Bof Endpoint Based On Artificial Neural Network Prediction Model

A new method to control end-point of basic oxygen furnace (BOF) is put forward.This method is based on the furnace flame light intensity and image information combinedwith artificial neural networks. The experimental system is established, including lightintensity collecting system, flame image capturing system and the data analysis andprocessing system. The hardware and software systems for is tested and used in the fieldexperiments. The experimental results show that some characters of light intensity andflame image off mouth of furnace are viewed in the process of steel-making.Further,the BP(Error Back Propagation) Neural Network Prediction Model based on the flamelight intensity and image information are built. The BP neural network is well trained bysamples. The end-point time of basic oxygen furnace (BOF) can be predicted by the BPneural network based on the flame light intensity and image information. The predictionresults meet the demand theoretically. Therefore, The results of experiment show that theBP(Error Back Propagation) Neural Network Prediction Model based on the flame lightintensity and image information can be used to predicted the end-point time of BOF.