The Blast Furnace Temperature Prediction Method Based On The Furnace Heat Index And ST-PLS And Study On The Fuzzy Inference System

Blast furnace (BF) temperature is an important parameter of blast furnace whichdetermines the quality and high yield. Neither blast furnace temperature is too high nor toolow is not conducive to the production. If the furnace temperature is too high, the cokeratio will rise and the production of pig iron will reduce, and trigger hanging accident; onthe contrary, if the temperature is too low, the furnace heat of reaction will be insufficient,and lead to the occurrence of blast furnace cool. Therefore, accurate prediction of thefurnace temperature is the key factor to ensure the smooth operation of blast furnace.Because blast furnace temperature is the result of the combined effects of multiple factors,using only [Si] or Temperature in Molten Iron to react furnace temperature is not accurateenough. Therefore, this selection of [Si],[S], iron reduction rate, temperature in molteniron,Wu andT ffurnace heat index to comprehensive represent of blast furnacetemperature.First of all, two models are presented in this paper to calculate the lag time of themain parameters and predict the parameters related to furnace temperature ([Si],[S], IronReduction Rate and Temperature in Molten Iron). The delay analysis model is based on theGeneralized Correlation Coefficient of Shannon Entropy, and the parameters predictionmodel of the blast furnace temperature is based on a Nonlinear ST-PLS. And theindependent variables that are used in the modeling process of ST-PLS: Feed Rate, AirVolume, the Thermal Wind Pressure, Permeability, Temperature of Hot Air,Oxygen-Enriched Pressure, Oxygen-Enriched Flow, the Amount of Pulverized Coal, IronAmount Difference and a previous furnace [Si] and [S]. Experiments illustrate that thepresented model has a high forecast precision. When the relative errors are0.11and0.18,the hit rate of [Si],[S], Iron Reduction Rate and Temperature in Molten Iron are [0.7143,0.7347,0.6122] and [0.9184,0.9184,0.8163]. When the relative errors are0.005and0.01,the hit rate of Temperature in Molten Iron are0.8163and0.9388. What’s more, the HeatIndex ofW_u andT_frespectively represent the hot state of the lower part of blast furnace and air outlet. According to the experience of BF Long, they can be a good reactiontemperature. Therefore, this paper calculated the Heat Index ofW_u andT_fusing thecomponent data of the charge, the gas, the hot metal as well as the coke.Because the Thermal State of the blast furnace temperature is a visual representationof blast furnace temperature, the aim of this paper is to predict the Thermal State of theblast furnace, i.e."partial low","normal" and the "partial high", under the influence ofmultiple factors. In order to quantify the qualitative variables, this paper defined acomposite index to refer to the thermal state of blast furnace temperature. And six reactionfurnace temperature parameters ([Si],[S], Iron Reduction Rate, Temperature of MoltenIron,Wu andT fFurnace Heat Index) were divided into three categories ("low ","normal" and "high") base on Fuzzy C-Means Clustering method, and determined themembership function of each parameter in each class. In order to predict the Thermal Stateof blast furnace temperature, this paper established a Fuzzy Inference System and the inputof the system are six reaction furnace temperature parameters, and the output of the systemis the composite index of blast furnace temperature. In summary, the model has a certainguiding role in the multiple factor comprehensive prediction of blast furnace heat.