WO2021203912A1 - 氧气底吹炉铜吹炼过程参数在线预测方法 - Google Patents
氧气底吹炉铜吹炼过程参数在线预测方法 Download PDFInfo
- Publication number
- WO2021203912A1 WO2021203912A1 PCT/CN2021/080600 CN2021080600W WO2021203912A1 WO 2021203912 A1 WO2021203912 A1 WO 2021203912A1 CN 2021080600 W CN2021080600 W CN 2021080600W WO 2021203912 A1 WO2021203912 A1 WO 2021203912A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- model
- slag
- prediction
- data
- neural network
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 137
- RYGMFSIKBFXOCR-UHFFFAOYSA-N Copper Chemical compound [Cu] RYGMFSIKBFXOCR-UHFFFAOYSA-N 0.000 title claims abstract description 119
- 238000007664 blowing Methods 0.000 title claims abstract description 111
- 230000008569 process Effects 0.000 title claims abstract description 82
- 229910052802 copper Inorganic materials 0.000 title claims abstract description 69
- 239000010949 copper Substances 0.000 title claims abstract description 69
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 title claims abstract description 39
- 239000001301 oxygen Substances 0.000 title claims abstract description 39
- 229910052760 oxygen Inorganic materials 0.000 title claims abstract description 39
- 239000002893 slag Substances 0.000 claims abstract description 96
- 230000007246 mechanism Effects 0.000 claims abstract description 54
- 239000000463 material Substances 0.000 claims abstract description 44
- 238000004519 manufacturing process Methods 0.000 claims abstract description 21
- 238000003062 neural network model Methods 0.000 claims abstract description 18
- 239000002994 raw material Substances 0.000 claims abstract description 8
- XEEYBQQBJWHFJM-UHFFFAOYSA-N iron Substances [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 claims description 49
- 229910052742 iron Inorganic materials 0.000 claims description 38
- 238000013528 artificial neural network Methods 0.000 claims description 31
- 210000002569 neuron Anatomy 0.000 claims description 17
- 229910052710 silicon Inorganic materials 0.000 claims description 10
- 239000010703 silicon Substances 0.000 claims description 10
- 238000009776 industrial production Methods 0.000 claims description 8
- 230000004907 flux Effects 0.000 claims description 7
- 230000002829 reductive effect Effects 0.000 claims description 5
- 230000015572 biosynthetic process Effects 0.000 claims description 2
- 238000003786 synthesis reaction Methods 0.000 claims description 2
- 230000000694 effects Effects 0.000 abstract description 10
- 229910000519 Ferrosilicon Inorganic materials 0.000 abstract description 5
- 230000006870 function Effects 0.000 description 14
- XUIMIQQOPSSXEZ-UHFFFAOYSA-N Silicon Chemical compound [Si] XUIMIQQOPSSXEZ-UHFFFAOYSA-N 0.000 description 9
- 238000006243 chemical reaction Methods 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 4
- XWHPIFXRKKHEKR-UHFFFAOYSA-N iron silicon Chemical compound [Si].[Fe] XWHPIFXRKKHEKR-UHFFFAOYSA-N 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 239000000155 melt Substances 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 238000012546 transfer Methods 0.000 description 3
- VYPSYNLAJGMNEJ-UHFFFAOYSA-N Silicium dioxide Chemical compound O=[Si]=O VYPSYNLAJGMNEJ-UHFFFAOYSA-N 0.000 description 2
- 238000003723 Smelting Methods 0.000 description 2
- 210000004556 brain Anatomy 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000012512 characterization method Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000013213 extrapolation Methods 0.000 description 2
- 239000000446 fuel Substances 0.000 description 2
- 239000007789 gas Substances 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000007670 refining Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 210000004027 cell Anatomy 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000008094 contradictory effect Effects 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000005868 electrolysis reaction Methods 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 230000000670 limiting effect Effects 0.000 description 1
- 238000005272 metallurgy Methods 0.000 description 1
- 230000001590 oxidative effect Effects 0.000 description 1
- 230000036961 partial effect Effects 0.000 description 1
- 230000001902 propagating effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 235000012239 silicon dioxide Nutrition 0.000 description 1
- 239000000377 silicon dioxide Substances 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- C—CHEMISTRY; METALLURGY
- C22—METALLURGY; FERROUS OR NON-FERROUS ALLOYS; TREATMENT OF ALLOYS OR NON-FERROUS METALS
- C22B—PRODUCTION AND REFINING OF METALS; PRETREATMENT OF RAW MATERIALS
- C22B15/00—Obtaining copper
- C22B15/0026—Pyrometallurgy
- C22B15/0028—Smelting or converting
- C22B15/003—Bath smelting or converting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Definitions
- This application relates to the field of metallurgy, and specifically, to an online prediction method of copper blowing process parameters in an oxygen bottom blowing furnace.
- the blowing process is an extremely important process in the copper smelting process. It takes hot copper matte, flux and other raw materials (electrolysis residue) through the oxygen-enriched air at the bottom to generate oxidative exothermic reaction to form blister copper.
- the refining process provides raw materials.
- the oxygen bottom-blowing copper conversion furnace is the core equipment of the "new oxygen-enriched bottom-blowing copper conversion process" with complete independent intellectual property rights in China, and it is also the equipment for the copper conversion process commonly used in China, and the conversion process is continuous. The characteristics of instantaneity require corresponding control technology to further exert its technological characteristics, and to ensure stable product quality and optimized furnace conditions.
- the copper blowing process is a very complex process. It is a multi-input and multi-output system.
- the three parameters of blister copper grade, iron-silicon ratio in slag and slag temperature are important parameters for investigating the copper blowing process of oxygen bottom blowing furnace, reflecting the state of oxygen bottom blowing copper blowing process.
- the blister copper grade refers to the mass fraction of copper in the blister copper
- the slag iron to silicon ratio refers to the mass ratio of iron to silicon dioxide in the slag.
- Their detection methods all use sampling during the blister copper and slag discharge process. Then, a fluorescence analyzer was used to test the composition of the sample, and the slag iron to silicon ratio and the blister copper grade value were calculated.
- the key parameter in the oxygen bottom blowing copper blowing process is the melt temperature. However, due to its harsh production environment, the melt temperature cannot be directly measured online.
- the slag temperature is generally used as an indirect characterization parameter of the melt temperature in production.
- the copper blowing process requires the three important parameters of blister copper grade, iron-silicon ratio in the blowing slag, and slag temperature to be kept in an appropriate range and the fluctuations are as small as possible. Therefore, it is necessary to establish a corresponding and reliable prediction model for bottom blowing Three important parameters in the blowing process are predicted to provide guidance for the on-site staff in their decision-making and operation.
- This application aims to solve one of the technical problems in the related technology at least to a certain extent. For this reason, one purpose of this application is to propose an online method for predicting the parameters of the copper blowing process in an oxygen bottom blowing furnace.
- the prediction method can significantly improve the accuracy of the prediction results, and effectively solve the problems of poor adaptability of existing prediction models and methods and unsatisfactory actual operation effects.
- the research methods for the parameter prediction model of the oxygen bottom blowing copper blowing process mainly adopt the mechanism modeling method or the data-driven intelligent modeling method, but these two methods have their own shortcomings: 1) The model is dedicated In terms of performance, as long as the objects of the model are different, the structure and parameters of the mechanism model are very different, and the portability of the model is poor; 2) The entire modeling process of the mechanism model requires a lot of manpower and material resources, regardless of Is it the study of the essential kinetics of the reaction, the determination of various equipment models, the characterization of the heat and mass transfer effects of the device in practical applications, the estimation of a large number of parameters (including test equipment and devices), each step is very difficult; 3 ) Mechanism models are generally composed of algebraic equations, differential equations, and even partial differential equations.
- the present application proposes an online prediction method for the parameters of the copper blowing process in an oxygen bottom blowing furnace.
