Method of gas turbine malfunction diagnostics using hybrid intellectual systems

Methods of improving the safety and efficiency of gas turbines are analyzed. The existing diagnostic methodologies and software for compression stations are reviewed. Lists of significant diagnostics parameters, typical faults and their causes are compiled. A model of a hybrid intellectual system of gas turbine fault diagnosis based on artificial neural network and fuzzy inference is introduced. The fuzzy inference method which allows processing an arbitrary number of intermediate variables and transitive relations is described. Moreover, the method makes it possible to take into account the inaccuracies in expert knowledge not only in describing the facts of an application domain but also in cause-and-effect relations between them. The implementation of the proposed model and the method in the system of taking decisions that makes it possible to reveal both the presence and character of the faults and their possible causes is described. The proposed system makes it possible to improve the accuracy and completeness of gas turbine diagnostics, which, in its turn, will increase the labor safety for the compression station staff as well as the promptness of repair and maintenance. Thus, the application of the system may have a positive effect on the service life and economic feasibility of gas turbines.