CN109935280B - Blast furnace molten iron quality prediction system and method based on ensemble learning - Google Patents
Blast furnace molten iron quality prediction system and method based on ensemble learning Download PDFInfo
- Publication number
- CN109935280B CN109935280B CN201910163699.6A CN201910163699A CN109935280B CN 109935280 B CN109935280 B CN 109935280B CN 201910163699 A CN201910163699 A CN 201910163699A CN 109935280 B CN109935280 B CN 109935280B
- Authority
- CN
- China
- Prior art keywords
- blast furnace
- molten iron
- data
- furnace
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Landscapes
- Manufacture Of Iron (AREA)
Abstract
The invention provides a blast furnace molten iron quality prediction system and method based on ensemble learning, which comprises the following steps: the system comprises a blast furnace, a hot blast furnace, a first flowmeter, a second flowmeter, a third flowmeter, a thermometer, a pressure gauge, a hygrometer, a furnace belly gas measurement analyzer, an oxygen enrichment rate measurement analyzer, a data acquisition device and a computer; the acquired real-time blast furnace data to be tested are preprocessed to acquire input and output parameters, and an online forecasting model of the quality of the multi-element molten iron of the established root mean square error probability weighted and integrated random weight neural network is used to acquire an online forecasting result. The invention can avoid the hysteresis of off-line test and the uncertainty caused by manual operation, realize real-time on-line accurate soft measurement, provide key indexes for field operators to accurately judge the internal running state of the blast furnace in time, and simultaneously update the parameters of the soft measurement model by using the latest process data according to the change of working conditions, thereby avoiding the limitation of a time-invariant model and having high practical value.
Description
Technical Field
The invention belongs to the technical field of blast furnace smelting automation control, and particularly relates to a blast furnace molten iron quality prediction system and method based on integrated learning.
Background
Blast furnace is a large convection reactor and heat exchanger in the iron-making process, and blast furnace iron-making is an important link in social development. However, the smelting environment in the blast furnace is extremely harsh, the temperature of the most severe reaction region reaches more than 2000 ℃, the pressure reaches about 4 times of the standard atmospheric pressure, and the internal state of the blast furnace is difficult to monitor in real time along with the coexistence state of solid, liquid and gas phases, so that the blast furnace is difficult to optimally control. At present, the indexes widely used for indirectly reflecting the internal state of the blast furnace are molten iron quality parameters, the comprehensive molten iron quality indexes are usually measured by adopting Si content, P content, S content and molten iron temperature, the measurement of the molten iron quality parameters is generally measured by adopting an off-line assay method, and the measurement result lags 1-2 hours, so the result cannot reflect the internal state of the blast furnace in real time. In order to realize real-time and comprehensive monitoring of the internal operation state of the blast furnace, an online soft measurement model of blast furnace multi-element molten iron quality parameters is required to be established so as to realize real-time online soft measurement of molten iron quality, and a data-driven online soft measurement model of blast furnace multi-element molten iron quality is established by fully utilizing detectable operation data in the blast furnace ironmaking process.
Patent publication No. CN102031319A discloses a method for forecasting the silicon content of blast furnace molten iron, which comprises data parameter selection and pretreatment, a prediction algorithm, result output and operation guidance, wherein the data parameter selection adopts five parameters of a short-term average value of silicon content, a medium-term average value of silicon content, a long-term average value of silicon content, a theoretical combustion temperature of a tuyere corresponding to the former molten iron and the sulfur content of the former molten iron, and the silicon content is forecasted by a neural network algorithm.
Patent publication No. CN101211383A discloses a feature analysis forecasting method for the silicon content in blast furnace molten iron, which forecasts the silicon (Si) content in the blast furnace molten iron by using a least square support vector machine introduced with genetic algorithm optimization. According to the method, the characteristics of sample data are extracted by performing exponential weighted moving average filtering and normalization preprocessing on an original sample, and a dynamic recursive model for forecasting the silicon content of the molten iron in the blast furnace is established.
Patent publication No. CN103320559A discloses a method for forecasting sulfur content of blast furnace molten iron, which forecasts the sulfur content of next molten iron based on a radial basis function neural network modeling technology by taking a short-term average value of sulfur content, a medium-term average value of sulfur content, a long-term average value of sulfur content, slag alkalinity corresponding to previous molten iron, silicon content of previous molten iron, coke S content entering a furnace and coal powder S content entering the furnace as input variables for forecasting the sulfur content of the molten iron.
The method reported in the above patent and the methods in other related documents are soft measurement methods for single molten iron quality parameters, such as molten iron temperature, si content, S content, etc., and the single molten iron quality parameters cannot comprehensively reflect the complex state in the blast furnace, cannot provide comprehensive guidance for field operators, and have low practical application value. Moreover, models established by the existing molten iron quality soft measurement method are all parameter time-invariant models and are not suitable for the slow time-varying characteristics of the working conditions of the blast furnace ironmaking process, so that the prediction results are inaccurate when the actual working conditions of the models change. In addition, most of the existing soft measurement methods are complex in model structure, the soft measurement model is long in calculation time, and a simple and quick soft measurement method is needed in an industrial field, so that the method is low in efficiency in practical application. In summary, at present, no method for performing multi-element dynamic rapid online soft measurement aiming at the quality parameters of the molten iron, namely Si content, P content, S content and molten iron temperature in the blast furnace smelting process exists at home and abroad.
Disclosure of Invention
Aiming at the technical defects, the invention provides a blast furnace molten iron quality prediction system and method based on ensemble learning; specifically, a blast furnace molten iron quality prediction system based on ensemble learning includes: the system comprises a blast furnace, a hot blast furnace, a first flowmeter, a second flowmeter, a third flowmeter, a thermometer, a pressure gauge, a hygrometer, a furnace belly gas measurement analyzer, an oxygen enrichment rate measurement analyzer, a data acquisition device and a computer;
a sample to be tested is placed into the blast furnace from a blast furnace inlet, pulverized coal injection is carried out from a blast furnace belly tuyere, a first flowmeter is arranged at the pulverized coal injection position, and the first flowmeter is respectively connected with a data acquisition device and a belly coal gas measurement analyzer; the thermometer is arranged at the air outlet of the hot blast stove and is connected with the data acquisition device; the pressure gauge is arranged at the air outlet of the hot blast stove and is connected with the data acquisition device; the second flowmeter, the third flowmeter and the hygrometer are respectively installed at an air inlet of the hot blast stove, the second flowmeter is respectively connected with the furnace belly gas measurement analyzer and the oxygen enrichment rate measurement analyzer, and the third flowmeter is respectively connected with the furnace belly gas measurement analyzer and the oxygen enrichment rate measurement analyzer; the hygrometer is respectively connected with the furnace belly gas measurement analyzer and the data acquisition device; the furnace belly gas measurement analyzer and the oxygen enrichment rate measurement analyzer are respectively connected with the data acquisition device; the data acquisition device is connected with the computer through a communication bus;
the blast furnace is used for completing a blast furnace smelting process;
the hot blast stove is used for conveying oxygen-enriched cold air to the hot blast stove and conveying hot air to the interior of the blast furnace;
the first flowmeter is arranged at the coal powder injection position and used for measuring the coal powder injection amount on line and transmitting the coal powder injection amount measured value to the furnace belly coal gas measurement analyzer and the data acquisition device respectively;
the second flowmeter is arranged at an inlet of the hot blast stove and is used for measuring the oxygen-enriched flow on line and transmitting the measured value of the oxygen-enriched flow to the furnace belly gas measurement analyzer and the oxygen enrichment rate measurement analyzer respectively;
the third flowmeter is arranged at an inlet of the hot blast stove and is used for measuring the flow of cold air on line and transmitting the measured value of the flow of the cold air to the furnace belly coal gas measuring analyzer and the oxygen enrichment rate measuring analyzer respectively;
the thermometer is arranged at the outlet of the hot blast stove and used for measuring the hot blast temperature of the hot blast stove on line and transmitting the measured value of the hot blast temperature to the data acquisition device;
the pressure gauge is arranged at the outlet of the hot blast stove and used for measuring the hot blast pressure of the hot blast stove on line and transmitting the measured value of the hot blast pressure to the data acquisition device;
the hygrometer is arranged at an inlet of the hot blast stove and used for measuring the blast humidity of the hot blast stove on line and transmitting the blast humidity measured value to the furnace belly gas measurement analyzer, the oxygen enrichment rate measurement analyzer and the data acquisition device respectively;
the furnace bosh gas measuring analyzer analyzes a coal powder injection amount measured value measured by the first flowmeter, an oxygen-enriched flow amount measured value measured by the second flowmeter, a cold air flow measurement value measured by the third flowmeter and an air blast humidity measured value measured by the hygrometer, calculates a furnace bosh gas amount parameter and transmits the furnace bosh gas amount parameter to the data acquisition device;
the oxygen enrichment rate measurement analyzer analyzes an oxygen enrichment flow measurement value measured by the second flowmeter, a cold air flow measurement value measured by the third flowmeter and an air blowing humidity measurement value measured by the hygrometer, calculates an oxygen enrichment rate parameter and transmits the oxygen enrichment rate parameter to the data acquisition device;
the data acquisition device preprocesses the coal powder injection quantity measured value transmitted by the first flowmeter, the furnace belly coal gas quantity parameter transmitted by the furnace belly coal gas measurement analyzer, the hot air temperature measured value transmitted by the thermometer, the hot air pressure measured value transmitted by the pressure gauge, the oxygen enrichment rate parameter transmitted by the oxygen enrichment rate measurement analyzer and the air blast humidity measured value transmitted by the hygrometer in the data acquisition device, and transmits the preprocessed data result to the computer;
the computer uses the preprocessed data result transmitted by the data acquisition device and stores the data result in the computer according to a time sequence, adopts a blast furnace molten iron quality prediction method based on ensemble learning to predict, and performs online prediction on the quality index of the multi-element molten iron by establishing a multi-element molten iron quality online prediction model of a root mean square error probability weighted integrated Random weighted neural network (RVFLNs), so as to obtain the predicted value of the quality index of the multi-element molten iron.