- the prediction method includes:
- the slag silicon-to-iron ratio neural network model and the slag temperature neural network model between the target parameters and the input parameters, the predicted value of the blister copper grade, the predicted value of the slag silicon-to-iron ratio and the slag temperature neural network model are established.
- the intelligent coordinator is suitable for blister copper grade prediction value, slag silicon-to-iron ratio prediction value, slag temperature prediction value and blister copper grade, slag-silicon-to-iron ratio and output based on the respective output of the mechanism model and the data-driven model.
- the deviation between the actual measured value of the slag temperature, the weighting coefficient of the mechanism model and the data-driven model in the mixed model is calculated, and the weighting coefficient, the predicted value of the mechanism model, and the data-driven Model prediction value, output the final prediction value of blister copper grade, slag silicon-to-iron ratio and slag temperature.
- the three important parameters of blister copper grade, iron-silicon ratio in the blowing slag, and slag temperature during the bottom-blowing blowing process are investigated to establish respectively
- the mechanism model and data-driven model of the bottom blowing furnace predict three important parameters, and on this basis, design a suitable intelligent coordinator to integrate the two, and compare the prediction results of the integrated hybrid model with the actual production results.
- the prediction method can fully combine the advantages of the mechanism model and the data-driven model, maximize the strengths and avoid the weaknesses, significantly improve the accuracy of the prediction results, and effectively solve the problems of poor adaptability and unsatisfactory actual operation effects caused by the modeling of existing prediction methods. It has great significance and value in theory and practical application.
- the method for online prediction of copper blowing process parameters in an oxygen bottom blowing furnace may also have the following additional technical features:
- the material balance model is established based on a material balance equation
- the energy balance model is established based on an energy balance equation
- the multiphase balance model is established based on a multiphase balance equation
- the METCAL software or the METSIM software is used to simultaneously solve the material balance equation, the energy balance equation and the multiphase balance equation
- the mechanism model is established in combination with the process characteristics of the copper bottom blowing process.
- the blister copper grade neural network, the slag silicon-to-iron ratio neural network, and the slag temperature neural network each independently include a plurality of artificial neurons, and the artificial neurons include but are not limited to Copper matte grade, copper matte temperature, oxygen matte ratio, flux rate, electrolytic residual rate and oxygen enrichment rate in the copper bottom blowing process.
- the modeling mechanism of BP neural network is used in combination with the industrial production big data of the copper bottom blowing process and the blister copper grade neural network, the slag silicon to iron ratio neural network, and the slag temperature neural network to establish the method.
- the data-driven model is used in combination with the industrial production big data of the copper bottom blowing process and the blister copper grade neural network, the slag silicon to iron ratio neural network, and the slag temperature neural network to establish the method.
- the hybrid model is corrected periodically or in real time based on actual production results.
- calibrating the hybrid model includes: comparing the final predicted value of the blister copper grade, slag silicon-to-iron ratio, and slag temperature output by the hybrid model with actual measured values: if the error is Within the expected range, keep the weighting coefficient in the hybrid model unchanged; if the error is outside the expected range, return the final predicted value to the smart coordinator and adjust the weighting coefficient, repeat the above operation, Until the error is reduced to the expected range.
- the intelligent coordinator adopts a method of fuzzy division of input variable regions and synthesis to calculate the weighting coefficients of the mechanism model and the data-driven model prediction method.
- f 1 is used to represent the prediction result output by the mechanism model
- f 2 is used to represent the prediction result output by the data-driven model
- ⁇ (x) is used to represent the data-driven model in the
- the weighting coefficient in the hybrid model uses (1- ⁇ (x)) to represent the weighting coefficient of the mechanism model in the hybrid model
- the prediction result of the output of the intelligent coordinator is:
- y represents the prediction result
- the prediction result includes the prediction value of blister copper grade, the prediction value of slag silicon-to-iron ratio and the prediction value of slag temperature
- the weighting coefficient ⁇ (x) of the data-driven model in the hybrid model is:
- the selection range of the input variable includes but not limited to copper matte grade, copper matte temperature, oxygen matte ratio, flux rate, electrolytic residual rate and oxygen enrichment rate; a, b, c, d are based on actual conditions
- the characteristic parameter corresponding to the input variable obtained from the technical data of industrial production, and the characteristic parameter determines the membership function of the input variable.
- the weighting coefficient ⁇ (x) of the data-driven model in the hybrid model is:
- ⁇ i is the membership function calculated from the input variable i and its corresponding characteristic parameters a, b, c, d, ⁇ i is the weight coefficient of the input variable i in the j input variables, and ⁇ i is the empirical value Sure.
- Fig. 1 is a schematic diagram of a hybrid model of an online prediction method for blister copper grade, slag silicon-to-iron ratio and slag temperature according to an embodiment of the present application.
- Fig. 2 is a schematic diagram of a neuron model structure according to an embodiment of the present application.
- Fig. 3 is a schematic diagram of a BP neural network model structure according to an embodiment of the present application.
- Fig. 4 is a flowchart of an online prediction method for parameters of copper blowing process in an oxygen bottom blowing furnace according to an embodiment of the present application.
- this application proposes an online method for predicting parameters of the copper blowing process in an oxygen bottom blowing furnace.
- the prediction method includes: establishing the prediction value of blister copper grade and the prediction of slag-iron-silicon ratio based on raw material input conditions and based on a material balance model, an energy balance model, and a multi-phase balance model.
- Bottom-blown converting furnace mechanism model based on the predicted value and slag temperature; according to actual production data and based on the blister copper grade neural network model, the slag silicon-to-iron ratio neural network model and the slag temperature neural network model between the target parameters and the input parameters, Establish a bottom-blown converting furnace data-driven model about the predicted value of blister copper grade, the predicted value of slag silicon-to-iron ratio, and the predicted value of slag temperature; use the intelligent coordinator to integrate the mechanism model and the data-driven model to obtain the predicted value of blister copper grade , Bottom-blown converting furnace mixing model with predicted value of slag ferrosilicon ratio and predicted value of slag temperature, using the hybrid model to output the final predicted value of blister copper grade, slag ferrosilicon ratio and slag temperature during the copper bottom blowing process.
- the intelligent coordinator is suitable for the predicted value of blister copper grade, the predicted value of slag silicon to iron ratio, the predicted value of slag temperature and the actual measured value of blister copper grade, the slag silicon to iron ratio and the slag temperature based on the respective output of the mechanism model and the data-driven model.
- the deviation between the computer model and the data-driven model in the mixed model, and according to the weighted coefficient, the predicted value of the mechanism model and the predicted value of the data-driven model, the final output of blister copper grade, slag silicon-to-iron ratio and slag temperature Predictive value.
- the prediction method can fully combine the advantages of the mechanism model and the data-driven model, maximize the strengths and avoid the weaknesses, significantly improve the accuracy of the prediction results, and effectively solve the problems of poor adaptability and unsatisfactory actual operation effects caused by the modeling of existing prediction methods. It is of great significance and value in practical application.
- the material balance model is established based on the material balance equation
- the energy balance model is established based on the energy balance equation
- the multiphase balance model is established based on the multiphase balance equation, using METCAL software or METSIM software
- the material balance equation, energy balance equation and multi-phase balance equation are solved simultaneously and combined with the process characteristics of the copper bottom blowing process to establish a bottom blowing furnace mechanism model.
- the material balance equation is:
- Var represents a variable
- Con represents a constant
- M represents a material
- C represents a component of the material
- E represents an element of the component
- X represents a mole fraction
- E c, e represents a specific element in a specific component
- the energy balance equation is:
- This equation represents that the heat of the input material and the heat of the output material are equal in the bottom blowing process, that is, energy conservation.
- the multiphase balance equation is:
- G is the total Gibbs free energy of the system, Is the standard generating Gibbs free energy of pure substance c component in phase p; ⁇ pc is the activity factor of component c in phase p; ⁇ pc is the mole fraction of component c in phase p ; C p is p The number of components in the phase; T is the temperature; R is the universal constant of the gas; N pc is the number of moles of the c component in the p phase.
