Skip to Main Content
As the convergence speed of the back-propagation neural network (BPNN) usually is slow and the information from the sensor is too much, if BPNN is used in the fault diagnosis of the steam turbine directly, and the training time will be long. Rough set theory can make it up better by getting rid of redundant information. In this paper, rough set theory is used to deal with the input data of BPNN, the input dimensionality of the neural network and the training time are reduced, and the convergence speed is picked up too. Finally, the method is applied in fault diagnosis of steam turbines and good results are achieved.