The computer system is provided with OPC communication software for data bidirectional communication between the computer and the data acquisition device.
A blast furnace molten iron quality prediction method based on ensemble learning is realized by using a blast furnace molten iron quality prediction system based on ensemble learning, and comprises the following specific steps:
step 1: acquiring original historical data of a blast furnace, and preprocessing the data, wherein the preprocessing comprises the following steps: unifying time granularity of data, eliminating damping-down data, abnormal data and normalized data, and specifically comprising the following steps:
step 1.1: marking the collected data according to the time sequence, and manually matching according to the nearest neighbor time principle, namely according to the time sequence to obtain the historical data of the blast furnace ironmaking process with consistent time granularity;
step 1.2: eliminating damping data and abnormal data: eliminating damping data in a specific mode: determining a planned maintenance time period of the blast furnace according to the shift-changing record, and eliminating damping-down data of the blast furnace body in the time period, wherein the damping-down data specifically refers to data that the hot blast furnace does not blow air to the blast furnace; eliminating abnormal data, namely eliminating abnormal values by adopting a Lauda criterion, namely a 3 sigma criterion, namely eliminating data with data deviation larger than 3 sigma; σ is a standard deviation of the screened blast furnace body data, and is shown by the following formula:
step 1.3: carrying out data normalization processing on blast furnace data to obtain normalized blast furnace historical data serving as a sample data set, wherein the formula is as follows:
wherein x is i 、
Respectively before and after normalization of the ith variable, max (x) i )、min(x i ) The maximum value and the minimum value of the ith variable are respectively, and the data after normalization processing is in the following range: x is a radical of a fluorine atom i ∈(0,1)；
Step 2: acquiring input and output parameters required by blast furnace multi-element molten iron quality index soft measurement, and extracting a sample data set from preprocessed blast furnace original historical data according to the input and output parameters;
step 2.1: determining the output parameter of the quality of the blast furnace molten iron needing soft measurement as Si (silicon) content y according to the process mechanism of the blast furnace 1 (%), P (phosphorus) content y 2 (%), S (Sulfur) content y 3 (%) and the temperature y of the molten iron 4 (℃)；
Step 2.2: a grey correlation analysis method is adopted, and the first k blast furnace body parameters with the highest correlation degree are extracted to serve as auxiliary variables of soft measurement, and the method comprises the following steps: gas flow u of furnace chamber 1 (m 3 ) Temperature u of hot air 2 (° c), hot air pressure u 3 (KPa) and oxygen enrichment ratio u 4 Humidity of blast airu 5 (RH) amount of coal injected u 6 (m 3 /h)。
Step 2.3: according to the process dynamic characteristics, a non-linear autoregressive model (NARX) is introduced based on the 6 auxiliary variables, and 16 variables are determined as input variables of a soft measurement model as follows:
gas flow u of furnace chamber at current moment 1 (t)(m 3 ) (ii) a Hot air temperature u at present 2 (t) (° c); hot air pressure u at present 3 (t) (KPa); oxygen enrichment rate u at the present moment 4 (t); current moment blast humidity u 5 (t) (RH); setting coal injection quantity u at current moment 6 (t)(m 3 H); si content y at the last moment 1 (t-1) (%); p content y at the last moment 2 (t-1) (%); last moment furnace bosh coal gas quantity u 1 (t-1)(m 3 ) (ii) a Last moment hot air temperature u 2 (t-1) (. Degree.C.); last moment hot air pressure u 3 (t-1) (KPa); last moment oxygen enrichment rate u 4 (t-1); last moment blast air humidity u 5 (t-1) (RH); setting the coal injection amount u at last moment 6 (t-1)(m 3 H); s content y at the last moment 3 (t-1) (%); molten iron temperature y at last moment 4 (t-1)(℃)；
Step 2.4: extracting a sample data set from the preprocessed original blast furnace historical data according to the input and output parameters;
and step 3: establishing a multi-element molten iron quality online forecasting model of weighted integration RVFLNs of root mean square error probability based on input and output parameters required by soft measurement of multi-element molten iron quality indexes of the blast furnace;
step 3.1: and establishing a multi-element molten iron quality online forecasting model of the weighted integration of the root mean square error probability RVFLNs.
Step 3.2: if the multi-element molten iron quality online forecasting models of the RVFLNs are weighted and integrated by the root mean square error probability, the absolute error between the blast furnace molten iron quality output parameters obtained by model training and the corresponding actual blast furnace molten iron quality output parameters is less than or equal to a set absolute error value, and the model training is finished to obtain the final multi-element molten iron quality online forecasting models of the RVFLNs; if the absolute error between the blast furnace molten iron quality output parameter obtained by the model training and the corresponding actual blast furnace molten iron quality output parameter is larger than the set absolute error value, retraining the multi-element molten iron quality online forecasting model of the root mean square error probability weighting integration RVFLNs, turning to the step 3.1 until the absolute error between the blast furnace molten iron quality output parameter obtained by the model training and the corresponding actual blast furnace molten iron quality output parameter is smaller than or equal to the set absolute error value, finishing the model training, and obtaining the final multi-element molten iron quality online forecasting model of the root mean square error probability weighting integration RVFLNs;
the step 3.1 of establishing a weighted integrated RVFLNs multi-element molten iron quality online forecasting model comprises the following specific steps:
step 3.1.1: dividing the M groups of data in the sample data set into two sample sets D 1 、D 2 Wherein D is 1 For training the sample set, take the top M 1 Group data, D 2 To test the sample set, take the remaining M 2 Group data.