- This equation indicates that the Gibbs free energy of the system reaches the minimum during the blowing process, that is, the system reaches a steady state.
- Gaussian method can be used to solve the multivariate linear equations for each element of each component, for example, for a specific element of a specific component
- the reliable third-party software such as METCAL, METSIM, etc. can be used to establish the bottom blowing furnace mechanism model, and calculate and predict the blister copper grade and blowing
- the ratio of iron to silicon in the slag and the temperature of the slag are the three important parameters of the copper bottom blowing process.
- the blister copper grade neural network, the slag silicon-to-iron ratio neural network, and the slag temperature neural network each independently include a plurality of artificial neurons
- the artificial neurons include but are not limited to the copper bottom blowing process Copper matte grade, copper matte temperature, oxygen matte ratio, flux rate, electrolytic residual electrode rate and oxygen enrichment rate.
- the data-driven model can be processed by the method of artificial neural network.
- Artificial neural network is a nonlinear information processing system composed of a large number of interconnected processing units that imitates the biological structure and related functions of the human brain. It consists of artificial neurons.
- the schematic structure of the neuron model is shown in Figure 2.
- the neuron model has R inputs, and each input is connected to the next layer through a weight w, where f represents the input/output relationship
- w represents the input/output relationship
- S j is the output function
- b j is the threshold
- w j,i are the connection weights
- x 0 b j
- w j,0 -1
- p and y are the input and output of the neuron, respectively.
- f(wp+b) is the transfer function.
- the modeling mechanism of the BP neural network can be used in combination with the industrial production big data of the copper bottom blowing process and the blister copper grade neural network, the slag silicon to iron ratio neural network, and the slag temperature neural network.
- BP neural network is currently one of the most widely used and successful artificial neural networks. It is a reverse learning algorithm for multi-layer networks. Its learning process consists of two processes: forward propagation of signals and backward propagation of errors.
- BP neural network can learn a large number of input-output mapping relationships without revealing the explicit mathematical equations of this mapping relationship.
- the BP neural network is composed of an input layer, an intermediate layer or a hidden layer and an output layer.
- the input sample data is passed in from the input layer, and then passed through the hidden layer processing, information transformation, and learning. Output layer. If the output does not match expectations, the error will be passed back to the hidden layer and input layer in some form, and the error will be allocated to each layer in the process, as the basis for modifying the weights and thresholds of various places.
- the weights and thresholds are continuously adjusted until the output error is reduced to an acceptable level or reaches the predetermined number of learning times.
- the modeling mechanism of the BP neural network and combining the big data of industrial production of the copper bottom blowing process to construct a neural network model of the copper bottom blowing process, and using the production big data to train and optimize the model, it can further improve the
- the three important parameters of blister copper grade, the ratio of iron to silicon in the slag and the temperature of the slag are the reliability of prediction.
- the intelligent coordinator can use the method of fuzzy dividing the input variable area and comprehensively calculate the weighting coefficient of the bottom blowing furnace mechanism model and the bottom blowing furnace data-driven model prediction method, that is, through The intelligent coordinator integrates the two prediction models based on fuzzy division.
- the neural network model obtains greater weight, and its compensation effect is more ensured Prediction accuracy; when the bottom-blowing furnace copper blowing balance is disturbed and the working conditions are unstable, the mechanism model gets more weight, so that the intelligent integrated hybrid prediction model can guarantee the global fitting ability of the bottom-blowing furnace copper blowing process .
- the intelligent coordinator to intelligently integrate the two models, not only the reliability of the integrated hybrid model is greatly enhanced, but also the artificial neural network formed by the combination of a large number of neurons will also show similarity to the human brain.
- the characteristics make the hybrid model have adaptive and self-organizing capabilities, which can significantly improve the accuracy of the prediction results.
- f 1 can be used to represent the predicted results of the bottom-blown converting furnace mechanism model output
- f 2 can be used to represent the predicted results of the bottom-blown converting furnace data-driven model output
- ⁇ (x) can be used to represent the bottom-blown converting furnace data-driven
- y represents the prediction result.
- the prediction results include the predicted value of blister copper grade, the predicted value of slag silicon-to-iron ratio and the predicted value of slag temperature.
- the value of the weighting coefficient ⁇ (x) of the data-driven model in the mixed model is obtained by the membership function.
- the relationship between ⁇ (x) and the membership function is
- x represents a certain input variable, such as copper matte grade, copper matte temperature, oxygen matte ratio, flux rate, electrolytic residual rate and oxygen enrichment rate, etc.;
- a, b, c, d are characteristic parameters, and different input variables correspond to different The characteristic parameters a, b, c, and d of each characteristic parameter can be obtained by referring to the historical record of actual industrial production to obtain the reference initial value of the membership function parameter of each input variable, and using industrial data to optimize it, for example, when using copper matte grade
- the copper matte grade and its corresponding characteristic parameters a, b, c, d can be:
- characteristic parameters a, b, c, and d are not fixed and can be adjusted according to actual production.
- the values of characteristic parameters a, b, c, and d are different for different production conditions. It's not all the same.
- the weighting coefficient of the data-driven model in the mixed model can be used to predetermine the input variables and their corresponding characteristic parameters a, b, c, d, obtain the membership function ⁇ i value of each input variable, and then use the weighted average method to calculate The final weighting coefficient ⁇ (x) value, namely
- ⁇ i is the membership function calculated from the input variable i and its corresponding characteristic parameters a, b, c, d, ⁇ i is the weight coefficient of the input variable i in the j input variables, and ⁇ i is the empirical value Sure.
- the weighting coefficient ⁇ (x) of the data-driven model in the mixed model is
- the mixing model can be corrected regularly or in real time based on actual production results, thereby further improving the reliability and accuracy of the online prediction method for the copper blowing process parameters of the oxygen bottom blowing furnace of the present application.
- calibrating the hybrid model may include: comparing the final predicted values of the blister copper grade, slag silicon-to-iron ratio, and slag temperature output by the hybrid model with the actual measured values: if the error is within the expected range If the error is outside the expected range, return the final predicted value to the smart coordinator and adjust the weighting coefficient. Repeat the above operation until the error is reduced to the expected range.
- This method is especially suitable In order to correct the preliminarily established mixing model, the reliability and accuracy of the online prediction method for the copper blowing process parameters of the oxygen bottom blowing furnace can be further improved.
- the flow chart of the method for online prediction of oxygen-enriched bottom blowing process parameters based on the hybrid model can be shown in Figure 4, where the real-time collection of key process parameters is performed by detecting sensors (load cells, flow Sensors, etc.) measure the actual parameters such as the amount of raw fuel and oxygen-enriched gas flow during the oxygen-enriched bottom blowing process, and transmit them to the prediction model; the key parameters for offline detection are input and output in the human-machine interface Key process parameters such as copper flow rate and composition, slag flow rate, composition and temperature; the establishment of an online prediction model refers to the use of the principles of material balance, energy balance and phase balance to establish the mechanism model of the oxygen-enriched bottom blowing process and use production Big data constructs a neural network model (data-driven model), and uses an intelligent coordinator to integrate the two to obtain a mixed model for the prediction of three important parameters of the bottom-blowing furnace blowing process.
- sensors load cells, flow Sensors, etc.
- the three major factors namely the grade of blister copper in the bottom blowing process, the ratio of iron to silicon in the blowing slag, and the slag temperature, are mainly investigated.
- the bottom blowing furnace mechanism model and data-driven model are respectively established to predict the three important parameters.
- it uses the advantages of the mechanism model of good extrapolation and strong interpretability, and on the other hand, it adopts the neural network analysis method. Analyze and predict the three important parameters of the bottom blowing process with big data.
