Research On Fault Detection And Diagnosis Method Of Hot Strip Mill Process

The hot strip mill process is an important part of steel production.The normal operation of the production process can guarantee the safety of production and product quality.For the fault detection and diagnosis of the hot strip mill process,accurate mathematical models cannot be established due to the complexity of the system.Therefore,the fault detection and diagnosis method based on multivariate statistical analysis is used.The method does not require a clear system structure and principle.It is only necessary to analyze the process data to determine the operating status of the system.Due to the influence of various factors,the actual measured hot strip mill process data is non-gaussian and nonlinear,while the independent element analysis method does not require the data to obey the gaussian distribution,so the fault detection based on independent element analysis method will be focused.By introducing a kernel function,the nuclear independent element analysis method was used to detect the fault of the hot strip mill process.The simulation results show that the fault detection accuracy based on improved method is increased by 35.26% on average.In order to improve the accuracy of fault detection,a wavelet packet combining improved kernel independent component analysis method is proposed.The wavelet packet method can remove the noise interference and provide more accurate data.By introducing the over-relaxation factor,the problem of random assignment of initial weights was solved to improve the stability of convergence.The simulation results show that the average false alarm rate of the improved method is reduced by 7.24%.Aiming at the problem of difficult diagnosis of nonlinear data in hot strip mill process,an improved kernel Fisher discriminant analysis method is proposed.The distance function was introduced to weight the class spacing to solve the problem of large classification gap that led to poor classification results.The simulation results show that the fault diagnosis accuracy of improved Fisher discriminant analysis method is increased by 6.38% on average.Figure 25;Table 4;Reference 55...