Billet Temperature Modeling Approach Of Reheating Furance Based On Data-driven Methods

As the research on billet temperature estimated model of reheating furnace becomes a hot topic, establishing a reasonable model is of great significance for improving the heating quality of billets and reducing energy consumption. Usually the prediction model can be established by analysing the parameters, such as heat radiation, heat transfer mechanism in the furnace and so on, but this method is difficultly achieved, because it is based on many assumed conditions and parameters.There is a brief overview of reheating furnace, tring to use the methods of data-driven to create the billet temperature models in this paper. From statistica modeling and intelligent modeling two aspects respectively, we use the swarm intelligece optimization methods to adjust the parameters.From the statistical modeling of view, it estimates the billet temperature by using the Least Square Support Vector Machine regression modeling method. Through analyzing the model of LSSVM, the emulated results have proved that the normalization factor and the kernel function have a great effect on the accuracy and precision of the models. In the meantime, the limiting dataes can not fully stand for the process of heating, so it needs to be improved about the estimated results.About the intelligent modeling perspective, it establishes the billet temperature model by using the BP neural network. Through analyzing the models, the results of BP network model are better than the LSSVM. But BP is great dependent on the weights and thresholds, so there is a large effect on the estimated results.Against the dependence on the parameters of LSSVM and BP network, by using swarm intelligence optimization algorithms, Particle Swarm Optimization (PSO) and Quantum Particle Swarm Optimization (QPSO), it adjusts the relevant parameters, and establishes the optimal models. Comparing the results of the simulations, it has been proved that the BP network model using the QPSO optimized method has achieved better results on the accuracy and precision.The stuy shows that, based on data-driven modeling approaches are feasible on estimated the billet trmperature, and can achieve better results.