- the prediction method can fully combine the advantages of the mechanism model and the data-driven model, maximize the strengths and avoid the weaknesses, significantly improve the accuracy of the prediction results, and effectively solve the problems of poor adaptability and unsatisfactory actual operation effects caused by the modeling of existing prediction methods. It has great significance and value in theory and practical application.
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Physics & Mathematics (AREA)
- Manufacturing & Machinery (AREA)
- Chemical & Material Sciences (AREA)
- Marketing (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Mechanical Engineering (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Materials Engineering (AREA)
- Development Economics (AREA)
- Metallurgy (AREA)
- Organic Chemistry (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Manufacture And Refinement Of Metals (AREA)
Abstract
本申请公开了氧气底吹炉铜吹炼过程参数在线预测方法。该方法包括:根据原料输入条件并基于物料平衡模型、能量平衡模型及多相平衡模型建立底吹吹炼炉机理模型;根据实际生产数据并基于目标参数和输入参数之间的粗铜品位神经网络模型、渣硅铁比神经网络模型及渣温度神经网络模型建立底吹吹炼炉数据驱动模型;利用智能协调器对机理模型和数据驱动模型进行集成,得到关于粗铜品位预测值、渣硅铁比预测值和渣温度预测值的底吹吹炼炉混合模型,利用所述混合模型输出铜底吹吹炼过程中粗铜品位、渣硅铁比和渣温度的最终预测值。该预测方法可以有效解决现有预测模型及方法适应能力差、实际运行效果不理想的问题,显著提高预测结果的准确性。
Description
相关申请的交叉引用
本申请要求申请号为202010281411.8、申请日为2020年04月10日的中国专利申请的优先权和权益,上述中国专利申请的全部内容在此通过引用并入本申请。
本申请涉冶金领域,具体而言,涉及氧气底吹炉铜吹炼过程参数在线预测方法。
吹炼过程是铜冶炼过程中一个极为重要的工序,它将热铜锍、熔剂及其他原料(电解残极)通过底部富氧空气的吹入发生氧化放热反应形成粗铜的过程,为后续精炼过程提供原料。氧气底吹铜吹炼炉是我国完全自主知识产权的“富氧底吹炼铜新工艺”的吹炼的核心设备,也是目前国内常用的铜吹炼过程的设备,其吹炼过程具有连续性、瞬时性的特点,需要有相应的控制技术能进一步发挥其工艺特点,并确保产品质量稳定、炉况处于优化状态。同时,该铜吹炼工艺是一个十分复杂的工艺过程,是一个多输入和多输出的***,每个变量之间具有强耦合、时变、分布式参数和显著的不确定性等特点,而吹炼过程大部分关键参数难于实时检测,存在时间滞后性,仅仅依靠操作员生产经验进行操作,吹炼终点不宜控制,造成炉况不顺及产品质量波动,制约了该工艺优势的发挥。
其中,粗铜品位、渣中铁硅比及渣温度这三个参数是考察氧气底吹炉铜吹炼过程中的重要参数,反映了氧气底吹铜吹炼过程中的状态。粗铜品位指的是粗铜当中铜元素的质量分数,炉渣铁硅比指的是炉渣当中铁元素和二氧化硅的质量比,它们的检测方法均采用放粗铜及放渣过程中取样,然后采用荧光分析仪对样品成分进行化验,计算得到炉渣铁硅比及粗铜品位值。氧气底吹铜吹炼过程中的关键参数有熔体温度,但是由于其恶劣的生产环境,熔体温度无法在线直接测量,生产中一般采用炉渣温度作为熔体温度的间接表征参数。也就是说,在底吹炉铜吹炼实际生产中是无法实时监测到产物粗铜品位、渣中铁硅比及渣温度这三个关键参数。一般铜吹炼过程要求粗铜品位、吹炼渣中铁硅比及渣温度这三大重要参数保持在适宜的范围及波动尽可能小,因此,有必要通过建立相应且可靠的预测模型对底吹吹炼过程中三大重要参数进行预测,为现场工作人员的决策与操作提供指导意见。
申请内容
本申请旨在至少在一定程度上解决相关技术中的技术问题之一。为此,本申请的一个目的在于提出氧气底吹炉铜吹炼过程参数在线预测方法。该预测方法通过对机理模型和数据驱动模型进行集成,可以显著提高预测结果的准确性,有效解决现有预测模型及方法适应能力差,实际运行效果不理想的问题。
本申请是申请人基于以下问题和发现提出的:
目前对于氧气底吹铜吹炼过程参数预测模型的研究方法主要采用机理建模方法或是基于数据驱动的智能建模方法,但这两种方法各有其不足之处:1)就模型的专用性来说,只要模型的对象不同,其机理模型的结构和参数都存在非常大的差别,模型的可移植性较差;2)机理模型的整个建模过程需花费的人力物力很大,无论是反应本质动力学的研究、各种设备模型的确定,还是实际应用中装置传热传质效果的表征、大量参数(包括试验设备和装置)的估计,每一个步骤都是十分困难的;3)机理模型一般都是由代数方程组、微分方程组,甚至是偏微分方程组组成,当模型结构较为庞大时,求解时将要面对大量的数学计算,收敛慢,很难达到满足在线的实时估计;4)机理模型的建立通常是基于一定的假设条件的,而这些假设条件与实际情况存在一定的差别,难以保证模型的精确性。而基于数据驱动所构建的智能建模方法,例如神经网络,其性能不但受训练样本的质量、空间分布和训练算法的影响,且其外推性能较差,这种模型具有不可解释性。并且目前虽然有涉及基于混合模型的参数在线预测方法,但其实际上是在建立复杂机理模型的基础上仅仅通过实际生产数据对机理模型进行修正,实质上还是采用单一机理模型,存在构建模型过程复杂,计算量大,实际数据对机理模型作用小等缺点。而对于底吹炉吹炼而言,针对原燃料条件等参数对粗铜品位、渣铁硅比、渣温度具有显著影响,因此需通过采取配料制度优化及采取操作制度优化来减少原料波动对吹炼三大重要参数影响。申请人设想,可以将机理模型与数据驱动模型结合起来,开发出一种集成模型对炼铜吹炼中的三大重要参数进行预测,以便通过输入投料物及吹炼过程中的参数预测产物的重要参数,并与产物的实际检测结果对比修正预测模型,提高预测精度,使得集成模型的预测准确度能够满足指导实际生产。
为此,根据本申请的一个方面,本申请提出了一种氧气底吹炉铜吹炼过程参数在线预测方法。根据本申请的实施例,该预测方法包括:
根据原料输入条件并基于物料平衡模型、能量平衡模型及多相平衡模型,建立关于粗 铜品位预测值、渣铁硅比预测值和渣温度预测值的底吹吹炼炉机理模型;
根据实际生产数据并基于目标参数和输入参数之间的粗铜品位神经网络模型、渣硅铁比神经网络模型及渣温度神经网络模型,建立关于粗铜品位预测值、渣硅铁比预测值和渣温度预测值的底吹吹炼炉数据驱动模型;
利用智能协调器对所述机理模型和所述数据驱动模型进行集成,得到关于粗铜品位预测值、渣硅铁比预测值和渣温度预测值的底吹吹炼炉混合模型,利用所述混合模型输出铜底吹吹炼过程中粗铜品位、渣硅铁比和渣温度的最终预测值,
其中,所述智能协调器适于基于所述机理模型和所述数据驱动模型各自输出的粗铜品位预测值、渣硅铁比预测值、渣温度预测值与粗铜品位、渣硅铁比和渣温度的实际测量值之间的偏差,计算所述机理模型和所述数据驱动模型在所述混合模型中的加权系数,并根据所述加权系数、所述机理模型预测值和所述数据驱动模型预测值,输出粗铜品位、渣硅铁比和渣温度的最终预测值。
