Study of signal of roll eccentricity based on wavelet thresholding de-noising

With the ever-increasing demand for the quality of steel plate and strip products, roll eccentricity has become an important factor affecting the product quality. In this paper, the roll eccentricity signal is modeled based on modern signal processing theory. As the data gained from the sensor contain interference, the wavelet threshold de-noising method is developed. By using wavelet analysis, the step in the signal can also be detected. The signal preprocessing is realized. The simulation results to find the optimal threshold of quantitative methods and the most suitable wavelet, to achieve a high-quality signal de-noising. The method has been applied to the scene in the actual production, far superior than the traditional method of noise filter, rolling greatly enhance the accuracy.