Step 3.1.2: the method comprises the following steps of performing a sample-back experiment on a training sample set based on a Bootstrap idea to obtain N sub-training sample sets, wherein the specific method comprises the following steps: at M 1 In the group training sample set, performing m times of random sampling experiments with replacement, and considering modeling efficiency, performing N groups of random sampling experiments with replacement to obtain N sub-sample sets, wherein the number of data of the sub-sample sets is m;
step 3.1.3: and establishing a sub-model by using the sub-sample set, and modeling by using a random weight neural network (RVFLNs) model as the sub-model.
Step 3.1.4: calculating the root mean square error RMSE of each sub-model j As shown in the following formula,
wherein, y ji For the actual value of the ith output variable in the jth sub-model,
is the jth sub-modelEstimated values of i output variables, RMSE j The root mean square error of the jth sub-model is obtained;
step 3.1.5: estimating a Root Mean Square Error (RMSE) probability density curve of each submodel by using a kernel density estimation method, and solving a root mean square error probability distribution curve of each submodel;
the method of kernel density estimation is as follows:
suppose z i e.R, i =1, …, n is an independent same distribution random variable subject to a distribution density function of f (z), z is e.R, then f (z) kernel density estimate
Is defined as:
where φ (·) is called a kernel function, h p Usually called window width or smoothing parameter, is a positive number that is manually pre-specified.
Using KDE method for root mean square error sample set { RMSE i I =1,2 … K } to obtain an estimated error probability density function gamma RMSE Comprises the following steps:
selection of the phi (-) kernel function: the following conditions are satisfied:
a.φ(z)≥0
b.∫φ(z)du＝1
selecting a Gaussian kernel function, wherein the table formula is as follows:
h p selection of the window width: width h of window p Is set to h p ＝1.06θK -1/5 Where θ is defined by min { S,0.746Q } estimate, S represents the sample standard deviation, Q is the interquartile range, K is the RMSE sample set number;
step 3.1.6: distributing weights to the submodels, and obtaining a multi-element molten iron quality online forecasting model integrating RVFLNs in a weighted manner by root mean square error probability: taking the probability of each sub-model in the root mean square error probability curve of the sub-models as the respective weight, then carrying out weighted summation to obtain the multielement molten iron quality online forecasting model of the root mean square error probability weighted integration RVFLNs, and solving as follows:
wherein, w i The weights determined for the proposed method and the condition met is that the sum of all weights is 1.RVFLNs i And representing the ith sub-model, N represents the number of the sub-models, and the RVFLNs are multi-element molten iron quality online forecasting models of the weighted integration RVFLNs with the obtained root mean square error probability.
And 4, step 4: online prediction is carried out on real-time collected blast furnace data according to a multi-element molten iron quality online prediction model of weighted and integrated RVFLNs (root mean square error probabilities) to obtain an online prediction result: the method comprises the steps of preprocessing acquired real-time blast furnace data to be tested in step 1, acquiring input and output parameters required by blast furnace multi-element molten iron quality index soft measurement in step 2, extracting the data to be tested from preprocessed blast furnace original historical data according to the input and output parameters, and using an established multi-element molten iron quality online forecasting model of root mean square error probability weighted integration RVFLNs to obtain an online forecasting result.
The beneficial technical effects are as follows:
in order to solve the defects of the online soft measurement method of the molten iron quality parameters in the blast furnace smelting process, the modeling is carried out on the basis of the weighted integration RVFLNs algorithm of root mean square error probability; then, in order to improve the modeling efficiency and reduce the calculation complexity, extracting the key variable with the strongest correlation degree with the quality index of the multi-element molten iron by adopting a grey correlation analysis method as an input variable of modeling; secondly, in order to better reflect the Nonlinear dynamic characteristics of blast furnace ironmaking, an online estimation NARX (Nonlinear autoregressive exogenous) model is introduced; meanwhile, in order to improve the precision and efficiency of modeling, a root mean square error probability weighted-integrated Random-weight neural network (RVFLNs) method is provided for establishing a multi-element molten iron quality model based on data driving. The method realizes the simultaneous multi-element online dynamic soft measurement of comprehensive molten iron quality indexes, namely Si content, P content, S content and molten iron temperature, and can update the model parameters online through the latest measurement data along with the change of the working condition of the blast furnace. Based on the online detection value of the blast furnace body parameter provided by the conventional online detection equipment of the steel mill, the online soft measurement value of the multi-element molten iron quality index at the current moment can be obtained, and the key molten iron quality index is provided for the optimized operation control of the blast furnace ironmaking process. The method is based on the fact that blast furnace body parameters obtained by real-time measurement of conventional detection equipment on an industrial site are used as input data of a model, the time sequence relation between the hysteresis characteristic and the input and output variables of the blast furnace smelting process is fully considered, and dynamic online soft measurement of the quality indexes of the multielement molten iron, including Si content, P content and S content and the molten iron temperature in the blast furnace smelting process is achieved. Compared with the existing method for offline manual detection of the quality indexes of the multi-element molten iron, the method can avoid the hysteresis of offline assay and the uncertainty caused by manual operation, realize real-time online accurate soft measurement, and provide key indexes for field operators to timely and accurately judge the internal operating state of the blast furnace. Meanwhile, the method can update the parameters of the soft measurement model by using the latest process data according to the change of the working condition, avoids the limitation of a time-invariant model, and has high practical value. In addition, the method is favorable for further realizing the operation optimization control of the blast furnace ironmaking process.
Drawings
FIG. 1 is a diagram of a detection apparatus for a blast furnace ironmaking process according to an embodiment of the present invention;
FIG. 2 is a flowchart of a method for predicting the quality of molten iron in a blast furnace based on ensemble learning according to an embodiment of the present invention;
FIG. 3 is a flow chart of a RMS error probability weighted integration RVFLNs modeling process of an embodiment of the present invention;
fig. 4 is a diagram showing the effect of the soft measurement result of the quality index of the multi-element molten iron of the blast furnace according to the embodiment of the present invention, wherein (a) is a comparison curve of the predicted value of the silicon content at the current moment and the actual value; (b) A comparison curve of the predicted value of the phosphorus content at the current moment and the actual value is obtained; (c) A comparison curve of the predicted value of the sulfur content at the current moment and the actual value is obtained; (d) A comparison curve of the predicted value and the actual value of the molten iron temperature at the current moment is obtained;
FIG. 5 is a plot of a probability density function of root mean square error for an embodiment of the present invention;
FIG. 6 is a plot of a probability function of root mean square error for an embodiment of the present invention;
in the figure: 1-blast furnace, 2-hot blast furnace, 3-1-first flowmeter, 3-2-second flowmeter, 3-3 third flowmeter, 4-thermometer, 5-pressure gauge, 6-hygrometer, 7-belly coal gas measuring analyzer, 8-oxygen enrichment rate measuring analyzer, 9-data acquisition device, 10-computer, 11-ore to be tested, coke and solvent, 12-pulverized coal injection, 13-hot blast, 14-oxygen enrichment cold air, 15-residue after smelting, mu 1 The furnace gas flow parameter, μ 2 Measured value of temperature of hot air, μ 3 Hot air pressure measurement, μ 4 -oxygen enrichment factor parameter, μ 5 Blast humidity measurement, μ 6 -a measurement of the amount of coal dust injected.