根据本申请上述实施例的氧气底吹炉铜吹炼过程参数在线预测方法,通过重点考察底吹吹炼过程中粗铜品位、吹炼渣中铁硅比及渣温度这三大重要参数,分别建立底吹吹炼炉机理模型和数据驱动模型对三个重要参数进行预测,在此基础上设计适宜的智能协调器对二者进行集成,并将集成后混合模型的预测结果与实际生产结果对比修正,不断完善机理模型模型与数据驱动模型,并修正智能协调器参数,使其预测结果更加满足实际生产结果,从而显著提高其对氧气底吹炉铜吹炼过程中三大参数的预测精度。综上,该预测方法可以充分结合机理模型和数据驱动模型的优点,扬长避短,显著提高预测结果的准确性,有效解决现有预测方法建模造成的适应能力差、实际运行效果不理想的问题,在理论与实际应用上都具有重大意义与价值。
另外,根据本申请上述实施例的氧气底吹炉铜吹炼过程参数在线预测方法还可以具有如下附加的技术特征:
在本申请的一些实施例中,所述物料平衡模型是基于物料平衡方程建立的,所述能量平衡模型是基于能量平衡方程建立的,所述多相平衡模型是基于多相平衡方程建立的,采用METCAL软件或METSIM软件对所述物料平衡方程、所述能量平衡方程和所述多相平衡方程进行联立求解并结合铜底吹吹炼过程的工艺特征建立所述机理模型。
在本申请的一些实施例中所述粗铜品位神经网络、所述渣硅铁比神经网络和所述渣温 度神经网络分别独立地包括多个人工神经元,所述人工神经元包括但不限于铜底吹吹炼过程中的铜锍品位、铜锍温度、氧锍比、熔剂率、电解残极率和富氧率。
在本申请的一些实施例中,利用BP神经网络的建模机理并结合铜底吹吹炼过程的工业生产大数据和粗铜品位神经网络、渣硅铁比神经网络及渣温度神经网络建立所述数据驱动模型。
在本申请的一些实施例中,基于实际生产结果定期或实时对所述混合模型进行校正。
在本申请的一些实施例中,对所述混合模型进行校正包括:将所述混合模型输出的粗铜品位、渣硅铁比和渣温度的最终预测值与实际测量值进行对比:若误差在预期范围内,保持所述混合模型中所述加权系数不变;若误差在预期范围外,将所述最终预测值返回至所述智能协调器并对所述加权系数进行调整,重复上述操作,直至误差降低至预期范围内。
在本申请的一些实施例中,所述智能协调器采用模糊划分输入的变量区域并综合的方法计算所述机理模型和所述数据驱动模型预测方法的加权系数。
在本申请的一些实施例中,利用f
1表示所述机理模型输出的预测结果,利用f
2表示所述数据驱动模型输出的预测结果,利用μ(x)表示所述数据驱动模型在所述混合模型中的加权系数,利用(1-μ(x))表示所述机理模型在所述混合模型中的加权系数,所述智能协调器的输出的预测结果为:
y=f
2×μ(x)+f
1×(1-μ(x)),
其中,y代表预测结果,所述预测结果包括粗铜品位预测值、渣硅铁比预测值和渣温度预测值,所述数据驱动模型在所述混合模型中的加权系数μ(x)为:
x代表输入变量,所述输入变量的选择范围包括但不限于铜锍品位、铜锍温度、氧锍比、熔剂率、电解残极率和富氧率;a、b、c、d为根据实际工业生产的技术数据得到的与所述输入变量对应的特征参数,所述特征参数决定所述输入变量的隶属函数。
在本申请的一些实施例中,所述数据驱动模型在所述混合模型中的加权系数μ(x)为:
其中,μ
i为输入变量i及其对应的特征参数a、b、c、d计算得到的隶属函数,λ
i为输入变量i在j个输入变量中所占的权重系数,λ
i由经验值确定。
本申请的附加方面和优点将在下面的描述中部分给出,部分将从下面的描述中变得明显,或通过本申请的实践了解到。
本申请的上述和/或附加的方面和优点从结合下面附图对实施例的描述中将变得明显和容易理解,其中:
图1是根据本申请一个实施例的关于粗铜品位、渣硅铁比和渣温度的在线预测方法混合模型示意图。
图2是是根据本申请一个实施例的神经元模型结构示意图。
图3是是根据本申请一个实施例的BP神经网络模型结构示意图。
图4是是根据本申请一个实施例的氧气底吹炉铜吹炼过程参数在线预测方法流程图。
下面详细描述本申请的实施例,所述实施例的示例在附图中示出,其中自始至终相同或类似的标号表示相同或类似的元件或具有相同或类似功能的元件。下面通过参考附图描述的实施例是示例性的,旨在用于解释本申请,而不能理解为对本申请的限制。
根据本申请的一个方面,本申请提出了一种氧气底吹炉铜吹炼过程参数在线预测方法。根据本申请的实施例,参考图1所示,该预测方法包括:根据原料输入条件并基于物料平衡模型、能量平衡模型及多相平衡模型,建立关于粗铜品位预测值、渣铁硅比预测值和渣温度预测值的底吹吹炼炉机理模型;根据实际生产数据并基于目标参数和输入参数之间的粗铜品位神经网络模型、渣硅铁比神经网络模型及渣温度神经网络模型,建立关于粗铜品位预测值、渣硅铁比预测值和渣温度预测值的底吹吹炼炉数据驱动模型;利用智能协调器对机理模型和数据驱动模型进行集成,得到关于粗铜品位预测值、渣硅铁比预测值和渣温 度预测值的底吹吹炼炉混合模型,利用混合模型输出铜底吹吹炼过程中粗铜品位、渣硅铁比和渣温度的最终预测值。其中,智能协调器适于基于机理模型和数据驱动模型各自输出的粗铜品位预测值、渣硅铁比预测值、渣温度预测值与粗铜品位、渣硅铁比和渣温度的实际测量值之间的偏差,计算机理模型和数据驱动模型在混合模型中的加权系数,并根据加权系数、机理模型预测值和数据驱动模型预测值,输出粗铜品位、渣硅铁比和渣温度的最终预测值。该预测方法可以充分结合机理模型和数据驱动模型的优点,扬长避短,显著提高预测结果的准确性,有效解决现有预测方法建模造成的适应能力差、实际运行效果不理想的问题,在理论与实际应用上都具有重大意义与价值。
下面参考图1~4对本申请上述实施例的氧气底吹炉铜吹炼过程参数在线预测方法进行详细描述。
根据本申请的一个具体实施例,物料平衡模型是基于物料平衡方程建立的,能量平衡模型是基于能量平衡方程建立的,多相平衡模型是基于多相平衡方程建立的,采用METCAL软件或METSIM软件对物料平衡方程、能量平衡方程和多相平衡方程进行联立求解并结合铜底吹吹炼过程的工艺特征建立底吹吹炼炉机理模型,其中:
物料平衡方程为:
其中,
Var代表变量,
Con代表常量,M代表物料,C代表物料具有的组分,E代表组分具有的元素,X代表摩尔分数,E
c,e代表特定组分中的特定元素;
和
分别代表输入项物料和输入项物料中特定物料特定组分的摩尔分数;
和
分别代表输入项物料和输入项物料中特定物料的特定组分;
和
分别代表输入项物料和输入项物料中特定物料的摩尔分数;
和
分别代表输入项物料和输入项物料中的特定物料;
和
分别表示吹炼过程中输入项物料和输出项物料中各组分各元素之和;
和
分别表示吹炼过程中输入项物料和输出项物料中各组分之和;
和
分别表示吹炼过程中输入项物料和输出项物料中各组分各元素质量之和,该方程代表底吹吹炼过程中 输入项物料和输出项物料元素、组分及质量守恒,即物料守恒。
能量平衡方程为:
其中,ΔH
298,Ai为输入项A
i标准生成焓;ΔH
298,Bj为输出项B
j标准生成焓;Cp
Ai为输入项A
i的热容;Cp
Bj为输出项B
j的热容;Q
Loss为吹炼过程中的热损失。该方程代表底吹吹炼过程中输入项物料热量和输出项物料热量相等,即能量守恒。
多相平衡方程为:
其中,G为体系总吉布斯自由能,
为p相中纯物质c组分的标准生成吉布斯自由能;γ
pc为p相中c组分的活度因子;χ
pc为p相中c组分的摩尔数分数;C
p为p相中的组分数;T为温度;R为气体普适常数;N
pc为p相中c组分的摩尔数。该方程表示吹炼过程中***的吉布斯自由能达到最小,即***达到稳定状态。
对物料平衡方程、能量平衡方程和多相平衡方程进行联立求解时,可以采用高斯法求解得到针对每一组分每一元素的多元一次线性方程组,例如,针对某一特定组分特定元素得到的多元一次线性方程组为Ax=b:
对方程组进行初等行变换,将非奇异矩阵A逐步消元化为上三解阵:
回代求解,逐步代入计算可得方程组的解:
可以在上述计算原理基础上,结合铜底吹吹炼过程的工艺特征,采用可靠性较高的第三方软件如METCAL、METSIM等建立底吹吹炼炉机理模型,并计算预测粗铜品位、吹炼渣中铁硅比及渣温度这铜底吹吹炼过程三大重要参数。
根据本申请的再一个具体实施例,粗铜品位神经网络、渣硅铁比神经网络和渣温度神经网络分别独立地包括多个人工神经元,人工神经元包括但不限于铜底吹吹炼过程中的铜锍品位、铜锍温度、氧锍比、熔剂率、电解残极率和富氧率。其中,数据驱动模型可以采用人工神经网络的方法进行处理,人工神经网络是模仿人脑生物结构及相关功能的由大量处理单元相互联系组成的非线性的信息处理***,每个人工神经网络由多个人工神经元组成,其中神经元的模型的示意结构如图2所示,神经元模型有R个输入,每个输入都通过一个权值w和下一层相连,f为表示输入/输出关系的传递函数,为方便理解输入/输出关系的传递函数,我们以第j个神经元模型为例,第j个神经元模型的输入输出关系为:
y
j=f(s
j)=f(wp+b)
其中,S
j为输出函数,b
j为阈值,w
j,i为连接权值,x
0=b
j,w
