The Design Of Breakout Prediction System For The No.4Slab Caster In Tianjin Iron&Steel Co. Ltd

Breakout is the most detrimental event than others events happened in continuous casting. To avoid or decrease breakout, it need to research the form theory on sticker breakout deeply and to develop the breakout prediction system of continuous casting. During the caster producing, the breakout prediction system can detect the breakout symptom in the mold, then to take certain measures such as reducing the casting speed. The thesis have studied deeply on the slab sticker breakout, and the contents are as follows:Firstly, the breakout prediction model of slab continuous casting is designed to solve the breakout of slab caster No.4in Tianjing Iron and Steel CO. LTD, then we carefully researched and analyzed the type of breakout based on the labour of the Sticker-type breakout which is the most common type of breakouts from mechanism of the breakout in detail.The software and hardware and communication systems and so on are all included in the design of the breakout prediction system based on the principle of thermocouple temperature measurement in the slab caster No.4. Especially, it have been particularly descripted that the sampling data from thermocouple communicated to the manage computer and also displayed in graphical form on each client operator station in the breakout prediction system.On the basis of analysing and researching the spread abroad speed mode of the sticker-type breakout, we present the breakout prediction model based on logic. The breakout is juided by the warp of temperature, the changing speed of temperature, the changing delay of temperature, the temperature dropped off in begin time and so on.In analyzing the prediction model based on logic judgments, and analyzing the characteristics of BP neural network and its learning algorithm, and identifying temperature changing characteristics of thermocouple during breakout, a prediction model based on BP neural network was established. The results of the simulation show that the prediction model based on BP neural network can predict more effectively and quickly than the original system in the plan which base on logic judgments.