Detailed Description
The invention is further explained by combining the attached drawings and the concrete embodiment, and provides a blast furnace molten iron quality prediction system and a blast furnace molten iron quality prediction method based on integrated learning; specifically, a blast furnace molten iron quality prediction system based on ensemble learning, as shown in fig. 1, includes: the system comprises a blast furnace 1, a hot blast furnace 2, a first flowmeter 3-1, a second flowmeter 3-2, a third flowmeter 3-3, a thermometer 4, a pressure gauge 5, a hygrometer 6, a furnace belly gas measurement analyzer 7, an oxygen enrichment rate measurement analyzer 8, a data acquisition device 9 and a computer 10;
ore, coke and solvent 11 to be tested are placed into a blast furnace 1 from an inlet of the blast furnace 1, pulverized coal injection 12 is carried out from a tuyere of a furnace belly of the blast furnace 1, a first flow meter 3-1 is installed at the pulverized coal injection 12, and the first flow meter 3-1 is respectively connected with a data acquisition device 9 and a furnace belly coal gas measurement analyzer 7; the thermometer 4 is arranged at the air outlet of the hot blast stove 2, and the thermometer 4 is connected with the data acquisition device 9; the pressure gauge 5 is arranged at the air outlet of the hot blast stove 2, and the pressure gauge 5 is connected with the data acquisition device 9; the second flowmeter 3-2, the third flowmeter 3-3 and the hygrometer 6 are respectively installed at an air inlet of the hot blast stove 2, the second flowmeter 3-2 is respectively connected with the furnace belly gas measurement analyzer 7 and the oxygen enrichment rate measurement analyzer 8, and the third flowmeter 3-3 is respectively connected with the furnace belly gas measurement analyzer 7 and the oxygen enrichment rate measurement analyzer 8; the hygrometer 6 is respectively connected with the furnace belly gas measurement analyzer 7 and the data acquisition device 9; the furnace bosh gas measuring analyzer 7 and the oxygen enrichment rate measuring analyzer 8 are respectively connected with a data acquisition device 9; the data acquisition device 9 is connected with a computer 10 through a communication bus;
the hot blast stove 2 conveys oxygen-enriched cold air 14 to the hot blast stove 2, and conveys hot air 13 to the interior of the blast furnace 1 through the hot blast stove 2;
the first flowmeter 3-1 is arranged at the coal powder injection position 12 and is used for measuring the coal powder injection amount on line and measuring the coal powder injection amount value mu 6 Respectively transmitted to a furnace bosh gas measuring analyzer 7 and a data acquisition device 9;
the second flowmeter 3-2 is arranged at an inlet of the hot blast stove 2 and is used for measuring the oxygen-enriched flow on line and transmitting the measured value of the oxygen-enriched flow to the furnace belly gas measuring analyzer 7 and the oxygen enrichment rate measuring analyzer 8 respectively;
the third flow meter 3-3 is arranged at an inlet of the hot blast stove 2 and is used for measuring the flow of cold air on line and transmitting the measured value of the flow of the cold air to the furnace belly coal gas measuring analyzer 7 and the oxygen enrichment rate measuring analyzer 8 respectively;
the thermometer 4 is arranged at the outlet of the hot blast stove 2 and is used for measuring the hot blast temperature of the hot blast stove 2 on line and measuring the value mu of the measured hot blast temperature 2 Transmitted to the data acquisition device 9;
the pressure gauge 5 is arranged at the outlet of the hot blast stove 2 and is used for measuring the hot blast pressure of the hot blast stove 2 on line and measuring the hot blast pressure value mu 3 Transmitting the data to a data acquisition device;
the hygrometer 6 is arranged at the inlet of the hot blast stove 2 and is used for measuring the blast humidity of the hot blast stove on line and measuring the blast humidity value mu 5 Respectively transmitted to a furnace belly gas measurement analyzer 7, an oxygen enrichment rate measurement analyzer 8 and a data acquisition device 9;
the coal gas measuring analyzer 7 measures the measured value mu of the coal powder injection amount measured by the first flowmeter 3-1 6 Oxygen-enriched flow measurement value measured by the second flowmeter 3-2, cool air flow measurement value measured by the third flowmeter 3-3, and blast air humidity measurement value mu measured by the hygrometer 5 Analyzing, the gas quantity parameter mu of the furnace bosh is calculated by the gas measuring analyzer 7 of the furnace bosh 1 And the gas quantity at the furnace bosh is measured 1 The parameters are transmitted to the data acquisition device 9;
the oxygen enrichment rate measurement analyzer 8 analyzes the oxygen enrichment flow measurement value measured by the second flow meter 3-2, the cold air flow measurement value measured by the third flow meter 3-3 and the blast air humidity measurement value measured by the hygrometer, and calculates an oxygen enrichment rate parameter mu 4 And the oxygen enrichment rate parameter mu is measured 4 Transmitted to the data acquisition device 9;
the data acquisition device 9 is used for measuring the coal powder injection amount mu transmitted by the first flowmeter 3-1 6 The gas quantity parameter of the furnace hearth transmitted by the gas measuring analyzer of the furnace hearthSeveral mu 1 The temperature measurement value of the hot air transmitted from the thermometer 4 and the pressure measurement value mu of the hot air transmitted from the pressure gauge 5 3 Oxygen enrichment rate parameter mu transmitted by the oxygen enrichment rate measurement analyzer 8 4 The blast air humidity measurement value transmitted by the hygrometer 6 is preprocessed in the data acquisition device 9, and the preprocessed data result is transmitted to the computer 10;
the computer 10 uses the preprocessed data results transmitted by the data acquisition device 9 and stores the data results in the computer according to the time sequence, adopts a blast furnace molten iron quality prediction method based on ensemble learning to predict, and performs online prediction on the quality indexes of the multi-element molten iron by establishing a multi-element molten iron quality online prediction model of root mean square error probability weighted integration RVFLNs to obtain the predicted values of the quality indexes of the multi-element molten iron.
The computer system is provided with OPC communication software for data bidirectional communication between the computer and the data acquisition device.
