METHOD AND APPARATUS FOR PREDICTING TURBINE OUTLET TEMPERATURE IN GAS TURBINE

A method for predicting the turbine outlet temperature of a gas turbine during future use on the basis of past usage records for the gas turbine, wherein parameters for environment/operation conditions scheduled for future use and a fan rotational frequency scheduled for future use are used as explanatory variables to predict the turbine outlet temperature during future use, as an objective variable, by means of a turbine outlet temperature model, and a coefficient for the explanatory variables is learned every time a learning trigger condition is fulfilled, thereby identifying the coefficient. During learning of the coefficient, the coefficient for the explanatory variables is identified on the basis of the result of regression learning of the explanatory variables and the objective variable of the turbine outlet temperature model performed using the aforementioned parameters, the fan rotational frequency, and the turbine outlet temperature during past use of the gas turbine.