many real world application problems because they are found to be universal approximators. The gradient descent algorithm is used for the optimization of the parameters of the interval type-2 intuitionistic fuzzy logic
system membership and non-membership functions. The capability to optimize the parameters of the fuzzy logic systems increases performance and robustness of the engine and leads to increased engine production and reliability thereby reducing system cost and mitigating environmental risks. In this study, we investigate the performance of an industrial power plant gas turbine used in oil fields to produce power especially for plants that are far away on oil fields and offshore sites where there is no possibility to connect to the general electricity network. Analysis of the gas turbine data is carried out using two types of fuzzy logic systems. Specifically, we evaluate the performance of interval type-2 intuitionistic fuzzy logic system, a fuzzy logic system that enables hesitation with fuzzy membership and non-membership functions with the classical interval type-2 fuzzy logic system (with no hesitation) and compare results. The root mean squared error is used as the performance metric and comparison made between actual data and predicted values. Results of analysis demonstrate how promising the combined method of intuitionistic interval type-2 fuzzy logic systems with hesitation indices is as compared to the classical interval type-2 fuzzy logic system with no hesitation in this application domain.