Blind system identification and fault diagnosis

Blind system identification (BSI) is a fundamental si gnal processing technology, which retrieves a system's unknown information from it s output only. The basic idea of blind system identification and its algorithms are discussed. The method of machine fault diagnosis based on BSI are presented . Through experiment research of a cracked rotor, the bispectrum characteristi cs of the cracked rotor with different crack depth and crack location were inves tiga ted using blind identification method of time series model. Some valuable concl u sions were obtained, and a new approach for the diagnosis of cracke d rotors was provided. The experiment results showed that this approach is very efficient. Th e bispectrum analysis based on blind identification of time series model is sens itive to fault; displays the first, second and higher harmonics of the signal, a nd describes nonlinear phenomena such as second phase coupling.