Temperature sensor fault diagnosing in heavy duty gas turbines using Laguerre network-based hierarchical fuzzy systems

This study present an application of Laguerre network-based hierarchical fuzzy modeling approach in fault diagnosis of the temperature sensors in industrial heavy duty gas turbines. The recorded experimental data from the performances of a V94.2 gas turbine unit were employed in modeling stage. A comparison between the responses of the models and real data indicate the capability of the model for long-term prediction of the turbine outlet temperature at different operating conditions. The differences between the models and measured values were defined as the residuals. To deal with uncertainties and disturbances, the thresholds bounds were considered for the residuals. The residuals deviations with respect to threshold boundaries yield to symptoms, which were analyzed in a Takagi-Sugeno fuzzy inference expert system. The performances of fault detection and fault diagnosis system were evaluated by subjecting the sensors to faults. The obtained results show that the faults are successfully detected and diagnosed.