Aiming at the problem that the equipment in power plant are complex and difficultto extract their main fault features and diagnose their faults, an approach of faultdiagnosis based on nonlinear principal component analysis (NLPCA) neural networksand probabilistic neural networks(PNN) is presented in this paper. At first, the nonlinearprincipal component analysis neural networks is employed to extract main features fromhigh dimension patterns in order to simplify the diagnosing process and ensurediagnosing accuracy as well. Consequently, the probabilistic neural networks is utilizedto obtain the final diagnosed results. The proposed scheme is applied to diagnose thefaults in the rotor bearing system and condenser system of turbo-generator unit andcompared with other schemes, the diagnosis results show that it is effective and easy tobe put into practice.
|