The Research On Galvanizing Production's Annealing Furnace Of Seam Tracking

With the development of national economy in our country,as anticorrosion plating products and economic material which is commonly used in the national economy,the dosage of the galvanized steel is more and more bigger.And if the seam tracking precision of strip steel galvanizing production line is too low,that will lead to production of waste and the increase of production cost.This thesis involves a steel galvanizing production line,whose construction time is earlier.Due to space limitation,we can't install advanced welding seam tracking device at the scene.To this problem,we need to combine galvanizing production line,by establishing the mathematical model for weld prediction,to realize the weld position accurately predict.Through mechanism analysis and simulation experiment,this thesis get a conclusion that relationship between various influence factors of weld position is complex which lead to multiple linear regression equation based on a simple linear relation is difficult to reflect them effectively.So it is necessary to choose a model which can be fitting of nonlinear relationship.In addition,because of the field data can be used for modeling is limited,so we'd better choose the model which can apply small model.Support vector machine(SVM)has good generalization ability.Under the condition of the statistical sample size is less,the support vector machine(SVM)can also obtain good statistical law.Therefore,this thesis's weld prediction model is established based on support vector machine,and the prediction accuracy of the model is tested by simulation.We found that the scheme of weld tracking accuracy can reach the requirement of the project.Particle swarm optimization algorithm(PSO)which has simple structure,is easy to implement and has good convergence rate.We may use PSO for SVM parameters optimization,in order to get more appropriate SVM parameters.Joining a priori knowledge to a mechanism can increase its "transparency".And then the model's generalization ability and forecasting accuracy can be increased.In this thesis,a priori knowledge is added to the weld in the model,the PSO is added for SVM parameters'optimization.On this basis,this thesis establish a mechanism of a priori knowledge +PSO + welding seams of the SVM prediction model.Then the more accurate prediction of the weld is got,and practical application scheme is put forward.The proposed scheme can effectively improve the accuracy of seam tracking.it is suitable for practical production.If the proposed scheme is applied to the practice,we will save the cost for the enterprise,and achieved good economic and social benefits.