On-line Monitoring and Fault Diagnosis of Air Gap and Stator Bar Vibration of Large-scale Hydro-generator by Using RBf

摘要

In this paper, the mathematics' model of RBF Neural Network that is special cases of FLANN is investigated. FLANN is that unknown functions are expressed by a combination of a series of base functions φ_i(x) that consist of input variables. The 1# generator air gap and stator bar vibration of Ge Heyan Hydropower Station are measured on-line using the ZOOM system and SBV system of VibroSystem Inc. The stating graphs of ZOOM system and SBV system include Signature, Pole, Sample, and Trend pictures of generator air gap and stator bar vibration. The fault trees of generator including 15 fault eigenvectors are generated. The simulation of RBF is launched by function simurb. The training of RBF network is carried on with two functions: solverbe or solverb.