Abstract
Using a new real-time signal analysis, any signal drawn from motion and vibration may be expressed in time series by a complete set of orthonormal bases with their amplitudes to reflect the signal's dynamic characteristics, such as velocity, acceleration and power. Expressing the signal dynamics from a low- to a high-frequency region becomes possible while observing the dynamic similarity expressed in power. These bases are really wavelets, whose usage is different from that in wavelet analysis. They can reconstruct the state or phase spaces in which to study the system dynamics. The bases then become various analyzing windows to observe the dynamic signal comprising many events localized in time and frequency. As a simple example of its real-time use, some fundamental analyses are first shown using the signals from an electronic oscillator. Then how an abnormal running condition of vehicles on wheels may be diagnosed is discussed. Its simulation test was done with a small mechanical rotor rotating at about 3000 rpm in an axial air flow.