In the present embodiment, the following measurement system is installed in an iron-making blast furnace object having a volume of 2600m3, and includes:
the first flowmeter is an HDLWG-06 pulverized coal flowmeter and is used for measuring the pulverized coal injection amount; the second flowmeter is an A + K balance flowmeter and is used for measuring the oxygen-enriched flow; the third flow meter is a model LB differential pressure flow meter and is used for measuring the flow of cold air; the thermometer is a YHIT infrared thermometer and is used for measuring the temperature of hot air; the pressure gauge is a harp EJA series pressure transmitter and is used for measuring the hot air pressure of a blast furnace hot air system; the hygrometer type JWSK-6CWDA air humidity sensor is used for measuring the blast air humidity;
in addition, two measurement analyzers were included:
the device comprises a furnace bosh gas quantity measurement analyzer and an oxygen enrichment rate measurement analyzer, wherein the furnace bosh gas quantity measurement analyzer is arranged as follows:
the amount of coal gas in the furnace belly =1.21 × cold blast flow rate/60 + (2 × oxygen-enriched flow rate/60) + (44.8 × blast humidity (cold blast flow rate/60 + (oxygen-enriched flow rate/60))/18000) + (22.4 × hour coal injection amount × 1000 × hydrogen content of pulverized coal/12000);
the oxygen enrichment ratio measurement analyzer is set as follows:
an oxygen enrichment rate = ((oxygen enrichment flow rate = 0.98/60+ ((0.21 + (0.29 × blast humidity/8/100))) cold blast flow rate/60))/(cold blast flow rate/60 + (oxygen enrichment flow rate/60)) - (0.21 + (0.29 × blast humidity/8/100)))/100;
a blast furnace molten iron quality prediction method based on ensemble learning is realized by using a blast furnace molten iron quality prediction system based on ensemble learning, and as shown in figure 2, the method comprises the following specific steps:
step 1: acquiring original historical data of a blast furnace, and preprocessing the data, wherein the preprocessing comprises the following steps: unifying time granularity of data, eliminating damping-down data, abnormal data and normalized data, and specifically comprising the following steps of:
step 1.1: because each process variable is measured by the sensors with different sampling frequencies, the time granularity of the collected data is inconsistent, and the tapping time of the blast furnace molten iron must be corresponding to the time point of the body parameter, specifically: marking the collected data according to the time sequence, and manually matching according to the nearest neighbor time principle, namely according to the time sequence to obtain historical data of the blast furnace ironmaking process with consistent time granularity;
the ontology parameters include 19 key process variables, specifically: cold air flow, air supply ratio, hot air pressure, top pressure, pressure difference, top pressure air quantity ratio, air permeability, resistance coefficient, hot air temperature, oxygen enrichment flow, oxygen enrichment rate, coal injection quantity, blast humidity, theoretical combustion temperature, standard air speed, actual air speed, blast kinetic energy, furnace belly coal gas quantity and furnace belly coal gas index;
step 1.2: eliminating damping data and abnormal data: eliminating damping down data in a specific mode: determining a planned maintenance time period of the blast furnace according to the shift-changing record, and eliminating damping-down data of the blast furnace body in the time period, wherein the damping-down data specifically refers to data that the hot blast furnace does not blow air to the blast furnace; eliminating abnormal data, namely eliminating abnormal values by adopting a Lauda criterion, namely a 3 sigma criterion, namely eliminating data with data deviation larger than 3 sigma; σ is a standard deviation of the screened blast furnace body data, and is shown by the following formula:
step 1.3: in order to effectively eliminate the problem of dimension inconsistency among features, accelerate the convergence performance of an algorithm and reduce model errors to a certain extent, a minimum normalization method and a maximum normalization method are used for carrying out data normalization on blast furnace data to obtain normalized blast furnace historical data serving as a sample data set, and the following formula is shown:
wherein x is i 、
Respectively before and after normalization of the ith variable, max (x) i )、min(x i ) The maximum value and the minimum value of the ith variable are respectively, and the data after normalization processing is in the following range: x is a radical of a fluorine atom i ∈(0,1)；
Step 2: acquiring input and output parameters required by blast furnace multi-element molten iron quality index soft measurement, and extracting a sample data set from preprocessed blast furnace original historical data according to the input and output parameters;
step 2.1: determining the output parameter of the quality of the blast furnace molten iron needing soft measurement as Si (silicon) content y according to the process mechanism of the blast furnace 1 (%), P (phosphorus) content y 2 (%), S (Sulfur) content y 3 (%) and molten iron temperature y 4 (℃)；
Step 2.2: in order to improve the efficiency and the accuracy of modeling, a grey correlation analysis method is adopted, the first k blast furnace body parameters with the highest correlation degree are extracted as auxiliary variables of soft measurement, and the 6 process variables with the strongest correlation are extracted from the 19 blast furnace body parameters influencing the quality indexes of the multi-element molten iron to serve as the auxiliary variables of the soft measurement. The method comprises the following steps: gas flow u of furnace chamber 1 (m 3 ) Temperature u of hot air 2 (° c), hot air pressure u 3 (KPa) and oxygen enrichment ratio u 4 Blast air humidity u 5 (RH) Coal injection amount u 6 (m 3 /h)。
Step 2.3: according to the process dynamic characteristics, a non-linear autoregressive model (NARX) is introduced based on the 6 auxiliary variables, and the following 16 variables are determined as input parameters of a soft measurement model:
gas flow u of furnace chamber at current moment 1 (t)(m 3 ) (ii) a Hot air temperature u at present 2 (t) (° c); hot air pressure u at present 3 (t) (KPa); oxygen enrichment rate u at the present moment 4 (t); blast humidity u at present 5 (t) (RH); setting coal injection quantity u at current moment 6 (t) (m 3/h); si content y at the last moment 1 (t-1) (%); p content y at the last moment 2 (t-1) (%); last moment furnace bosh coal gas volume u 1 (t-1)(m 3 ) (ii) a Last moment hot air temperature u 2 (t-1) (. Degree.C.); last moment hot air pressure u 3 (t-1) (KPa); last moment oxygen enrichment rate u 4 (t-1); last moment blast air humidity u 5 (t-1) (RH); setting the coal injection amount u at last moment 6 (t-1)(m 3 H); s content y at the last moment 3 (t-1) (%); last moment molten iron temperature y 4 (t-1)(℃)；
Step 2.4: extracting a sample data set from the preprocessed original blast furnace historical data according to the input and output parameters;
and 3, step 3: establishing a multi-element molten iron quality online forecasting model of weighted integration RVFLNs of root mean square error probability based on input and output parameters required by soft measurement of multi-element molten iron quality indexes of the blast furnace;
step 3.1: and establishing a multi-element molten iron quality online forecasting model of weighted RVFLNs (relevance vector regression) of root mean square error probability.
Step 3.2: if the multi-element molten iron quality online forecasting model of the RMS error probability weighted integration RVFLNs is adopted, the absolute error between the blast furnace molten iron quality output parameter obtained by model training and the corresponding actual blast furnace molten iron quality output parameter is less than or equal to a set absolute error value, the model training is finished, and the final multi-element molten iron quality online forecasting model of the RMS error probability weighted integration RVFLNs is obtained; in this embodiment, the following are set: si (silicon) content y 1 (％)、P (phosphorus) content y 2 (%), S (Sulfur) content y 3 (%) and molten iron temperature y 4 The absolute error value of the convergence condition (deg.C) is: 0.1, 0.006, 0.003 and 5; if the absolute error between the blast furnace molten iron quality output parameter obtained by model training and the corresponding actual blast furnace molten iron quality output parameter is larger than the set absolute error, retraining the multi-element molten iron quality online forecasting model of the root mean square error probability weighting integration RVFLNs, turning to the step 3.1 until the absolute error between the blast furnace molten iron quality output parameter obtained by model training and the corresponding actual blast furnace molten iron quality output parameter is smaller than or equal to the set absolute error, finishing the model training, and obtaining the final multi-element molten iron quality online forecasting model of the root mean square error probability weighting integration RVFLNs;
the step 3.1 is to establish a multi-element molten iron quality online forecasting model of the weighted integrated RVFLNs, as shown in fig. 3, and comprises the following specific steps:
step 3.1.1: dividing M =520 groups of data in the sample data set into two sample sets D 1 、D 2 In which D is 1 Training and establishing molten iron quality index model for training sample set, taking M 1 =420 groups of data, D 2 For testing the sample set, testing the molten iron quality index model, and taking the rest M 2 =100 sets of data.
Step 3.1.2: the method comprises the following steps of performing a sample-back experiment on a training sample set based on a Bootstrap idea to obtain N sub-training sample sets, wherein the specific method comprises the following steps: in 420 training sample sets, performing m =400 times of random sampling experiments with putting back, and considering modeling efficiency, performing N =9 sets of such random sampling experiments with putting back to obtain N =9 sub-sample sets, wherein the number of data of the sub-sample sets is m = 400;
step 3.1.3: and establishing a sub-model by using the sub-sample set, and modeling by using a random weighted neural network (RVFLNs) model with a fast learning speed as the sub-model in order to improve the modeling efficiency.
Step 3.1.4: calculating the root mean square error RMSE of each sub-model j As shown in the following formula,
wherein, y ji For the actual value of the ith output variable in the jth sub-model,
is an estimate of the ith output variable, RMSE, in the jth submodel j The root mean square error of the jth sub-model is obtained;
step 3.1.5: estimating a Root Mean Square Error (RMSE) probability density curve of each sub-model by using a kernel density estimation method, as shown in fig. 5, and solving a root mean square error probability distribution curve of each sub-model, as shown in fig. 6;
the principle of the method for estimating the nuclear density is as follows:
kernel Density Estimation (KDE) is a non-parametric Estimation method to solve the problem of distribution Density for a given set of random variables, first proposed by Parzen.
