Exploration And Application Of Constrained Dynamic Programming In Iron Ore Sintering

Iron ore sintering is the first process of blast furnace smelting.The blast furnace ironmaking process is the main method of iron production in the modern steel industry.It can be seen that the quality of sinter ore is closely related to the overall benefits of the steel industry.Therefore,it is of great significance to study the mechanism model in the iron ore sintering process including establishing a mathematical simulation model and a sinter quality prediction model.These models are conducive to improving the output of sintered ore and promoting the rapid development of the steel industry.The reasonable distribution of the thermal state of the sintering process is an important condition of producing high-quality sintered ore.It is very important for the prediction and control of Burn Through Point in the iron ore sintering process.In this paper,we will construct the temperature distribution model of the bellows,which fitting the linear curve of the temperature of the last eight bellows.It estimates the position of the virtual burn-through point online in real time.Furthermore,it can predict the future trend of the virtual burn-through point,guide the sintering production process online,and improve the quality and output of sintered ore.Based on the sintering theory of iron ore,this paper divides the sintering process into four layers:Sinter,Burning,Preheating and Wet.It establishes a piecewise linear regression model of the temperature in Burning,Preheating and Burn through point.Then we use the least squares estimate to optimize the model for the multivariate ob-jective function.According to the actual physical characteristics and additional con-straints,we establish a multi-objective function hierarchical optimization model that estimating the virtual burn-through point and the highest accurate temperature value of the sintering process.Finally,this paper selects actual data to discuss the location of the virtual burn-through point(BTP)in the iron ore sintering process,and predicts the future trend of BTP.In the empirical part of this paper,we select a section of steady-state sintering data from a steel plant in Shandong in December 2018 as sample data,and use a mobile smoothing algorithm to smooth the original data.Then it uses the finite difference iterative algorithm to find the best fitting parameters in the sintering process in layer-s,and simulates the specific location of BTP during this period.Finally,we conduct prediction analysis of Burn Through Point in frequency domain and time domain.In the frequency domain,this paper uses fast Fourier transform and wavelet transform to extract the frequency domain signal characteristics of BTP.The reconstructed signal based on spectral characteristics represents the basic trend of BTP.In the time domain,this paper analyzes the data of the virtual Burn Through Point through the ARIMA model and the ARIMA combination model based on wavelet.By comparing and ana-lyzing the results of the two models,it is concluded that the combined model fits better than the single ARIMA model.Therefore,this paper chooses the ARIMA combination model based on wavelet analysis to predict the future trend of Burn Through Point,and makes corresponding adjustments and controls.