j,0=-1,p、y分别为神经元的输入和输出。f(wp+b)为传递函数。将神经元模型对输入加权求和后
与阈值b
j比较,如果加权和超过阈值,则该神经元被激活,输出为1,否则该神经元未被激活,输出为0。
根据本申请的又一个具体实施例,可以利用BP神经网络的建模机理并结合铜底吹吹炼过程的工业生产大数据和粗铜品位神经网络、渣硅铁比神经网络及渣温度神经网络建立底吹吹炼炉数据驱动模型。BP神经网络是目前应用最广泛、成功的人工神经网络之一,它是一种多层网络的逆推学习算法,其学习过程由信号的正向传播和误差的反向传递两个过程组成,BP神经网络能学习大量的输入输出映射关系而无需揭示这种映射关系的明确的数学方程。如图3所示,BP神经网络由输入层、中间层或隐层及输出层组成,信号正向传播时,输入样本数据从输入层传入,经过隐层处理、信息变换、学习后传到输出层。若输出与期 望不符,则将误差以某种形式反向传递至隐层、输入层,并在此过程中将误差分摊给各层,作为修改各处权值、阈值的依据。随着学习过程的多次进行,权值、阈值得到不断的调整,直到输出的误差减少到可接受程度或达到预定学习次数为止。由此,通过利用BP神经网络的建模机理并结合铜底吹吹炼过程工业生产大数据构建铜底吹吹炼过程神经网络模型,利用生产大数据对模型进行训练与优化,可以进一步提高对粗铜品位、吹炼渣中铁硅比及渣温度这铜底吹吹炼三大重要参数进行预测的可靠性。
根据本申请的又一个具体实施例,智能协调器可以采用模糊划分输入的变量区域并综合的方法计算底吹吹炼炉机理模型和底吹吹炼炉数据驱动模型预测方法的加权系数,即通过智能协调器对两个预测模型基于模糊划分的集成,当底吹炉铜吹炼过程工业生产参数变化平稳、工况正常时,神经网络模型获得更大的权重,其具有的补偿作用更加确保了预测的精度;当底吹炉铜吹炼平衡受到干扰、工况不稳时,机理模型获得更大的权重,使得智能集成混合预测模型对底吹炉铜吹炼过程的全局拟合能力得到保障。由此,通过利用智能协调器将两种模型进行智能集成,不仅使得集成后的混合模型可靠性大大增强,而且利用大量神经元相互组合而成的人工神经网络还将显示出与人脑相似的特征,使混合模型具有自适应与自组织能力,由此可以显著提高预测结果的准确性。进一步地,可以利用f
1表示底吹吹炼炉机理模型输出的预测结果,利用f
2表示底吹吹炼炉数据驱动模型输出的预测结果,利用μ(x)表示底吹吹炼炉数据驱动模型在混合模型中的加权系数,利用(1-μ(x))表示底吹吹炼炉机理模型在混合模型中的加权系数,则智能协调器的输出的预测结果为:
y=f
2×μ(x)+f
1×(1-μ(x)),
其中,y代表预测结果,预测结果包括粗铜品位预测值、渣硅铁比预测值和渣温度预测值,数据驱动模型在混合模型中的加权系数μ(x)的值由隶属函数求得,μ(x)与隶属函数的关系为
x代表某一输入变量,如铜锍品位、铜锍温度、氧锍比、熔剂率、电解残极率和富氧率等;a、b、c、d为特征参数,不同的输入变量对应不同的特征参数a、b、c、d,各个特征参数可以参照实际工业生产的历史记录得到各输入变量的隶属函数参数的参考初始值,并利用工业数据进行优化得到,例如,当以铜锍品位为输入变量时,铜锍品位及其对应的特征参数a、b、c、d可以为:
{铜锍品位|a,b,c,d}={60%,67%,72%,75%}
需要说明的是,特征参数a、b、c、d的取值并非是固定不变的,可以根据实际生产进行调节,例如生产工况的不同,特征参数a、b、c、d的取值也不尽相同。进一步地,数据驱动模型在混合模型中的加权系数可以预先确定输入变量及其对应的特征参数a、b、c、d,得到各输入变量的隶属函数μ
i值,然后采用加权平均的方法计算最终的加权系数μ(x)的值,即
其中,μ
i为输入变量i及其对应的特征参数a、b、c、d计算得到的隶属函数,λ
i为输入变量i在j个输入变量中所占的权重系数,λ
i由经验值确定。
根据本申请的又一个具体实施例,当有6个输入变量时,数据驱动模型在混合模型中的加权系数μ(x)为
其中,μ
i(i=1,2,…,6)为根据某一输入变量及其对应的特征参数a、b、c、d计算得到的隶属函数,该隶属函数为
或0,λ
i(i=1,2,…,6)为某一输入变量在所有输入变量中所占的权重系数,可以参考经验值确定。
根据本申请的又一个具体实施例,可以基于实际生产结果定期或实时对混合模型进行校正,由此可以进一步提高本申请氧气底吹炉铜吹炼过程参数在线预测方法的可靠性和准确性。
根据本申请的又一个具体实施例,对混合模型进行校正可以包括:将混合模型输出的粗铜品位、渣硅铁比和渣温度的最终预测值与实际测量值进行对比:若误差在预期范围内, 保持混合模型中加权系数不变;若误差在预期范围外,将最终预测值返回至智能协调器并对加权系数进行调整,重复上述操作,直至误差降低至预期范围内,该方法尤其适用于对初步建立的混合模型进行校正,由此可以进一步提高本申请氧气底吹炉铜吹炼过程参数在线预测方法的可靠性和准确性。
根据本申请的又一个具体实施例,基于混合模型的富氧底吹吹炼过程参数在线预测方法流程图可以如图4所示,其中实时采集关键工艺参数是通过检测传感器(称重传感器、流量传感器等)对富氧底吹吹炼过程中现场原燃料的进料量和富氧气体流量等实际参数进行测量,并传输至预测模型;输入离线检测的关键参数是在人机交互界面输入出炉铜流量及成分、出炉渣流量及成分和温度等关键过程参数;建立在线预测模型指的是利用物料平衡、能量平衡及物相平衡的原理建立富氧底吹吹炼过程的机理模型及利用生产大数据构建神经网络模型(数据驱动模型),并利用智能协调器将两者进行集成,得到底吹炉吹炼过程三大重要参数预测的混合模型。利用初步建立的底吹炉吹炼过程三大重要参数预测模型进行三大重要参数的预测,并将最终预测结果与实际测量值进行对比,假如误差在要求的范围内,将模型确立,并将底吹吹炼三大重要参数预测值输入至服务器数据库存储,并在电脑界面显示;假如误差较大超过了要求的范围,返回重新对模型进行修正,并相应调整智能协调系数,重复上述步骤,直至误差减少至要求的范围内。
综上所述,根据本申请上述实施例的氧气底吹炉铜吹炼过程参数在线预测方法,通过重点考察底吹吹炼过程中粗铜品位、吹炼渣中铁硅比及渣温度这三大重要参数,分别建立底吹吹炼炉机理模型和数据驱动模型对三个重要参数进行预测,一方面利用机理模型外推性好、可解释性较强的优点,另一方面采用神经网络分析法对底吹吹炼过程三大重要参数进行大数据分析并进行预测,在此基础上设计适宜的智能协调器对二者进行集成,并将集成后混合模型的预测结果与实际生产结果对比修正,不断完善机理模型模型与数据驱动模型,并修正智能协调器参数,使其预测结果更加满足实际生产结果,从而显著提高其对氧气底吹炉铜吹炼过程中三大参数的预测精度。综上,该预测方法可以充分结合机理模型和数据驱动模型的优点,扬长避短,显著提高预测结果的准确性,有效解决现有预测方法建模造成的适应能力差、实际运行效果不理想的问题,在理论与实际应用上都具有重大意义与价值。
在本说明书的描述中,参考术语“一个实施例”、“一些实施例”、“示例”、“具体示例”、或“一些示例”等的描述意指结合该实施例或示例描述的具体特征、结构、材料或者特点包含于本申请的至少一个实施例或示例中。在本说明书中,对上述术语的示意性表述不必针对的是相同的实施例或示例。而且,描述的具体特征、结构、材料或者特点可以在任一个或多个实施例或示例中以合适的方式结合。此外,在不相互矛盾的情况下,本领域的技术人员可以将本说明书中描述的不同实施例或示例以及不同实施例或示例的特征进行结合和组合。
尽管上面已经示出和描述了本申请的实施例,可以理解的是,上述实施例是示例性的,不能理解为对本申请的限制,本领域的普通技术人员在本申请的范围内可以对上述实施例进行变化、修改、替换和变型。
Claims (9)
- 一种氧气底吹炉铜吹炼过程参数在线预测方法,其特征在于,包括:根据原料输入条件并基于物料平衡模型、能量平衡模型及多相平衡模型,建立关于粗铜品位预测值、渣铁硅比预测值和渣温度预测值的底吹吹炼炉机理模型;根据实际生产数据并基于目标参数和输入参数之间的粗铜品位神经网络模型、渣硅铁比神经网络模型及渣温度神经网络模型,建立关于粗铜品位预测值、渣硅铁比预测值和渣温度预测值的底吹吹炼炉数据驱动模型;利用智能协调器对所述机理模型和所述数据驱动模型进行集成,得到关于粗铜品位预测值、渣硅铁比预测值和渣温度预测值的底吹吹炼炉混合模型,利用所述混合模型输出铜底吹吹炼过程中粗铜品位、渣硅铁比和渣温度的最终预测值,其中,所述智能协调器适于基于所述机理模型和所述数据驱动模型各自输出的粗铜品位预测值、渣硅铁比预测值、渣温度预测值与粗铜品位、渣硅铁比和渣温度的实际测量值之间的偏差,计算所述机理模型和所述数据驱动模型在所述混合模型中的加权系数,并根据所述加权系数、所述机理模型预测值和所述数据驱动模型预测值,输出粗铜品位、渣硅铁比和渣温度的最终预测值。