Suppose z i E.r, i =1, …, n is an independent same distribution random variable subject to a distribution density function of f (z), z e R, then f (x) kernel density estimate
Is defined as:
where φ (·) is called a kernel function, h p Usually called window width or smoothing parameter, is a positive number that is manually pre-given. From the above definition, the kernel density estimation of the distribution density function f (z)
Not only with respect to a given set of data samples, but also with respect to the selection of the kernel function and the window width parameter h p Is relevant to the selection of (2).
Using KDE method for root mean square errorSample set { RMSE i I =1,2 … K } to obtain an estimated error probability density function gamma RMSE Comprises the following steps:
selection of the phi (-) kernel function: there are many methods for selecting kernel functions when estimating the unknown probability density function of the random variable. Theoretically, the kernel function does not need to be a density function, but from a practical point of view, since the function to be estimated is a probability density function, the kernel function is required to meet the property of the probability density function, that is, the following conditions are satisfied:
a.φ(z)≥0
b.∫φ(z)du＝1
commonly used kernel functions include gaussian kernel functions, rectangular window kernel functions, epanechnikov kernel functions, and the like. Different choices of kernel functions are insensitive in KDE, and when sample data is large, the influence on the result of kernel function density estimation is not large. The invention selects a Gaussian kernel function, and the table formula is as follows:
h p selection of the window width: width h of window p Has a locally smooth effect on the density estimation of the kernel function if h p Too large will make the model error PDF shape very smooth, make some characteristics of its main part, such as the multimodality, be covered, thus increase the deviation of the estimator; if h p Too small, the overall density function is rough, and particularly the tails of the density estimates are subject to large disturbances. Thus, the window width h p Is set as h p ＝1.06θK -1/5 Where θ is estimated by min { S,0.746Q }, S represents the sample standard deviation, Q is the interquartile range, and K is the number of RMSE sample sets.
Step 3.1.6: distributing weights to the sub-models, and obtaining a multi-element molten iron quality online forecasting model integrating RVFLNs in the weighted mode of the root mean square error probability: taking the probability of each sub-model in the root mean square error probability curve of the sub-models as the respective weight, then carrying out weighted summation to obtain the multielement molten iron quality online forecasting model of the root mean square error probability weighted integration RVFLNs, and solving as follows:
wherein, w i The weights determined for the proposed method and the condition met is that the sum of all weights is 1.RVFLNs i And representing the ith sub-model, N represents the number of the sub-models, and the RVFLNs are multi-element molten iron quality online forecasting models of the weighted integration RVFLNs with the obtained root mean square error probability.
And 4, step 4: online prediction is carried out on real-time collected blast furnace data according to a multi-element molten iron quality online prediction model of weighted and integrated RVFLNs (root mean square error probabilities) to obtain an online prediction result: the method comprises the steps of preprocessing acquired real-time blast furnace data to be tested in step 1, acquiring input and output parameters required by blast furnace multi-element molten iron quality index soft measurement in step 2, extracting the data to be tested from preprocessed blast furnace original historical data according to the input and output parameters, and using an established multi-element molten iron quality online forecasting model of root mean square error probability weighted integration RVFLNs to obtain an online forecasting result.
Acquiring 400 groups of data from historical data as model training sample data, and taking 100 groups of data as model test sample data, wherein fig. 4 is a comparison result of a soft measurement value and an actual value of a quality index of the multi-element molten iron obtained by a soft measurement system within a period of time, and (a) is a comparison curve of a predicted value and an actual value of the silicon content at the current moment; (b) A comparison curve of the predicted value of the phosphorus content at the current moment and the actual value is obtained; (c) A comparison curve of the predicted value of the sulfur content at the current moment and the actual value is obtained; (d) A comparison curve of the predicted value and the actual value of the molten iron temperature at the current moment is obtained; it can be seen that the soft measurement result of the quality of the multi-element molten iron has high precision, the soft measurement error is small, and the variation trend of the soft measurement error is consistent with the actual value. In addition, the soft measurement model of the method has the advantages of simple structure, low model complexity, high operation speed, high measurement precision and strong generalization capability, and has higher practicability and superiority compared with other existing soft measurement methods for the quality index of the molten iron. Therefore, the invention is a low-cost, high-efficiency and practical multi-element measuring means for the quality of the molten iron in the blast furnace ironmaking process.
Claims (4)
1. The utility model provides a blast furnace molten iron quality prediction system based on ensemble learning which characterized in that includes: the system comprises a blast furnace, a hot blast furnace, a first flowmeter, a second flowmeter, a third flowmeter, a thermometer, a pressure gauge, a hygrometer, a furnace belly gas measurement analyzer, an oxygen enrichment rate measurement analyzer, a data acquisition device and a computer;
a sample to be tested is placed into the blast furnace from a blast furnace inlet, pulverized coal injection is carried out from a blast furnace belly tuyere, a first flowmeter is arranged at the pulverized coal injection position, and the first flowmeter is respectively connected with a data acquisition device and a belly coal gas measurement analyzer; the thermometer is arranged at the air outlet of the hot blast stove and is connected with the data acquisition device; the pressure gauge is arranged at the air outlet of the hot blast stove and is connected with the data acquisition device; the second flowmeter, the third flowmeter and the hygrometer are respectively arranged at an air inlet of the hot blast stove, the second flowmeter is respectively connected with the furnace belly gas measurement analyzer and the oxygen enrichment rate measurement analyzer, and the third flowmeter is respectively connected with the furnace belly gas measurement analyzer and the oxygen enrichment rate measurement analyzer; the hygrometer is respectively connected with the furnace belly gas measurement analyzer and the data acquisition device; the furnace belly gas measurement analyzer and the oxygen enrichment rate measurement analyzer are respectively connected with the data acquisition device; the data acquisition device is connected with the computer through a communication bus;
the blast furnace is used for finishing a blast furnace smelting process;
the hot blast stove is used for conveying oxygen-enriched cold air to the hot blast stove and conveying hot air to the interior of the blast furnace;
the first flowmeter is arranged at the coal powder injection position and used for measuring the coal powder injection amount on line and transmitting the coal powder injection amount measured value to the furnace belly coal gas measurement analyzer and the data acquisition device respectively;
the second flowmeter is arranged at an inlet of the hot blast stove and is used for measuring the oxygen-enriched flow on line and transmitting the measured value of the oxygen-enriched flow to the furnace belly gas measurement analyzer and the oxygen enrichment rate measurement analyzer respectively;
the third flow meter is arranged at an inlet of the hot blast stove and used for measuring the flow of cold air on line and transmitting the measured value of the flow of the cold air to the furnace belly gas measurement analyzer and the oxygen enrichment rate measurement analyzer respectively;
the thermometer is arranged at the outlet of the hot blast stove and used for measuring the hot blast temperature of the hot blast stove on line and transmitting the measured value of the hot blast temperature to the data acquisition device;
the pressure gauge is arranged at the outlet of the hot blast stove and is used for measuring the hot blast pressure of the hot blast stove on line and transmitting the measured value of the hot blast pressure to the data acquisition device;
the hygrometer is arranged at an inlet of the hot blast stove and used for measuring the blast humidity of the hot blast stove on line and transmitting the blast humidity measured value to the furnace belly gas measurement analyzer, the oxygen enrichment rate measurement analyzer and the data acquisition device respectively;
the furnace bosh gas measuring analyzer analyzes a coal powder injection amount measured value measured by the first flowmeter, an oxygen-enriched flow amount measured value measured by the second flowmeter, a cold air flow measurement value measured by the third flowmeter and an air blast humidity measured value measured by the hygrometer, calculates a furnace bosh gas amount parameter and transmits the furnace bosh gas amount parameter to the data acquisition device;
the oxygen enrichment rate measurement analyzer calculates an oxygen enrichment rate parameter through an oxygen enrichment rate measurement value measured by the second flowmeter, a cold air flow measurement value measured by the third flowmeter and a blast humidity measurement value measured by the hygrometer, and transmits the oxygen enrichment rate parameter to the data acquisition device;
the data acquisition device preprocesses the coal powder injection quantity measured value transmitted by the first flowmeter, the furnace belly coal gas quantity parameter transmitted by the furnace belly coal gas measurement analyzer, the hot air temperature measured value transmitted by the thermometer, the hot air pressure measured value transmitted by the pressure gauge, the oxygen enrichment rate parameter transmitted by the oxygen enrichment rate measurement analyzer and the air blast humidity measured value transmitted by the hygrometer in the data acquisition device, and transmits the preprocessed data result to the computer;
the computer uses the preprocessed data results transmitted by the data acquisition device and stores the data results in the computer according to a time sequence, a blast furnace molten iron quality prediction method based on ensemble learning is adopted for prediction, and the multivariate molten iron quality indexes are predicted online by establishing a multivariate molten iron quality online prediction model of the root mean square error probability weighted ensemble RVFLNs, so that the multivariate molten iron quality index prediction value is obtained.
