Simulate The Heating Stress Of ROF Bottom Plate And Optimize The Structure

Rotary Annealing Furnace is used for high temperature anneal of Oriented Silicon steel. ROF bottom plate is one of the main consumable parts for the ROF. It bears the heat treatment work pieces (Oriented Silicon Strip) in the process of high temperature anneal. The service life of bottom plate of annealing furnace is connected with the circular stress. By means of finite element method,the values of temperature and distribution of heating stress are given in the paper. In the process of high temperature anneal, the biggest heating stress is produced in the fringe, which dues to the asymmetry temperature of the bottom plate .Draught stress is the ultimate cause of the crack. In the process of cooling , the fringe of the bottom plate bears the draught stress, at the same time the limit of intensity is lower under the effect of high temperature . The bottom plate crack easily.Under the same load of temperature, the biggest heating circular stresses are simulated of different structure parameters and then search the optimal parameter. The wave fringe and the reduction of the radius of pole can decrease the heating stress in the fringe. Compared the heating stress in different structures of furnace bottom plate, the optimal parameters are obtained. the inner radius of wave is 0.440m ,the extra radius is 0.6m,the number of wave is24 and the radius of pole is 0.286m. The ratio of the biggest heating circular stresses is 0.948. So the improved effect isn't big.Then a new structure is given. This bottom plate can release the stress to reduce the circular stress. By comparing of different structures, the effect of the parameter is very complex. BP network bases on erro back-promulgated algorithmic and it is Multi-layered Feed-forward Neural Network. It is one of the most successful manual neural networks. It needs no test formula and can conclude regulation automatically.It has the strong ability of non-linearity mapping. The text uses the BP network which has three layers to build the regress model and the optimal parameters are 12, 0.0825 and 0.3775. The ratio of the biggest heating circular stresses is 0.6198, so the improvement is evident.