Research On The Head Bending Of Plate Based On Artificial Neural Network

Plate is an important product which is essential for the national economy development. In recent years, a lot of plate factories in our country have carried on renewal of their outmoded equipments to enhance their products'quality and function. However, following the upgrade of the apparatus some questions have appeared, one of the most outstanding problems is the bending on the heads of the work-pieces. It will cause such questions as diminishing productivity and reducing the products'quality and outputs.In essence, the reason of the bending on the heads of the work-pieces is that the rolling asymmetry of the upper and lower departments while rolling. The factors that cause rolling asymmetry is numerous, such as different temperatures between upper and lower surface of work-piece, different friction conditions on the surface of the upper and lower rollers , different linear velocity of the upper and lower rollers, different diameters of the upper and lower rollers and so on. In the actual rolling process, these factors often interweave together, make the head warping more complicated.Based on data measured in production scene, this text aims at the head bending problem in the plate production in Shaoguan Steel and Iron Group Company NO.2 Plate Factory and employs BP neural network to train the data measured, then sets up work-pieces'ends bending mathematical model. Using the built mathematical models, this article analyzes the factors which influence work-pieces'ends bending in production and finds out the relations between the influencing factors and head bending of the plate and offers the basis for formulation of the rolling craft system. Meanwhile, when the difference in temperature between the upper and lower work-piece, the difference in velocity between the upper and lower rollers, thickness of supplied materials and finished thickness and other conditions are clear in production, using these models can also predict the extent of work-pieces'ends bending, then adjust one or two parameters to control bending according to the predicted head bending. The research work of this thesis not only can be used in Shaoguan Steel and Iron Group Company NO.2 Plate Factory, but also can be used to solve the similar problem in other plate factory.