2. The system for predicting the quality of the molten iron in the blast furnace based on the ensemble learning of claim 1, wherein the computer system is provided with OPC communication software for data bidirectional communication between the computer and the data acquisition device.
3. A blast furnace molten iron quality prediction method based on ensemble learning is realized by adopting the blast furnace molten iron quality prediction system based on ensemble learning in claim 1, and is characterized by comprising the following specific steps:
step 1: acquiring original historical data of a blast furnace, and preprocessing the data, wherein the preprocessing comprises the following steps: unifying time granularity of data, eliminating damping-down data, abnormal data and normalized data, and specifically comprising the following steps:
step 1.1: marking the collected data according to the time sequence, and manually matching according to the nearest neighbor time principle, namely according to the time sequence to obtain historical data of the blast furnace ironmaking process with consistent time granularity;
step 1.2: eliminating damping data and abnormal data: eliminating damping down data in a specific mode: determining a planned maintenance time period of the blast furnace according to the shift change record, and eliminating damping down data of a blast furnace body in the time period, wherein the damping down data specifically refers to data of the hot blast furnace which does not blow air to the blast furnace; eliminating abnormal data, namely eliminating abnormal values by adopting a Lauda criterion, namely a 3 sigma criterion, namely eliminating data with data deviation larger than 3 sigma; σ is the standard deviation of the screened blast furnace body data, and is shown in the following formula:
step 1.3: carrying out data normalization processing on blast furnace data to obtain normalized blast furnace historical data serving as a sample data set, wherein the formula is as follows:
wherein x is i 、
Respectively before and after the normalization of the ith variable, max (x) i )、min(x i ) The maximum value and the minimum value of the ith variable are respectively, and the data after normalization processing is in the following range: x is the number of i ∈(0,1)；
Step 2: acquiring input and output parameters required by blast furnace multi-element molten iron quality index soft measurement, and extracting a sample data set from preprocessed blast furnace original historical data according to the input and output parameters;
step 2.1: determining the output parameters of the quality of the blast furnace molten iron needing soft measurement according to the blast furnace process mechanism as follows: silicon content y 1 Phosphorus content y 2 Sulfur content y 3 And the temperature y of the molten iron 4 ；
Step 2.2: a grey correlation analysis method is adopted, and the first k blast furnace body parameters with the highest correlation degree are extracted to serve as auxiliary variables of soft measurement, and the method comprises the following steps: gas flow u of furnace chamber 1 Temperature u of hot air 2 Pressure u of hot air 3 Oxygen-enriched airRate u 4 Blast air humidity u 5 Coal injection amount u 6 ；
Step 2.3: according to the process dynamic characteristics, a nonlinear autoregressive model is introduced based on the 6 auxiliary variables, and the following 16 variables are determined as input variables of the soft measurement model:
gas flow u of furnace chamber at current moment 1 (t); hot air temperature u at present 2 (t); hot air pressure u at present 3 (t); oxygen enrichment rate u at present 4 (t); blast humidity u at present 5 (t); setting coal injection quantity u at current moment 6 (t); si content y at the last moment 1 (t-1); p content y at the last moment 2 (t-1); last moment furnace bosh coal gas volume u 1 (t-1); last moment hot air temperature u 2 (t-1); last moment hot air pressure u 3 (t-1); last moment oxygen enrichment rate u 4 (t-1); last moment blast air humidity u 5 (t-1); setting the coal injection amount u at last moment 6 (t-1); s content y at the previous moment 3 (t-1); molten iron temperature y at last moment 4 (t-1)；
Step 2.4: extracting a sample data set from the preprocessed original blast furnace historical data according to the input and output parameters;
and step 3: establishing a multi-element molten iron quality online forecasting model of weighted integration RVFLNs of root mean square error probability based on input and output parameters required by soft measurement of multi-element molten iron quality indexes of the blast furnace;
step 3.1: establishing a multi-element molten iron quality online forecasting model of weighted integration of root mean square error probability (RMS) and RVFLNs;
step 3.2: if the multi-element molten iron quality online forecasting model of the RMS error probability weighted integration RVFLNs is adopted, the absolute error between the blast furnace molten iron quality output parameter obtained by model training and the corresponding actual blast furnace molten iron quality output parameter is less than or equal to a set absolute error value, the model training is finished, and the final multi-element molten iron quality online forecasting model of the RMS error probability weighted integration RVFLNs is obtained; if the absolute error between the blast furnace molten iron quality output parameter obtained by model training and the corresponding actual blast furnace molten iron quality output parameter is larger than the set absolute error, retraining the multi-element molten iron quality online forecasting model of the root mean square error probability weighting integration RVFLNs, turning to the step 3.1 until the absolute error between the blast furnace molten iron quality output parameter obtained by model training and the corresponding actual blast furnace molten iron quality output parameter is smaller than or equal to the set absolute error, finishing the model training, and obtaining the final multi-element molten iron quality online forecasting model of the root mean square error probability weighting integration RVFLNs;
and 4, step 4: online prediction is carried out on real-time collected blast furnace data according to a multi-element molten iron quality online prediction model of weighted integrated RVFLNs (root mean square error probabilities) to obtain an online prediction result: the method comprises the steps of preprocessing acquired real-time blast furnace data to be tested in step 1, acquiring input and output parameters required by blast furnace multi-element molten iron quality index soft measurement in step 2, extracting the data to be tested from preprocessed blast furnace original historical data according to the input and output parameters, and using an established multi-element molten iron quality online forecasting model of root mean square error probability weighted integration RVFLNs to obtain an online forecasting result.
