The Numerical Simulation Of Head Bending In Medium Plate Rolling And The Study On Fuzzy Control

Head bending phenomenon in plate rolling is one of important technique problem in rolling field which hasn't be fully solved.During the process,it is good to see that the head is upward slightly.In the actual production,the degree of head bending is serious.Bending too much will cause the problem of low rate of products,product quality and production decline and etc.During the process of plate rolling,the primary cause of the head bending when the rolled piece come out of the roller is due to the uneven surface deformation of up and down surface.And the uneven deformation is caused by uneven deformation condition.There are lots of reasons would lead to uneven rolling.In the actual production,all those factors often affect each other and make the rule of head bending more complex.Technical personnel home and abroad have a long-term research in this field.But this problem still exists in many steel rolling mills and theory of which haven't been explained correctly.In order to analyse the problem of head bending,this paper uses the asymmetrical conditions of roller speed.As in all the causes,the condition of the roller speed and the temperature of plate is controllable and the speed condition is more stable than the other.Before confirm the influence law of the roller speed condition and the shape parameter of plate,we have ascertained the influence factors of other causes which were not easy to control.Then we proceed the numerical simulation of shape parameter such as the entry thickness of plate,reduction rate and speed ratio of down and up roller to confirm the rule of influence.In order to achieve real-time control of the head bending height in actual production,we workout the head bending control system which based on the simulation data used in the fuzzy neural network.Fuzzy control is to express a complex system which is hard to describe using the existing rules by inaccurate and fuzzy conditional statement.Fuzzy neural network is using the training ways of artificial neural network to train the input data after it is blurred.After the training,we achieve the mapping relation between input and output.The head bending controller of plate have three input variables,including speed ratio of down and up roller,entry thickness of plate and reduction rate.The output variable is the head bending height when the rolling length is 1 meter.The fuzzy rules can be generated after training the data obtained from numerical simulation.Then build the fuzzy control relation among the influence causes of head bending.Finally we got the error is less than 4.8099×10-5,which means that the system is feasible.