Process Monitoring And Quality Prediction For Perforation Production Of Seamless Tube

As the first process of seamless pipe,the perforation process has a very important influence on the quality of the steel pipe.Perforation process is a complicated process which has a number of the process variables.And the relationship between the variables and the quality of steel pipe is complex.So its mechanism model is difficult to establish.In this thesis,the process fault monitoring and diagnosis model and the quality prediction model is established,based on the data of the SWW mill of Baosteel Tube Branch.Finally,a finite element model of the perforation process is established by using the actual parameters of the SWW punch.Perforation process is a typical batch process,which has many characteristics such as multi-stage,multi-variable and high-coupling.So we can analyze perforation process by the modeling method of batch process.Based on the traditional MPCA method,in this thesis,the multi-stage MPCA model is established considering the multi-period characteristic of the perforation process.We divide the stages by process,K-means algorithm and genetic algorithm,respectively,to establish the multi-stage MPCA model.Simulation results show that the traditional MPCA model is not good for the perforation process,it needs improvements.The multi-stage MPCA model is fit for the multi-stage feature of the perforation process.The monitoring effect of multi-stage MPCA model based on optimized segmentation is much more accurate.In this thesis,the ELM ensemble learning algorithm is used to establish the quality prediction model of the perforation process.Three ELM ensemble learning approaches are:(1)the mean of ELM models;(2)the mean of stochastic optimized ELM models;(3)the mean of ELM models of new activation function after optimizing weighs by GA.Then prediction model is established by the approaches.Simulation results reveal that the prediction error of model(2)and model(3)is lower than model(1).And the prediction error of model(3)is much lower,which can predict longitudinal wall thickness of capillary.Finally,the finite element analysis model of perforation process is established based on the finite element theory and LS-DYNA software.Through the finite element model we can emulate practical condition of field production,study the mechanism theory of perforation process further for relationship between the parameters and the quality of capillary.