- 根据权利要求1所述的在线预测方法,其特征在于,所述物料平衡模型是基于物料平衡方程建立的,所述能量平衡模型是基于能量平衡方程建立的,所述多相平衡模型是基于多相平衡方程建立的,采用METCAL软件或METSIM软件对所述物料平衡方程、所述能量平衡方程和所述多相平衡方程进行联立求解并结合铜底吹吹炼过程的工艺特征建立所述机理模型。
- 根据权利要求1所述的在线预测方法,其特征在于,所述粗铜品位神经网络、所述渣硅铁比神经网络和所述渣温度神经网络分别独立地包括多个人工神经元,所述人工神经元包括但不限于铜底吹吹炼过程中的铜锍品位、铜锍温度、氧锍比、熔剂率、电解残极率和富氧率。
- 根据权利要求1所述的在线预测方法,其特征在于,利用BP神经网络的建模机理并结合铜底吹吹炼过程的工业生产大数据和粗铜品位神经网络、渣硅铁比神经网络及渣温度神经网络建立所述数据驱动模型。
- 根据权利要求1所述的在线预测方法,其特征在于,基于实际生产结果定期或实时对所述混合模型进行校正。
- 根据权利要求1所述的在线预测方法,其特征在于,对所述混合模型进行校正包括:将所述混合模型输出的粗铜品位、渣硅铁比和渣温度的最终预测值与实际测量值进行对比:若误差在预期范围内,保持所述混合模型中所述加权系数不变;若误差在预期范围外,将所述最终预测值返回至所述智能协调器并对所述加权系数进行调整,重复上述操作,直至误差降低至预期范围内。
- 根据权利要求1~6中任一项所述的在线预测方法,其特征在于,所述智能协调器采用模糊划分输入的变量区域并综合的方法计算所述机理模型和所述数据驱动模型预测方法的加权系数。
- 根据权利要求7所述的在线预测方法,其特征在于,利用f 1表示所述机理模型输出的预测结果,利用f 2表示所述数据驱动模型输出的预测结果,利用μ(x)表示所述数据驱动模型在所述混合模型中的加权系数,利用(1-μ(x))表示所述机理模型在所述混合模型中的加权系数,所述智能协调器的输出的预测结果为:y=f 2×μ(x)+f 1×(1-μ(x)),其中,y代表预测结果,所述预测结果包括粗铜品位预测值、渣硅铁比预测值和渣温度预测值,所述数据驱动模型在所述混合模型中的加权系数μ(x)为:x代表输入变量,所述输入变量的选择范围包括但不限于铜锍品位、铜锍温度、氧锍比、熔剂率、电解残极率和富氧率;a、b、c、d为根据实际工业生产的技术数据得到的与所述输入变量对应的特征参数,所述特征参数决定所述输入变量的隶属函数。
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010281411.8 | 2020-04-10 | ||
CN202010281411.8A CN111598293A (zh) | 2020-04-10 | 2020-04-10 | 氧气底吹炉铜吹炼过程参数在线预测方法 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2021203912A1 true WO2021203912A1 (zh) | 2021-10-14 |
Family
ID=72184921
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2021/080600 WO2021203912A1 (zh) | 2020-04-10 | 2021-03-12 | 氧气底吹炉铜吹炼过程参数在线预测方法 |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN111598293A (zh) |
WO (1) | WO2021203912A1 (zh) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114219175A (zh) * | 2021-12-28 | 2022-03-22 | 鞍钢集团自动化有限公司 | 一种集装箱卷板的力学性能预测方法 |
CN114593663A (zh) * | 2022-02-23 | 2022-06-07 | 本钢板材股份有限公司 | 一种基于副边电流模型的精炼lf炉渣厚测量方法 |
CN116258087A (zh) * | 2023-05-15 | 2023-06-13 | 矿冶科技集团有限公司 | 冰铜品位软测量方法、装置、电子设备及存储介质 |
CN117690505A (zh) * | 2024-01-25 | 2024-03-12 | 昆明理工大学 | 一种lstm融合机理模型预测富氧底吹铜熔炼关键参数的方法 |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111598293A (zh) * | 2020-04-10 | 2020-08-28 | 中国恩菲工程技术有限公司 | 氧气底吹炉铜吹炼过程参数在线预测方法 |
CN112434961B (zh) * | 2020-12-01 | 2022-04-15 | 内蒙古科技大学 | 一种铁钢界面铁水温降预测方法、装置和终端设备 |
CN114217592B (zh) * | 2021-12-17 | 2024-06-28 | 北京瑞太智联技术有限公司 | 一种用于铜顶吹生产过程的控制***及方法 |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
RU2297463C1 (ru) * | 2006-02-09 | 2007-04-20 | Государственное образовательное учреждение высшего профессионального образования "Санкт-Петербургский государственный горный институт имени Г.В. Плеханова (технический университет)" | Способ управления газовым режимом конвертерного передела цветной металлургии |
CN101482750A (zh) * | 2009-02-06 | 2009-07-15 | 北京矿冶研究总院 | 湿法冶金合成过程草酸钴粒度预测方法 |
CN104296801A (zh) * | 2014-06-12 | 2015-01-21 | 东北大学 | 湿法冶金浓密洗涤过程关键变量检测方法 |
CN104328285A (zh) * | 2014-10-29 | 2015-02-04 | 中国科学院沈阳自动化研究所 | 基于混合模型的富氧底吹炼铜过程参数在线估计方法 |
CN111554353A (zh) * | 2020-04-10 | 2020-08-18 | 中国恩菲工程技术有限公司 | 氧气底吹炉铜熔炼过程参数在线预测方法 |
CN111598293A (zh) * | 2020-04-10 | 2020-08-28 | 中国恩菲工程技术有限公司 | 氧气底吹炉铜吹炼过程参数在线预测方法 |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101353729A (zh) * | 2008-07-18 | 2009-01-28 | 中南大学 | 一种基于工况判断的智能集成建模方法 |
CN105624425B (zh) * | 2014-11-05 | 2017-09-22 | 中国科学院沈阳自动化研究所 | 一种氧气底吹铜熔炼过程智能控制方法 |
-
2020
- 2020-04-10 CN CN202010281411.8A patent/CN111598293A/zh active Pending
-
2021
- 2021-03-12 WO PCT/CN2021/080600 patent/WO2021203912A1/zh active Application Filing
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
RU2297463C1 (ru) * | 2006-02-09 | 2007-04-20 | Государственное образовательное учреждение высшего профессионального образования "Санкт-Петербургский государственный горный институт имени Г.В. Плеханова (технический университет)" | Способ управления газовым режимом конвертерного передела цветной металлургии |
