The Application Of Genetic Algorithm And BP Neural Network Technique In LF End Temperature Prediction Of Molten Steel

In order to keep the continuous casting operation in order.the steel tapping operation with high temperature is often performed by many domestic steel Plants That is a very urgent problem to develop a good temperature system and accurately control molten steel temperature.It is an important that premise to predict accurately the end temperature of molten steel in ladle furnace for organizing production, improving steel quality, cutting down cost and controlling the temperature of steel water.The factors of influencing molten steel temperature of LF were deeply think about during building up the model of GA-BP newtork. See the whole steel the system as a system considering of the energy balance.Therefore, the factors used in model were confimred as follows:the supplying power to heat molten steel, initial temperature, ladle states,absorbing and emanating heat of metal alloy and slag, argon blowing quantity and sokas time. The model of pridicting molten steel terminal tenrperature in LF was established by using the artificial intelligence technique aimed at the production technique of 180t LF in the first steel Steel-making Plant at Benxi Iron and Steel (Group) Co.Using the MATLAB toolbox software,we can simulate the model and get a good network parameter. The GA-BP network was compared with standard BP network and improved BP network.The GA-BP program was developed by VC.The system sotfware with the complete function and the friendly interface was built up. The softwear has the function, of data editor, data output, the end temperature predicton.train the netwok system etc.The network was trained by 400 heats data in January, February and March,2009.The model was examined by 100 heats data in March,2009.The radio of the prediction error in the range of±5℃to all heats can reach 77%.