Decision Support System Based On The Prediction Model Of The Converter Steelmaking

LD-process is a very complex process with multi-element,multi-phase,and high-temperature reaction.It includes periodically warming,carbon reducing and impurities eliminating.It is very important to forecast the carbon content and temperature at end point accurately,for it is the prerequisite of reasonable production organization,the quality of the molten steel improvement and the cost reducing of steelmaking.Neural network is an effective way to solve the problem of nonlinear systems.It worth to make the research with neural network on the control of carbon content and temperature at end point.This paper is based on the productive practice and the datas of the converter.It established the auxiliary decision-making system of the steel-making according to BP neural network method to help determine the final program.The prediction model was build for achieving the prediction function.The data-processing effort is enhanced by analyzing the characteristics of the data source. According to the laws corresponding with the different datas,it splits the train datas to two cross sections and build the corresponding classification forecast model.The simulation shows that the network capacity was enhanced significantly by this method.Meanwhile, it analyzed the weakness of the standard BP algorithm and used the improved algorithm—LM algorithm to meet the needs of online learning in this system.For improving the generalization ability of the prediction model,it tried a varieties of ways.This paper build the multi-objective optimal control models of steelmaking control at the end-point,for meeting the controll variable optimization and recommendation. Based on the full analysis of the extreme region of the controll variables,it improved the search speed by introducing the batch computing under the premise of ensuring search quality.By tested with the historical data and the online data,the result shows that the system runs a good performance and play a supporting role in auxiliary decision-making of the LD-process.It could improve the hit rate of the end-point carbon content and temperature...