Turbine fault diagnosis expert system

The invention discloses a steam turbine fault diagnosis expert system. The system comprises the steps that real-time monitoring and data acquisition are conducted on the running state of a steam turbine; preprocessing the collected data, and selecting effective features related to fault diagnosis; determining possible fault types and corresponding characteristic modes thereof, and modeling the dependency relationship between the fault types and the characteristics of the steam turbine; learning the probability distribution of each fault mode from the data, calculating the probability distribution of each fault mode under the current characteristic, and carrying out the estimation of the basic probability assignment according to the probability distribution; and calculating the confidence coefficient of each fault mode according to the existing basic probability assignment and the current feature information, and fusing the confidence coefficients of various fault modes to obtain a final diagnosis result. Acco