Comprehensive Evaluation Of Blast Furnace Conditions Based On The Combination Of Expert Knowledge And Data

Blast furnace is the most complicated single equipments in industrial production process.It is a complex process with large time delay,polyphase,strong coupling and non-linearity.The essence of the blast furnace smelting process is to control the blast furnace by intermittently feeding the upper part and continuously blowing hot air and pulverized coal powder with the lower part.The index parameters of blast furnace have dynamic parameter distribution characteristics.During the ironmaking process,due to the closed blast furnace and the harsh internal operating environment,the process parameters are difficult to detect accurately online in real time.At present,in blast furnace smelting process,external factors are difficult to measure accurately in real time.The furnace chief s only use their own expert experience,accumulated ironmaking expertise and the actual blast furnace operating process status measured by the human eye Judgment and decision-making,commanding multiple processes for compound control,which will make each unit operate in isolation,lacking coordination.At the same time,this empirical operation mode has blindness,roughness,and subjectivity.Although it can temporarily ensure that the blast furnace enters a stable operation state,the smelting goal of excellent blast furnace products and low energy consumption is still very difficult.The main raw material for blast furnace production is coke.During the ironmaking process,blast furnace operators have adopted a series of optimization operations to reduce the cost of pig iron.The most effective measure is to replace coke with pulverized coal.The second requirement is the so-called replacement ratio.Due to the complexity of the furnace conditions and the variability of the conditions during the blast furnace smelting process,the production conditions of the blast furnace are highly volatile,and the premise of optimization is to ensure the stability of the furnace conditions.When the blast furnace runs smoothly near the working point,the subjective experience of the blast furnace operator usually makes the blindness and ambiguity in the timing and quantity of the increase and decrease of materials,and the assessment of the stability of the furnace condition is extremely urgent at this time.This subject takes the smelting blast furnace of asteel plant as the main research object,and realizes the high quality products and low energy consumption as the control objectives.A fuzzy comprehensive evaluation model of blast furnace conditions based on the combination of expert knowledge and data is proposed,combining expert experience and practical furnace condition analysis of blast furnace ironmaking data.The main research content includes the following three aspects:(a)The monitoring of the existing blast furnace data has a three-dimensional distribution,but its application is insufficient.The data of the temperature difference of the blast furnace cooling wall water is typical.Here,the water temperature difference of the main cooling wall sections of the blast furnace are selected to establish the blast furnace cooling wall water based on the time series fuzzy comprehensive evaluation model of temperature difference.During the modeling process,the time-lag effect of the water temperature difference of the cooling wall sections of blast furnace on the cooling wall sections were analyzed in turn by calculating the correlation coefficients.In the established evaluation model,the calculation of index parameters' weight was determined by combining expert experience and objective calculation.At the same time,by analyzing the relationship between the state of the blast furnace cooling stave water temperature difference and the current furnace temperature,combined with expert knowledge,the validity of the established model can be verified.(b)The traditional prediction of the thermal state of the hearth is only characterized by a single parameter or a prediction model established by a single parameter,and the prediction model established by the two-dimensional detection information to predict the furnace thermal state has limitations.Based on this,this paper defines the concept of the three-dimensional water temperature difference of the blast furnace cooling walls.This paper proposes a new method for predicting the thermal state of the furnace hearth based on time series and multi-dimensional blast furnace.The innovation in this article is that the prediction model built incorporates the three-dimensional water temperature difference of the blast furnace cooling stave and considers the influence of historical furnace temperature.By comprehensively using the parameters and related historical information ofthe three-dimensional blast furnace,this paper establishes the BP neural network(BP-NN)and PSO-LSSVM prediction model of the furnace hearth based on the time series and multi-dimensional blast furnace.Compared with the prediction model results of water temperature difference,the multi-dimensional change of the three-dimensional water temperature difference of the blast furnace cooling stave can assist the blast furnace operator to accurately predict the change trend of the furnace thermal state.(c)This paper combines the prediction model and the judgment model to finally establish a fuzzy comprehensive evaluation model for the thermal state of furnace hearth based on time series and multi-dimensional blast furnace.In the modeling process,the weight of the index parameters is determined by combining expert experience and objective calculation,and the level interval in which the index parameter data is located should be divided by expert experience and statistical knowledge.The high matching rate of the evaluation model indicates that the established evaluation model can assist the furnace chief to predict the change trend of the furnace condition in advance,and guide the furnace chief to adjust the operation direction of the blast furnace.This paper combines a prediction model and a two-level fuzzy comprehensive evaluation model,and proposes a new fuzzy comprehensive evaluation model based on the time series and multi-dimensional blast furnace thermal state of furnace.Based on the longitudinal material flow motion law,the hitting rate of BP-NN and PSO-LS SVM prediction models verifies that there is a lag in the response of the multidimensional change of TDWD to the hearth thermal state.The multidimensional change of TDWD can accurately predict the hearth thermal state.The judging model built in this paper has practical significance for judging the thermal state of the blast furnace hearth,and provided direction for the blast furnace operator to control the subsequent blast furnace.However,this topic only provides a guide method.Due to the large differences in the operating conditions,parameters of controll and blast furnace equipment,specific analysis is required for different blast furnaces.