Find in Library

While diagnosing or locating the source of equipment fault, noise is often superimposed on the sampled vibration signal data. This study proposes an oversampling and two-stage autocorrelation model for signal de-noising. Simulation shows the effect of denoising is not proportional to the number of autocorrelator. Using this model, the useful feature information submerged in noise is extracted effectively, and the signal data are restored to the original vibration state as far as possible.