The Research Of Vibration Signal Analysis And Processing Of 300 Cold Rolling Mill Based On Wavelet Transform

Because the rolling mill is the most critical parts in strip rolling production line, and the working roll of rolling mill is the main working parts when rolling mill is carrying on rolling, so once the working roll appears wearing occurs, will seriously damage the thickness accuracy of plate strip products. If appears a significant event may hurt the personal safety of staff. Causing incalculable losses to the enterprise. So it is very necessary to analyze and processing the vibration signals of the working roll to determine the characteristics of the working roll of rolling mill.This paper based on the theory of wavelet transform, in 300 four high reversible cold-rolling mill as research object, using the method of wavelet basis adaptive selection based on genetic algorithm, analyzes and process the signal data collected.Obtaining the wavelet basis function matched the signal to denoise for working roll vibration signal and extract and separate feature inform ation for working roll vibration signal.Firstly, this paper by using the improved threshold function proposed reduced noise of the simulation signal. With the hard threshold function of the reduction noise effect and the soft threshold function of the reduction noise effect by comparison, it is concluded that the superiority of the threshold function. For decom position layers,reducting noise effect of another important factor, this paper use the theory of singular spectral and modulus maxima theory respectively to choose the adaptive selection of decomposition layers, and then with SNR calculation for its verification, it is concluded that the rationality of the algorithm. In the application of adjacent wavelet coefficients to reduce noise, this paper found that for different parameters, the the noise reduction effect is different. So putforward the parameter adaptive adjustment of adjacent wavelet coefficients for noise reduction. At the same time, its rationality was verified. Then let the method combine with the wavelet threshold shrinkage joint to reduce noise.It is conclude that the joint of the noise reduction result is better than either one of two noise reduction alone.Secondly, needing to choose the wavelet basis function matching the signal data, because only use the wavelet basis function similar to signal data, we can be to put forward the characteristic vector of signal data more effectively. This paper are based on the survival of the fittest principle of genetic algorithm, using wavelet filter parameter equation,and determine the highest fitness value individuals in the form of chromosome. So based on the param eters of the chromosome, calculating the wavelet filter coefficient of the wavelet basis function. Then Wavelet basis function is analyzed in the noise reduction for working roll vibration signals and used to extract and separate for the vibration of the working roll signal feature. And comparing with db3 wavelet basis function, from the results we can be concluded that the rationality of wavelet base.Finally, in the vibration test and signal analysis laboratory of teacher Dong, using high precision data acquisition instrument to do vibration testing experiments of working roll of rolling mill and collectvertical direction vibration data of working roll, processing the collecting data,then extracting characteristic vector to characterize the status of the working roll. Then using the theory of SVM to calculate classification decision function m odel, and testing the generalization ability of the model, we can be concluded that the rationality of the model.