Abstract: The dynamical principle of fault diagnosis is addressed, and the initial concept of nonlinear dynamical principle of mechanical fault diagnosis is introduced. The mechanical devices can be divided into two classes of systems, namely: possible-of-modeling systems and impossible-of-mo- deling systems. For possible-of-modeling systems, the analysis of fault mechanism based on the bifurcation theory may provide an instruction to mechanism, control and prediction of vibrating faults which are difficult to diagnose. The modeling method is described using the rotating machinery as an example. For impossible-of-modeling systems, using the chaotic dynamical theory, the phase space can be reconstructed in terms of the measuring vibration data, and the sources of generating faults can be justified through the form of singular spectrum which describes the energy distribution of the system. The studies on the nonlinear dynamical principle of fault diagnosis may significantly enhance the ability of our manufacturing industry of complicated devices and also make great contribution to the nonlinear dynamical theory.

Key words: Impossible-of-modeling systems, Nonlinear dynamics, Possible-of-modeling systems, Principle of fault diagnosis, Singular spectrum analysis