Multi-Method Combination Modelling And Double-Time-Scale Optimization Method For Carbon Efficiency In Sintering Process

With the implementation of energy reduction and green manufacturing policies,the iron and steel industry is seeking to improve fuel utilization efficiency to achieve energy conservation and emission reduction.As the second energy consuming process in the steel industry,iron ore sintering process is the process to produce raw materials for blast furnace.The combustion of coke,mainly consists of carbon,provides the major energy source for the process.Therefore,improving the utilization efficiency of carbon in the sintering process is of great significance for optimizing carbon efficiency and reducing energy consumption.Sintering process has the characteristics of long process flow,multiple kinds of process variables,strong coupling and many working conditions,resulting in lacking of carbon efficiency modeling and optimization methods,which consider the actual production conditions.Aiming at the problem of modeling and optimization for the carbon efficiency,and taking the characteristics of the sintering process,this paper selects comprehensive coke ratio（CCR）and the ratio of CO and CO₂（CO/CO₂）as indexes to measure the carbon efficiency,and focuses on related problems such as multi-level modeling,multi model combination modeling,semi-supervised data-driven modeling,multi-objective optimization and so on.The main research results and innovations of the thesis are as follows:（1）A CO/CO₂prediction model based on semi-supervised model method is proposed.The carbon efficiency index CO/CO₂cannot be detected by fixed sensors and needs to be measured manually,which makes it difficult to obtain the data.Therefore,it is difficult to meet the requirements of high-precision prediction by using the traditional prediction models.To solve this problem,a CO/CO₂prediction model based on a small amount of labeled data and a large amount of unlabeled data is developed to improve the prediction accuracy of the model.Firstly,considering the complexity of sintering process,an improved just-in-time learning is proposed.Combined with the multiple NNs mean combination strategy,the state variable prediction model is established.Secondly,a CO/CO₂prediction model based on deep belief network（DBN）is developed.Using the extract features ability of the DBN,the features are extracted from a large number of unlabeled data,and then the CO/CO₂prediction value is obtained by the regression models.At last,the experimental results show that the proposed method can accurately predict the CO/CO₂when there are a few labeled data.（2）A linear combination prediction model is proposed to achieve high accuracy of the carbon efficinecy indexes.Based on the mechanism analysis and correlation analysis of the sintering process,a carbon efficiency prediction model based on linear combination is designed.Through the analysis of material and energy flow,the input variables of the prediction model are determined.Then,the structure and the inputs of the model are finely divided by correlation analysis method.On this basis,a linear combination-based multi-method combination strategy is proposed.Three widely used neural networks（NNs）are employed to build sub-models,and the evaluation indexes are used to evaluate the different sub-models.According to the evaluation results,the combination weights of each sub-models are obtained.The final prediction results are achieved by combining the sub-models with their combination weights.The simulation results show that the proposed method is superior and can meet the actual production requirements.（3）A carbon efficiency performance assessment strategy based on fuzzy synthetic evaluation is proposed.To realize optimization for the carbon efficiency,an important prerequisite is to monitor the sintering process and judge whether the sintering process is in a stable operation state or whether it needs to be optimized.According to the actual production needs,a carbon efficiency performance assessment strategy is proposed.Firstly,the variables reflecting the sintering state are determined through mechanism analysis.On this basis,the fuzzy synthetic evaluation method is used to assess the sintering state,which is the premise of the optimization for the carbon efficiency.Secondly,in order to improve the assessment accuracy,a carbon efficiency performance assessment method based on two-level combination is proposed.The carbon efficiency index prediction model is established by multiple NN model combination,and then the carbon efficiency performance assessment method is established by using the fuzzy synthetic evaluation method and majority voting strategy.Based on the actual sintering data,the simulation shows that the proposed method can effectively evaluate the sintering state and carbon efficiency,and the proposed carbon efficiency performance assessment method can improve the accuracy of performance assessment.（4）A double-time-scale based optimization strategy for the carbon efficiency is proposed.In order to improve the carbon efficiency in sintering process,considering the different adjustment periods of the raw material variables and operating variables,which affect the carbon efficiency,a double-time-scale optimization strategy is proposed.In short-time scale,a weighted optimization strategy based on particle swarm optimization（PSO）is developed by taking operating variables as decision variables,CCR and CO/CO₂as optimization objectives.In long-time scale,taking the operating variables and raw material variables as decision variables,CCR and CO/CO₂as optimization objectives,a multi-objective optimization strategy based on NSGA-II is developed.Based on the actual sintering data simulation experiment,it shows that the proposed method can optimize the carbon efficiency in different time scales,and give the best setting values of the raw material variables and operating variables.This method considers the actual process demand and adopts different optimization strategies to realize the optimization for the carbon efficiency,which provides a new idea for the engineering application of the optimization in actual sintering production.By studying the data-driven modeling and multi-objective optimization for carbon efficiency in sintering process,it provides new theoretical technology and solutions for data-driven modeling and optimization of sintering process,and lays an important foundation for energy conservation,emission reduction and intelligent manufacturing in sintering process.