Research On The Dynamic Process Modeling Of BOF Steelmaking Based On Entropy

Dynamic process is an important step in the process of BOF steelmaking, whose control effect is the direct decision that whether the endpoint temperature and carbon content can meet the process requirements. The dynamic process of BOF steelmaking mainly consists of calculation of blowing oxygen, calculation of the amount of refrigerant and the endpoint forecast of temperature and carbon content. Because the BOF steelmaking process is a complex, nonlinear, time-varying delay industrial process, a thorough analysis of the mechanism of industrial process and the establishment of a precise mathematical model become increasingly difficult. In order to establish an accurate dynamic model, research work related to information entropy method is carried out in this paper as a background of the dynamic process of BOF steelmaking, which is applied to the calculation of blowing oxygen, the calculation of the amount of refrigerant and the endpoint forecast of temperature and carbon content in the dynamic process of BOF steelmaking.First of all, as problem of ignoring information between the problem properties and solution properties in the traditional of case retrieval, mutual information is introduced the process of determining attributes weights. At the same time as mechanism and intelligent methods are commonly used to calculate the oxygen blowing directly in the traditional model, it is difficult to improve the accuracy of the model. To solve this problem a new prediction method of calculation of oxygen blowing is proposed, in which the oxygen decarburization efficiency is the solution property of case-based reasoning. The prediction model of oxygen decarburization efficiency based on mutual information case-based reasoning is proposed and blowing oxygen of the static and dynamic phase of BOF steelmaking is calculated according to the forecasting results, which can ensure the calculation accuracy of the two phases of blowing oxygen of BOF steelmaking. The validity of the model can be verified by the actual data of a steel plant. The accuracy of two-stage blowing oxygen can be Improved by the actual data of a steel plant.Secondly, in the process of calculation the amount of refrigerant, for the problem of a poor predictor of a single model, integrating the advantages of multi-model, a combination forecasting model of the coolant amount of dynamic phase of BOF steelmaking based on conditional entropy is proposed. The conditional entropy is the basis to determine the contribution of a single model for the forecast result. Fuzzy comprehensive evaluation is used to determine the corresponding model weights. The validity of the model can be verified by the actual data of a steel plant.Finally, a radial basis function neural network model based on particle swarm optimization algorithm and independent component analysis is proposed in the paper, in order to forecast the endpoint in BOF steelmaking. The model adopts the global features of particle swarm optimization algorithm and the local optimizing capacity of fast fixed-point algorithm to improve the traditional algorithm for independent component analysis, and solves the issues of the objective function falling into a local optimum and the uncertain sequence of independent characteristics, also compresses the redundant information and reduces the input dimensions. The extracted independent features are introduced into the radial basic function neural network, so as to predict the endpoint temperature and carbon content. The simulation shows that the model can effectively improve the accuracy and stability of endpoint prediction.