4. The method for predicting the quality of the molten iron in the blast furnace based on the ensemble learning of claim 3, wherein the step 3.1 of establishing the online forecasting model of the quality of the multi-element molten iron of the weighted integrated RVFLNs comprises the following specific steps:
step 3.1.1: dividing M groups of data in the sample data set into two sample sets D 1 、D 2 In which D is 1 For training the sample set, take the top M 1 Group data, D 2 To test the sample set, take the remaining M 2 Group data;
step 3.1.2: based on Bootstrap idea, putting back a sampling experiment on a training sample set to obtain N sub-training sample sets, wherein the method specifically comprises the following steps: at M 1 In the group training sample set, performing m times of random sampling experiments with replacement, and considering modeling efficiency, performing N groups of random sampling experiments with replacement to obtain N sub-sample sets, wherein the number of data of the sub-sample sets is m;
step 3.1.3: establishing a sub-model by using the sub-sample set, and modeling by using a random weight neural network (RVFLNs) model as the sub-model;
step 3.1.4: calculating the root mean square error RMSE of each sub-model j As shown in the following formula,
wherein, y ji For the actual value of the ith output variable in the jth sub-model,
is an estimate of the ith output variable in the jth sub-model, RMSE j The root mean square error of the jth sub-model is obtained;
step 3.1.5: estimating a root mean square error probability density curve of each sub-model by using a kernel density estimation method, and solving a root mean square error probability distribution curve of each sub-model;
the method of kernel density estimation is as follows:
suppose z i E R, i =1, …, n is an independent same distribution random variable subject to a distribution density function of f (z), z is e R, and then f (z) is estimated by kernel density
Is defined as:
where φ (-) is referred to as the kernel function, h p Commonly referred to as window width or smoothing parameter, is a positive number that is manually pre-specified;
using KDE method for root mean square error sample set { RMSE i I =1,2 … K } to obtain an estimated error probability density function gamma RMSE Comprises the following steps:
selection of the phi (-) kernel function: the following conditions are satisfied:
a.φ(z)≥0
b.∫φ(z)du＝1
selecting a Gaussian kernel function, wherein the table formula is as follows:
h p selection of the window width: width h of window p Is set to h p ＝1.06θK -1/5 Wherein θ is estimated by min { S,0.746Q }, S represents the sample standard deviation, Q is the interquartile range, and K is the number of RMSE sample sets;
step 3.1.6: distributing weights to the sub-models, and obtaining a multi-element molten iron quality online forecasting model integrating RVFLNs in the weighted mode of the root mean square error probability: taking the probability of each sub-model in the root mean square error probability curve of the sub-models as the respective weight, then carrying out weighted summation to obtain the multielement molten iron quality online forecasting model of the root mean square error probability weighted integration RVFLNs, and solving as follows:
wherein, w i Weights determined for the proposed method and satisfying the condition that the sum of all weights is 1,RVFLNs i And representing the ith sub-model, N represents the number of the sub-models, and the RVFLNs are multi-element molten iron quality online forecasting models of the weighted integration RVFLNs with the obtained root mean square error probability.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910163699.6A CN109935280B (en) | 2019-03-05 | 2019-03-05 | Blast furnace molten iron quality prediction system and method based on ensemble learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910163699.6A CN109935280B (en) | 2019-03-05 | 2019-03-05 | Blast furnace molten iron quality prediction system and method based on ensemble learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109935280A CN109935280A (en) | 2019-06-25 |
| CN109935280B true CN109935280B (en) | 2022-11-15 |
Family
ID=66986295
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910163699.6A Active CN109935280B (en) | 2019-03-05 | 2019-03-05 | Blast furnace molten iron quality prediction system and method based on ensemble learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109935280B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110400007A (en) * | 2019-07-05 | 2019-11-01 | 浙江大学 | Based on the molten steel quality forecasting procedure for improving gating cycle neural network |
| CN112350878A (en) * | 2019-08-07 | 2021-02-09 | 阿里巴巴集团控股有限公司 | Pressure test system |
| CN111593155B (en) * | 2020-05-28 | 2022-09-20 | 江苏省沙钢钢铁研究院有限公司 | Blast furnace diagnosis system and method |
| CN116502769A (en) * | 2020-05-28 | 2023-07-28 | 江苏省沙钢钢铁研究院有限公司 | Scoring method and scoring system for blast furnace conditions |
| CN111607673B (en) * | 2020-05-28 | 2022-09-20 | 江苏省沙钢钢铁研究院有限公司 | Pre-iron diagnostic system and method |
| JP7384150B2 (en) * | 2020-12-09 | 2023-11-21 | Ｊｆｅスチール株式会社 | Operation guidance method, blast furnace operation method, hot metal production method and operation guidance device |
| CN114879514B (en) * | 2022-07-07 | 2022-10-11 | 大唐环境产业集团股份有限公司 | Multi-model wet desulphurization efficiency prediction method based on optimal training sample |
| CN115456264B (en) * | 2022-09-01 | 2023-07-25 | 北京科技大学 | Endpoint carbon content and endpoint temperature prediction method for small and medium-sized converter |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH07305105A (en) * | 1994-05-02 | 1995-11-21 | Nippon Steel Corp | Method for evaluating raceway condition of blast furnace |
| CN104651559A (en) * | 2015-01-30 | 2015-05-27 | 东北大学 | Blast furnace liquid iron quality online forecasting system and method based on multivariable online sequential extreme learning machine |
| CN105608492A (en) * | 2016-03-02 | 2016-05-25 | 东北大学 | Robust random-weight neural network-based molten-iron quality multi-dimensional soft measurement method |
| JP2019019385A (en) * | 2017-07-19 | 2019-02-07 | Ｊｆｅスチール株式会社 | Method and device for predicting molten iron temperature, operation method of blast furnace, operation guidance device, and method and device for controlling molten iron temperature |
-
2019
- 2019-03-05 CN CN201910163699.6A patent/CN109935280B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH07305105A (en) * | 1994-05-02 | 1995-11-21 | Nippon Steel Corp | Method for evaluating raceway condition of blast furnace |
| CN104651559A (en) * | 2015-01-30 | 2015-05-27 | 东北大学 | Blast furnace liquid iron quality online forecasting system and method based on multivariable online sequential extreme learning machine |
| CN105608492A (en) * | 2016-03-02 | 2016-05-25 | 东北大学 | Robust random-weight neural network-based molten-iron quality multi-dimensional soft measurement method |
| JP2019019385A (en) * | 2017-07-19 | 2019-02-07 | Ｊｆｅスチール株式会社 | Method and device for predicting molten iron temperature, operation method of blast furnace, operation guidance device, and method and device for controlling molten iron temperature |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109935280A (en) | 2019-06-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109935280B (en) | Blast furnace molten iron quality prediction system and method based on ensemble learning | |
| CN110066895B (en) | Stacking-based blast furnace molten iron quality interval prediction method | |
| CN105608492B (en) | A kind of polynary molten steel quality flexible measurement method based on robust random weight neutral net | |
| CN107526927B (en) | Blast furnace molten iron quality online robust soft measurement method | |
| CN104651559B (en) | Blast furnace liquid iron quality online forecasting system and method based on multivariable online sequential extreme learning machine | |
| CN109447346B (en) | Converter oxygen consumption prediction method based on gray prediction and neural network combined model | |
| CN109001979B (en) | Blast furnace molten iron quality self-adaptive robust prediction control method based on lazy learning | |
| Fang et al. | Soft sensors based on adaptive stacked polymorphic model for silicon content prediction in ironmaking process | |
| US11900255B2 (en) | Multi-mode smelting method based on the classification system of molten iron | |
| CN104750902B (en) | Molten steel quality dynamic soft-measuring method based on multi output support vector regression | |
| CN105821170A (en) | Soft measuring system and method for quality indexes of multielement molten iron of blast furnace | |
| Li et al. | Dynamic time features expanding and extracting method for prediction model of sintering process quality index | |
| CN104915518A (en) | Establishing method and application of two-dimensional prediction model of silicon content in hot metal in blast furnace | |
| CN105574297B (en) | Self adaptation blast furnace molten iron silicon content trend prediction method | |
| CN113919559A (en) | Ultra-short-term prediction method and device for equipment parameters of comprehensive energy system | |
| Wu et al. | Integrated soft sensing of coke-oven temperature | |
| CN113761787A (en) | Blast furnace molten iron silicon content online prediction method and system based on deep migration network | |
| CN111915080A (en) | Raw fuel cost optimal proportioning method based on molten iron quality constraint | |
| CN110533247B (en) | Monthly power consumption prediction method adopting temperature data anomaly point compensation | |
| CN111518980B (en) | Correction method and system for converter end point carbon content prediction model | |
| CN113517037A (en) | Method and system for predicting sintering ore FeO by fusing data and knowledge | |
| CN116757078A (en) | Method and system for measuring flow velocity of pulverized coal based on acting force | |
| CN114525372B (en) | Multi-mode fusion based blast furnace state monitoring method and device | |
| CN107832880B (en) | Blast furnace state variable prediction method based on material distribution parameters | |
| CN115927770A (en) | Real-time tracking method for tapping state of blast furnace |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |