Method for predicting temperature of armature winding of hydraulic generator

The invention discloses a hydro-generator armature winding temperature prediction method, which comprises the following steps of: establishing a regression model by adopting an artificial neural network, and carrying out space-time correlation modeling on the change of historical temperature data at previous and later moments; according to the method, fitting association and data mining in statistical significance are carried out on continuously-changing temperature data with a relatively complex generation mechanism in a large-scale system, and a temperature model of a hydro-generator is established through a finite element analysis method to carry out analogue simulation on heating temperatures of monitoring points of a resistor winding in physical space. And finally, the corresponding temperature of the fault point of the armature winding sensor is predicted through comprehensive modeling of the simulation data set and the historical data set of the hydro-generator. The temperature detection problem that o