基于小波分解的燃煤电厂磨煤机磨辊检测研究OA

In order to predict the wear degree of the grinding roller in the operation of the coal mill,this paper uses the wavelet signal processing technology to decompose the unit consumption signal of the coal mill.The low-frequency unit consumption signal reconstructed by this method can initially reflect the wear condition of the grinding roller.Further-more,in order to solve the problem that the reconstructed low-frequency unit consumption signal may fluctuate within a certain range due to the influence of the working conditions of the coal mill,and based on the fixed threshold of the pre-diction model established by this method,the principal component analysis method is used to screen the coal mill vari-ables closely related to the wear of the grinding roller.Based on the normal operation data,an adaptive neural network benchmark model for reconstructing the unit consumption signal is established.The simulation results show that after us-ing the adaptive threshold,compared with the fixed threshold,the accuracy of the model increases from 91.3%to 97.4%,the false negative rate decreases from 9.4%to 1.5%,and the average detection time decreases from 413 s to 230 s.It shows that the adaptive neural network method can effectively characterize the wear state of the grinding roller,effectively monitor and warn the reconstructed signal,increase the accuracy of the alarm,and reduce the false negative rate and detection time,thus providing a more reliable and efficient method for the predictive maintenance of the grind-ing roller of the coal mill.