CN101482750A (zh) * | 2009-02-06 | 2009-07-15 | 北京矿冶研究总院 | 湿法冶金合成过程草酸钴粒度预测方法 |
CN104296801A (zh) * | 2014-06-12 | 2015-01-21 | 东北大学 | 湿法冶金浓密洗涤过程关键变量检测方法 |
CN104328285A (zh) * | 2014-10-29 | 2015-02-04 | 中国科学院沈阳自动化研究所 | 基于混合模型的富氧底吹炼铜过程参数在线估计方法 |
CN111554353A (zh) * | 2020-04-10 | 2020-08-18 | 中国恩菲工程技术有限公司 | 氧气底吹炉铜熔炼过程参数在线预测方法 |
CN111598293A (zh) * | 2020-04-10 | 2020-08-28 | 中国恩菲工程技术有限公司 | 氧气底吹炉铜吹炼过程参数在线预测方法 |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114219175A (zh) * | 2021-12-28 | 2022-03-22 | 鞍钢集团自动化有限公司 | 一种集装箱卷板的力学性能预测方法 |
CN114593663A (zh) * | 2022-02-23 | 2022-06-07 | 本钢板材股份有限公司 | 一种基于副边电流模型的精炼lf炉渣厚测量方法 |
CN114593663B (zh) * | 2022-02-23 | 2023-10-03 | 本钢板材股份有限公司 | 一种基于副边电流模型的精炼lf炉渣厚测量方法 |
CN116258087A (zh) * | 2023-05-15 | 2023-06-13 | 矿冶科技集团有限公司 | 冰铜品位软测量方法、装置、电子设备及存储介质 |
CN117690505A (zh) * | 2024-01-25 | 2024-03-12 | 昆明理工大学 | 一种lstm融合机理模型预测富氧底吹铜熔炼关键参数的方法 |
Also Published As
Publication number | Publication date |
---|---|
CN111598293A (zh) | 2020-08-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2021203912A1 (zh) | 氧气底吹炉铜吹炼过程参数在线预测方法 | |
CN111554353A (zh) | 氧气底吹炉铜熔炼过程参数在线预测方法 | |
Chen et al. | Modelling and optimization of fed-batch fermentation processes using dynamic neural networks and genetic algorithms | |
CN108676955A (zh) | 一种转炉炼钢终点碳含量和温度控制方法 | |
CN102540879A (zh) | 基于群决策检索策略的多目标评价优化方法 | |
Jiang et al. | Real-time moisture control in sintering process using offline–online NARX neural networks | |
CN110322014A (zh) | 一种基于bp神经网络的水泥成品比表面积预测方法 | |
CN109359320B (zh) | 基于多采样率自回归分布滞后模型的高炉指标预测方法 | |
CN110097929A (zh) | 一种高炉铁水硅含量在线预测方法 | |
Wang et al. | Strip hardness prediction in continuous annealing using multiobjective sparse nonlinear ensemble learning with evolutionary feature selection | |
CN115620846A (zh) | 一种活性锰材料制备及控制方法 | |
Wang et al. | Optimization of aluminum fluoride addition in aluminum electrolysis process based on pruned sparse fuzzy neural network | |
CN112881601A (zh) | 一种基于云平台的水分检测*** | |
CN112083694A (zh) | 氧气底吹铜吹炼过程反馈控制方法、装置及电子设备 | |
Li et al. | Dual ensemble online modeling for dynamic estimation of hot metal silicon content in blast furnace system | |
CN101285816A (zh) | 铜锍吹炼的过程参数软测量仪表及其软测量方法 | |
Liang et al. | A transfer predictive control method based on inter-domain mapping learning with application to industrial roasting process | |
Yang et al. | Modeling and optimal-setting control of blending process in a metallurgical industry | |
Tang et al. | A constrained multi-objective deep reinforcement learning approach for temperature field optimization of zinc oxide rotary volatile kiln | |
Zhu et al. | Temperature prediction of aluminum reduction cell based on integration of dual attention LSTM for non-stationary sub-sequence and ARMA for stationary sub-sequences | |
Xu et al. | Soft sensor for ammonia concentration at the ammonia converter outlet based on an improved particle swarm optimization and BP neural network | |
Mi et al. | Prediction of accumulated temperature in vegetation period using artificial neural network | |
US20240002964A1 (en) | Method and system for determining converter tapping quantity | |
Wu et al. | An expert control system using neural networks for the electrolytic process in zinc hydrometallurgy | |
Ye et al. | Novel weight-adaptive fusion grey prediction model based on interval sequences and its applications |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 21784750 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 21784750 Country of ref document: EP Kind code of ref document: A1 |