Research On Burning Through Point Prediction Based On Moving Pattern

The sintering production process is important link of the blast furnace ironmaking process.The quality of sinter ore is directly related to the smooth production of blast furnace smelting.The burning through point(BTP)is the correspondent bellows position when the burning mixture is fully burned.It is important index to judge whether the burning process is successful or not.It is the important process parameter closely related with the sintering production,quality,cost and energy consumption and it is main evidence with the sintering machine operation.When the BTP position is normal,the yield of sintering ore could be improved and the effective space of the sintering machine should be fully used.With the premise of maintaining the quality and output of sintering ore,it reduces the consumption and pollution.Since there is obvious nonlinearity,strong coupling,uncertainty and large time-delay in the process of sintering production,the BTP is difficult to determine.Hence,the prediction model of the BTP should be built up to have a prediction to the BTP in advance.The paper introduces the process of sintering production in detail.Besides,the characteristics of sintering production process and the main factors of influencing the BTP are further explored.Since there has no instrument to test directly the BTP at the site,the bellows exhaust gas temperature is used to characterize indirectly the BTP after several methods to judge the BTP are analyzed comparatively.Meanwhile,the soft measurement model of the BTP is established and the model is corrected with standard value of large flue gas temperature.The collected actual working data of the BTP is eliminated abnormal data processing,filtering analyzed and normalized to get the dimensionless data.Finally,the information of main variations is gained with the principal component analysis.These lay a foundation to build up the BTP prediction model based on moving pattern.In addition,the paper designs process of prediction model of the BTP based on moving pattern modeling.The quadtree particle swarm optimization algorithm is employed to build the two-dimensional pattern moving space with the pretreatment actual working data.Two-dimensional interval autoregressive model with exogenous input(IARX)is proposed structure.The factors influencing the BTP is input and the bellows exhaust gas temperature is output of the model to build up the primary prediction of the BTP.By using the least squares method to identify the interval parameter of the primary prediction modeland through the K nearer classification to obtain the final prediction model of the BTP,so the BTP prediction based on moving pattern is achieved.The actual working data of the BTP collected in the spot verified the model and the experiment results shows the feasibility and effectiveness of the model.