Experimental study of time-frequency characteristic of wavelet transform

A health monitoring and damage detection method using the time-frequency characteristic of the wavelet transform was presented based on the results of free vibration and random vibration tests of a system consisting of a cantilever plate and several springs. During vibration of the system, burning out the string connecting the spring to the plate resulted sudden damage to the structure. Discrete wavelet transform was used to recognize the damage at the moment it occurs. Subsequently, the shifts of frequencies and damping ratios were identified by continuous wavelet transform to study the amplitude of stiffness loss of the structure and make sure that the sudden change or impulse in the signal from the discrete wavelet transform was resulted from the damage but not the impulse noise. Test results showed that the proposed method is effective for health monitoring and damage detection of the considered structure. For the random forced vibration, the random decrement technique can be performed on the original signal to obtain the free decaying responses, and then the continuous wavelet transform can be applied to identify the structural parameters. The present method is free of modelling error and can be used for on-line health monitoring.