Factor Analysis And Feature Extraction On Rolling Mill Chatter

The trouble of the rolling mill chatter is a universal problem in the field of metallurgy, and it is also a desirable problem. The control of mill chatter is playing an important role in improving the strip flatness and productive efficiency.The clue of this paper is mill chatter, and it analyzes the reasons of mill chatter and brings forward improvement measures.It studies on the mill chatter based on wavelet packet analysis and grey cluster, gets the fault classification.Based on fractal theory, using the fractal dimension shows the state of the mill chatter.The studies contents on mill chatter is summed up, analyzes fault reason of the vertical vibration of the mill and rollers. It focuses on study of chatter phenomenon and analysis, analyzes emulsion factor, and putts forward the improved measures of mill chatter control.Based on wavelet packet analysis, the feature extraction method of energy distribution is studied by chatter signal.Mill feature extraction is obtained by this method. The mill chatter signals are mainly non-stationary signal, therefore this paper introduces wavelet packet analysis as a tool to analyze non-stationary signal, which is better than wavelet analysis. This paper obtains a good recognition by using grey cluster to classify the results of the wavelet packet analysis.It calculates the fractal dimension of mechanic system, by using phase space reconstruction of the time-domain sequence. Monitoring the run-state of the equipment by using fractal dimension is putted forward at this time. Nonlinear science, such as fractal theory, has been applied into fault diagnosis in mechanism science. Fractal theory has application in mill chatter analysis in this paper. And it also focuses on the study of parameters selection in correlation dimension of fractal dimension, such as delay time, embedding dimension, no-scale interval, etc.