No abstract

Based on the time series, a new concept of spectrum representing the system state, called State Spectrum, is developed through the analysis of the varying principles of self-regression coefficients in this paper. The physical explanation of the spectrum, the meanings of the spectrum’s longitudinal and transverse coordinate and the way to use them in the state recognition and fault diagnosis of mechanical devices have been discussed in details. some experiments have carried out and the results indicate that the proposed new approach is very promising in fault diagnosis of mechanical devices.

In this paper, the application of the principal-component analysis method in fault diagnosis is explored. Characterized as fast and precise, this method can be directly used for analyzing gear noise and vibration signals in time domain. The principal component method and its1 error occur-ring in the calculation are theoretically discussed in detail. A program for implementing this method has been developed and the experiments for gear fault diagnosis have been carried out with satisfactory results.