In this paper

we will try to identify the chaotic signals of the rub-impacting rotor system using learning vector quantization (LVQ) which is based on competitive learning and clustering analysis.The theory analysis and results of the computation are given.We have mainly studied the situations of the different classification when the response signals are in the different ratio of signal to noise.Our research shows that the LVQ neural network not only can identify rub-impacting chaotic signals but also work well for the signals with noise.So the network supplies a direct method for classifying such nonlinear signals of the rub-impacting rotor system and a possibility of practical application for a real rotor system.